Macrocyclic inhibitors of penicillopepsin. 1. Design, synthesis, and evaluation of an inhibitor bridged between P1 and P3

Article abstract of DOI:10.1021/ja973715j

The macrocyclic peptidyl phosphonate 1-L was designed on the basis of the conformation of an acyclic analogue (4) bound to the aspartic protease penicillopepsin. This material and the two acyclic comparison compounds 2-L and 3 were synthesized and evaluated as inhibitors; their binding affinity was found to be inversely related to the degree of conformational flexibility across the series: 3 (K(i) = 110 μM), 2- L (K(i) = 7.6 μM), 1-L (K(i) = 0.80 μM). NMR methods in conjunction with molecular modeling were used to assign the stereochemical configurations of the precursor 16-L and its diastereomer 16-D and to determine the solution conformations of the macrocyclic ring systems. The conformation of the peptide backbone in 1-L closely approximates that desired for a mimic of the lead inhibitor 4, and it appears that the low-energy conformation of 1-L can be accommodated in the pencillopepsin active site without significant distortion.

Full text of DOI:10.1021/ja973715j

4600
J. Am. Chem. Soc. 1998, 120, 4600-4609
Macrocyclic Inhibitors of Penicillopepsin. 1. Design, Synthesis, and
Evaluation of an Inhibitor Bridged between P1 and P3
J. Hoyt Meyer and Paul A. Bartlett*
Contribution from the Department of Chemistry, UniVersity of California, Berkeley, California 94720-1460
ReceiVed October 27, 1997
Abstract: The macrocyclic peptidyl phosphonate 1-L was designed on the basis of the conformation of an
acyclic analogue (4) bound to the aspartic protease penicillopepsin. This material and the two acyclic comparison
compounds 2-L and 3 were synthesized and evaluated as inhibitors; their binding affinity was found to be
inversely related to the degree of conformational flexibility across the series: 3 (Ki ) 110 µM), 2-L (Ki ) 7.6
µM), 1-L (Ki ) 0.80 µM). NMR methods in conjunction with molecular modeling were used to assign the
stereochemical configurations of the precursor 16-L and its diastereomer 16-D and to determine the solution
conformations of the macrocyclic ring systems. The conformation of the peptide backbone in 1-L closely
approximates that desired for a mimic of the lead inhibitor 4, and it appears that the low-energy conformation
of 1-L can be accommodated in the pencillopepsin active site without significant distortion.
An important step in the rational design of enzyme inhibitors
is the optimization of a known ligand using structural informa-
tion from the enzyme-ligand complex. The aim of such
“second-generation” designs may be to enhance the hydrophobic
or electrostatic interactions between inhibitor and target, or to
reduce the conformational flexibility of the ligand. If the
predominant conformation of the free ligand is the same as it
adopts in the binding site, the entropic disadvantage of complex
formation is reduced and its affinity enhanced in comparison
to a more conformationally mobile analogue. Conformational
constraints have often been introduced in peptidase inhibitors
by macrocyclization between the side chains, with recent
examples in both zinc and aspartic proteases.1,2 These con-
straints can lead to analogues of enhanced potency or selectivity;
however, in only a few cases have the effects of the macrocyclic
constraints on binding affinity or conformation been evaluated
explicitly, and structurally related acyclic analogues are not often
available to allow a quantitative comparison to be made. Both
the structural and energetic effects of a macrocyclic constraint
need to be determined to assess the success of the design. This
report describes the design and synthesis of the macrocyclic
phosphonate 1 and the acyclic analogues 2 and 3, their
evaluation as inhibitors of the aspartic protease penicillopepsin,
and the determination of the solution conformations of the
macrocyclic ring system in 1. Accompanying papers report
crystallographic analyses of the complexes of all three inhibitors
with the enzyme3 and the parallel investigation of a different
macrocyclic inhibitor design for the same peptidase.4
Design. Phosphorus-containing peptide analogues are known
as potent inhibitors of the aspartic proteases.5-7 Some phos-
(3) Ding, J.; Fraser, M.; Meyer, J. H.; Bartlett, P. A.; James, M. N. G.
J. Am. Chem. Soc. 1998, 120, 4610.
(4) Smith, W. W.; Bartlett, P. A. 1998, 120, 4622.
(1) For leading references, see: Ksander, G. M.; de Jesus, R.; Yuan, A.;
Ghai, R. D.; Trapani, A.; McMartin, C.; Bohacek, R. J. Med. Chem. 1997,
40, 495. Ksander, G. M.; de Jesus, R.; Yuan, A.; Ghai, R. D.; McMartin,
C.; Bohacek, R. J. Med. Chem. 1997, 40, 506. March, D. R.; Abbenante,
G.; Bergman, D. A.; Brinkworth, R. I.; Wickramasinghe, W.; Begun, J.;
Martin, J. L.; Fairlie, D. P. J. Am. Chem. Soc. 1996, 118, 3375. Ettmayer,
P.; Billich, A.; Hecht, P.; Rosenwirth, B.; Gstach, H. J. Med. Chem. 1996,
39, 3291. Podlogar, B. L.; Farr, R. A.; Friedrich, D.; Tarnus, C.; Huber, E.
W.; Cregge, R. J.; Schirlin, D. J. Med. Chem. 1994, 37, 3684. Weber, A.
E.; Steiner, M. G.; Krieter, P. A.; Colletti, A. E.; Tata, J. R.; Halgren, T.
A.; Ball, R. G.; Doyle, J. J.; Schorn, T. W.; Stearns, R. A.; Miller, R. R.;
Siegl, P. K. S.; Greenlee, W. J.; Patchett, A. A. J. Med. Chem. 1992, 35,
3755.
(5) Bartlett, P. A.; Kezer, W. B. J. Am. Chem. Soc. 1984, 106, 4282.
Williams, P. D.; Perlow, D. S.; Payne, L. S.; Holloway, M. K.; Siegl, P. K.
S.; Schorn, T. W.; Lynch, R. J.; Doyle, J. J.; Strouse, J. F.; Vlasuk, G. P.;
Hoogsteen, K.; Springer, J. P.; Bush, B. L.; Halgren, T. A.; Richards, A.
D.; Kay, J.; Veber, D. F. J. Med. Chem. 1991, 34, 887. Thorsett, E. D.;
Harris, E. E.; Aster, S. D.; Peterson, E. R.; Snyder, J. P. J. Med. Chem.
1986, 29, 251. Allen, M. C.; Fuhrer, W.; Tuck, B.; Wade, R.; Wood, J. M.
J. Med. Chem. 1989, 32, 1652. Grobelny, D.; Wondrak, E. M.; Galardy, R.
E.; Oroszlan, S. Biochem. Biophys. Res. Commun. 1990, 169, 1111. Peyman,
A.; Budt, K.-H.; Spanig, J.; Stowasser, B.; Ruppert, D. Tetrahedron Lett.
1992, 33, 4549. Abdel-Meguid, S. S.; Zhao, B.; Murthy, K. H. M.;
Winborne, E.; Choi, J.-K.; DesJarlais, R. L.; Minnich, M.; Culp, J. S.;
Debouck, C.; Tomaszek, T. A., Jr.; Meek, T. D.; Dreyer, G. B. Biochemistry
1993, 32, 7972. Raddatz, P.; Minck, K.-O.; Rippmann, F.; Schmitges, C. J.
J. Med. Chem. 1994, 37, 486.
(2) Morgan, B. P.; Bartlett, P. A.; Holland, D. R.; Matthews, B. W. J.
Am. Chem. Soc. 1994, 116, 3251.
S0002-7863(97)03715-3 CCC: $15.00 © 1998 American Chemical Society
Published on Web 04/30/1998

Macrocyclic Inhibitors of Penicillopepsin. 1
J. Am. Chem. Soc., Vol. 120, No. 19, 1998 4601
phonate penta- and hexapeptides inhibit penicillopepsin and
pepsin with Ki values as low as 250 pM,6 and a series of
phosphonate tetrapeptides were recently shown to be transition-
state analogues for pepsin.7 The X-ray crystal structure of the
complex between penicillopepsin and the phosphonate pen-
tapeptide 4 (Ki ) 2.8 nM) was determined by Fraser et al. and
refined to an R value of 0.131 at 1.7 Å resolution.8 The
tetrahedral phosphonate of 4 is oriented between the catalytic
aspartates of the enzyme, in imitation of the tetrahedral
intermediate for substrate hydrolysis, and the backbone of the
inhibitor is bound in the extended conformation that is typical
for aspartic peptidase ligands. A salient feature of this
conformation is the proximity of the side chains on alternating
residues (e.g., P3 and P1, and P2 and P1′); these side chains
share extended, hydrophobic binding pockets on opposite sides
of the peptide, and there is room to connect them together.9
To design a bridge to link the P1 and P3 residues of 4,
CAVEAT was used to identify structural frameworks that match
the CR-Câ bonds,10 and standard modeling and minimization
programs were used to elaborate and assess different possibili-
ties. The naphthalene-bridged macrocycle of 1 was selected
not only for its ability to adopt the desired conformation but
also because of its rigidity and synthetic accessibility. The
acyclic comparison compounds 2 and 3 were designed to
distinguish the effect of macrocyclization from that of the
hydrophobic bridging unit itself. These control compounds arise
from replacement of methylene units with hydrogens at two
positions in the macrocycle. The intent of this substitution was
to enable the acyclic analogues to adopt conformations in the
active site similar to the macrocycle, in contrast to the situation
if one of the ring bonds were simply replaced with hydrogens.
