Total synthesis and evaluation of [Ψ[CH2NH]Tpg 4]vancomycin aglycon: Reengineering vancomycin for dual D-Ala-D-Ala and D-Ala-D-Lac binding

Article abstract of DOI:10.1021/ja0572912

An effective synthesis of [Ψ[CH₂NH]Tpg⁴] vancomycin aglycon (5) is detailed in which the residue 4 amide carbonyl of vancomycin aglycon has been replaced with a methylene. This removal of a single atom was conducted to enhance bindin

Full text of DOI:10.1021/ja0572912

Published on Web 02/04/2006
Total Synthesis and Evaluation of
[Ψ[CH2NH]Tpg4]Vancomycin Aglycon: Reengineering
Vancomycin for Dual D-Ala-D-Ala and D-Ala-D-Lac Binding
Brendan M. Crowley and Dale L. Boger*
Contribution from the Department of Chemistry and the Skaggs Institute for Chemical Biology,
The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037
Received October 25, 2005; E-mail: boger@scripps.edu
Abstract: An effective synthesis of [Ψ[CH2NH]Tpg4]vancomycin aglycon (5) is detailed in which the residue
4 amide carbonyl of vancomycin aglycon has been replaced with a methylene. This removal of a single
atom was conducted to enhance binding to D-Ala-D-Lac, countering resistance endowed to bacteria that
remodel their D-Ala-D-Ala peptidoglycan cell wall precursor by a similar single atom change (ester O for
amide NH). Key elements of the approach include a synthesis of the modified vancomycin ABCD ring
system featuring a reductive amination coupling of residues 4 and 5 for installation of the deep-seated
amide modification, the first of two diaryl ether closures for formation of the modified CD ring system (76%,
2.5-3:1 kinetic atropodiastereoselectivity), a Suzuki coupling for installation of the hindered AB biaryl bond
(90%) on which the atropisomer stereochemistry could be thermally adjusted, and a macrolactamization
closure of the AB ring system (70%). Subsequent DE ring system introduction enlisted a room-temperature
aromatic nucleophilic substitution reaction for formation of the remaining diaryl ether (86%, 6-7:1 kinetic
atropodiastereoselectivity), completing the carbon skeleton of 5. Consistent with expectations and relative
to the vancomycin aglycon, 5 exhibited a 40-fold increase in affinity for D-Ala-D-Lac (Ka ) 5.2 × 103 M-1
)
and a 35-fold reduction in affinity for D-Ala-D-Ala (Ka ) 4.8 × 103 M-1), providing a glycopeptide analogue
with balanced, dual binding characteristics. Beautifully, 5 exhibited antimicrobial activity (MIC ) 31 µg/mL)
against a VanA-resistant organism that remodels its D-Ala-D-Ala cell wall precursor to D-Ala-D-Lac upon
glycopeptide antibiotic challenge, displaying a potency that reflects these binding characteristics.
The most common strains of enterococci resistant to vanco-
mycin (1), VanA and VanB, possess an inducible resistance
pathway in which the terminal dipeptide of the cell wall
peptidoglycan precursor is modified from D-Ala-D-Ala to D-Ala-
D-Lac.1 Binding of the antibiotic to this modified ligand is
reduced 1000-fold, leading to a 1000-fold drop in antimicrobial
activity.1d,k In a recent disclosure,2 we provided the first
experimental study on the origin of this loss in binding affinity,
partitioning the effect into lost H-bond and repulsive lone-pair
contributions (Figure 1). Thus, the binding affinity of vanco-
mycin for 3, which incorporates a methylene (CH2) in place of
the linking amide NH of Ac2-L-Lys-D-Ala-D-Ala, was compared
with that of Ac2-L-Lys-D-Ala-D-Ala (2) and Ac2-L-Lys-D-Ala-
D-Lac (4). The vancomycin affinity for 3 was approximately
(1) Reviews: (a) Kahne, D.; Leimkuhler, C.; Lu, W.; Walsh, C. Chem. ReV.
2004, 105, 425. (b) Hubbard, B. K.; Walsh, C. T. Angew. Chem., Int. Ed.
2003, 42, 730. (c) Nicolaou, K. C.; Boddy, C. N. C.; Bra¨se, S.; Winssinger,
N. Angew. Chem., Int. Ed. 1999, 38, 2096. (d) Williams, D. H.; Bardsley,
B. Angew. Chem., Int. Ed. 1999, 38, 1172. (e) Malabarba, A.; Nicas, T. I.;
Thompson, R. C. Med. Res. ReV. 1997, 17, 69. Glycopeptide resistance
and analogues: (f) Malabarba, A.; Ciabatti, R. Curr. Med. Chem. 2001, 8,
1759. (g) Pootoolal, J.; Neu, J.; Wright, G. D. Annu. ReV. Pharmacol.
Toxicol. 2002, 42, 381. (h) Van Bambeke, F. V.; Laethem, Y. V.; Courvalin,
P.; Tulkens, P. M. Drugs 2004, 64, 913. (i) Sussmuth, R. D. ChemBioChem
2002, 3, 295. (j) Gao, Y. Nat. Prod. Rep. 2002, 19, 100. (k) Healy, V. L.;
Lessard, I. A. D.; Roper, D. I.; Knox, J. R.; Walsh, C. T. Chem. Biol.
2000, 7, R109.
Figure 1. Binding constants of D-Ala-D-Ala, D-Ala-D-Lac, and a peptide
analogue with vancomycin.
(2) McComas, C. C.; Crowley, B. M.; Boger, D. L. J. Am. Chem. Soc. 2003,
125, 9314.
9
10.1021/ja0572912 CCC: $33.50 © 2006 American Chemical Society
J. AM. CHEM. SOC. 2006, 128, 2885-2892
2885

A R T I C L E S
Crowley and Boger
10-fold less than that of 2, but 100-fold greater than that of 4.
