Macrocyclic inhibitors of penicillopepsin. 3. Design, synthesis and evaluation of an inhibitor bridged between P2 and P1'

Article abstract of DOI:10.1021/ja973713z

Through application of a thorough design and modeling protocol, the macrocyclic peptidyl phosphonate 5 was derived from the bound structure of an acyclic inhibitor of penicillopepsin, 4, by linking the P1' and P2 side chains; the acyclic analogues 6 and 7

Full text of DOI:10.1021/ja973713z

4622
J. Am. Chem. Soc. 1998, 120, 4622-4628
Macrocyclic Inhibitors of Penicillopepsin. 3. Design, Synthesis, and
Evaluation of an Inhibitor Bridged between P2 and P1′
Whitney W. Smith and Paul A. Bartlett*
Contribution from the Department of Chemistry, UniVersity of California, Berkeley, California 94720-1460
ReceiVed October 27, 1997
Abstract: Through application of a thorough design and modeling protocol, the macrocyclic peptidyl
phosphonate 5 was derived from the bound structure of an acyclic inhibitor of penicillopepsin, 4, by linking
the P1′ and P2 side chains; the acyclic analogues 6 and 7 were also synthesized and evaluated for comparison.
As observed for a related set of inhibitors, 1-3 [Meyer, J. H.; Bartlett, P. A. J. Am. Chem. Soc. 1998, 120,
xxxx], the binding affinity at pH 4.5 was found to be inversely related to the degree of conformational flexibility
across the series, 7 (Ki ) 1300 nM), 6 (Ki ) 42 nM), and 5 (Ki ) 0.10 nM), although the differences are much
greater. The structure of the macrocyclic ring of 5 in solution was determined by a combined NMR and
molecular modeling method and found to correspond closely to the designed structure and to the backbone
conformation of the original target, 4. The binding enhancement conveyed by the macrocyclic constraint
corresponds to 0.9-1.4 kcal/mol/deg of rotational freedom lost, indicating what can be gained from this structure-
based design strategy.
An accompanying paper described the structure-based design
of an inhibitor of the aspartic protease penicillopepsin in which
conformational constraint was effected by linking the P1 and
P3 side chains of a peptidyl phosphonate in a macrocyclic
structure, 1.1 This strategy enhanced the affinity of the inhibitor
by 1-2 orders of magnitude in comparison to the acyclic control
compounds, 2 and 3. The conformations adopted by the
macrocycle in solution were shown to correspond to the original
design with respect to the peptide backbone; moreover, the
solution conformation proved to be the same as that found in
the X-ray crystal structure of the enzyme/inhibitor complex.2
Figure 1. Conformation of isovaleryl-L-Val-L-Val-L-Leu-{PO2--O}-
L-Phe-OMe (4) in its complex with penicillopepsin.4
affinity resulting from macrocyclic constraint. In this paper,
we describe another series of macrocyclic and acyclic penicil-
lopepsin inhibitors from which quantitative comparisons may
be more valid.
The P1 and P3 side chains of a peptidic ligand are not the
only ones brought into proximity as it binds in the extended
conformation to an aspartic protease; for example, the side
chains of the alternate residues P2 and P1′ share a binding pocket
along the other wall of the active site.3 In the case of
penicillopepsin, the crystal structure of its complex with the
phosphonate pentapeptide 4 shows a separation of only 3.3 Å
between the phenyl group of the P1′ residue and a γ-methyl of
the P2 valine (Figure 1).4 In this paper, we describe the
macrocyclic phosphonate 5, in which the P2 and P1′ side chains
have been bridged. The synthesis and evaluation of this
inhibitor, and the acyclic controls 6 and 7, were undertaken to
explore further the generality of macrocyclization as a strategy
for structure-based design and to provide additional insight into
However, the bound conformations of the comparison com-
pounds, 2 and 3, are sufficiently different from that of 1 that
this series cannot be used to quantitate the increase in binding
(3) 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 BV: Amsterdam, 1986; pp 321-334.
(4) Fraser, M. E.; Strynadka, N. C. J.; Bartlett, P. A.; Hanson, J. E.;
James, M. N. G. Biochemistry 1992, 31, 5201.
(1) Meyer, J. H.; Bartlett, P. A. J. Am. Chem. Soc. 1998, 120, 4600.
(2) Ding, J.; Fraser, M.; Meyer, J. H.; Bartlett, P. A.; James, M. N. G.
J. Am. Chem. Soc. 1998, 120, 4610.
S0002-7863(97)03713-X CCC: $15.00 © 1998 American Chemical Society
Published on Web 04/30/1998

Macrocyclic Inhibitors of Penicillopepsin. 3.
J. Am. Chem. Soc., Vol. 120, No. 19, 1998 4623
Figure 2. Use of CAVEAT to identify potential side chain-side chain
connecting units for phosphonate 4.
the quantitative enhancement in binding affinity that can be
obtained by this approach.
Design. A CAVEAT5 search for a unit that would link the
CR-Câ bond of the P2-valine side chain with the meta C-H
bond of the P1′ phenyl moiety, in the orientation they adopt in
the bound complex of 4 with penicillopepsin, readily identified
a 4-atom chain with a planar sp2-sp2 bond (Figure 2). We
evaluated a number of possible linkers in the following
fashion: models were constructed with acetyl in place of the
isovaleryl-valine (Iva-Val) tail and with methyl in place of the
leucine side chain and subjected to Monte Carlo conformational
searches to estimate the likely low-energy conformations of the
macrocyclic rings. Unique conformations within 3 kcal/mol of
the lowest energy conformation were superimposed on the
bound form of the acyclic inhibitor 4, according to the backbone
atoms of the leucine phosphonate moiety, and similarities in
orientations of the rest of the backbone, the aromatic ring, and
the side chain substituents were assessed. Ring systems were
sought that adopted relatively few conformations overall, but
with a high percentage matching the acyclic target.
Figure 3. Relationship between structure and conformational properties
of model macrocycles. (a) Number of discrete conformers within 3
kcal/mol of lowest energy. (b) Percentage of low-energy conformers
that match backbone of inhibitor 4 in bound conformation. (c) Analogue
without â-methyl at P2 position.
The behavior of a number of linking groups is shown in
Figure 3, which reveals some important and unanticipated trends.
For example, the alkene 1 is clearly distinct from the corre-
sponding amide 8, despite their geometric resemblance. Most
of the analogues were modeled with the â-methyl group
corresponding to the Val pro-S methyl, on the expectation that
this group would help control the conformation of the ring.
Among those analogues, the amide-containing linkage 8 has
the lowest number of conformers and the highest percentage
that match the bound target. When the methyl is removed (9),
the number of conformers remains the same, but many more of
them position the backbone in the desired fashion.
Fully elaborated models of the most promising macrocycles
were constructed, and their conformational behavior within the
active site of the protein was explored by high-temperature (600
K) dynamics simulations. Representative structures along the
trajectories were minimized, and selection of the final candidates
was made by comparison to a model of 4 that had itself been
minimized in the binding site. Macrocycle 5, with a carbox-
amide unit for the sp2-sp2 bond, remained the most attractive
analogue, not only because of its conformational behavior
(Figure 4) but also because of its synthetic accessibility.
