Enantiomerically pure trans -β-lactams from α-amino acids via compact fluorescent light (CFL) continuous-flow photolysis

Article abstract of DOI:10.1021/ja1050023

Photolysis of α-diazo-N-methoxy-N-methyl (Weinreb) β-ketoamides derived from enantiomerically pure (EP) α-amino acids affords the corresponding EP β-lactams via an intramolecular Wolff rearrangement. The photochemistry is promoted with either standard UV irradiation or through the use of a 100 W compact fluorescent light; the latter affords a safe and environmentally friendly alternative to standard photolysis conditions. A continuous-flow photochemical reactor made from inexpensive laboratory equipment reduced reaction times and was amenable to scale-up. The diastereoselectivity (cis or trans) of the product β-lactams has been shown to vary from modest to nearly complete. An extremely facile, atom-economical method for the epimerization of the product mixture to the trans isomer, which is generally highly crystalline, has been developed. Evidence for C3 epimerization of Weinreb amide structures via a nonbasic, purely thermal route is presented. Subsequent transformations of both the Weinreb amide at C3 (β-lactam numbering) and the amino acid side chain at C4 are well-tolerated, allowing for a versatile approach to diverse β-lactam structures. The technology is showcased in the synthesis of a common intermediate used toward several carbapenem-derived structures starting from unfunctionalized aspartic acid.

Full text of DOI:10.1021/ja1050023

Published on Web 07/22/2010
Enantiomerically Pure trans-ꢀ-Lactams from r-Amino Acids
via Compact Fluorescent Light (CFL) Continuous-Flow
Photolysis
Yvette S. Mimieux Vaske, Maximillian E. Mahoney, Joseph P. Konopelski,*
David L. Rogow, and William J. McDonald
Department of Chemistry and Biochemistry, UniVersity of California, Santa Cruz, California 95064
Received June 18, 2010; E-mail: joek@chemistry.ucsc.edu
Abstract: Photolysis of R-diazo-N-methoxy-N-methyl (Weinreb) ꢀ-ketoamides derived from enantiomerically
pure (EP) R-amino acids affords the corresponding EP ꢀ-lactams via an intramolecular Wolff rearrangement.
The photochemistry is promoted with either standard UV irradiation or through the use of a 100 W compact
fluorescent light; the latter affords a safe and environmentally friendly alternative to standard photolysis
conditions. A continuous-flow photochemical reactor made from inexpensive laboratory equipment reduced
reaction times and was amenable to scale-up. The diastereoselectivity (cis or trans) of the product ꢀ-lactams
has been shown to vary from modest to nearly complete. An extremely facile, atom-economical method
for the epimerization of the product mixture to the trans isomer, which is generally highly crystalline, has
been developed. Evidence for C3 epimerization of Weinreb amide structures via a nonbasic, purely thermal
route is presented. Subsequent transformations of both the Weinreb amide at C3 (ꢀ-lactam numbering)
and the amino acid side chain at C4 are well-tolerated, allowing for a versatile approach to diverse ꢀ-lactam
structures. The technology is showcased in the synthesis of a common intermediate used toward several
carbapenem-derived structures starting from unfunctionalized aspartic acid.
Introduction
Azetidin-2-ones (ꢀ-lactams) are among the most investigated
of all heterocyclic ring systems because of their well-
documented impact on small-molecule drug discovery. When
suitably functionalized, the ꢀ-lactam ring in enantiomerically
pure (EP) form represents the core of a vast array of antibiotics,1
including some used to treat drug-resistant bacterial strains.2
Included in this array are carbapenem structures inspired by the
natural product thienamycin and monocyclic ꢀ-lactam com-
pounds exemplified by aztreonam, each possessing trans ster-
eochemistry at the two chiral centers of the azetidinone ring
(Figure 1). Additional uses for EP ꢀ-lactams in medicinal
chemistry include inhibition of such diverse targets as intestinal
cholesterol absorption3 and viral proteases.4 However, the
ꢀ-lactam ring system is a “privileged structure” in more than
the usual medicinal chemistry sense.5 A ready route to EP
ꢀ-lactams provides access to ꢀ-amino acids6 by the well-
Figure 1. Representative ꢀ-lactam structures.
established synthetic dynamic between the two functionalities,7
and interest in polymers of EP ꢀ-amino acids (“ꢀ-foldamers”8,9
)
remains high. In addition, studies of the use of EP ꢀ-lactams as
multifunctional organic motifs for the elaboration of more
complex targets10 and as rigid ꢀ-turn scaffolds11 ensure that
research into this strained four-membered ring system will
continue unabated for years to come.
In view of the triple utilization of ꢀ-lactams in biomedical,
synthetic, and structural chemistry, it is not surprising to find
new methods for the preparation of both the racemic and EP
(1) Von Nussbaum, F.; Brands, M.; Hinzen, B.; Weigand, S.; Ha¨bich, D.
Angew. Chem., Int. Ed. 2006, 45, 5072–5129.
(2) Hugonnet, J.-E.; Tremblay, L. W.; Boshoff, H. I.; Barry, C. E., 3rd.;
Blanchard, J. S. Science 2009, 323, 1215–1218.
(7) (a) Palomo, C.; Aizpurua, J. M.; Ganboa, I. In EnantioselectiVe
Synthesis of ꢀ-Amino Acids; Juraisti, E., Ed.; Wiley-VCH: New York,
1997; Chapter 14. (b) Palomo, C.; Aizpurua, J. M.; Ganboa, I.;
Oiarbide, M. In EnantioselectiVe Synthesis of ꢀ-Amino Acids, 2nd ed.;
Juraisti, E., Soloshonok, V., Eds.; Wiley: Hoboken, NJ, 2005; Chapter
20.
(3) Kværnø, L.; Werder, M.; Hauser, H.; Carreira, E. M. J. Med. Chem.
2005, 48, 6035–6053.
(4) For a review, see: Xing, B.; Rao, J.; Liu, R. Mini-ReV. Med. Chem.
2008, 8, 455–471.
(5) Patchett, A. A.; Nargund, R. P. Annu. Rep. Med. Chem. 2000, 35,
289–298, and references therein.
(8) Horne, W. S.; Gellman, S. H. Acc. Chem. Res. 2008, 41, 1399–1408.
(9) Seebach, D.; Gardiner, J. Acc. Chem. Res. 2008, 41, 1366–1375.
(10) Alcaide, B.; Almendros, P.; Aragoncillo, C. Chem. ReV. 2007, 107,
4437–4492.
(6) (a) EnantioselectiVe Synthesis of ꢀ-Amino Acids; Juraisti, E., Ed.;
Wiley-VCH: New York, 1997. (b) EnantioselectiVe Synthesis of
ꢀ-Amino Acids, 2nd ed.; Juraisti, E., Soloshonok, V., Eds.; Wiley:
Hoboken, NJ, 2005.
(11) Palomo, C.; Aizpurua, J. M.; Balentova´, E.; Jimenez, A.; Oyarbide,
J.; Fratila, R. M.; Miranda, J. I. Org. Lett. 2007, 9, 101–104.
9
10.1021/ja1050023  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 11379–11385 11379