This indicated that the reduced binding affinity of 4 (4.1 kcal/
mol) may be attributed to both the loss of a key H-bond and a
destabilizing lone pair/lone pair interaction introduced with the
ester oxygen of 4 (2.6 kcal/mol), with the latter, not the H-bond,
being responsible for the greater share (100-fold) of the 1000-
fold binding reduction. These observations have significant
ramifications on the reengineering of vancomycin to bind D-Ala-
D-Lac, suggesting that the design could focus principally on
removing the destabilizing lone-pair interaction rather than
reintroducing a H-bond and that this may be sufficient to
compensate for two of the three orders of magnitude in binding
affinity lost with D-Ala-D-Lac. Thus, synthesis of a vancomycin
analogue with removal of the residue 4 carbonyl and its
destabilizing lone-pair interaction could potentially restore much
of the binding affinity of the antibiotic for the modified ligand.
At present, such a deep-seated change in the molecule can only
be achieved by total synthesis.3 Having developed a convergent
and efficient synthesis of the vancomycin aglycon,4 and having
extended its effectiveness in the synthesis of the teicoplanin5
and ristocetin aglycons,6 we embarked on the total synthesis of
[Ψ[CH2NH]Tpg4]vancomycin aglycon (5), in which the residue
4 carbonyl has been replaced with a methylene. Through such
efforts, we were able to establish the effect this structural
modification has on the binding affinity for both D-Ala-D-Ala
and D-Ala-D-Lac and probe its impact on the antimicrobial
activity of the analogue.
Challenges and Synthetic Plan for [Ψ[CH2NH]Tpg4]-
Vancomycin Aglycon. The desired analogue 5 was anticipated
to be prepared by a route analogous to that developed for
vancomycin,4 with notable modifications. Thus, two aromatic
nucleophilic substitution reactions with formation of the biaryl
ethers would be enlisted for CD and DE macrocyclization, a
key macrolactamization reaction would be employed for cy-
clization of the AB ring system, and the defined order of CD,
AB, and DE ring closures was expected to permit sequential
control of the atropisomer stereochemistry of each of the newly
formed ring systems or their immediate precursors (Figure 2).
Thus, in addition to any kinetic diastereoselection that may be
achieved in the ring closures, this order was expected to permit
the recycling of any undesired atropisomer for each newly
introduced ring system by thermal equilibration, providing a
predictable control of the stereochemistry and dependably
funneling all synthetic material into one of eight possible
atropodiastereomers. Key to recognition of this preferential order
of ring closures was our establishment of the thermodynamic
parameters of atropisomerism for the individual vancomycin
Figure 2. Plan for [Ψ[CH2NH]Tpg4]vancomycin aglycon.
ring systems: DE ring system7 (Ea ) 18.7 kcal/mol) < AB
biaryl precursor4 (Ea ) 25.1 kcal/mol) < CD ring system7 (Ea
) 30.4 kcal/mol). Thus, the molecule was to be assembled by
coupling the modified and fully functionalized ABCD ring
system 27 with the E ring tripeptide 28, followed by a
diastereoselective aromatic nucleophilic substitution reaction for
closure of the 16-membered DE ring system with formation of
the biaryl ether linkage. Notably, the activating nitro substituent
additionally serves as the precursor functionality for aryl chloride
introduction, and the analogous vancomycin ring closures4,8-10
were effected with preferential formation of the natural (P)-
atropisomer. The E ring tripeptide 28 would be derived in the
manner described for vancomycin,4 except that the E ring
subunit would be prepared enlisting an improved route devel-
oped during our more recent total synthesis of the ristocetin
(3) Efforts to modify the residue 4 carbonyl by selective reaction of the amide
linking residues 4 and 5 within vancomycin aglycon derivatives have not
yet been successful in our efforts, and we are unaware of any reports of
such modifications. Synthetic reviews: (a) Boger, D. L. Med. Res. ReV.
2001, 21, 356. (b) Lloyd-Williams, P.; Giralt, E. Chem. Soc. ReV. 2001,
30, 145. (c) Zhang, A. J.; Burgess, K. Angew. Chem., Int. Ed. 1999, 38,
634. (d) Rao, A. V. R.; Gurjar, M. K.; Reddy, K. L.; Rao, A. S. Chem.
Rev. 1995, 95, 2135. (e) Evans, D. A.; DeVries, K. M. Drugs Pharm. Sci.
1994, 63, 63.
(4) (a) Boger, D. L.; Miyazaki, S.; Kim, S. H.; Wu, J. H.; Loiseleur, O.; Castle,
S. L. J. Am. Chem. Soc. 1999, 121, 3226. (b) Boger, D. L.; Miyazaki, S.;
Kim, S. H.; Wu, J. H.; Castle, S. L.; Loiseleur, O.; Jin, Q. J. Am. Chem.
Soc. 1999, 121, 10004.
(5) (a) Boger, D. L.; Kim, S. H.; Miyazaki, S.; Strittmatter, H.; Weng, J.-H.;
Mori, Y.; Rogel, O.; Castle, S. L.; McAtee, J. J. J. Am. Chem. Soc. 2000,
122, 7416. (b) Boger, D. L.; Kim, S. H.; Mori, Y.; Weng, J.-H.; Rogel, O.;
Castle, S. L.; McAtee, J. J. J. Am. Chem. Soc. 2001, 123, 1862.
(6) Crowley, B. M.; Mori, Y.; McComas, C. C.; Tang, D.; Boger, D. L. J.
Am. Chem. Soc. 2004, 126, 4310.
(7) (a) Boger, D. L.; Borzilleri, R. M.; Nukui, S.; Beresis, R. T. J. Org. Chem.
1997, 62, 4721. (b) Boger, D. L.; Borzilleri, B.; Nukui, S. J. Org. Chem.
1996, 61, 3561. (c) Boger, D. L.; Castle, S. L.; Miyazaki, S.; Wu, J. H.;
Beresis, R. T.; Loiseleur, O. J. Org. Chem. 1999, 64, 70. (d) Boger, D. L.;
Loiseleur, O.; Castle, S. L.; Beresis, R. T.; Wu, J. H. Bioorg. Med. Chem.