The difference in the C-C bond distance and the C-C
separation of two C-H groups at van der Waals contact requires
that the carbon atoms move apart by ca. 2.5 Å and could result
in a different binding orientation in the active site, as is indeed
observed for a macrocyclic thermolysin inhibitor and its seco
analogue.2 Even replacement of the methylene unit with two
hydrogens leads to an increase in the distance between the
remaining carbons of ca. 1.4 Å; however, the macrocyclic
structure of 1 cannot lose two adjacent carbons and maintain
the necessary connectivity between the remaining functionality.
Synthesis. The key unit for construction of the macrocyclic
inhibitor 1 was the R-amino phosphonic acid derivative 11,
which was prepared by the Oleksyszyn procedure11 from a
suitably functionalized naphthaleneacetaldehyde precursor as
shown in Scheme 1. Addition of ethyl lithioacetate to 7-bromo-
2-tetralone 5 followed by dehydration of the tertiary alcohol
afforded the endo- and exo-alkenes 6a and 6b. The mixture
could be aromatized by dehydrogenation with palladium;
however, this procedure resulted in considerable debromination.
A two-step process involving bromine addition followed by
double elimination with 1,8-diazabicyclo[5.4.0]undec-2-ene
(DBU) is an effective alternative for producing the 7-bro-
monaphthalene-1-acetate ester 7; both isomers of 6 can be
Scheme 1
carried through the addition, elimination, isomerization se-
quence, although the product is cleaner starting with the endo
isomer. Replacement of the bromine with an allyl group
followed by two-step oxidative cleavage provided the naphtha-
leneacetaldehyde precursor 10 in good overall yield. Condensa-
tion with benzyl carbamate and triphenyl phosphite11 followed
by exchange of all esters for methyl was accomplished in modest
yield to give the key intermediate 11.
Phosphonate 11 was incorporated into the macrocycle as
shown in Scheme 2. Standard procedures were employed to
replace the Cbz group with Boc-valine and to convert the
carboxylate methyl ester to the pentafluorophenyl ester 14. After
cleavage of the Boc group with trifluoroacetic acid (TFA), slow
addition of the activated ester to dioxane and pyridine at 50 °C
gave the macrocycle 15 as a mixture of diastereomers. After
chromatographic separation, the two isomers were individually
hydrolyzed to the monoesters 16-L and 16-D and coupled with
methyl â-phenyllactate to give the corresponding diastereomers
of 18. Several procedures were investigated for this coupling
step, including activation of 16 with diphenyl phosphoryl azide
or benzotriazole-1-yloxy-tris-pyrrolidinophosphonium hexafluo-
rophosphate (PyBOP), or condensation with methyl D-â-
phenyllactate under Mitsunobu conditions.12 Displacement of
the triflate ester of methyl D-â-phenyllactate (17) with the
tetramethylammonium salt of 16 proved to be the most
reproducible and highest yielding. Synthesis of the macrocyclic
inhibitors 1-L and 1-D was completed by cleavage of the
phosphonate ester with trimethylsilyl bromide13 and cation
exchange. The absolute configurations of these analogues were
assigned by NMR methods as described below.
(6) Bartlett, P. A.; Hanson, J. E.; Giannousis, P. P. J. Org. Chem. 1990,
55, 6268.
Syntheses of the two acyclic comparison compounds are
depicted in Scheme 3. The â-2-naphthylalanine phosphonate
unit was built up from tert-butyl (dimethylphosphono)acetate
and 2-(bromomethyl)naphthalene, with Schmidt degradation of
the carboxyl group to provide the amino moiety. After
introduction of the valine residue and formylation, the L- and
(7) Bartlett, P. A.; Giangiordano, M. A. J. Org. Chem. 1996, 61, 3433.
(8) Fraser, M. E.; Strynadka, N. C. J.; Bartlett, P. A.; Hanson, J. E.;
James, M. N. G. Biochemistry 1992, 31, 5201.
(9) Blundell, T. L.; Sibanda, B. L.; Hemmings, A.; Foundling, S. F.;
Tickle, I. J.; Pearl, L. H.; Wood, S. P. In Molecular Graphics and Drug
Design; Burgen, A. S. V., Roberts, G. C. K. , Tute, M. S., Eds.; Elsevier
Science Publishers: Amsterdam, The Netherlands, 1986; p 321.
(10) Lauri, G.; Bartlett, P. A. J. Comput.-Aided Mol. Design 1994, 8,
51.
(12) Campbell, D. A.; Bermak, J. C. J. Org. Chem. 1994, 59, 658.
(13) McKenna, C. E.; Higa, M. T.; Cheung, N. H.; McKenna, M.-E.
Tetrahedron Lett. 1977, 155.
(11) Oleksyszyn, J.; Subotkowska, L.; Mastalerz, P. Synth. Commun.
1979, 985.

4602 J. Am. Chem. Soc., Vol. 120, No. 19, 1998
Meyer and Bartlett
Scheme 2
Scheme 3
D-diastereomers of 24 were separated by chromatography.
Deformylation competes with hydrolysis of the phosphonate
ester under alkaline conditions; however, the ester is cleaved
selectively under nucleophilic conditions with lithium pro-
panethiolate in HMPA.14 Formation of the â-phenyllactate ester,
phosphonate ester cleavage, and isolation of the acyclic
analogues 2-L and 2-D were accomplished as described above
for the macrocyclic inhibitor. Deformylation was also a
significant side reaction in the triflate alkylation step, but it could
be minimized by using only a slight excess of triflate, limiting
the reaction time, and purifying the product immediately after
workup. The configurations of the diastereomers of 2 were
inferred from their inhibition constants, as described below. The
other comparison compound, 3, was prepared from the phos-
phonate analogue of glycine, 26, and 1-naphthaleneacetic acid
using similar procedures (Scheme 3).
established assay procedures using Ac-Ala-Ala-Lys-(p-nitro)-
Phe-Ala-Ala-NH2 as substrate (Table 1).6 Although each
inhibitor carries a significant amount of hydrophobic functional-
ity, the phosphonate anion conveys adequate solubility at the
assay pH of 3.5. Not surprisingly, there is a large difference in
activity between the diastereomeric inhibitors 1-L and 1-D and
between 2-L and 2-D. Our inference that the more inhibitory
analogues have the natural L-configuration at the stereocenter
adjacent to phosphorus was confirmed in the case of the
Inhibition of Penicillopepsin. The inhibition constants Ki
were determined for the cyclic and acyclic phosphonates by
(14) Bartlett, P. A.; Johnson, W. S. Tetrahedron Lett. 1970, 4459.

Macrocyclic Inhibitors of Penicillopepsin. 1
J. Am. Chem. Soc., Vol. 120, No. 19, 1998 4603
Table 1. Inhibition of Penicillopepsin by Acyclic and
P1-P3-Linked Cyclic Phosphonatesa
This difference is equivalent to that between the cyclic and
acyclic inhibitors, showing that the latter have higher hydro-
phobic surface areas than the former. Thus, the enhanced
binding of the macrocycle 1-L over the controls 2-L and 3
cannot be attributed to differences in hydrophobic surface area;
indeed, this difference should favor the acyclic analogues by
0.5-1 kcal/mol.15
inhibitor
Ki (µM)
1-D
16-D
16-L
2-D
3
>1000
>1000
>1000
130
110
NMR Structure Determination. The degree to which the
conformation of an unbound inhibitor corresponds to that in
the binding site is also important, since the entropic advantage
of a constrained inhibitor may be negated by the enthalpic
penalty from a high-energy bound conformation. The solution
structures of the macrocycles 16 and 1 were therefore deter-
mined using a combined 1D 1H NMR and molecular modeling
approach,16 both to determine their conformations as well as to
assign the configuration adjacent to phosphorus. 1D 1H NMR
spectra were obtained at several temperatures in pH 3.5 buffer
to simulate the assay conditions. The nonaromatic resonances
corresponding to the backbone hydrogens were assigned on the
basis of coupling patterns and chemical shift (Table 2). A
TOCSY experiment with DANTE water suppression was used
to assign the amide hydrogens of 1-D and 16-D, since coupling
of the NH of naphthylalanine phosphonate (NalaP) to phosphorus
was not observed. The diastereotopic hydrogens at the naph-
thalene benzylic positions could not be assigned stereospecifi-
cally.
2-L
1-L
7.6
0.80
a Determined at 25 °C, pH 3.5, in 100 mM NaOAc.
Figure 1. Bonds with significant conformational constraint.
macrocycles by NMR studies of the precursors 16-L and 16-
D. Of the active inhibitors, the macrocycle 1-L binds an order
of magnitude more strongly than the acyclic comparison
compound 2-L and more than 2 orders of magnitude tighter
than 3. The importance of the P′ residues is underscored by
the absence of inhibition observed for the macrocyclic analogues
16 which lack the â-phenyllactate residue.
Analysis. The inhibitory potency across the series 3 f 2-L
f 1-L parallels the degree to which conformational freedom is
reduced in the progression from linear to branched to cyclic
topologies. Attachment of the 2-naphthylmethyl group to the
middle of the chain, as in 2-L, restricts its conformational
mobility, as well as that of the neighboring bonds in the peptide
backbone, to a much greater extent than occurs when this group
is attached at the end of the chain, as in 3. Cyclization of the
molecule to 1-L thus results in a larger loss of flexibility for 3
than for 2-L (Figure 1).