To allow us to gauge the significance of the macrocyclic
constraint on the binding affinity of the peptidyl phosphonate,
we prepared the acyclic analogues 6 and 7 for comparison. These
control compounds differ from the macrocycle by loss of a
carbon atom (replacement of a methylene group by two
Figure 4. Lowest energy conformer predicted for macrocycle 5 (line
and label) and bound conformation of phosphonate 4 (ball-and-stick)
superimposed by backbone atoms of P2-P1′ residues.
hydrogens);1 thus, in principle, they are able to adopt the same
conformation in the active site as the macrocycle itself.
Synthesis. The key steps in the synthesis of the macrocyclic
phosphonate 5 were preparation of methyl L-â-(m-(amino-
methyl)phenyl)lactate, 12, elaboration of the acyclic precursor,
and cyclization. Several asymmetric approaches to the hy-
droxyacid 11 were explored, including alkylation of N-(diphen-
ylmethylene)glycine tert-butyl ester6 followed by diazotization
of the amino acid7 and alkylation of a chiral derivative of
glycolic acid.8 The most effective proved to be asymmetric
hydroxylation of the chiral oxazolidinone 9, as described by
Evans et al.9 m-Cyanohydrocinnamic acid, 8, was prepared from
m-cyanobenzaldehyde by Horner-Emmons reaction, hydroge-
nation, and hydrolysis, activated as the mixed anhydride with
pivalic acid, and condensed with the lithium salt of benzylox-
(6) O’Donnell, M. J.; Bennett, W. D.; Wu, S. J. Am. Chem. Soc. 1989,
111, 2353.
(7) Li, W.; Ewing, W.; Harris, B.; Joullie, M. J. Am. Chem. Soc. 1990,
112, 7659.
(8) Evans, D. A.; Bender, S. L.; Morris, J. J. Am. Chem. Soc. 1988,
110, 2506.
(9) Evans, D. A.; Morrissey, M. M.; Dorow, R. L. J. Am. Chem. Soc.
1985, 107, 4346.
(5) Lauri, G.; Bartlett, P. A. J. Comput.-Aided Mol. Des. 1994, 8, 51.

4624 J. Am. Chem. Soc., Vol. 120, No. 19, 1998
Smith and Bartlett
Scheme 1
Scheme 2
azolidinone to give imide 9 (Scheme 1). The enolate of this
material was generated at -95 °C in the presence of 2-ben-
zenesulfonyl-3-phenyloxaziridine by slow addition of sodium
hexamethylsilazide,10 and the resulting alkoxide was quickly
quenched at low temperature by addition of a THF solution of
camphorsulfonic acid prior to workup. The hydroxy ester 11
was liberated from the chiral auxiliary as described previously,9
and the cyano moiety was reduced catalytically by transfer
hydrogenation to afford the key intermediate 12.
The sometimes capricious nature of phosphonate esterifica-
tions with hindered components suggested that the phosphonate
linkage should be established early in the synthesis.11 Sequences
were therefore explored in which the Asn-LeuP amide or the
side chain-to-side chain linkages were formed in the macrocy-
clization step (Scheme 2). Route A was evaluated first. The
protected precursor 15 was assembled by acylating the amine
of intermediate 12 with an aspartic acid derivative and then
acylating the hydroxyl group with the protected L-leucine-
phosphonochloridate 14.12,13 Simultaneous deprotection of the
benzyl ester and Cbz group proceeded in poor yield, and material
of sufficient purity for subsequent cyclization proved difficult
to isolate. We therefore pursued Route B, assembling the
peptide backbone prior to ring closure.
Scheme 3
The aminomethyl group of 12 was protected as the Boc
derivative, and the hydroxyl moiety was phosphonylated to give
17 (Scheme 3). Elaboration of the peptide backbone proceeded
straightforwardly to give the acyclic intermediate 20.14 Double
deprotection of 20 was accomplished with trifluoroacetic acid
(TFA) in the presence of dimethyl sulfide, and the TFA salt
was converted to the zwitterion 21 by washing an ethyl acetate
solution with small portions of pH 7.5 phosphate buffer. Several
peptide coupling reagents were explored for cyclization of 21
to 22; the reaction proceeded slowly but in reproducible yield
(10) Davis, F. A.; Stringer, O. D. J. Org. Chem. 1982, 47, 1774.
(11) Hirschmann, R.; Yager, K. M.; Taylor, C. M.; Moore, W.;
Sprengeler, P. A.; Witherington, J.; Phillips, B. W.; Smith, A. B., III J.
Am. Chem. Soc. 1995, 117, 6370.
(12) Oleksyszyn, J.; Subotkowska, L.; Mastalerz, P. Synth. Commun.
1979, 985. Kafarski, P.; Leijczak, B.; Szewczyk, J. Can. J. Chem. 1983,
61, 2425.
(13) Bartlett, P. A.; Hanson, J. E.; Giannousis, P. P. J. Org. Chem. 1990,
55, 6268.
(14) Deprotection of the side chain functionality and macrocyclization
were evaluated at two points in this sequence, prior to and after introduction
of the Iva-Val group. The cyclization itself proceeded in lower yield when
carried out at the earlier stage, and it was difficult to acylate the macrocycle
(H in place of Iva-Val in 22) without racemization.
with ethyl dimethylaminoethyl carbodiimide (EDC), hydroxy-
benzotriazole (HOBt), and a tertiary amine in dilute THF
solution (0.001 M) at room temperature. Clear differences were
observed in the rates of reaction of the phosphonate diastere-
omers in both the cyclization (21 f 22) as well as earlier

Macrocyclic Inhibitors of Penicillopepsin. 3.
J. Am. Chem. Soc., Vol. 120, No. 19, 1998 4625
Table 1. Inhibition of Penicillopepsin by Constrained and
Unconstrained Phosphonate Analogues
Table 2. NH-RH Coupling Constants and Dihedral Constraints
for Macrocycle 5a
Ki (nM)
δ
∆ppb/
J
dihedral range
for N-CR bond
resonance
Asn RNH
(ppm)
K
(Hz)
inhibitor
at pH 4.5
at pH 3.5
8.55
8.3
8.1
+60° ( 20°
-80° ( 20°
-160° ( 20°
(not assigned)
+60° ( 20°
-75° ( 20°
-165° ( 20°
-120° ( 20°
7
6
4
5
1300 ( 155
42 ( 2.9
860 ( 3
14 ( 5.6
2.8a
NH of Asn â-amide
Val NH
8.5
8.08
8.3
9.4
5.4, 7.3b
7.85
0.10 ( 0.002
a From ref 13.
LeuP NH
7.55
8.9
9.75
acylation (18 f 19) steps, although loss of the stereocenter at
phosphorus at the end of the synthesis made these differences
unimportant. As the last step, the phosphonate methyl ester
was cleaved selectively with trimethylsilyl bromide and the
macrocycle 5 was purified by reverse phase chromatography.