Lett. 1997, 7, 3199. (e) Boger, D. L.; Beresis, R. T.; Loiseleur, O.; Wu, J.
H.; Castle, S. L. Bioorg. Med. Chem. Lett. 1998, 8, 721. (f) Boger, D. L.;
Miyazaki, S.; Loiseleur, O.; Wu, J. H.; Jin, Q. J. Am. Chem. Soc. 1998,
120, 8920.
(8) Boger, D. L.; Borzilleri, R. M.; Nukui, S. Bioorg. Med. Chem. Lett. 1995,
5, 3091.
(9) Evans, D. A.; Watson, P. S. Tetrahedron Lett. 1996, 37, 3251.
(10) (a) Evans, D. A.; Wood, M. R.; Trotter, B. W.; Richardson, T. I.; Barrow,
J. C.; Katz, J. L. Angew. Chem., Int. Ed. 1998, 37, 2700. (b) Evans, D. A.;
Dinsmore, C. J.; Watson, P. S.; Wood, M. R.; Richardson, T. I.; Trotter,
B. W.; Katz, J. L. Angew. Chem., Int. Ed. 1998, 37, 2704.
9
2886 J. AM. CHEM. SOC. VOL. 128, NO. 9, 2006

[
Ψ
[CH2NH]Tpg4]Vancomycin Aglycon
A R T I C L E S
aglycon,6 employing an R-hydroxypinanone11 as the chiral
auxiliary for a diastereoselective aldol addition. The most
significant deviations rest with the required modifications in
the preparation of the ABCD subunit, which houses the modified
amide, and include the use of a reductive amination coupling
of residues 4 and 5 (D and B rings) with protection of the newly
generated amine as a methyl carbamate and an experimentally
derived altered order to the assembly of the BCD tripeptide. A
relatively small and robust amine protecting group was chosen
to avoid the introduction of unfavorable steric interactions that
may affect the CD macrocyclic ring closure and that would be
stable throughout the synthesis, yet still be compatible with a
final stage global deprotection. CD macrocyclization, enlisting
a key aromatic nucleophilic substitution reaction for formation
of 16-membered biaryl ether, followed by Suzuki coupling of
the A ring subunit and AB macrolactamization, was anticipated
to complete the preparation of the modified ABCD ring system
27, enlisting a ring closure order that would permit the sequential
and selective thermal adjustment of the CD and AB ring system
atropisomer stereochemistry. Key unknown features of the
approach include the feasibility of conducting the critical CD
ring closure enlisting the residue 4 protected amine versus
amide, the resulting unknown atropisomer stereochemical issues
(kinetic and thermodynamic diastereoselectivity), and the impact
the deep-seated structural change would have on the confor-
mational features of the CD or ABCD ring systems and those
of the final molecule. Finally, the subtle choice of a nitrile as
a precursor to the residue 3 side-chain carboxamide permits a
final stage amide deprotection yet conveys stability to any
projected thermal atropisomer equilibrations in its presence,7f
and the use of a protected hydroxymethyl precursor (vs a methyl
ester) to the C-terminus carboxylic acid enhances the rate of
the projected AB macrolactamization4 and precludes inadvertent
epimerization throughout the synthesis.
Scheme 1
tection14 (Raney nickel, CH3OH, 0 °C, 5 h, 98%) and saponi-
fication (3.0 equiv of LiOH, THF-H2O, 0 °C, 6 h, 100%),
provided 12. Unexpectedly, the order of these latter two
deprotections proved important. Saponification of 1015 under a
variety of conditions (LiOH, THF-H2O or t-BuOH-H2O, -10
to 0 °C; LiOOH, THF-H2O; Me3SnOH, 1,2-dichloroethane,
70 °C) led to variable amounts of an epimer (5-20%) that was
difficult to separate from the product. In contrast, benzyl ether
deprotection of 10, followed by saponification of 11, reduced
the amount of epimer (0-3%), presumably due to preferential
deprotonation of the phenols such that subsequent CR deproto-
nation at residue 5 was less facile.15 Coupling of 12 with 1316
(3.0 equiv of DEPBT,17 3.0 equiv of NaHCO3, DMF, 0-25
°C, 8 h) gave tripeptide 14 in good yield (70%) and excellent
diastereoselectivity (14:1). A range of other more conventional
coupling reagents (EDCI-HOAt, HATU, FDPP) also provided
good conversions (65-80%) but suffered from considerable
competitive racemization.
Synthesis of the ABCD Ring System. The stage was now
set for a detailed examination of one of the critical reactions in
the approach to 5, entailing the cyclization of 14. After
considerable optimization (Supporting Information Tables S1
and S2), cyclization of 14 (20 equiv of K2CO3, 20 equiv of
CaCO3, 3 wt equiv of 3-Å molecular sieves, 12 mM THF, 75
°C bath temperature, 12 h) afforded 15 in good yield (54%)
and good atropodiastereoselectivity (2.5:1, 15 (54%) and 16
(22%)), even when conducted on a large scale (2.7 g) (Scheme
Synthesis of the BCD Tripeptide. The B and D subunits 6
and 7 were prepared following previously optimized proce-
dures.6,7 Oxidation of alcohol 712 (2.0 equiv of Dess-Martin
periodinane, CH2Cl2, 0-25 °C, 1 h, 100%) was followed by
immediate reductive amination coupling of the sensitive alde-
hyde 8 with 613 (1.1 equiv, CH3OH, 3-Å molecular sieves, 0
°C, 45 min; 3.0 equiv of AcOH, 3.0 equiv of NaBH3CN, -20
°C, 2 d) to afford amine 9 in good yield (75%) and excellent
diastereoselectivity (12:1) (Scheme 1). Shorter reaction times
(14-20 h) at higher temperatures (-15 to -5 °C) led to
substandard selectivities (4:1 to 9:1), and the use of less NaBH3-
CN (1.5-2.0 equiv) or lower temperatures (-30 °C) led to
incomplete reactions. Longer reaction times (3-8 d) led to only
marginal increases in yield (82% after 8 d) and roughly equal
diastereoselectivities. Initial efforts to prepare 9 directly by
displacement of the mesylate derived from alcohol 7 were
ineffective, as were attempts to conduct the reductive amination
with the BC dipeptide and 8. Amine protection of 9 as the
methyl carbamate (10 equiv of MeOCOCl, 10 equiv of K2CO3,
THF, 0-25 °C, 18 h, 85%), followed by benzyl ether depro-
(14) Benzyl ether deprotection at higher temperatures (25 °C) may lead to
competitive aryl bromide reduction, although this was observed in
appreciable amounts only when excess Raney nickel was employed.