Implicit in any quantitative assessment of the entropic factor
in these binding differences is the assumption that the inhibitors
bind in the same (congruent) orientation and share the same
interactions with the enzyme and with solvent. In designing
these analogues, we expected that the backbone atoms of the
phosphonate peptides would adopt the same orientation in the
active site as that found for the linear phosphonate 4 and related
inhibitors.8 Since the hydrophobic S1 and S3 binding pockets
are well-defined, we considered it unlikely that the 1-L or 2-L
side chains would bind in abnormal orientations. We were less
confident of the bound conformation of the linear inhibitor 3,
since the penicillopepsin active site opens up near the S3 site
and the naphthyl moiety could extend outward, rather than
folding around toward the S1 pocket. As reported in the
accompanying structural paper,3 the conformations of the
macrocycle 1-L and acycle 2-L in the enzyme active site proved
to be comparable, but not identical, while that of 3 turned out
to be quite different. Thus, the observed differences in affinity
between the three inhibitors represent lower limits to the true
differences were they to adopt congruent orientations.
1
From the 1D H coupling constants, possible φ angles were
calculated from the appropriate Karplus relationship (Figure 3).
The RH-âH coupling patterns of the NalaP moieties in both
16-L and 16-D indicate well-defined conformations for the
macrocyclic rings with a ø1 angle of +60° or -60°; the large
coupling constants between valine RH and âH in turn suggest
a predominant anti relationship between these hydrogens. The
chemical shift dependencies (∆δ/∆T) for the amide hydrogens
indicate that they are not significantly shielded from solvent,
as would be expected if they participate in intramolecular
hydrogen bonds.
Molecular modeling was used to generate potential solution
conformations for 16-L and 16-D that were energetically
reasonable and in accord with the observed NMR spectra. Since
the configuration adjacent to phosphorus was ambiguous, two
separate Monte Carlo conformational searches were conducted
for each diastereomer, applying each set of the experimentally
determined φ angles to the (R)- and (S)-configurations. The
trans geometry was deduced for both of the amide bonds in
each macrocycle, since searches with the cis configurations in
either or both of them yielded structures from 7 to 20 kcal/mol
higher in energy than the minima for the all-trans conformers.
The conformations generated from the four searches were
clustered according to the backbone atoms, amide oxygens, and
phosphorus atom. Three distinct families of structures within
5 kcal/mol of the global minimum and which satisfy the NMR
data were obtained for the compound that proved to be 16-L
(Figure 4a); similarly, six families were found for 16-D (Figure
4b). However, the configurations of the two isomers could not
be determined unambiguously from these data alone: with each
set of NMR data, low-energy conformations were found for both
the (R)- and (S)-structures.
The stereochemical ambiguity was resolved by the NOESY
spectra of the two isomers (Table 3). In the spectrum of 16-L,
The acyclic inhibitors 2 and 3 differ in composition from
the cyclic analogue 1 by a carbon atom; that is, a methylene
unit is replaced by two hydrogens. For comparison, the van
der Waals surface of ethylbenzene is 183 Å2, while benzene +
methane have 194 Å2 of surface area when they are in van der
Waals contact and 209 Å2 when they are separated (Figure 2).
(15) Eriksson, A. E.; Baase, W. A.; Zhang, X.-J.; Heinz, D. W.; Blaber,
M.; Baldwin, E. P.; Matthews, B. W. Science 1992, 255, 178. Sharp, K.
A.; Nicholls, A.; Friedman, R.; Honig, B. Biochemistry 1991, 30, 9686.
(16) Sefler, A. M.; Lauri, G.; Bartlett, P. A. Int. J. Pept. Protein Res.
1996, 48, 129.

4604 J. Am. Chem. Soc., Vol. 120, No. 19, 1998
Meyer and Bartlett
Figure 3. Dihedral constraints within the macrocyclic rings.
Figure 2. Effect on solvent-accessible surface area from removal of a
methylene linkage.
Figure 4. Low-energy conformer families for 16-L that satisfy NMR
constraints.
Table 2. 1H NMR Characterization of 16-L and 16-Da
δ (ppm)
J (Hz)
∆δ/∆T (ppb/K)
coupled
Table 3. Distances (Å) from Low-Energy Conformationsa and
NOE Cross-peaks Observed for 16-L and 16-D
hydrogen 16-L 16-D nucleus 16-L 16-D 16-L
16-D
NalaP NH 8.16b 7.74
Val NH 6.62 8.76
NalaP RH 3.92 3.96
RH
RH
NH
âH
â′H
P
NH
âH
H′
10
10
10
2
13
17
10
11
18
18
2
10
8
8.0
4.4
9.0
5.4
structureb
NalaP NHc to Val RH NalaP NH to Val âH
c
16-L, (R)-configuration
16-L, (S)-configuration
NOE observed for 16-L
2.2
3.6
+
3.8-4.7
2.3-4.0
-
2c
13c
13c
8
11
15
15
2
16-D, (R)-configuration
16-D, (S)-configuration
NOE observed for 16-D
2.2-2.3
3.8-4.7
2.1-4.3
+
Val RH
4.18 3.84
3.6
-
Nac H
Nac H′
4.21 4.23
4.11 3.79
a Distance ranges encompass the separate families of minimum
energy structures in Figures 2 & 3. b 16-L, (R)-configuration indicates
minimum energy conformations determined for (R)-configuration
adjacent to phosphorus using 1D 1H NMR data obtained with compound
that proved to be 16-L. c NalaP ) 2-naphthyl-1-aminoethyl phosphonate
moiety.
H
NalaP âH 3.32 3.22
RH
â′H
P
RH
âH
P
13
4
13
13
5
13
4
13
13
5
NalaP âH′ 2.82 3.09
using the same protocol, and the chemical shifts and coupling
constants for the macrocyclic backbone hydrogens were very
similar in the two series. A constrained Monte Carlo search of
10 000 steps was conducted for each isomer of 1 as described
above, assuming the trans configuration of the amides and the
(R)- and (S)-configurations adjacent to phosphorus for 1-L and
1-D, respectively. The structures generated were clustered and
filtered against the experimentally determined ø1 angles as
described above. Six families of conformers within 5 kcal/mol
of the global minimum were found for 1-L, of which all but
one were virtually superimposable with respect to the ring
systems; furthermore, this ring conformation was identical to
a In 90% H2O/10% D2O, 100 mM CD3CO2D, 1 mN KH2PO4, pH
3.5 at 29 °C. NalaP ) 2-naphthyl-1-aminoethyl phosphonate moiety;
Nac ) naphthylacetamide CH2 unit. b Chemical shift at 0 °C. c Mea-
sured in 100% D2O since resonance was obscured in H2O.
an NOE cross-peak is observed from NalaP NH to Val RH, but
none to Val âH; for 16-D, the pattern is reversed, with a cross-
peak for NalaP NH to Val âH but not to Val RH. The low-
energy conformers with the (R)-stereochemistry fit this pattern
for 16-L, and conversely, those with the (S)-configuration fit
that for 16-D.
The structures of the inhibitors 1-L and 1-D were determined

Macrocyclic Inhibitors of Penicillopepsin. 1
J. Am. Chem. Soc., Vol. 120, No. 19, 1998 4605
Figure 7. Comparison of original modeled structure (bold) with
experimental conformations of 1-L.
Figure 5. Low-energy conformer familes for 1-L and 1-D that satisfy
NMR constraints.
Figure 8. Conformational change on minimization in active site of
penicillopepsin (minimized structures shown in bold with highlighted
atoms).
(-106.4 f -106.5 kcal/mol). Thus, it does not appear that
there is any significant enthalpic penalty for 1-L to pay on
binding to the enzyme. On the other hand, the higher energy
solution conformer, 1-Lb, is not as compatible with the active
site and on minimization adopts a new conformation resulting
from rotation about the NalaP CR-Câ bond and movement of
the aromatic linker away from one wall of the binding pocket
(Figure 8b). The NMR data suggest that this conformation for
1-L is not populated in solution; however, it is calculated to be
1.6 kcal/mol lower in energy than 1-Lb itself (-103.4 f -105.0
kcal/mol).
In view of the close correspondence between the conformation
determined for the macrocyclic inhibitor in solution (1-La) and
that observed in the crystal structure of the complex with
penicillopepsin, described in a following paper,3 it appears that
the design of 1-L represents a successful application of the
structure-based design strategy. On the other hand, since these
structural studies show that the bound conformations for the
cyclic and acyclic inhibitors differ in significant respects, we
can only draw qualitative conclusions regarding the impact of
conformational restriction from the observed differences in
binding affinity.
Figure 6. Comparison of bound form of acyclic peptide phosphonate
4 (bold) with solution conformations of 1-L.
that in 16-L (Figure 5a). In the case of diastereomer 1-D, all
eight families found were essentially the same, and the rings
were superimposable with 16-D (Figure 5b).