Synthesis of the acyclic comparison compound 6 was
uneventful, in that similar coupling reactions were employed
to link commercially available amino acids and hydroxy esters
with the L-leucine phosphonate 14. In contrast, synthesis of
the other acyclic analogue 7 was unexpectedly difficult (Scheme
3). The glycolate ester 24 was conveniently prepared by
esterification of monoacid 23 with methyl diazoacetate and
readily elaborated to the full-length derivative 25; however, we
were not able to remove the phosphonate methyl ester selec-
tively. Trimethylsilyl bromide15 cleaved both P-O bonds at
similar rates, and, in model experiments with dipeptide analogue
24, refluxing tert-butylamine16 generated the phosphonate
monoester but transaminated the glycolate residue at the same
time to give amide 26. We were also frustrated in attempts to
cleave only one of the phosphonate esters from symmetrical
diester 27 with trimethylsilyl bromide, and no reaction was
observed with tert-butylamine. However, 1,4-diazabicyclo-
[2.2.2]octane (DABCO) gave the desired monoanion 28 as the
major product.17 After elaboration of 29, treatment with
DABCO in refluxing toluene for 1.5 h afforded 55% of the
desired monoanion 7, 19% of the dianion 30, and 20% of
recovered starting material after fractionation.
Inhibition of Penicillopepsin. The inhibition constants Ki
for the acyclic comparison inhibitors were determined using the
established assay procedure at the pH optimum of 3.5 as well
as at pH 4.5 (Table 1). The Ki value for the macrocyclic
phosphonate 5 was difficult to measure, since its high affinity
required low enzyme and inhibitor concentrations for equilib-
rium experiments, with prolonged incubation times as a
consequence.18 By stabilizing the enzyme for overnight incuba-
tions with bovine serum albumin (BSA) and using a Henderson
analysis to compensate for inhibitor depletion,19 we were able
to determine a Ki value at pH 4.5, where inhibitor affinity is
reduced; tighter binding and enzyme instability prevented us
from measuring an inhibition constant at pH 3.5.
a Determined at a concentration of 10 mM 5 in 0.1 M sodium acetate-
d3/1 mN KH2PO4 at pH 3.5. b Diastereotopic benzylic hydrogens not
assigned.
chemical shifts (8-9 ppb/K) indicated that the amide hydrogens
are not shielded from solvent by intramolecular hydrogen bonds.
Proton assignments were made from a TOCSY spectrum, and
NH-RH coupling constants were used to define torsional
constraints for a Monte Carlo conformational search (Table 2).21
Several ranges for the N-CR dihedral angles are consistent with
the observed coupling constants for the valine and asparagine
residues, but only one fits the large coupling constant of the
leucine-phosphonate moiety.
These constraints were applied in a Monte Carlo conforma-
tional analysis of the complete inhibitor with a search protocol
modified to include multiple angle constraints.20 Some 20 000
different starting conformers were generated and minimized
using an Amber force field with the GB/SA solvation treatment
of MacroModel22 and modified to include parameters for the
phosphonate moiety developed by Merz and Kollman.23,24 All
minimizations reached convergence before 5000 steps, and
multiple occurrences of the lowest energy structures suggested
that the conformational ensemble was adequately sampled. The
search resulted in 5469 distinct conformations that were
clustered into only 240 groups when compared by the backbone
atoms of the macrocycle (rmsd < 0.3 Å).25 The lowest energy
cluster, which contains 537 structures and 47 of the 48 lowest
energy conformations, is 2.4 kcal/mol lower than the next
ranked;26 a representative set of these conformations is shown
in Figure 5a. Of the other two clusters within 3 kcal/mol of
cluster 1, only in cluster 2 (272 members) is the macrocyclic
ring significantly different (Figure 5b); it may therefore represent
a minor component of the ensemble.
The same force field and solvation treatment were used to
predict the conformation of the macrocycle in advance of
synthesis and to model it with constraints derived from the NMR
data; the predicted conformation and that modeled with the
NMR data are compared in Figure 5c. Finding the same
structure in both approaches suggests that these protocols
provide a realistic indication of the solution structure of 5.
Analysis of Design Strategy: Quantitative Impact of
Macrocyclization. In the absence of other torsional constraints,
a linear chain of n-nodes has n - 1 independent rotations;
connecting the termini to form a macrocycle adds another
Conformation of Macrocycle 5. The solution structure of
the macrocyclic ring in phosphonate 5 was determined by a
combined NMR-molecular modeling approach.20 1H NMR
spectra were obtained in aqueous buffer at 0, 27, and 40 °C,
and there was no evidence of slow interconversion of confor-
mational isomers. The temperature dependence of the N-H
(21) Bistrov, V. F. Prog. NMR Spectrosc. 1976, 10, 41.
(22) 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. Still, W. C.; Tempczyk, A.; Hawly, R. C.;
Hendrickson, T. J. Am. Chem. Soc. 1990, 112, 6127.
(23) Merz, K. M., Jr.; Kollman, P. A. J. Am. Chem. Soc. 1989, 111,
5649.
(24) Ferguson, D. M.; Kollman, P. A. J. Comput. Chem. 1991, 12, 620.
(25) Lauri, G.; Bartlett, P. A. J. Comput.-Aided Mol. Des. 1994, 8, 51.
(26) The “energy” of a cluster is considered to be that of the lowest
energy single conformation contained within it.
(15) McKenna, C. E.; Higa, M. T.; Cheung, N. H.; McKenna, M.
Tetrahedron Lett. 1977, 155.
(16) Gray, M. D. M.; Smith, D. J. H. Tetrahedron Lett. 1980, 21, 859.
(17) Saady, M.; Lebeau, L.; Mioskowski, C. J. Org. Chem. 1995, 60,
2946.
(18) Morrison, J. F.; Walsh, C. T. AdV. Enzymol. Relat. Areas Mol. Biol.
1987, 61, 201.
(19) Henderson, P. J. F. Biochem. J. 1972, 127, 321.
(20) Sefler, A. M.; Lauri, G.; Bartlett, P. A. Int. J. Pept. Protein Res.
1996, 48, 129.

4626 J. Am. Chem. Soc., Vol. 120, No. 19, 1998
Smith and Bartlett
The presence of the macrocyclic ring has a dramatic effect
on the binding of the inhibitors described in this report,
enhancing the affinity of 5 by 3-4 orders of magnitude over
the acyclic analogues 6 and 7 (Table 1). Moreover, 5 is a
significantly better inhibitor than 4, the original structure from
which it was derived. The 3.6-5.6 kcal/mol in free energy
represented by the Ki differences between macrocycle 5 and
acycles 6 and 7 are among the highest values reported for a
conformational constraint. Nevertheless, the per-bond effects
are consistent with analyses from a variety of systems28 and
demonstrate the value of this structure-based design strategy.