(15) Saponification of 11 was considerably slower than that of 10 and
occasionally required additional LiOH for complete conversion to 12, with
little effect on the amount of epimer generated in the reaction.
(16) Compound 13 is available in three steps (45% overall) from 4-fluoro-3-
nitrobenzaldehyde and was scaled to 30 g, ref 6.
(11) Solladie´-Cavallo, A.; Nsenda, T. Tetrahedron Lett. 1998, 39, 2191.
(12) Compound 7 is available in six steps (37% overall) from methyl gallate
using three recrystallizations and was scaled to 300 g, ref 6.
(13) Compound
6
is available in five steps (55% overall) from (R)-4-
(17) Fan, C.-X.; Hao, X.-L.; Ye, Y.-H. Synth. Commun. 1996, 26, 1455. Li, H.;
Jiang, X.; Ye, Y.-H.; Fan, C.; Romoff, T.; Goodman, M. Org. Lett. 1999,
1, 91.
hydroxyphenyl-glycine using two recrystallizations and was scaled to 60
g, ref 7a.
9
J. AM. CHEM. SOC. VOL. 128, NO. 9, 2006 2887

A R T I C L E S
Scheme 2
Crowley and Boger
Scheme 3
CD ring system, in which the atropisomers could be thermally
equilibrated at 120-140 °C, permitting the recycling and
productive use of the unnatural atropisomer, the atropisomers
15 and 16 could not be thermally interconverted, even at
temperatures as high as 210-230 °C (Scheme 3).
Reduction of the nitro group (Raney nickel, 0 °C, CH3OH, 1
h), followed by diazotization (1.3 equiv of HBF4, 1.3 equiv of
t-BuONO, CH3CN, 0 °C, 30 min) and Sandmeyer substitution
(50 equiv of CuCl, 60 equiv of CuCl2, H2O, 0-25 °C, 1 h,
70% from 15), cleanly provided 17 without loss of the
atropisomer stereochemistry inherent in starting 15. The un-
natural atropisomer 16 was also subjected to these conditions
to cleanly give 18 (75%) (Scheme 3). The stereochemical
assignments of these two compounds and their relationship as
atropisomers (vs epimers) were established by 2D ROESY 1H-
1H NMR experiments and confirmed chemically by their
reductive dechlorination (H2, 10%, Pd/C) to afford the identical
product 19 (Scheme 3). Diagnostic NOE cross-peaks for 17 were
2). The inclusion of CaCO3 in the reaction mixture is critical
and serves to trap the liberated fluoride arising from the aromatic
nucleophilic substitution as an insoluble byproduct (CaF2),
preventing TBS ether deprotection and a subsequent competitive
base-catalyzed retro aldol reaction of the free alcohol. Nearly
comparable results were obtained by promoting the ring closure
of 15 with the stronger base t-BuOK (1.0 equiv, THF, -20 °C,
18 h), providing 15 and its atropisomer 16 in 57% and 19%
yields (3:1 atropodiastereoselectivity), respectively, under re-
markably mild reaction conditions (-20 °C, THF). However,
the use of t-BuOK proved more sensitive to the reaction
parameters, suffered competitive racemization if excess base
was employed, and proved more difficult to implement on a
large scale than the reaction enlisting K2CO3/CaCO3. The
cyclization of 14 represents a considerable improvement over
the analogous ring closure reaction enlisted in our original
synthesis of vancomycin (50-65%, 1:1 atropisomers vs 76-
87%, 2.5-3:1 atropisomers), where both the overall conversion
and the atropodiastereoselectivity were lower, illustrating that
the closure of 14 may benefit from the increased conformational
flexibility of the cyclization substrate.18 Unlike the vancomycin
6
6
6
observed between C5a6-H/C3 -H (s), C5a6-H/C2 -H (s), C5b
-
5
6
6
5
H/C6b6-H (s), C5 -H/C4b5-H (s), C3 -H/C2 -H (s), C2 -H/
C4a5-H (w), C2 -H/C4b5-H (w), C2 -H/C1b4-H (s), C2 -H/
5
4
4
C4b4-H (s), C1a4-H/C1b4-H (s), C5 -OH/C6 -OMe (s), and
4
4
5
5
C5 -H/C6 -OMe (m), and no NOE cross-peaks were observed
between C5b6-H/C3 -H and C5b6-H/C2 -H. Diagnostic NOE
6
6
cross-peaks for 18 were observed between C5b6-H/C6b6-H (s),
C5b6-H/C3 -H (s), C5b6-H/C2 -H (s), C5 -H/C4b5-H (s),
6
6
5
C3 -H/C2 -H (s), C2 /N1 -H (w), C2 -H/C4a5-H (m), C2 -
6
6
6
6
5
5
4
4
H/C4b5-H (m), C2 -H/C1b4-H (s), C1b4-H/C4a5-H (w), C1b
-
-
H/C4b5-H (w), C1a4-H/C4b4-H (s), C1b4-H/C1a4-H (s), C4b
4
4
4
4
5
5
H/C5 -OH (m), C5 -OH/C6 -OMe (w), C5 -H/C6 -OMe
(w), C2 -H/N1 -H (w), and no NOE cross-peaks were observed
4
4
6
6
between C5a6-H/C3 -H and C5a6-H/C2 -H.