Success of the Design Strategy. Overlap of the solution
conformation of the highest affinity inhibitor 1-L and the bound
conformations of the linear phosphonate 4 shows a high degree
of similarity throughout the backbone atoms of these peptide
analogues (Figure 6).17 Thus, as a means of constraining the
peptide backbone in these inhibitors to the conformation adopted
in the binding site of penicillopepsin, the strategy appears to
have been successful. On the other hand, the orientation adopted
by the naphthalene moiety itself, relative to the peptide
backbone, is displaced from that in our original model (Figure
7). To estimate the impact that this difference might have on
the binding affinity, we docked each of the two low-energy
solution conformations for 1-L in the active site of penicil-
lopepsin and minimized the structure while holding the protein
fixed. The macrocyclic ring of the lower energy conformer,
1-La, underwent very little movement and the energetic
consequence was small (Figure 8a); the solvated energies
calculated for the two ring conformers (with methyl ester in
place of the phenyllactate ester) differed by only 0.1 kcal/mol
Experimental Section18
Ethyl 7-Bromo-3,4-tetrahydro-1-naphthaleneacetate (6a) and
Ethyl 7-Bromo-1,2,3,4-tetrahydronaphth-1-ylideneacetate (6b). A
solution of diisopropylamine (5.8 mL, 29.6 mmol) in 15 mL of THF
was cooled to -78 °C, and n-butyllithium (12.1 mL, 2.42 M in hexanes)
was added. After 5 min, a solution of EtOAc (2.86 mL, 29.3 mmol)
in 15 mL of THF was added to the preformed lithium diisopropylamide
(LDA) solution by cannula over 10 min. The resulting mixture was
stirred at -78 °C for 15 min before addition of tetralone 5 (6.28 g,
27.9 mmol) in 20 mL of THF over 20 min. After 2.5 h, the mixture,
which had warmed to -25 °C, was recooled to -78 °C and 6 mL of
12 M HCl was added over 5 min with simultaneous precipitation of a
(17) The conformational differences depicted among the phosphonate
ester and phenyllactate moieties are not significant; the conformers of 1-L
generated in the Monte Carlo search were clustered according to the
macrocycle backbone and phosphorus atoms only. Hence, each conformer
family contains many different conformers for the acyclic region of the
structure.

4606 J. Am. Chem. Soc., Vol. 120, No. 19, 1998
Meyer and Bartlett
white solid. The mixture was allowed to warm to 25 °C and partitioned
between 50 mL of H2O and 75 mL of ether. The aqueous layer was
extracted with 50 mL of ether, and the organic layers were combined,
washed with 40 mL of 1 N HCl, 40 mL of H2O, 40 mL of saturated
NaHCO3, and 40 mL of brine, dried, and concentrated to afford 8.49
g of the crude tertiary alcohol as a yellow oil. This oil (8.49 g) and
350 mg (1.8 mmol) of p-toluenesulfonic acid monohydrate were
dissolved in 30 mL of benzene, and the resulting solution was heated
at reflux for 1.5 h with azeotropic removal of H2O using a Dean-
Stark trap. The cooled solution was partitioned between 75 mL of
ether and 25 mL of H2O, and the organic layer was washed with 50
mL of H2O, 50 mL of saturated NaHCO3, and 50 mL of brine, dried,
and concentrated. The resulting 7.70 g (96%) of amber oil proved to
be a 3.5:1 mixture of the endo- and exocyclic alkenes 6a (endo) and
6b (exo), previously prepared by a different method and reported
without complete characterization.20 The two isomers were separable
by column chromatography (10% EtOAc/hexanes). For 6a: 1H NMR
(300 MHz) δ 1.26 (t, 3, J ) 7), 2.31 (m, 2), 2.73 (t, 2, J ) 8), 3.39 (s,
2), 4.16 (q, 2, J ) 7), 6.04 (t, 1, J ) 5), 7.00 (d, 1, J ) 8), 7.25 (m,
1), 7.30 (d, 1, J ) 2); 13C NMR δ 14.1, 23.0, 27.4, 38.9, 60.8, 120.0,
125.7, 129.1, 129.4, 129.7, 130.4, 135.1, 136.2, 171.4; MS (FAB) m/z
294.0, 296.0 (M+), 295.0, 297.0 (MH+); a sample of this material was
distilled bulb-to-bulb at 185 °C/2 Torr. Anal. Calcd for C14H15O2Br:
C, 56.97; H, 5.12. Found: C, 57.16; H, 5.43. For 6b: 1H NMR (300
MHz) δ 1.32 (t, 3, J ) 7), 1.84 (tt, 2, J ) 6, 6), 2.73 (t, 2, J ) 6), 3.16
(dt, 2, J ) 2, 6), 4.20 (q, 2, J ) 7), 6.28 (t, 1, J ) 2), 7.02 (d, 1, J )
8), 7.37 (dd, 1, J ) 2, 8), 7.77 (d, 1, J ) 2); 13C NMR δ 14.3, 22.4,
27.6, 29.7, 59.8, 113.6, 119.9, 127.6, 130.7, 132.2, 136.2, 139.0, 153.0,
166.6; a sample of this material was distilled bulb-to-bulb at 110 °C/
0.25 Torr.
Ethyl 7-Bromo-1-naphthaleneacetate (7). Bromine in CCl4 (140
µL, 1.0 M, 0.14 mmol) was added to a stirred solution of the endo
alkene 6a (21 mg, 0.07 mmol) in 0.5 mL of CCl4 at 25 °C until a
red-orange color persisted. Stirring was continued for an additional 2
h at which time the solution was washed with 1 mL of 10% NaHSO3,
1 mL of 5% NaHCO3, and 1 mL of brine, dried, and concentrated.
The resulting oil was redissolved in 0.5 mL of benzene, DBU (25 µL,
0.17 mmol) was added, and the resulting green, cloudy suspension was
stirred at 25 °C for 12 h. The mixture was diluted with 2 mL of ether,
washed with 2 mL of H2O, 2 mL of 1 N HCl, 2 mL of 5% NaHCO3,
and 2 mL of brine, dried, and concentrated. The crude product was
purified by column chromatography (50% CHCl3/hexanes) to provide
20 mg (95%) of the aromatized product 7. For 7: 1H NMR (300 MHz)
δ 1.25 (t, 3, J ) 7), 4.00 (s, 2), 4.16 (q, 2, J ) 7), 7.42 (m, 2), 7.56 (m,
1), 7.71 (m, 2), 8.16 (m, 1); 13C NMR δ 14.1, 39.0, 61.1, 120.6-133.3
(10 peaks), 171.1; MS (FAB) m/z 292.0, 294.0 (M+), 293.0, 295.0
(MH+); a sample of this compound was further purified by sublimation
(mp 39-42 °C). Anal. Calcd for C14H13O2Br: C, 57.36; H, 4.47.
Found: C, 57.23; H, 4.62.
(2.3 mL, 7.4 mmol) was added by syringe to a solution of bromide 7
(2.16 g, 7.4 mmol) in 7 mL of benzene containing a catalytic amount
of Pd(PPh3)4 (0.43 g, 0.37 mmol), and the resulting mixture was heated
at reflux for 15 h. After cooling, the mixture was diluted with 50 mL
of ether and added to 50 mL of 10% aqueous KF; the biphasic mixture
was stirred vigorously for 30 min and filtered. The filter pad was
washed with 50 mL of ether, the aqueous layer was separated, and the
ethereal layer was dried and concentrated. The resulting cloudy yellow
oil was purified by chromatography on a short silica plug (50 × 70
mm, 10% ether/hexanes) to afford 1.53 g (82%) of the product 8 as a
pale yellow oil. For 8: 1H NMR δ 1.24 (t, 3, J ) 7), 3.59 (d, 2, J )
7), 4.05 (s, 2), 4.16 (q, 2, J ) 7), 5.13 (dd, 1, J ) 2, 10), 5.15 (dd, 1,
J ) 2, 17), 6.06 (dddd, 1, J ) 10, 17, 7, 7), 7.37 (m, 3), 7.79 (m, 3);
13C NMR δ 14.1, 39.2, 40.6, 60.9, 116.1, 122.8-132.5 (9 peaks), 137.3,
138.0, 171.6; MS (FAB) m/z 254.0 (M+), 255.0 (MH+). Anal. Calcd
for C17H18O2: C, 80.28; H, 7.13. Found: C, 80.12; H, 7.03.
Ethyl 7-(2,3-Dihydroxypropyl)-1-naphthaleneacetate (9). Os-
mium tetroxide (23 mg, 0.09 mmol) was added to a solution of the
alkene 8 (3.58 g, 14.1 mmol), N-methylmorpholine N-oxide (1.73 g,
14.8 mmol), and 3 mL of H2O in 17 mL of acetone. The black mixture
was stirred for 17 h and diluted with 150 mL of EtOAc, and the solution
was washed with 10% Na2S2O4 (100 mL and 25 mL), 100 mL of H2O,
100 mL of 1 M HCl, and 100 mL of brine. The organic layer was
dried and concentrated to leave a brown oil which was purified by
chromatography through a silica plug (5% MeOH/CH2Cl2) and crystal-
lization from ether to afford 2.25 g of the diol 9 as a light-brown solid
(mp 70-73 °C). The mother liquor was purified by column chroma-
tography to provide an additional 450 mg of product (combined yield
67%). For 9: 1H NMR (CDCl3/0.1% H2O) δ 1.23 (t, 3, J ) 7), 2.96
(dd, 1, J ) 8, 13), 3.00 (dd, 1, J ) 5, 13), 3.58 (dd, 1, J ) 7, 11), 3.73
(dd, 1, J ) 3, 11), 4.04 (m, 1), 4.05 (s, 1), 4.15 (q, 2, J ) 7), 7.38 (m,
3), 7.77 (m, 1), 7.84 (m, 2); 13C NMR δ 14.1, 39.1, 40.1, 61.0, 65.8,
72.9, 123.9-135.9 (10 peaks), 171.8; MS (FAB) m/z 288.0 (M+), 289.0
(MH+). Anal. Calcd for C17H20O4: C, 70.81; H, 6.99. Found: C,
70.45; H, 6.96.