Experimental Section29
(4S)-(3-(3-Cyanophenyl)propanoyl)-4-(phenylmethyl)-2-oxazoli-
dinone (9). To a stirring solution of 3-(3-cyanophenyl)propanoic acid
8 (4.0 g, 23 mmol) in 75 mL of THF at -78 °C was slowly added
triethylamine (TEA; 4.0 mL, 28.75 mmol) and pivaloyl chloride (3.1
mL, 25.3 mmol). After 10 min at -78 °C, the solution was brought
to room temperature for 1.5 h, and cooled again to -78 °C. During
this time, a solution of (S)-(-)-4-benzyl-2-oxazolidinone (4.4 g, 25.3
mmol) in 75 mL of THF was prepared and cooled to -78 °C,
n-butyllithium (9.7 mL, 2.60 M in hexanes, 25.3 mmol) was added,
and the mixture was stirred for 15 min. The lithium salt of the
oxazolidinone was slowly added as a slurry to the mixed-anhydride
solution via cannula. After 30 min, this mixture was allowed to warm
to room temperature and stirred for 3.5 h. Half-saturated aqueous NH4-
Cl (100 mL) was added, and most of the THF was removed by
evaporation. The residue was extracted with CH2Cl2 (3 × 50 mL),
and the combined organic layer was washed with saturated NaHCO3
(50 mL) and brine (50 mL), dried, and evaporated to provide 8.0 g of
a waxy yellow solid. Flash chromatography (20-50% EtOAc in
hexanes) of this material provided 6.5 g (86% yield) of 9 as a thick,
clear oil: 1H NMR δ 7.17-7.58 (m, 9), 4.65-4.71 (m, 2), 4.17-4.24
(m, 1), 3.20-3.37 (m, 3), 3.06 (t, 2, J ) 7.4), 2.78 (dd, 1, J ) 9.5,
13.4); 13C NMR δ 171.5, 153.3, 141.9, 135.0, 133.1, 132.1 130.0, 129.3,
129.2, 128.9, 127.3, 118.8, 112.4, 66.3, 55.0, 37.7, 36.6, 29.6. Anal.
Calcd for C20H18N2O3: C, 71.84; H, 5.43; N, 8.38. Found: C, 71.44;
H, 5.40; N, 8.23.
(2S,4S)-(3-(3-Cyanophenyl)-2-hydroxypropanoyl)-4-(phenylmethyl)-
2-oxazolidinone (10). An equivalent of sodium bis(trimethylsilyl)-
amide in THF was added via syringe pump over 1.5 h to a solution of
amide 9 (3.63 g, 10.9 mmol) and 2-(phenylsulfonyl)-3-phenyloxaziri-
dine (3.13 g, 12 mmol) in 55 mL of THF at -98 °C. During this
time, a solution of camphorsulfonic acid (10.1 g, 43.6 mmol) in 90
mL of THF was prepared, cooled to -98 °C, and then transferred
rapidly via cannula to the oxidation solution. The mixture was allowed
to warm to room temperature and poured into 600 mL of 3:1 petroleum
ether/CH2Cl2, and the cloudy suspension was washed with water until
clear (3 × 200 mL), 1 M HCl (200 mL), and brine. The organic layer
was dried and evaporated, and the residue was chromatographed (1:
3:6 CH2Cl2/EtOAc/hexanes) to give 2.12 g (56% yield) of pure 10 as
a clear oil, as well as 1.10 g of impure product. Repurification of the
latter material (10% ether in CH2Cl2) gave an additional 0.40 g (10%
yield) of hydroxyimide 10: 1H NMR δ 7.19-7.61 (m, 9), 5.16 (dd, 1,
J ) 3.6, 8.6), 4.66-4.71 (m, 1), 4.27-4.34 (m, 2), 3.29 (dd, 1, J )
3.3, 13.6), 3.20 (dd, 1, J ) 3.5, 13.9), 2.83-2.89 (m, 2); 13C NMR δ
73.4, 153.3, 138.7, 134.5, 134.2, 133.2, 130.5, 129.4, 129.1, 129.0,
127.6, 118.8, 112.4, 71.3, 67.1, 55.4, 39.7, 37.4; HRMS (FAB) calcd
for C20H19N2O4 (MH+) m/z 351.1345, found 351.1295.
Figure 5. (a) A selection of conformers from the lowest energy cluster
for macrocycle 5 consistent with the NMR data; (b) comparison of
ring conformations from the low energy conformers from clusters 1
(ball-and-stick) and 2 (solid line); (c) comparison of conformation of
5 predicted in advance of synthesis (solid line) and that determined
from NMR data (ball-and-stick).
Table 3. Effect of Constraining Bond Rotations
effect of constraint on
no. of independent
rotations in central
structurea
binding affinity per
independent rotation
(kcal/mol)
compd
1
2
3
5
6
7
7 - 6 ) 1
5
0.34
0.74
5
11 - 6 ) 5
9
9
0.90b
1.41b
a Central structure corresponds to macrocycle and corresponding
bonds in acyclic analogues, not including amides or bonds inside
aromatic rings. b Calculated from inhibition constants determined at 25
°C and pH 4.5.
rotatable bond, but 6 degrees of freedom are lost as the bond
rotations become dependent on each other.27 Thus, the con-
formational constraint introduced by the macrocyclic rings in
1 and 5 can be gauged on a per-bond basis as shown in Table
3.
Although the same structural perturbation was applied to
obtain each of the acyclic comparison compounds, the results
are dramatically different. For both systems, the impact of
cyclization is greater when comparing the cyclic inhibitor with
the more linear acyclic analogue, since the comparison com-
pounds are not equally unconstrained. Bond rotation is clearly
more restricted in branched analogues 2 and 6 than in linear
analogues 3 and 7. Moreover, there is a significant difference
between the two macrocyclic systems: the effect is larger for
5 versus 6 and 7 than for 1 versus 2 and 3. Since the latter
compounds do not all adopt the same active site conformations,2
the observed differences in their binding affinity only represent
lower limits to what would be found for congruent binding
modes.
Methyl (2S)-3-(3-Cyanophenyl)-2-hydroxypropanoate (11).
A
solution of bromomagnesium methoxide was prepared by slow addition
of methylmagnesium bromide (2.7 mL, 8.0 mmol) to 60 mL of
anhydrous methanol cooled to 0 °C. After 10 min, this solution was
transferred via cannula into a flask containing a solution of hydroxy-
(28) Andrews, P. R.; Craik, D. J.; Martin, J. L. J. Med. Chem. 1984, 27,
1648. Gerhard, U.; Searle, M. S.; Williams, D. H. Bioorg. Med. Chem.
Lett. 1993, 3, 803. Williams, D. H.; Searle, M. S.; Mackay, J. P.; Gerhard,
U.; Maplestone, R. A. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 1172.
(29) See footnote 18 in ref 1.
(27) Go, N.; Scheraga, H. A. Macromolecules 1970, 3, 178. Guarnieri,
F.; Cui, W.; Wilson, S. R. J. Chem. Soc., Chem. Commun. 1991, 1542.

Macrocyclic Inhibitors of Penicillopepsin. 3.
J. Am. Chem. Soc., Vol. 120, No. 19, 1998 4627
imide 10 (1.4 g, 4.0 mmol) in 60 mL of anhydrous methanol at 0 °C.