Suzuki coupling of 17 with the hindered A ring boronic acid
204 (0.3 equiv of Pd2(dba)3, 1.5 equiv of (o-tol)3P, toluene-
CH3OH-1 N aqueous Na2CO3 10:3:1, 80 °C, 30 min) pro-
ceeded in excellent yield (90%) under remarkably effective
(18) Recent efforts have led to improvements in the vancomycin CD ring closure,
and this method was utilized for the preparation of the analogous
intermediates bearing a residue 4/5 thioamide. Its use in the preparation of
5 along with additional vancomycin analogues will be disclosed in due
time.
9
2888 J. AM. CHEM. SOC. VOL. 128, NO. 9, 2006

[
Ψ
[CH2NH]Tpg4]Vancomycin Aglycon
A R T I C L E S
conditions4 given the steric constraints of the substrate 20,
providing a separable 1:1.3 mixture of atropisomers (21:22)
slightly favoring the unnatural configuration. Thermal equilibra-
tion of isolated 22 was carried out initially employing our
reported conditions for vancomycin (o-dichlorobenzene, 120 °C,
18 h, 81% recovery of material)7 to afford a 1:1.1 separable
mixture, permitting the recycling of this unnatural atropisomer.
An examination of the parameters for this isomerization (k )
0.12 h-1, t1/2 ) 5.9 h at 120 °C and k ) 0.36 h-1, t1/2 ) 1.8 h
at 135 °C) revealed that it proceeds with an energy of activation
(Ea) of 25.6 kcal/mol (∆Hq ) 24.8 kcal/mol, ∆Sq ) -0.26 eu,
∆Gq ) 24.9 kcal/mol) essentially indistinguishable from that
observed with the authentic vancomycin AB biaryl system, but
it does not result in the analogous 3:1 thermodynamic preference
for the natural atropisomer. However, the unusual and unex-
pected atropisomer stability of the CD ring system allowed us
to improve on the recycling conditions. Heating the mixture in
a microwave reactor at an elevated temperature (210 °C,
o-dichlorobenzene) shortened the reaction time significantly (5
min vs 18 h) and slightly improved the recovery of material
(88% vs 81%). This improvement impacted the efficiency of
the recycling of 22 by allowing multiple equilibrations to be
run in a single day rather than over the course of a week. Silyl
ether deprotection of 21 (1.2 equiv of Bu4NF, THF, 0 °C, 10
min), followed by N-Cbz removal (H2, 10% Pd/C, 1% Cl3-
CCO2H-CH3OH, 15 min, 95%) and methyl ester hydrolysis
(1.0 equiv of LiOH, THF-H2O, 0 °C, 1 h, 96%), gave amino
acid 25. Notably, N-Cbz removal in the absence of Cl3CCO2H5
was much slower (11 h), and these conditions led to competitive
chloride reduction.19 Macrolactamization with closure of the AB
ring system was effected by treatment of 25 with PyBOP (3.0
equiv, 6.0 equiv of NaHCO3, 0.001 M CH2Cl2-DMF 5:1, 0-25
°C, 12 h) to afford the fully functionalized bicyclic ABCD ring
system 26 in good yield (70%) with only trace amounts of
competitive epimerization (<3%). Alternative coupling reagents
(EDCI and HOAt or HOBt, HATU) and reaction conditions
(10-100% DMF-CH2Cl2, 3-5 equiv of Na2CO3, -5 to 0 °C)
led to lower conversions (30-52%) or required extended
reaction times (3 d). N-Boc deprotection (HCO2H-CHCl3 1:1,
10 h, 84%) gave the free amine 27 for coupling with the E ring
tripeptide. Confirmation of the atropisomer stereochemistry and
amide conformational assignments for 26 were established by
2D ROESY 1H-1H NMR. Diagnostic NOE cross-peaks for 26
H, N1 -H/C2 -H, and N1 -H/C4a5-H. Most important in this
spectroscopic assessment was not only the expected confirmation
of the CD and AB atropisomer stereochemistry, but also the
establishment of a vancomycin-like conformation for 26 bearing
6
5
6
5
a cis amide linking the residues 5 and 6 (strong diagnostic C2 -
6
H/C2 -H NOE), maintaining the spatial relationships and
5
orientations of the AB ring system (strong diagnostic C2 -H/
6
C4a5-H and C2 -H/C4a5-H NOEs) and the CD ring system
(diagnostic C6b6-H/C4a4-H NOE). Although this might be
considered unusual on the surface, even the natural atropisomer
of the isolated AB ring system of vancomycin, without the
surrounding CD ring system, adopts a conformation incorporat-
ing this cis amide structure, illustrating that it is the confines
of the AB ring system, not that of the CD ring system, that
defines this key cis amide conformational preference.4 The lack
of discernible NOEs to the methyl carbamate protecting the
amine of the modified amide established that it extends out and
away from the ABCD ring system binding pocket.
Synthesis of the Full Carbon Skeleton. Coupling of 27 and
28 (2.0 equiv of DEPBT,17 2.2 equiv of NaHCO3, THF, 0-25
°C, 14 h, 73%) afforded 29 with excellent diastereoselectivity
(12:1), arising from little competitive racemization (Scheme 4).