Ethyl 7-(3-Oxopropyl)-1-naphthaleneacetate (10). A mixture of
sodium periodate (200 mg, 0.93 mmol) and diol 9 (188 mg, 0.65 mmol)
in 3 mL of ether and 300 µL of H2O was stirred at 25 °C for 75 min
and diluted with 5 mL of ether and 5 mL of H2O. The aqueous layer
was extracted with 3 mL of ether, and the combined organic layer was
dried and concentrated to provide 163 mg (98%) of aldehyde 10 as an
amber oil, which was carried on to the next reaction without purification.
For 10: 1H NMR δ 1.22 (t, 3, J ) 7), 3.86 (d, 2, J ) 2), 4.04 (s, 2),
4.14 (q, 2, J ) 7), 7.25-7.43 (m, 3), 7.75-7.86 (m, 3), 9.81 (t, 1, J )
2); 13C NMR δ 14.1, 39.1, 51.0, 60.9, 124.6-132.9 (10 peaks), 171.3,
199.1; MS (FAB) m/z 257.1 (MH+).
Methyl 7-[2-((N-Phenylmethoxycarbonyl)amino)-2-((dimethoxy)-
phosphinyloxy)ethyl]-1-naphthaleneacetate (11). A solution of al-
dehyde 10 (163 mg, 0.64 mmol), benzyl carbamate (124 mg, 0.82
mmol), and triphenyl phosphite (262 mg, 0.85 mmol) in 0.7 mL of
glacial HOAc was stirred at 25 °C for 45 min and at 90 °C for 45 min,
cooled to 25 °C, and diluted with 8 mL of EtOAc. This solution was
washed with 5 mL of H2O, 3 mL of saturated NaHCO3, and 3 mL of
brine, dried, and concentrated, and the resulting yellow oil was purified
by silica gel chromatography (25% to 50% EtOAc/hexanes), to give
Ethyl 7-(2-Propenyl)-1-naphthaleneacetate (8). Allyltributyltin
(18) General. Unless otherwise noted, reagents and solvents were
obtained from commercial suppliers and used without further purification
except for drying. All moisture or oxygen sensitive reactions were carried
out under an atmosphere of N2 in dry solvents. Molecular sieves were oven-
dried before use. Unless otherwise noted, reaction workups culminated in
drying the solution over MgSO4 and concentrating under aspirator pressure
using a rotary evaporator, followed by further evacuation under high
vacuum. Flash chromatography was performed by the method of Still, Kahn,
and Mitra,19 using 60-mesh silica gel. Melting points were determined in
157 mg of crude diphenyl phosphonate as a gummy semisolid.
A
portion of this material (52 mg, 0.085 mmol) was dissolved in 1.0 mL
of MeOH with anhydrous K2CO3 (33 mg, 0.24 mmol), and the resulting
suspension was stirred at 25 °C for 2.75 h. The mixture was diluted
with 5 mL of ether and 5 mL of 0.5 N H2SO4, the aqueous fraction
was extracted with 5 mL of ether, and the combined organic layer was
washed with 5 mL of H2O and 5 mL of brine, dried, and concentrated.
The oily residue was purified by silica gel chromatography (10% to
30% ether/CH2Cl2), affording 29 mg (28% from 10) of trimethyl ester
11 as a colorless, gummy semisolid. For 11: 1H NMR δ 3.06 (m, 1),
3.42 (m, 1), 3.62 (s, 3), 3.71 (d, 3, J ) 11), 3.76 (d, 3, J ) 11), 4.01
(s, 2), 4.56 (m, 1), 4.94 (s, 2), 5.18 (d, 1), 7.07-7.23 (m, 5), 7.36-
7.43 (m, 3), 7.74-7.80 (m, 3); 13C NMR δ 36.1, 38.7, 48.2 (d), 52.0,
53.0 (d), 53.2 (d), 66.8, 123.9-136.2 (16 peaks), 155.8, 171.8; 31P NMR
δ 27.1 (major rotamer), 26.5 (minor rotamer); MS (FAB) m/z 486.3
1
Pyrex capillaries and are uncorrected. Unless otherwise indicated, H and
13C NMR spectra were obtained in CDCl3 solution; all chemical shifts are
reported in ppm relative to the solvent, except for 13C NMR spectra acquired
in D2O, which were referenced to CH3CN as internal standard (1.3 ppm).
An external standard of trimethyl phosphate (δ 3.086 ppm) in CDCl3 was
used for 31P spectra. 1H spectra were acquired at 400 or 500 MHz and 13
C
NMR spectra at 75 or 100 MHz, unless stated otherwise. NMR spectral
data are reported as (MHz, solvent) chemical shift (multiplicity, number of
hydrogens, coupling constants in Hz). Combustion analyses and mass spectra
were obtained from the Mass Spectrometry Laboratory and the Microana-
lytical Laboratory, Department of Chemistry, University of California,
Berkeley.
(19) Still, W. C.; Kahn, M.; Mitra, A. J. Org. Chem. 1978, 43, 2923.
(20) Gruber, R.; Cagniant, D.; Cagniant, P. Bull. Soc. Chim. Fr. 1983,
96.

Macrocyclic Inhibitors of Penicillopepsin. 1
J. Am. Chem. Soc., Vol. 120, No. 19, 1998 4607
(MH+). Anal. Calcd for C25H28NO7P: C, 61.85; H, 5.81; N, 2.89.
Found: C, 61.53; H, 5.92; N, 2.87.
salt as a white solid. This material in turn was dissolved in 5 mL of
dry 1,4-dioxane, and the solution was added by syringe pump over 9
h to a stirring mixture of 180 mL of 1,4-dioxane and 32 mL of pyridine
at 50 °C. The mixture was stirred for an additional 2.5 h at 50 °C and
then for 12 h at 25 °C. The solvent was evaporated and the glassy
residue was chromatographed (2% MeOH/CHCl3 to 1% HOAc/10%
MeOH/CHCl3) to yield 42 mg of cyclic diastereomer 15-L and 37 mg
of 15-D (65% total yield from 13). For 15-L: 1H NMR ((CD3)2SO)
δ 0.76 (d, 3, J ) 7), 0.82 (d, 3, J ) 7), 1.93 (m, 1), 2.96 (m, 1), 3.16
(m, 1), 3.57 (d, 3, J ) 11), 3.68 (d, 3, J ) 11), 3.89 (d, 1, J ) 17),
4.08 (d, 1, J ) 17), 4.14 (m, 1), 4.25 (m, 1), 7.38 (m, 3), 7.51 (d, 1, J
) 9), 7.58 (s, 1), 7.62 (d, 1, J ) 9), 7.79 (m, 2); 13C NMR ((CD3)2SO)
δ 18.5, 19.3, 27.1, 34.9, 42.0, 44.7, 46.2, 52.7, 53.1, 53.12, 59.3, 125.1,
127.0, 127.6, 128.3, 128.7, 131.1, 132.0, 132.3, 133.7, 133.8, 169.9,
171.7; 31P NMR (DMSO-d6) δ 26.9. Anal. Calcd for C21H27N2O5P:
C, 60.28; H, 6.51; N, 6.70. Found: C, 60.54; H, 6.56; N, 6.30.
For 15-D: 1H NMR ((CD3)2SO) δ 0.89 (d, 3, J ) 7), 0.99 (d, 3, J
) 7), 1.97 (m, 1), 3.08 (m, 2), 3.63 (d, 1, J ) 14), 3.71 (d, 3, J ) 11),
3.73 (d, 3, J ) 11), 3.76 (m, 1), 4.01 (m, 1), 4.05 (d, 1, J ) 14), 7.35
(m, 3), 7.74 (dd, 1, J ) 2, 9), 7.80 (d, 1, J ) 8), 8.02 (d, 1, J ) 9),
8.05 (s, 1), 8.26 (d, 1, J ) 10); 13C NMR ((CD3)2SO) δ 18.7, 19.8,
29.2, 33.4, 41.8, 45.0, 46.6, 52.8, 52.9, 53.0, 53.1, 62.7, 125.2, 125.24,
126.5, 127.4, 128.1, 128.4, 131.2, 132.1, 133.6, 134.1, 170.5, 170.9;
31P NMR ((CD3)2SO) δ 27.8. Anal. Calcd for C21H27N2O5P: C, 60.28;
H, 6.51; N, 6.70. Found: C, 60.48; H, 6.58; N, 6.49.
Cyclo-7-[(2R)-((N-valyl)amino)-2-(hydroxymethoxyphosphinyloxy)-
ethyl]-1-naphthaleneacetamide (16-L). A solution of dimethyl phos-
phonate 15-L (81 mg, 0.19 mmol) and LiOH (0.7 mL, 0.93 M) in 2
mL of DMSO was stirred at 25 °C for 20 h. The resulting white slurry
was diluted with 10 mL of H2O and lyophilized, and the residue was
passed over a Dowex/H+ column in 1:1 MeOH/H2O. The eluate was
partially concentrated and then lyophilized to provide 75 mg (96%) of
the desired phosphonic acid 16-L as a white powder. For characteriza-
tion, a sample was converted to the sodium salt using Dowex/Na+ and
1:1 MeOH/H2O as the eluant. For 16-L: 1H NMR (D2O) δ 0.62 (d,
3, J ) 7), 0.67 (d, 3, J ) 7), 1.48 (m, 1), 2.66 (ddd, 1, J ) 4, 13, 13),
3.20 (ddd, 1, J ) 2, 5, 13), 3.44 (d, 3, J ) 10), 3.80 (ddd, 1, J ) 2, 13,
17), 3.97 (d, 1, J ) 18), 4.06 (d, 1, J ) 18), 4.06 (d, 1, J ) 11), 7.17
(s, 1), 7.34 (m, 3), 7.77 (m, 1), 7.79 (d, 1, J ) 9); 13C NMR (D2O) δ
17.9, 18.7, 27.7, 37.5, 42.4, 48.1 (d, J ) 144), 52.7 (d, J ) 6), 61.5,
124.6, 125.8, 127.9, 128.8, 129.8, 130.0, 131.2, 132.9, 136.2, 136.3,
172.1 (d, J ) 7), 176.4; 31P NMR (D2O) δ 20.4. HRMS (FAB) m/z
calcd for C20H24N2O5PNa (MH+) 427.1399, found 427.1412.