The mixture was stirred for 20 min and acidified with 30 mL of 0.5 M
HCl, and the methanol was removed under reduced pressure. The
remaining aqueous solution was extracted with ether (3 × 30 mL),
and the combined organic layer was washed with brine, dried, and
evaporated. The residue was chromatographed (10% ether in CH2Cl2)
to afford 0.513 g (63% yield) of hydroxyester 11 as a clear oil: [R]D
3, J ) 6.6); 31P NMR δ 27.1. For both diastereomers: 13C NMR δ
170.8, 170.6, 156.1, 156.0, 155.8, 155.7, 139.6, 139.3, 136.4, 136.0,
128.9, 128.7, 128.6, 128.4, 128.3, 128.1, 128.0, 127.9, 127.9, 126.2,
79.5, 79.4, 74.7, 74.4, 67.0, 66.8, 53.3, 52.7, 52.4, 51.8, 51.7, 46.5 (d,
J ) 150), 46.3 (d, J ) 150), 44.4, 44.3, 38.9, 37.8, 28.3, 24.3, 24.2,
23.3, 23.1, 21.1, 20.9. Anal. Calcd for C30H43N2O9P: C, 59.40; H,
7.14; N, 4.62; P, 5.10. Found: C, 59.72; H, 7.21; N, 4.40; P, 5.09.
1
-23° (c ) 0.012, CHCl3); H NMR δ 7.36-7.55 (m, 4), 4.44 (dd, 1,
Methyl (2S)-2-[(1R)-1-(N-(L-N-Phenylmethoxycarbonyl-â-tert-bu-
tylaspartyl)amino)-3-methylbutylmethoxyphosphinyloxy]-3-(3-(tert-
butoxycarbonyl)aminomethyl)phenylpropanoate (18). A suspension
of dipeptide phosphonate 17 (270 mg, 0.45 mmol) and 10% Pd/C in
2.2 mL of EtOAc was stirred under a hydrogen atmosphere for 2 h
and then filtered through a 4.5-µm nylon syringe filter directly into a
solution of Cbz-aspartic acid, â-tert-butyl ester (173 mg, 0.54 mmol),
EDC (120 mg, 0.63 mmol), and HOBt (120 mg, 0.90 mmol) in 2.2
mL of DMF at 0 °C. Another 2 mL of EtOAc was used to rinse the
syringe filter then added to the reaction solution. The solution was
allowed to warm, kept at room temperature for 16 h, diluted with 15
mL of EtOAc, washed with half-saturated aqueous KHSO4 (2 mL),
saturated NaHCO3 (2 mL), and brine (2 mL), dried, and evaporated,
and the crude product was chromatographed (0-2% methanol in
CHCl3) to give 303 mg (88% yield) of the tripeptide phosphonate 18
as a 2:1 mixture of diastereomers at phosphorus. Further purification
allowed partial separation of the diastereomers for identification. For
diastereomer 1: 1H NMR δ 7.12-7.42 (m, 9), 6.93 (d, 1, J ) 9.8),
5.99 (br s, 1), 5.08-5.15 (m, 4), 4.54-4.60 (m, 2), 4.28 (d, 2, J )
5.3), 3.74 (s, 3), 3.25 (dd, 1, J ) 3.2, 14.4), 3.20 (d, 3, J ) 10.9), 3.03
(dd, 1, J ) 9.2, 14.2), 2.94 (dd, 1, J ) 4.9, 17.1), 2.58 (dd, 1, J ) 5.2,
17.1), 1.50-1.63 (m, 3), 1.45 (s, 9), 1.41 (s, 9), 0.90 (d, 3, J ) 6.4),
0.85 (d, 3, J ) 6.3); 13C NMR δ 171.1, 170.7, 170.2, 158.1, 155.8,
139.4, 136.1, 136.0, 128.8, 128.6, 128.5, 128.4, 128.2, 128.0, 126.3,
81.6, 79.4, 74.8 (d, J ) 7), 67.1, 52.6, 51.9, 51.5, 44.5, 44.5 (d, J )
125), 39.0, 38.7, 37.3, 28.4, 28.0, 24.2 (d, J ) 11), 23.3, 21.1; 31P
NMR δ 26.5. For diastereomer 2: 1H NMR δ 7.10-7.37 (m, 9), 6.46
(br s, 1), 6.46 (br s, 1), 5.49, (br s, 1), 5.13 (s, 2), 4.96-4.98 (m, 1),
4.20-4.40 (m, 4), 3.75 (s, 3), 3.74 (d, 3, J ≈ 10), 3.15 (dd, 1, J ) <2,
15.2), 2.90 (m, 2), 2.55 (m, 1), 1.43 (s, 9), 1.41 (s, 9), 1.19-1.50 (m,
2), 0.82-0.90 (m, 1), 0.75 (d, 3, J ) 8.6), 0.74 (d, 3, J ) 6.6); 13C
NMR δ 171.1, 170.5, 170.1, 156.0, 155.9, 139.6, 136.0, 135.9, 128.8,
128.5, 128.2, 126.3, 81.7, 79.2, 75.6, 67.2, 53.4 (d, J ) 6.4), 52.4,
51.2, 44.3, 44.2 (d, J ) 157), 39.0, 37.5, 37.0, 28.3, 27.9, 24.0 (d, J )
14), 23.1, 20.8, 18.8; 31P NMR δ 25.6; HRMS (FAB) calcd for
C38H57N3O12P (MH+) m/z 778.3680, found 778.3682.
J ) 4.2, 6.9), 3.79 (s, 3), 3.15 (dd, 1, J ) 4.1, 14.1), 2.96 (dd, 1, J )
6.9, 14.1), 2.85 (br, 1); 13C NMR δ 174.0, 138.1, 134.0, 133.0, 130.5,
129.0, 118.7, 112.3, 70.6, 52.6, 39.6. Anal. Calcd for C11H11NO3:
C, 67.38; H, 5.40; N, 6.83. Found: C, 64.01; H, 5.49; N, 6.64.
Methyl (2S)-3-(3-((tert-Butoxycarbonyl)aminomethyl)phenyl)-2-
hydroxypropanoate. A suspension of palladium on carbon (10%, 80
mg, dried under vacuum and kept under dry nitrogen) and formic acid
(0.20 mL) in ethanol (2 mL) was prepared, and a solution of
hydroxyester 11 (80 mg, 0.39 mmol) in 1.6 mL of methanol, along
with more formic acid (0.20 mL), was added. After 2 h of stirring,
the suspension was passed through a 4.5-µm syringe filter, and the
filtrate was evaporated to give 86 mg (87% yield) of the formate salt
of 12 as a light yellow oil; this material was carried on without
purification: 1H NMR (CD3OD) δ 8.27 (br s, 1), 7.27-7.36 (m, 4),
4.42 (dd, 1, J ) 4.6, 7.9), 4.08 (s, 2), 3.70 (s, 3), 3.10 (dd, 1, J ) 4.5,
13.9), 2.94 (dd, 1, J ) 7.9, 13.9); 13C NMR δ 175.5, 139.8, 134.3,
131.3, 131.2, 130.2, 128.2, 72.7, 52.5, 44.4, 41.2; LRMS (FAB) calcd
for C11H15NO2 (MH+) m/z 210.2, found 210.1.