These conditions were utilized on the basis of our experience
with the teicoplanin5 and ristocetin6 aglycons and are superior
to those originally reported for vancomycin4 (EDCI) in terms
of diastereoselectivity (12:1 vs 3:1). Closure of the DE ring
system with formation of the key biaryl ether was accomplished
by treatment of 29 with CsF (10 equiv, 20 equiv of CaCO3,20
3-Å molecular sieves, DMF, 25 °C, 17 h) to afford 30 in good
yield (74%) and good atropodiastereoselectivity (6-7:1). No-
tably, the closure of 30 was conducted under milder conditions
than those originally disclosed for vancomycin4,7-10 (DMF vs
DMSO at 25 °C with added 3-Å molecular sieves and CaCO3)
and approaches the kinetic atropisomer diastereoselectivity
observed in our efforts4 (8:1), while surpassing that detailed in
the related efforts by Evans10 (5:1) and contrasting the closure
detailed by Nicolaou21 (1:3), where the unnatural atropisomer
predominated with an alternative substrate and method of ring
closure. Thus, consistent with the adoption of a vancomycin-
like conformation by 26, the amide modification in the ABCD
ring system of 29 had little impact on the ease or diastereose-
lectivity of the DE ring closure. Reduction of the nitro group22
(H2, 10% Pd/C, THF, 8 h, 94%), followed by diazotization of
the resulting amine 32 (1.2 equiv of HBF4, 1.2 equiv of
t-BuONO, CH3CN, 0 °C, 20 min) and Sandmeyer substitution
4
4
4
were observed between C5 -OH/C4b4-H (s), C5 -OH/C6 -
OMe (s), N1 -H/C4a5-H (s), N1 -H/C2 -H (s), N1 -H/C3 -H
7
7
5
7
6
7
6
6
6
(m), N1 -H/C2 -H (m), C5a6-H/C3 -H (s), C5a6-H/C2 -H
(20) Both the added 3-Å molecular sieves and CaCO3 result in cleaner
conversions to product. It is not yet clear whether the soluble base under
these conditions is CsF or Cs2CO3 with precipitation of insoluble CaF2.
(21) (a) Nicolaou, K. C.; Takayanagi, M.; Jain, N. F.; Natarajan, S.; Koumbis,
A. E.; Bando, T.; Ramanjulu, J. M. Angew. Chem., Int. Ed. 1998, 37, 2717.
(b) Nicolaou, K. C.; Natarajan, S.; Li, H.; Jain, N. F.; Hughes, R.; Solomon,
M. E.; Ramanjulu, J. M.; Boddy, C. N. C.; Takayanagi, M. Angew. Chem.,
Int. Ed. 1998, 37, 2708. (c) Nicolaou, K. C.; Jain, N. F.; Natarajan, S.;
Hughes, R.; Solomon, M. E.; Li, H.; Ramanjulu, J. M.; Takayanagi, M.;
Koumbis, A. E.; Bando, T. Angew. Chem., Int. Ed. 1998, 37, 2714. (d)
Nicolaou, K. C.; Mitchell, H. J.; Jain, N. F.; Winssinger, N.; Hughes, R.;
Bando, T. Angew. Chem., Int. Ed. 1999, 38, 240. (e) Nicolaou, K. C.; Li,
H.; Boddy, C. N. C.; Ramanjulu, J. M.; Yue, T.-Y.; Natarajan, S.; Chu,
X.-J.; Brase, S.; Rubsam, F. Chem. Eur. J. 1999, 5, 2584. (f) Nicolaou, K.
C.; Boddy, C. N. C.; Li, H.; Koumbis, A. E.; Hughes, R.; Natarajan, S.;
Jain, N. F.; Ramajulu, J. M.; Brase, S.; Solomon, M. E. Chem. Eur. J.
1999, 5, 2602. (g) Nicolaou, K. C.; Koumbis, A. E.; Takayanagi, M.;
Natarajan, S.; Jain, N. F.; Bando, T.; Li, H.; Hughes, R. Chem. Eur. J.
1999, 5, 2622. (h) Nicolaou, K. C.; Mitchell, H. J.; Jain, N. F.; Bando, T.;
Hughes, R.; Winssinger, N.; Natarajan, S.; Koumbis, A. E. Chem. Eur. J.
1999, 5, 2648.
6
6
6
6
(s), C5b6-H/N1 -H (m), C3 -OH/N1 -H (s), C5b6-H/C3 -
OH (m), C6b6-H/C5b6-H (s), C6b6-H/C4a4-H (w), N1 -H/
4
4
5
6
C4b4-H (m), N1 -H/C4a4-H (w), C4b5-H/C5 -H (s), C2 -H/
7
C4a5-H, C4b5-H/C1b4-H (m), C4a5-H/C6 -H (w), C4a5-H/
5
5
5
7
7
7
C2 -H (s), C5 -H/C6 -OMe (s), C4 -H/C2 -H (s), C4 -H/
7
7
7
C1b7-H (s), C4 -H/C1a7-H (w), C4 -H/C5b7-OMe (s), C4 -
7
7
5
7
7
H/C6 -H (w), C6 -H/C2 -H (w), C6 -H/C5b7-OMe (s), C6 -
5
6
5
6
6
H/C5a7-OMe (s), C2 -H/C3 -H (m), C2 -H/C2 -H (s), C3 -
6
7
7
H/C2 -H (m), C1 -(MEM-CH2)1/C1a7-H (s), C1 -(MEM-
7
7
7
CH2)1/C1 -(MEM-CH2)2 (s), C2 -H/C1b7-H (s), and C2 -H/
C1a7-H (s), and no NOE cross-peaks were observed between
6
6
6
6
6
7
C5b6-H/C3 -H, C5b6/C2 -H, C2 -H/C3 -OH, N1 -H/N1 -
(19) Use of Raney nickel for N-Cbz removal was also successful, although lower
recoveries (84%) of the product were observed.
(22) Reduction of the nitro group was sensitive to the choice of solvent in terms
of recovery and observance of side products.
9
J. AM. CHEM. SOC. VOL. 128, NO. 9, 2006 2889

A R T I C L E S
Scheme 4
Crowley and Boger
Figure 3. Potential modifications.
carboxylic acid 36. Hydrolysis of the residue 3 nitrile with
formation of the carboxamide 37 (40 equiv of 30% aqueous
H2O2, 8.0 equiv of 10% aqueous K2CO3, DMSO, 25 °C, 3.5 h,
87%)4 set the stage for a final global deprotection.23 In a final
key reaction, 37 was treated with AlBr3 (35 equiv, EtSH, 25
°C, 5 h, 80%) to afford 5, arising from the remarkable
deprotection of four aryl methyl ethers, the two TBS ethers,
the N-terminus Boc group, and the critical residue 4 methyl
carbamate.