Methyl 7-[2-((N-tert-Butoxycarbonylvalyl)amino)-2-(dimethoxy-
phosphinyloxy)ethyl]-1-naphthaleneacetate (12). A suspension of
carbamate 11 (250 mg, 0.51 mmol) and palladium on carbon (214 mg
5%, 0.10 mmol Pd) in 7 mL of MeOH was stirred under a hydrogen
atmosphere at 25 °C for 3.5 h. The mixture was filtered through Celite
and 0.45-µm nylon filter paper, and the pad was washed with 25 mL
of MeOH. The combined filtrate was concentrated to give 154 mg of
the free amine as a pale yellow oil which was dissolved along with
Boc-valine (105 mg, 0.48 mmol) and 1-hydroxybenzotriazole (HOBt)
(64 mg, 0.47 mmol) in 2.5 mL of CH2Cl2. This solution was cooled
to 0 °C, EDC (102 mg, 0.53 mmol) was added, and the mixture was
allowed to warm gradually to 25 °C and was stirred overnight. The
solution was diluted with 40 mL of EtOAc, washed with 10 mL of 1
M HCl, 10 mL of H2O, 10 mL of saturated NaHCO3 and 10 mL of
brine, dried, and concentrated. The crude product was purified by
chromatography (70:28:2 EtOAc/hexanes/MeOH) and recrystallized
from ether to provide 195 mg (70%) of the product 12 as a mixture of
two diastereomers (mp 130-132 °C). For 12: 1H NMR (CD3CN) δ
0.06 (d, 1.35, J ) 7), 0.40 (d, 1.4, J ) 7), 0.52 (d, 1.2, J ) 7), 0.58 (d,
1.35, J ) 7), 0.69 (d, 0.6, J ) 7), 1.29 (s, 4.5), 1.33 (s, 4.5), 1.49 (m,
0.5), 1.78 (m, 0.5), 3.02 (m, 1), 3.36 (m, 1), 3.65 (s, 1.5), 3.67 (s, 1.5),
3.74 (d, 1.5, J ) 11), 3.75 (d, 1.5, J ) 11), 3.75 (d, 1.5, J ) 11), 3.77
(d, 1.5, J ) 11), 4.08 (m, 2), 4.73 (m, 1), 7.08 (d, 0.5, J ) 9), 7.17 (d,
0.5, J ) 9), 7.40 (m, 3), 7.77 (m, 3); 13C NMR (CD3CN) δ 16.4, 16.7,
18.5, 18.8, 28.1, 28.5, 30.4, 31.0, 35.7, 38.4, 38.6, 51.9, 52.0, 52.6,
52.7, 59.3, 59.7, 79.1, 124.1-134.4 (14 peaks), 155.6, 171.4, 171.8,
172.0; 31P NMR (CD3CN)
δ 27.5, 27.3. Anal. Calcd for
C27H39N2O8P: C, 58.90; H, 7.14; N, 5.09. Found: C, 58.95; H, 7.21;
N, 5.14.
7-[2-((N-tert-Butoxycarbonylvalyl)amino)-2-(dimethoxyphos-
phinyloxy)ethyl]-1-naphthaleneacetic Acid (13). A solution of methyl
ester 12 (183 mg, 0.33 mmol) and LiOH (450 µL, 0.87 M) in 1.1 mL
of MeOH was stirred at 25 °C for 2.25 h, diluted with 15 mL of H2O,
and washed with 10 mL of ether. The ether layer was back-extracted
with 10 mL of 10% NaHCO3, and the aqueous layers were combined,
acidified to pH 1 with 2 N H2SO4, and extracted with EtOAc (2 × 15
mL). The organic layers were combined, washed with 10 mL of brine,
dried, and evaporated, and the residue was freeze-dried from benzene
to provide 157 mg (89%) of the desired carboxylic acid 13 as a white
solid. For 13: 1H NMR ((CD3)2SO) δ 0.01 (d, 1.5, J ) 7), 0.21 (d, 3,
J ) 7), 0.21 (d, 1.5, J ) 7), 0.53 (d, 1.5, J ) 7), 1.30 (s, 4.5), 1.31 (s,
4.5), 1.44 (m, 0.5), 1.52 (m, 0.5), 2.93 (ddd, 0.5, J ) 8, 13, 13), 2.94
(m, 0.5), 3.22 (m, 0.5), 3.22 (m, 0.5), 3.60 (dd, 0.5, J ) 8, 9), 3.65 (m,
0.5), 3.68 (d, 1.5, J ) 10), 3.69 (d, 1.5, J ) 10), 3.87 (d, 0.5, J ) 16),
3.91 (d, 0.5, J ) 16), 4.07 (d, 0.5, J ) 16), 4.08 (d, 0.5, J ) 16), 4.53
(m, 0.5), 4.54 (m, 0.5), 6.33 (d, 0.5, J ) 9), 6.45 (d, 0.5, J ) 10), 7.36
(m, 3), 7.71 (m, 0.5), 7.73 (m, 0.5), 7.76 (d, 0.5, J ) 8) 7.77 (d, 0.5,
J ) 8), 7.85 (s, 0.5), 7.91 (s, 1), 8.16 (d, 0.5, J ) 9), 8.50 (d, 0.5, J )
9); 13C NMR ((CD3)2SO) δ 16.6, 18.1, 18.5, 18.7, 28.1, 28.1, 30.1,
30.7, 34.9, 34.9, 45.7 (d, J ) 155), 46.2 (J ) 154), 52.6 (d, J ) 7),
52.7 (d, J ) 7), 53. 1 (d, J ) 7), 53.3 (d, J ) 7), 59.1, 60.2, 77.8,
78.1, 124.6, 124.9, 125.0, 125.1, 1266, 126.7, 126.8, 126.9, 127.6, 127.8,
128.2, 128.3, 131.6, 131.9,, 131.9, 132.2, 132.2, 134.5, 134.6, 134.8,
154.9, 155.3, 171.0, 171.2, 173.0; 31P NMR ((CD3)2SO) δ 27.4, 27.7;
HRMS (FAB) m/z calcd C26H38N2O8P (MH+) 537.2366, found 537.2354.
Cyclo-7-[2-((N-valyl)amino)-2-(dimethoxyphosphinyloxy)ethyl]-
1-naphthaleneacetamide (15-L and 15-D). A solution of pentafluo-
rophenol (207 mg, 1.1 mmol), naphthaleneacetic acid 13 (157 mg, 0.29
mmol), and 1-(3-(dimethylamino)propyl)-3-ethylcarbodiimide hydro-
chloride (90 mg, 0.47 mmol) in 7 mL of dry CH2Cl2 was stirred at 25
°C for 10 h and then evaporated. The residue was dissolved in 21 mL
of an EtOAc/ether/MeOH mixture (15:5:1), and the solution was washed
with 10 mL each of 1 N H2SO4, H2O, 10% K2CO3, and brine, dried,
and evaporated. The product was purified by chromatography (60%
EtOAc/38% hexanes/2% MeOH) and lyophilization from benzene to
give 152 mg of pentafluorophenyl ester 14 as a white solid. This
activated ester was dissolved in 5 mL of 1,2-dichloroethane and 5 mL
of TFA and stirred at 25 °C for 2 h. The solution was evaporated and
the residue was lyophilized from benzene to afford 153 mg of the amine
Cyclo-7-[(2S)-((N-valyl)amino)-2-(hydroxymethoxyphosphinyloxy)-
ethyl]-1-naphthaleneacetamide (16-D). In a similar fashion, diaste-
reomer 15-D (16 mg, 0.038 mmol) was converted to 15 mg (100%) of
the sodium salt of acid 16-D. For 16-D: 1H NMR (D2O) δ 0.87 (d, 3,
J ) 7), 0.95 (d, 3, J ) 7), 1.90 (m, 1), 2.96 (ddd, 1, J ) 5, 13, 13),
3.09 (ddd, 1, J ) 2, 4, 13), 3.45 (d, 3, J ) 10), 3.62 (d, 1, J ) 15),
3.72 (d, 1, J ) 11), 3.83 (ddd, 1, J ) 2, 13, 13), 4.14 (d, 1, J ) 15),
7.29 (m, 3), 7.71 (m, 3); 13C NMR (D2O) δ 18.6, 19.6, 29.6, 35.1,
41.8, 48.1 (d, J ) 148), 52.4 (d, J ) 5), 64.4, 124.3, 125.9, 127.8,
128.1, 129.2, 129.4, 131.5, 132.6, 136.0, 136.1, 173.4 (d, J ) 4), 175.9;
31P NMR (D2O) δ 21.2.