A 5.3-mmol sample of the formate salt of 12 was converted to the
chloride salt by evaporation twice from a solution of ca. 5 equiv of
HCl in methanol (prepared by from acetyl chloride (1.5 mL, 26.5 mmol)
and 5 mL of methanol), and the material was dissolved in 50 mL of
DMF. 2-(tert-Butoxycarbonyloxyimino)-2-phenylacetonitrile (2 g, 8.0
mmol) and di(isopropyl)ethylamine (2.7 mL, 10.6 mmol) were added,
and the reaction mixture was stirred overnight. The solution was
evaporated to ca. 5 mL, diluted with 50 mL of EtOAc, and washed
with half-saturated aqueous KHSO4 (10 mL), saturated NaHCO3 (10
mL), and brine (10 mL), and dried. Evaporation provided 2.97 g of a
yellow oil, which was purified by flash chromatography to yield 1.1 g
(69% yield) of the Boc-protected derivative as a pale yellow oil: [R]D
) -7.1° (c ) 0.0065, CHCl3); 1H NMR δ 7.08-7.25 (m, 4), 4.91 (br
s, 1), 4.41 (dd, 2, J ) 4.2, 6.9) 4.25 (br s, 2), 3.74 (s, 3), 3.08 (dd, 1,
J ) 4.2, 13.9), 2.91 (dd, 1, J ) 7.0, 13.9), 2.75 (br s, 1), 1.43 (s, 9);
13C NMR δ 174.5, 155.8, 139.0, 136.7, 128.6, 128.5, 128.4, 126.0,
79.4, 71.2, 52.4, 44.5, 40.4, 28.3. Anal. Calcd for C16H23NO5: C,
62.12; H, 7.49; N, 4.53. Found: C, 61.86; H, 7.62; N, 4.49.
Methyl (2S)-2-[[(1R)-1-(N-(L-N-Phenylmethoxycarbonylvalyl-L-
â-tert-butylaspartyl)amino)-3-methylbutyl]methoxyphosphinyloxy]-
3-(3-(tert-butoxycarbonyl)aminomethyl)phenylpropanoate (19). A
suspension of tripeptide phosphonate 18 (328 mg, 0.42 mmol) and 10%
Pd/C in 4.2 mL of EtOAc was stirred under an atmosphere of hydrogen
for 4 h, filtered through a 4.5-µm nylon syringe filter, and evaporated
to give a quantitative yield (273 mg) of the free amine as a yellow oil.
To a stirring solution of this material in 4.2 mL of DMF at 0 °C were
added Cbz-valine (130 mg, 0.50 mmol), EDC (115 mg, 0.59 mmol),
and HOBt (115 mg, 0.84 mmol). After 12 h, the mixture was diluted
with 50 mL of EtOAc, washed with half-saturated aqueous KHSO4 (5
mL), saturated NaHCO3 (5 mL), and brine (5 mL), dried, and
evaporated. The crude product was chromatographed (0-3% MeOH
in CHCl3) to give 270 mg (73% yield) of phosphonate 19 as a 2:1
mixture of diastereomers at phosphorus. For diastereomer 1: 1H NMR
δ 7.10-7.33 (m, 10), 6.88 (d, 1, J ) 9.8), 5.26-5.49 (m, 2), 5.00-
5.15 (m, 3), 4.70-4.74 (m, 1), 4.45-4.60 (m, 1), 4.20-4.35 (m, 2),
4.00-4.10 (m, 1), 3.76 (s, 3), 3.69-3.73 (m, 1), 3.20-3.24 (m, 1),
3.17 (d, 3, J ) 10.9), 3.01 (dd, 1, J ) 9.1, 14.2), 2.55 (dd, 1, J ) 5.6,
16.9), 2.10-2.20 (m, 1), 1.42 (s, 9), 1.40 (s, 9), 1.23-1.57 (m, 3),
0.72-0.96 (m, 12); 13C NMR δ 170.9, 170.3, 169.7, 169.5, 156.4, 155.8,
139.4, 136.0, 128.6, 128.4, 128.2, 128.1, 127.9, 126.2, 81.5, 79.4, 74.9,
(d, J ≈ 7), 67.0, 60.4, 52.5, 51.8, 49.4, 44.5 (d, J ) 157), 44.4, 38.9,
38.2, 36.6, 36.4, 31.3, 30.7, 28.3, 27.8, 24.2 (d, J ) 14), 23.2, 20.9,
19.1, 17.4; 31P NMR δ 26.2. For diastereomer 2: 1H NMR δ 7.09-
7.31 (m, 9), 6.67 (d, 1, J ) 9.3), 5.56 (br s, 1), 5.43 (d, 1, J ) 7.2),
5.07 (s, 2), 5.00 (br, s, 1), 4.50-4.60 (m, 1), 4.10-4.30 (m, 4), 3.95-
4.05 (m, 1), 3.70 (s, 3), 3.70 (d, 3, J ) 10.3), 3.16 (dd, 1, J ) <3, 14),
Methyl (2S)-2-[(1R)-1-(N-(Phenylmethoxycarbonyl)amino)-3-
methylbutylmethoxyphosphinyloxy]-3-(3-(tert-butoxycarbonyl)ami-
nomethyl)phenylpropanoate (17). A solution of Cbz-leucine phos-
phonate monomethyl ester12 (208 mg, 0.66 mmol) in 2.2 mL of CH2Cl2
was cooled to 0 °C, and thionyl chloride (0.240 mL, 3.3 mmol) was
added dropwise over 2 min. After 30 min, the solution was allowed
to warm to room temperature, sparged with N2 for 20 min, and placed
under vacuum for 2 h to remove the remaining volatile material; the
residue was then redissolved in 2.2 mL of CH2Cl2 and cooled to 0 °C
under N2. Meanwhile, a solution of the Boc-protected amine (265 mg,
0.86 mmol) and TEA (0.185 mL, 1.3 mmol) in 2.2 mL of CH2Cl2 was
prepared in the presence of a few 4 Å molecular sieves and then added
to the phosphonochloridate solution over 2 min. After 1 h, the mixture
was allowed to warm to room temperature and stirred overnight. The