Assessment of [Ψ[CH2NH]Tpg4]Vancomycin Aglycon. A
subtle element in the design of 5 and choice of simply removing
the residue 4 carbonyl rests with the projected properties of the
molecule. In principle, one might consider reengineering the
capabilities of a reverse H-bond into the vancomycin structure,
removing the destabilizing lone-pair interaction with D-Ala-D-
Lac and reinstating the lost H-bond. Such opportunities include
amidine derivatives (e.g., [Ψ[C(dNH)NH]Tpg4]vancomycin
aglycon 39, Figure 3), and ongoing efforts have provided a
thioamide intermediate that we intend to use to access such
derivatives.18 Significantly, such derivatives should enhance
D-Ala-D-Lac binding, approaching the level of affinity observed
with vancomycin and D-Ala-D-Ala. However, these derivatives
would also accordingly reduce binding to D-Ala-D-Ala. Con-
sequently, they would be expected to gain antimicrobial activity
against constituitively resistant bacteria endowed with a D-Ala-
D-Lac peptidoglycan cell wall precursor (e.g., VanD) but be
inactive against sensitive and inducibly resistant bacteria (VanA
and VanB) that maintain or at least start with a D-Ala-D-Ala
peptidoglycan cell wall precursor. The closest modified van-
comycins that would be expected to reproduce the binding
results observed in Figure 1 are those that replace the amide
bond linking residues 4 and 5 with a methylene (CH2CH2) or
ethylene (CHdCH) linker. Such derivatives, by analogy with
the results in Figure 1, would be expected to enhance D-Ala-
D-Lac affinity 100-fold, missing only the contribution to binding
derived from the H-bond. However, they might be expected to
pay an additional penalty for binding D-Ala-D-Ala derived from
not only the loss of the H-bond (10-fold) but also for not
(50 equiv of CuCl, 60 equiv of CuCl2, H2O, 0-25 °C, 1 h,
55%), gave 33, which embodies the full carbon skeleton of 5.
Completion of the Synthesis. With the full carbon skeleton
in hand, we directed our attention toward completion of the
synthesis (Scheme 4). TBS ether protection of the secondary
alcohols (65 equiv of CF3CONMeTBS, CH3CN, 55 °C, 22 h;
aqueous citric acid, 25 °C, 13 h, 96%), followed by MEM ether
deprotection of 34 (12 equiv of B-bromocatecholborane (BCB),
CH2Cl2, 0 °C, 2 h; 5.1 equiv of Boc2O, 6.0 equiv of NaHCO3,
dioxane-H2O 2:1, 0-25 °C, 2.5 h, 80%) and two-step oxidation
of the resulting primary alcohol 35 (4.0 equiv of Dess-Martin
periodinane, CH2Cl2, 0 °C, 15 min then 25 °C, 1 h; 9.0 equiv
of 80% aqueous NaClO2, 7.0 equiv of NaH2PO4‚H2O, t-BuOH/
2-methyl-2-butene 4:1, 25 °C, 20 min, 82%), provided the
(23) Node, M.; Nishide, K.; Fuji, K.; Fujita, E. J. Org. Chem. 1980, 45, 4275.
Evans, D. A.; Ellman, J. A. J. Am. Chem. Soc. 1989, 111, 1063.
(24) Nieto, M.; Perkins, H. R. Biochem. J. 1971, 124, 845. Nieto, M.; Perkins,
H. R. Biochem. J. 1971, 124, 773. Nieto, M.; Perkins, H. R. Biochem. J.
1971, 124, 789.
(25) McAtee, J. J.; Castle, S. L.; Jin, Q.; Boger, D. L. Bioorg. Med. Chem.
Lett. 2002, 12, 1319.
9
2890 J. AM. CHEM. SOC. VOL. 128, NO. 9, 2006

[
Ψ
[CH2NH]Tpg4]Vancomycin Aglycon
A R T I C L E S
Table 1. Binding and Antimicrobial Properties
the anticipated results relative to the vancomycin aglycon, where
the enhancement for binding D-Ala-D-Lac is 43-fold (5.2 × 103
vs 1.2 × 102 M-1) and the reduction in binding affinity for
D-Ala-D-Ala is 37-fold (4.8 × 103 vs 1.7 × 105 M-1). In
addition, the comparison of 5 with 41 seems to reflect the
intuitive expectations of the impact of the polar amine (proto-
nated) versus its carbamate derivative, where the binding affinity
for D-Ala-D-Ala with 5 versus 41 increases 3-fold (4.8 × 103
vs 1.6 × 103 M-1), while the impact on D-Ala-D-Lac is a more
marginal 1.2-fold increase in affinity (5.2 × 103 vs 4.1 × 103
M-1). Although there are additional structural features in the
comparison of 5 and 41 that might impact the absolute affinities
measured, in both instances the binding increases with the free
amine 5, and it is with 5 that the dual binding is balanced.
1
Ka (M-
)
VanA,c MIC
g/mL)d
compound
2a
4b
Ka2/Ka4
(µ
1, vancomycin
2.0 × 105 1.8 × 102 1100
>500
(2000)e
>500
(640)e
31
38, vancomycin aglycon 1.7 × 105 1.2 × 102 1400
5
41
4.8 × 103 5.2 × 103 0.92
1.6 × 103 4.1 × 103 0.40
31
a Ac2-L-Lys-D-Ala-D-Ala. b Ac2-L-Lys-D-Ala-D-Lac. c Enterococcus faeca-
lis (VanA, BM4166). d Vancomycin and vancomycin aglycon exhibit MICs
of 1-2.5 µg/mL against wild-type E. faecalis. e Taken from ref 25.
satisfying a ligand H-bond donor (10-fold?, for a cumulative
100-fold reduction?). As such, the derivatives might be expected
to be marginally active against sensitive and inducibly resistant
bacteria growing with a D-Ala-D-Ala peptidoglycan cell wall
precursor (10- to 100-fold loss in activity) and more active
against constituitively resistant bacteria (10-fold loss in activity)
bearing the D-Ala-D-Lac precursor. Consequently, and while we
intend to examine such vancomycin analogues, they were not
the first to be targeted for synthesis.