Methyl 3-Phenyl-(2R)-trifluorosulfonyloxypropanoate (17).
A
catalytic amount of concentrated sulfuric acid (approximately 100 µL)
was added to a suspension of 3-phenyl-(2R)-hydroxypropanoic acid in
10 mL of MeOH. The mixture was heated at reflux overnight and
concentrated. The residue was redissolved in 50 mL of ethyl ether,
washed with 30 mL of saturated NaHCO3 and 30 mL of brine, dried,
and concentrated to leave 1.26 g of the methyl ester as a viscous oil,
which solidified upon standing. Concurrently, trifluoromethanesulfonic
anhydride was added to pyridine (360 µL, 4.47 mol) in 12 mL of CH2-
Cl2 at -20 °C, and the resulting white slurry was stirred for 5 min. A
portion of the methyl ester (706 mg, 3.92 mmol) was added, and the
mixture was stirred for 15 min and then warmed to 25 °C in a water
bath. The suspension was filtered through glass wool and evaporated,
and the residue was dissolved in 25 mL of ethyl ether and filtered
through a plug of silica gel. The filtrate was evaporated to give 1.04
g (75%) of the triflate 17 as a pale yellow oily liquid. Fpr 17: 1H
NMR (CDCl3) δ 3.22 (dd, 1, J ) 9, 15), 3.37 (dd, 1, J ) 4, 15), 3.84

4608 J. Am. Chem. Soc., Vol. 120, No. 19, 1998
Meyer and Bartlett
(s, 3), 5.27 (dd, 1, J ) 4, 9), 7.23 (d, 2, J ) 8), 7.31-7.35 (m, 3); 13
NMR (CDCl3) δ 38.1, 53.2, 83.7, 118.2 (q, J ) 320), 127.8, 128.7,
129.3, 133.2, 167.0.
C
) 7), 2.03 (m, 1), 2.92 (ddd, 1, J ) 4, 13, 13), 2.97 (dd, 1, J ) 6, 13),
3.10 (m, 2), 3.52 (s, 3), 3.67 (d, 1, J ) 15), 3.76 (d, 1, J ) 11), 3.84
(m, 1), 4.20 (d, 1, J ) 15), 4.80 (ddd, 1, J ) 6, 6, 6), 7.15-7.40 (m,
8), 7.73-7.77 (m, 3); 13C NMR (D2O) δ 20.9, 21.5, 31.6, 37.4, 41.9,
43.9, 51.1 (d, J ) 150), 55.0, 66.3, 76.7, 126.3, 127.9, 129.7, 130.2,
131.1, 131.2, 131.3, 132.1, 133.5, 134.6, 134.7, 138.0, 138.2, 175.3,
176.3, 177.8; 31P NMR (D2O) δ 19.0; HRMS (FAB) m/z calcd for
C29H33N2O7PNa (MH+) 575.1923, found 575.1933.
Enzyme Assays. General. All solutions were prepared with
distilled and deionized water. Stock solutions were filtered through
Rainin Nylon-66 0.45-µm filters. Assays were performed on a Kontron
Uvikon 860 UV-vis spectrophotometer equipped with a data station.
Temperature regulation ((0.2 °C) was provided by a Lauda Model
RM 20 circulating constant-temperature bath connected to a water-
jacketed sample cell holder. All assays were performed at 25 °C in
100 mM NaOAc at pH 3.5.
Enzyme Stock Solution. Penicillopepsin was obtained in the form
of a lyophilized powder as a gift from Theo Hofmann (University of
Toronto). A 0.8-mg sample of penicillopepsin was dissolved in 1.00
mL of 0.1 M KH2PO4 containing 10% (v/v) glycerol at 5 °C. The
resulting solution was centrifuged for 5 min on an Eppendorf Model
5414 centrifuge. The upper 900 µL of this solution was removed and
stored in a plastic tube at 5 °C. The concentration of penicillopepsin
in this stock (14.1 µM) was determined by measurement of the
absorbance at 280 nm. A working solution of the enzyme was prepared
by dilution of 100 µL of the stock to 5.00 mL with 0.1 M KH2PO4
containing 10% (v/v) glycerol ([E] ) 281 nM). The stock solutions
were stored at 5 °C and were stable for at least 1 week.
Substrate Stock Solution. The substrate Ac-Ala-Ala-Lys-(p-NO2)-
Phe-Ala-Ala-NH2 was prepared using an automated peptide synthesizer.
The resin-bound substrate was cleaved and isolated and used without
further purification. The substrate stock was prepared by dilution of
the substrate in buffer solution, and the substrate concentration was
determined from the absorbance at 277 nm (ꢀ277 ) 10 970 M-1 cm-1).
Inhibitor Stock Solutions. Certified weights and 31P combustion
analyses were obtained for samples of inhibitors 1-L, 2-L, 3, and 16-
D, each of which consisted of a single phosphorus-containing compound
by 31P NMR. Inhibitor concentrations in these stock solutions were
calculated from the sample weight corrected for analytical purity
according to phosphorus analysis. Extinction coefficients were mea-
sured for 1-L, 2-L, and 16-D and employed in calculating the
concentrations for stock solutions of the diastereomers 1-D, 2-D, and
16-L, respectively.
Assay Procedure. Assays were initiated by addition of enzyme
(typical [E] ) 6-10 nM) to an appropriate mixture of substrate,
inhibitor, and buffer. After a 3-min equilibration period, substrate
hydrolysis was followed at λ ) 306 nm (∆ꢀ ) -1710 M-1 cm-1);
assays were followed for less than 15% of the total reaction.
Determination of Inhibition Constants. The initial velocity of the
enzyme-catalyzed reaction was determined from a least-squares fit of
the raw absorbance data using the program Enzfitter.21 Velocity values,
obtained in duplicate for the four inhibitors with measurable activity,
were incorporated into Lineweaver-Burk plots. A replot of the slopes
(Km,app/Vmax) versus [I] was linear for each of the four active inhibitors,
and the Ki was taken from the absolute value of the [I] intercept.
Solution Structure Determination: NMR. All NMR samples were
5-20 mM in 90% H2O/D2O with 100 mM CD3CO2D and 1 mN KH2-
PO4. The buffer was adjusted to pH 3.5 with concentrated NaOD in
D2O. Chemical shifts were referenced to 1,4-dioxane at 3.73 ppm.
All spectra were obtained using a Bruker AM 500 spectrometer
equipped with an inverse probe and an Aspect 3000 computer. Variable
temperature 1D spectra were acquired at 0, 29, and 50 °C using a jump
and return selective excitation sequence, (π/2) - D2 - (π/2) -
acquire.22 The carrier frequency was set on the water signal and the
delay time, D2, was set to 200 µs so that maxima occurred at frequency
intervals 1.8 kHz from the water signal. All 2D experiments were
recorded in the phase-sensitive mode using the time-proportional phase
incrementation method23 for quadrature detection in f1. The spectral
Methyl [Cyclo-7-[(2R)-((N-valyl)amino)-2-(methoxy-(1S)-1-meth-
oxycarbonyl-2-phenylethoxy)phosphinyloxyethyl]-1-naphthalene-
acetamide] (18-L). The macrocyclic acid 16-L (53 mg, 0.13 mmol)
was converted to the tetramethylammonium salt by dissolving it with
an equivalent of tetramethylammonium hydroxide in 1 mL of MeOH,
evaporating, and drying under vacuum for 2 h. A solution of triflate
17 (122 mg, 0.39 mmol) in 1 mL of CH3CN was added to the residue
by cannula, and the mixture, which had become a suspension within 1
h, was stirred at 25 °C overnight. The mixture was diluted with 15
mL of EtOAc, washed with 10 mL each of 1 M HCl, saturated NaHCO3,
and brine, dried, and concentrated to leave an oily residue which was
purified by chromatography (50 to 100% EtOAc/ hexanes) to provide
57 mg (78%) of the desired phosphonate diester 18-L as a mixture
(1:1) of phosphonate diastereomers. For 18-L: 1H NMR δ 0.73 (d,
1.5, J ) 7), 0.74 (d, 1.5, J ) 7), 0.81 (d, 3, J ) 7), 1.67 (m, 1), 1.95
(ddd, 0.5, J ) 6, 13, 13), 2.86 (ddd, 0.5, J ) 6, 13, 13), 3.05 (m, 0.5),
3.13 (dd, 0.5, J ) 9, 14), 3.16 (d, 1.5, J ) 11), 3.32 (m, 1), 3.81 (d,
1.5, J ) 11), 3.82 (s, 1.5), 3.84 (s, 1.5), 3.94 (d, 0.5, J ) 18), 3.97 (d,
0.5, J ) 18), 4.19 (dd, J ) 5, 10), 4.21 (dd, 10, J ) 5), 4.25 (m, 0.5),
4.27 (d, 0.5, J ) 18), 4.28 (d, 0.5, J ) 18), 4.41 (m, 0.5), 5.17 (d, 0.5,
J ) 10), 5.12 (d, 0.5, J ) 10), 5.15 (m, 0.5), 5.25 (ddd, 0.5, J ) 3, 9,
9), 6.11 (d, 0.5, J ) 10), 6.84 (dd, 0.5, J ) 2, 10), 7.22-7.48 (m, 9),
7.76-7.83 (m, 2); 13C NMR δ 18.2, 18.2, 19.0, 19.1, 27.7, 35.8, 37.0,
38.9 (d, J ) 6), 39.1, 40.5, 43.1, 46.1 (d, J ) 159), 46.3 (d, J ) 160),
51.8 (d, J ) 8), 52.5, 52.8, 53.6 (d, J ) 7), 60.0, 60.4, 75.1 (d, J ) 7),
75.7 (d, J ) 7), 124.8-135.7 (28 peaks), 170.0, 170.0, 170.3, 170.7,
173.0, 173.3; 31P NMR δ 24.9, 25.7.