reaction mixture was diluted with 10 mL of CH2Cl2, washed with half-
saturated aqueous KHSO4 (2 mL), saturated NaHCO3 (2 mL), and brine
(2 mL), dried, and evaporated; the oily residue was chromatographed
(30% EtOAc in CH2Cl2) to give 287 mg (72% yield) of phosphonate
17 as a colorless oil as a ca. 70:30 ratio of diastereomers at phosphorus;
each diastereomer appeared as a mixture of rotamers (ca. 10:1). For
diastereomer 1: 1H NMR δ 7.10-7.32 (m, 9), 5.45 (d, 1, J ) 9.7),
4.98-5.15 (m, 4), 4.15-4.30 (m, 3), 3.73 (s, 3), 3.23 (dd, 1, J ) <2,
14.5), 3.13 (d, 3, J ) 10.9), 2.93-3.01 (m, 1), 1.44-1.70 (m, 3), 1.43
(s, 9), 0.90 (d, 6, J ) 6.3); 31P NMR δ 27.8. For diastereomer 2: 1H
NMR δ 7.10-7.32 (m, 9), 4.98-5.15 (m, 4), 4.71 (d, 1, J ) 10.5),
4.15-4.30 (m, 3), 3.73 (d, 3, J ) 10.9), 3.72 (s, 3), 3.07 (m, 1), 2.93-
3.01 (m, 1), 1.44-1.70 (m, 3), 1.43 (s, 9), 0.80 (d, 3, J ) 6.4), 0.79 (d,

4628 J. Am. Chem. Soc., Vol. 120, No. 19, 1998
Smith and Bartlett
3.01 (dd, 1, J ) 8.1, 14.1), 2.75-2.85 (m, 1), 2.45-2.60 (m, 1), 2.05-
2.20 (m, 1), 1.41 (s, 9), 1.40 (s, 9), 1.19-1.50 (m, 3), 0.92 (d, 3, J )
4), 1.47-1.73 (m, 3), 0.87-0.95 (m, 18); 31P NMR δ 26.4. For both
diastereomers: 13C NMR δ 172.9, 171.7, 171.5, 170.9, 170.8, 170.0,
169.9, 169.8, 169.9, 169.5, 138.7, 138.2, 135.8, 135.3, 130.2, 128.4,
128.3, 128.1, 127.9, 127.3, 126.4, 125.5, 74.6 (d, J ) 7), 74.3 (d, J )
7), 58.3, 58.1, 53.6, 52.4, 52.1, 51.2, 49.5, 48.8, 45.6, 44.9 (d, J )
172), 43.5, 43.3 (d, J ) 175), 39.0, 38.4, 38.0, 37.2, 31.1, 31.0, 30.9,
29.7, 28.0, 27.9, 26.1, 26.0, 24.6 (d, J ) 13), 24.3 (d, J ) 13), 23.0,
22.4, 22.4, 22.2, 21.0, 21.0, 19.2, 19.0, 18.1, 17.9, 14.3, 13.7; HRMS
(FAB) calcd for C31H50N4O9P (MH+) m/z 653.3319, found 653.3307.
Methyl Cyclo[(2S)-2-[[(1R)-1-(N-(L-N-(3-methylbutanoyl)valyl-
L-aspartyl)amino)-3-methylbutyl]hydroxyphosphinyloxy]-3-(3-ami-
nomethyl)phenylpropanoate (5). Isobutylene was bubbled through
350 µL of CH2Cl2 for 5 min, bromotrimethylsilane (12 µL) was added,
and, after 5 min, a solution of macrocycle 22 in 350 µL of CH2Cl2
was added. The solution was stirred for 4 h, evaporated from CHCl3
(0.5 mL) and methanol (2 × 0.5 mL), and the residue was dissolved in
1 mL of 1:1 H2O/methanol and eluted through a 1-mL syringe column
of Li+ Dowex with 5 mL of 1:1 H2O/methanol. After evaporation,
the residue was dissolved in 1 mL of H2O, filtered through a 4.5-µm
nylon syringe filter, and lyophilized to give 8.5 mg (75% yield) of the
lithium salt of phosphonate 5 as a fluffy white powder: 1H NMR (CD3-
OD) δ 7.42 (s, 1), 7.14 (t, 1, J ) 7.5), 7.05 (d, 1, J ) 7.6), 6.94 (d, 1,
J ) 7.4), 4.96-4.99 (m, 1), 4.80-4.84 (m, 1), 4.60 (d, 1, J ) 15.2),
4.23-4.30 (m, 1), 4.18 (d, 1, J ) 7.1), 4.08 (d, 1, J ) 15.2), 3.65-
3.75 (m, 1), 3.62 (s, 3), 3.14 (dd, 1, J ) 3.8, 14.1), 3.06 (dd, 1, J )
5.5, 14.2), 2.94 (dd, 1, J ) 10.3, 15.2), 2.62 (dd, 1, J ) <2, 12.7),
1.85-2.30 (m, 4), 1.35-1.62 (m, 3), 0.86-0.96 (m, 18); 13C NMR
(CD3OD) δ 175.7, 173.7, 173.1, 172.1, 171.6, 139.6, 138.2, 130.4,
129.0, 128.9, 126.9, 74.4 (d, J ) 6.4), 68.9, 60.2, 52.4, 51.5, 49.9,
47.2, 46.0, 43.9, 40.4, 39.7 (d, J ) 187), 36.6, 31.7, 29.1, 27.4, 25.7
(d, J ) 12.7), 24.1, 22.8, 22.7, 21.7, 19.9, 18.8; 31P NMR (CD3OD) δ
19.4; HRMS (FAB) calcd for C30H48N4O9P (MH+) m/z 639.315893,
found 639.315130. Anal. Calcd for C30H46N4LiO9P: P, 4.81.
Found: P, 2.45 (weight purity ) 51%).
6.6), 0.86 (d, 3, J ) 6.8), 0.75 (d, 3, J ) 6.2), 0.71 (d, 3, J ) 6.1); 13
C
NMR δ 171.1, 171.0, 170.5, 169.7, 156.5, 156.0, 139.4, 135.8, 128.7,
128.4, 128.3, 128.1, 128.0, 126.3, 81.7, 79.1, 76.7, 75.4, 67.1, 60.4,
53.2, 52.3, 49.3, 44.4, 44.3 (d, J ) 150), 39.0, 37.5, 36.5, 36.3, 31.3,
30.6, 28.3, 27.9, 24.1 (d, 15), 23.1, 20.8, 19.2, 17.4, 15.1; 31P NMR δ
24.8; HRMS (FAB) calcd for C43H66N4O13P (MH+) m/z 877.4381, found
877.4377.
Methyl (2S)-2-[[(1R)-1-(N-(L-N-(3-methylbutanoyl)valyl-(â-tert-
butyl ester)-L-aspartyl)amino)-3-methylbutyl]methoxyphosphinyl-
oxy]-3-(3-(tert-butoxycarbonyl)aminomethyl)phenylpropanoate (20).