The targeted analogue 5, incorporating an amine in the linkage
of residue 4 with residue 5, not only removes the offending
carbonyl and the destabilizing lone-pair interaction with D-Ala-
D-Lac but also maintains a local polar environment (protonated
amine) that may better accommodate the binding of an elec-
tronegative group or atom (NH of D-Ala-D-Ala amide or O of
D-Ala-D-Lac ester). Intuitively, we anticipated that, while this
might not bind D-Ala-D-Lac quite as well as derivatives such
as 40, it would be better than 40 at binding D-Ala-D-Ala. In the
best case, 5 might bind D-Ala-D-Ala and D-Ala-D-Lac with equal
affinity, making it effective for the treatment of sensitive or
resistant bacteria, regardless of the structure of the peptidoglycan
cell wall precursor. Notably, this dual binding results from a
reduction in the affinity for D-Ala-D-Ala and an enhancement
in the affinity for D-Ala-D-Lac, such that the affinity for either
splits that observed with vancomycin. Accordingly, while the
spectrum of antimicrobial activity increases relative to vanco-
mycin to include sensitive and resistant bacteria, the potency
of the derivative would be anticipated to be reduced 30- to 40-
fold.
The four compounds were compared in an antimicrobial assay
against VanA Enterococcus faecalis (BM4166) that is inducibly
resistant to treatment by glycopeptide antibiotics including
vancomycin and teicoplanin (Table 1). It is the most difficult
of the resistant organisms to treat (vs VanB), and, characteristic
of such organisms, they grow unchallenged enlisting a D-Ala-
D-Ala peptidoglycan cell wall precursor, but switch to D-Ala-
D-Lac upon glycopeptide treatment. As such, it represents a
superb test of whether 5 and related dual D-Ala-D-Ala/D-Lac
binding antibiotics might prove useful in the treatment of
resistant bacteria. Beautifully, 5 as well as 41 exhibited MICs
of 31 µg/mL, being roughly 40-fold more potent than vanco-
mycin or its aglycon (MICs ) 2000 and 640 µg/mL), correlating
well with the ca. 40-fold increase in binding affinity for D-Ala-
D-Lac. Moreover, this potency is roughly 30-fold weaker than
that observed with vancomycin and its aglycon against sensitive
E. faecalis (MICs ) 1-2.5 µg/mL), correlating with the 35- to
40-fold loss in binding affinity for D-Ala-D-Ala. These results
suggest that, regardless of the peptidoglycan cell wall precursor
utilized by the organism, it remains equally sensitive to treatment
by 5 and 41.
Conclusions. The modification of the dipeptide terminus of
peptidoglycan cell wall precursors from D-Ala-D-Ala to D-Ala-
D-Lac in resistant bacteria reduces the binding affinity of
vancomycin for the ligand by 1000-fold, leading to a 1000-
fold loss in biological activity. We had shown that a modified
peptide ligand possessing a methylene in place of the lactate
oxygen restores 100-fold of this binding affinity by removal of
a destabilizing lone-pair interaction. This suggested that removal
of the residue 4 carbonyl in the vancomycin aglycon would
produce an analogue with enhanced affinity for D-Ala-D-Lac
and might restore much of the biological activity of the molecule
that is lost with resistant bacteria. Moreover, and among the
range of potential modifications that could be envisioned, that
entailing the simple removal of the residue 4 carbonyl, providing
5, was anticipated to bind D-Ala-D-Ala and D-Ala-D-Lac with
similar affinities, providing an analogue that might be equiva-
lently effective against sensitive (D-Ala-D-Ala) and resistant (D-
Ala-D-Lac) bacteria. Consequently, we extended our efforts on
the preparation of glycopeptide antibiotics to a total synthesis
of the [Ψ[CH2NH]Tpg4]vancomycin aglycon (5), in which the
residue 4 carbonyl has been replaced with a methylene.
Consistent with expectations and relative to the vancomycin
aglycon, 5 exhibited a 40-fold increase in affinity for D-Ala-
D-Lac (Ka ) 5.2 × 103 M-1) and a corresponding 35-fold
reduction in affinity for D-Ala-D-Ala (Ka ) 4.8 × 103 M-1),
The results of our assessment of 5 alongside vancomycin (1)
and its aglycon 38 are compiled in Table 1.24,25 An additional
analogue 41, derived from N-Boc deprotection of the synthetic
intermediate 33 (eq 1), was also examined that bears the
methoxycarbonyl protecting group on the residue 4/5 linking
amine. The binding affinities of 5 for Ac2-L-Lys-D-Ala-D-Ala
(2) and Ac2-L-Lys-D-Ala-D-Lac (4) were essentially equivalent
(4.8 × 103 vs 5.2 × 103 M-1, respectively), with the D-Ala-D-
Lac binding being slightly better. Impressively, this represented
9
J. AM. CHEM. SOC. VOL. 128, NO. 9, 2006 2891

A R T I C L E S
Crowley and Boger
providing a molecule with balanced, dual binding characteristics.
Beautifully, 5 exhibited antimicrobial activity against a VanA-
resistant organism that remodels its D-Ala-D-Ala peptidoglycan
cell wall precursor to D-Ala-D-Lac upon glycopeptide challenge,
displaying a potency that reflects these binding characteristics.
the modified CD and ABCD ring systems and Professor Julius
Rebek for his discussions, interest, and encouragement through-
out the work. B.M.C. is a Skaggs Fellow and recipient of Bristol-
Myers Squibb (2004-2005) and Fletcher Jones Foundation
fellowships (2003-2004).
Acknowledgment. We gratefully acknowledge the financial
support of the National Institutes of Health (CA 41101) and
the Skaggs Institute for Chemical Biology. We thank Drs. M.
Lall, D. Horne, J. J. McAtee, O. Rogel, and C. C. McComas
for initiating and contributing to studies on the preparation of
Supporting Information Available: Full experimental details.
This material is available free of charge via the Internet at
http://pubs.acs.org.
JA0572912
9
2892 J. AM. CHEM. SOC. VOL. 128, NO. 9, 2006