Methyl [Cyclo-7-[(2S)-((N-valyl)amino)-2-(methoxy-(1S)-1-meth-
oxycarbonyl-2-phenylethoxy)phosphinyloxyethyl]-1-naphthalene-
acetamide] (18-D). In a similar fashion, the epimer 16-D (9 mg, 0.022
mmol) was converted to 8 mg (67%) of ester 18-D, isolated as a single
diastereomer at phosphorus. For 18-D: 1H NMR (9:1 CDCl3 /CD3-
OD) δ 0.79 (d, 3, J ) 7), 0.84 (d, 3, J ) 7), 1.86 (m, 1), 2.38 (ddd, 1,
J ) 4, 4, 14), 2.86 (ddd, 1, J ) 6, 13, 13), 3.01 (dd, 1, J ) 8, 14), 3.13
(m, 1), 3.50 (d, 1, J ) 14), 3.61 (s, 3), 3.61 (d, 3, J ) 11), 3.70 (ddd,
J ) 2, 13, 15), 3.79 (d, 1, J ) 11), 3.97 (d, 1, J ) 14), 5.01 (ddd, 1,
J ) 4, 7, 8), 6.80 (dd, 1, J ) 1, 8), 7.10 (m, 5), 7.16 (m, 3), 7.52 (m,
2); 13C NMR δ 18.8, 19.3, 28.9, 33.0, 39.1 (d, J ) 7), 42.4, 47.3 (d, J
) 158), 52.6, 54.2 (d, J ) 7), 62.6, 75.5 (d, J ) 8), 124.8, 125.4,
127.1, 127.14, 127.3, 128.5, 128.7, 129.3, 131.5, 132.6, 133.3, 133.5,
135.2, 171.3, 171.4, 172.9; 31P NMR (9:1 CDCl3/CD3OD) δ 26.3.
Methyl [Cyclo-7-[(2R)-((N-valyl)amino)-2-(hydroxy-(1S)-1-meth-
oxycarbonyl-2-phenylethoxy)phosphinyloxyethyl]-1-naphthalene-
acetamide], Sodium Salt (1-L). Isobutylene was bubbled through 0.8
mL of CH2Cl2 for 20 min, and bromotrimethylsilane (30 µL, 0.23 mmol)
was added. After 5 min, this solution was added to the methyl
phosphonate 18-L (25 mg, 0.043 mmol) by cannula, and the reaction
solution was stirred overnight at 25 °C. The solvent had evaporated,
leaving an off-white residue which was suspended in 3 mL of MeOH;
this solution was concentrated, and the residue was dissolved in 1:1
MeOH/H2O and passed over a Dowex/Na+ column. The eluate was
partially concentrated and lyophilized to provide 20 mg (80%) of the
sodium salt 1-L as a white powder. For 1-L: 1H NMR (D2O) δ 0.59
(d, 3, J ) 7), 0.66 (d, 3, J ) 7), 1.46 (m, 1), 2.49 (ddd, 1, J ) 5, 13,
13), 3.01 (d, 2, J ) 6), 3.10 (ddd, 1, J ) 3, 6, 13), 3.56 (s, 3), 3.74
(ddd, 1, J ) 3, 13, 17), 3.98 (d, 1, J ) 18), 4.02 (d, 1, J ) 10), 4.06
(d, 1, J ) 18), 4.85 (ddd, 1, J ) 6, 6, 8), 7.12-7.34 (m, 9), 7.77 (m,
2); 13C NMR (D2O) δ 18.0, 18.8, 27.7, 37.5, 39.7 (d, J ) 6), 42.3,
49.1 (d, J ) 150), 53.2, 61.4, 74.3 (d, J ) 6), 124.8, 125.8, 127.7,
128.0, 128.6, 128.7, 129.1, 129.5, 129.8, 130.2, 131.2, 132.9, 136.1,
136.3, 172.0 (d, J ) 8), 174.1 (d, J ) 3), 176.1; 31P NMR (D2O) δ
18.7; HRMS (FAB) m/z calcd for C29H33N2O7PNa (MH+) 575.1923,
found 575.1922.
Methyl [Cyclo-7-[(2S)-((N-valyl)amino)-2-(hydroxy-(1S)-1-meth-
oxycarbonyl-2-phenylethoxy)phosphinyloxyethyl]-1-naphthalene-
acetamide], Sodium Salt (1-D). In a similar fashion, phosphonate
18-D (5 mg, 0.009 mmol) was converted to 4 mg (77%) of the sodium
salt 1-D. For 1-D: 1H NMR (D2O) δ 0.90 (d, 3, J ) 7), 0.97 (d, 3, J
(21) Leatherbarrow, R. J. Enzfitter, Elsevier Science Publishers BV:
Amsterdam, 1987.
(22) Plateau, P.; Gueron, M. J. Am. Chem. Soc. 1982, 104, 7310.

Macrocyclic Inhibitors of Penicillopepsin. 1
J. Am. Chem. Soc., Vol. 120, No. 19, 1998 4609
widths used were the same as those in the 1D experiments. Typically
between 8 and 32 transients of 2048 data points each were recorded in
the f2 dimension and 256-512 t1 increments were recorded and zero-
filled to 1024 data points. In cases where high digital resolution was
required in f1, up to 1024 t1 increments were collected and zero-filled
to 2048 data points. The NOESY 2D data were acquired at 30, 32, or
50 °C, and the TOCSY 2D data were acquired at 32 °C. All data
were processed using the Bruker UXNMR software package. Prior to
Fourier transformation, spectra were zero-filled once in the f1 dimension
and subjected to a phase-shifted sine bell weighting function in both
dimensions.
searches (5000 iterations for 16-L and 16-D, 10 000 iterations for 1-L
and 1-D) and minimizations were run with either double- or triple-
well Φ angle torsional constraints, calculated from the NH-RH
coupling constants.29 Energy minimizations were performed with
Macromodel’s TNCG method, the Amber force field with modified
phosphorus parameters,30 and GB/SA water solvation,31 until gradient
convergence was achieved. The structures were clustered10 into families
of rmsd < 0.3 Å by comparing all of the backbone atoms, amide
oxygens and NHs, and the phosphorus atom. Conformer families
violating the Nap ø1 angle torsional constraints were identified by visual
inspection and discarded.
Phase sensitive NOESY24 spectra were recorded using a jump and
return22 read pulse so that the overall sequence was (π/2) - t1 - (π/2)
- τm - (π/2) - D2 - (π/2) - acquire. The delay D2 was set to 200
µs. Appropriate phase cycling was used to select z-magnetization
during the mixing time, τm. Mixing times (τm) of 700 ms were used
with a 20 ms z-filter to remove zero quantum artifacts. The initial
value of t1 was set to 3 µs.
TOCSY25 spectra were recorded with a 50 ms and a 2 kHz MLEV-
17 spin-locking pulse.26 Phase coherent solvent suppression was
achieved using a DANTE27 sequence (pulse width 25 µs and inter-
pulse delay of 250 µs) applied for 1.7 s before the start of each transient.
Acknowledgment. We thank Professor Theo Hoffman
(University of Toronto) for providing the penicillopepsin, and
Dr. David King (UC Berkeley, HHMI) for synthesis of the assay
substrate. This work was supported by a grant from the National
Institutes of Health (GM-30759); P.A.B. was on appointment
as a Miller Research Professor in the Miller Institute for Basic
Research in Science (1996). We also thank Jason Hataye for
preparation of synthetic intermediates.
Supporting Information Available: Synthesis and charac-
terization of acyclic analogues 2 and 3 and kinetic plots for Ki
determinations (8 pages, print/PDF). See any current masthead
page for ordering information and Web access instructions.
Solution Structure Determination: Molecular Modeling. All
modeling was performed with Macromodel/Batchmin28 software version
4.5 including previously described modifications.16 Monte Carlo
JA973715J
(23) Marion, D.; Wu¨thrich, K. Biochem. Biophys. Res. Commun. 1983,
113, 967.
(24) Bodenhausen, G.; Kogler, H.; Ernst, R. R. J. Magn. Reson. 1984,
58, 370.
(28) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.;
Lipton, M.; Caufield, C.; Chang, G.; Hendrickson, T.; Still, W. C. J. Comput.
Chem. 1990, 11, 440.
(25) Braunschweiler, L.; Ernst, R. R. J. Magn. Reson. 1983, 53, 521.
(26) Davis, D. G.; Bax, A. D. J. Magn. Reson. 1985, 65, 355.
(27) Zuiderweg, E. R. P.; Hallenga, K.; Olejniczak, E. T. J. Magn. Reson.
1986, 70, 336. Morris, G. A.; Freeman, R. J. Magn. Reson. 1978, 29, 433.
(29) Bystrov, V. F. Prog. NMR Spectrosc. 1976, 10, 41.
(30) Ferguson, D. M.; Kollman, P. A. J. Comput. Chem. 1991, 12, 620.
(31) Still, W. C.; Tempczyk, A.; Hawly, R. C.; Hendrickson, T. J. Am.
Chem. Soc. 1990, 112, 6127.