This compound was prepared by a procedure similar to that above;
thus, tetrapeptide phosphonate 19 (100 mg, 0.11 mmol) was deprotected
by hydrogenolysis using 20 mg of 10% Pd/C in EtOAc (1.1 mL) and
coupled to isovaleric acid (17 µL, 0.154 mmol) using EDC (35 mg,
0.176 mmol) in DMF (1.1 mL). The product was purified by
chromatography (0-3% methanol in CHCl3) to give 81 mg (86% yield)
of phosphonate pentapeptide 20 as a 2:1 mixture of phosphorus
diastereomers. For diastereomer 1: 1H NMR δ 7.06-7.55 (m, 5), 6.97
(d, 1, J ) 9.3), 5.25-5.33 (m, 1), 4.95-5.05 (m, 2), 4.68-4.73 (m,
1), 4.51-4.59 (m, 1), 4.21-4.25 (m, 2), 3.73 (s, 3), 3.66-3.71 (m, 1),
3.15-3.26 (m, 1), 3.16 (d, 3, J ) 10.8), 2.95-3.08 (m, 1), 2.70-2.77
(m, 1), 2.54 (dd, 1, J ) 5.7, 16.9), 2.05-2.20 (m, 4), 1.39 (s, 9), 1.35
(s, 9), 1.23-1.89 (m, 3), 0.75-0.95 (m, 18); 31P NMR δ 25.4. For
diastereomer 2: 1H NMR δ 7.01-7.62 (m, 4), 6.68 (br s, 1), 6.17 (br
s, 1), 5.53 (br s, 1), 4.95-5.02 (m, 1), 4.55-4.60 (m, 1), 4.11-4.28
(m, 4), 3.79 (s, 3), 3.73 (d, 3, J ) 10.9), 3.20 (dd, 1, J ) <2, 11.6),
3.04 (dd, 1, J ) 8.1, 14.2), 2.68-2.72 (m, 1), 2.30-2.58 (m, 1), 2.05-
2.28 (m, 4), 1.42 (s, 9), 1.41 (s, 9), 1.31-1.63 (m, 3), 0.74-0.95 (m,
18); 31P NMR δ 24.9. For both diastereomers: 13C NMR δ 173.0,
172.7, 171.0, 170.9, 170.3, 169.7, 169.6, 169.5, 169.3, 155.8, 155.5,
139.4, 136.4, 135.8, 128.6, 128.6, 128.3, 128.0, 126.5, 126.3, 126.0,
81.7, 81.5, 79.1, 79.0, 75.9 (d, J ≈ 7), 75.3 (d, J ≈ 7), 58.5, 58.4,
53.2, 52.6, 52.3, 52.0, 49.4, 49.3, 45.7, 44.5 (d, J ) 160), 44.4, 44.2
(d, J ) 158), 39.0, 38.9, 38.3, 38.0, 36.6, 36.3, 30.8, 30.5, 28.3, 27.9,
27.8, 26.0, 24.2 (d, J ) 14), 24.1 (d, J ) 14), 23.2, 23.1, 22.4, 22.3,
20.9, 20.8, 19.2, 19.1, 17.9, 14.1, 13.7; HRMS (FAB) calcd for
C40H68N4O12P (MH+) m/z 827.4571, found 827.4592.
Methyl Cyclo[(2S)-2-[[(1R)-1-(N-(L-N-(3-methylbutanoyl)valyl-
L-aspartyl)amino)-3-methylbutyl]methoxyphosphinyloxy]-3-(3-ami-
nomethyl)phenylpropanoate (22). A solution of phosphonate pen-
tapeptide 20 (40 mg, 0.048 mmol) in 540 mL of 4:4:1 TFA/CH2Cl2/
DMS was stirred for 30 min, diluted with 1,2-dichloroethane, and
evaporated. This residue was dissolved in 3 mL of pH 7.5 phosphate
buffer and extracted with EtOAc (3 × 10 mL). The EtOAc solution
was dried with MgSO4 and evaporated to give 28 mg (88% yield) of
the zwitterion 21, which was further dried by lyophilization from
benzene. A solution of this material in 48 mL of THF (0.001 M) was
cooled to 0 °C, di(isopropyl)ethylamine (63 mL, 0.24 mmol), EDC
(37 mg, 0.192 mmol), and HOBt (26 mg, 0.192 mmol) were added,
and the mixture was allowed to warm to room temperature. After 5 d
of stirring, the solution was evaporated, the residue was dissolved in
30 mL of EtOAc, and the mixture was washed with 1 M HCl (5 mL),
saturated NaHCO3 (5 mL), and brine (5 mL) and dried. After
evaporation, the crude product was chromatographed (0-6% methanol
in CHCl3) to give 12 mg (38% yield) of macrocycle 22 as an opaque
wax. For diastereomer 1: 1H NMR δ 7.03-7.40 (m, 6), 6.76 (br s,
1), 6.21 (d, 1, J ) 8.0), 5.08 (td, 1, J ) 3.8, 7.5), 5.01 (dd, 1, J ) 3.8,
7.5), 4.90-4.94 (m, 1), 4.69-4.74 (m, 1), 4.20 (dd, 1, J ) 6.9, 7.6),
3.89 (dd, 1, J ) 3.8, 15.7), 3.71 (s, 3), 3.28 (d, 3, J ) 10.8), 3.18-
3.22 (m, 1), 3.03-3.10 (m, 1), 2.94 (dd, 1, J ) 10.8, 15.8), 2.78 (dd,
1, J ) 2.1, 15.5), 2.03-2.15 (m, 4), 1.43-1.72 (m, 3), 0.89-1.00 (m,
18); 31P NMR δ 26.5. For diastereomer 2: 1H NMR δ 7.55 (d, 1, J )
9.2), 7.50 (s, 1), 7.28 (d, 1, J ) 5.7), 7.16 (t, 1, J ) 7.7), 7.03 (d, 1,
J ) 7.6), 6.94 (d, 1, J ) 7.7), 6.68 (br s, 1), 6.16 (d, 1, J ) 8.5), 5.15
(dt, 1, J ) 4.4, 8.8), 4.87-4.95 (m, 2), 4.32-4.41 (m, 2), 3.85 (dd, 1,
J ) 5.0, 14.9), 3.70 (d, 3, J ) 11.0), 3.67 (s, 3), 3.15-3.20 (m, 2),
2.84 (dd, 1, J ) <2, 16.4), 2.61 (dd, 1, J ) 9.8, 16.6), 2.06-2.14 (m,
Determination of Inhibition Constants. Inhibition constants for
acyclic inhibitors 6 and 7 were determined at 25 °C as described in
the accompanying report.1 Assay of the macrocyclic analogue 5
involved overnight incubation of inhibitor (1-5 nM) with enzyme (3-5
nM) to achieve equilibrium prior to addition of substrate.30 The enzyme
could be stabilized during these long incubations at pH 4.5 (but not at
pH 3.5) by inclusion of BSA (0.1 mg/mL) in the buffer. Enzyme
velocities were determined by duplicate or triplicate assays at 140 or
280 µM substrate, with the assumption that inhibitor dissociation was
negligible during the course of the assay. The method of Henderson
was employed to analyze the data.19
NMR Structure Determination. NMR spectra were obtained and
analyzed according to the same protocols that are described in the
preceding paper.1
Acknowledgment. We thank Georges Lauri for suggesting
the approach to analyzing the impact of cyclization on a per-
bond basis, 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).
Supporting Information Available: Synthesis and charac-
terization of acyclic comparison compounds 6 and 7; preparation
of 8; kinetic plots for determination of inhibition constants (11
pages, print/PDF). See any current masthead page for ordering
information and Web access instructions.
JA973713Z
(30) Hanson, J. E.; Kaplan, A. P.; Bartlett, P. A. Biochemistry 1989, 28,
6294.