A R T I C L E S
Vaske et al.
Scheme 1. Wolff Rearrangement Pathway to ꢀ-Lactams
Scheme 2
Scheme 3. Synthesis and Reaction of R-Diazo-ꢀ-ketoester 7
ring systems continuously appearing in the literature.12-14
Indeed, this laboratory recently reported the serendipitous
transformation of R-diazo-ꢀ-ketoamide 2 derived from L-serine
into ꢀ-lactam 3 (Scheme 1).15 The results are consistent with a
mechanism involving a stereospecific intramolecular Wolff
rearrangement with full retention of absolute configuration.16
Intramolecular attack by the trityl-protected amine on the
intermediate ketene A, formally a disfavored 4-exo-dig process
under Baldwin’s rules,17-19 affords the product ꢀ-lactam. Other
key observations from this work were (1) the stability and
crystallinity of imidazolides 1, which translate into handling
ease in the sequence; (2) the observation of 3 as the major
isolable product from rhodium(II) catalysis, silver(I) catalysis,
and light (somewhat rare,20 fully expected,21 and best yields,
respectively); and (3) the stereospecificity of the Wolff rear-
rangement, as gauged by the clean formation of ent-3 when
starting with D-serine-derived ent-1.
for easy isolation of the major product by crystallization from
the reaction mixture.
Results and Discussion
Synthesis and Related Studies. Our previous work15 led to
the isolation of N-phenyl-N-benzyl amide 3, which, while
attractive for our initial goals, was deemed cumbersome for
manipulation into desired C3 ꢀ-lactam functionalities. For
example, treatment of pure 3 with Gassman’s “anhydrous
hydroxide22” for 24 h resulted in near-quantitative recovery of
the starting structure, now as an approximately 4:1 trans/cis
(3/4) mixture (Scheme 2). Thus, the trityl group, which has been
explored only briefly as a ꢀ-lactam ring nitrogen protection
group,23 offers extreme base stability to the normally reactive
ꢀ-lactam nucleus. Conversely, treatment with trifluoroacetic acid
(TFA) resulted in removal of the trityl group to give 5 in good
yield.
The desire for a more malleable functionality at C3 led
initially to the ester group as a replacement for the fully
substituted C3 amide of 3. The requisite R-diazo-ꢀ-ketoester 7
was prepared under mild acylation conditions (Scheme 3).24 In
an adaptation of a modified procedure,25 potassium methyl
malonate was treated with NEt3 and MgCl2 to form the
corresponding magnesium salt, which was subsequently added
to a solution of imidazolide ent-1 derived from D-serine. The
resulting ꢀ-ketoester 6 was obtained in 82% yield. Subsequent
treatment with 4-acetamidobenzenesulfonyl azide and 1,8-
Herein are presented further studies of this reaction that
document the Versatility of the route through the facile produc-
tion of various C3 functionalities and incorporation of different
amino acid side chains; the economy of the route through the
direct incorporation of the native R-amino acid into the reaction
scheme; the safety of the chemistry, which requires neither metal
nor organic catalysis to provide the strained ring system but
rather can be performed with a compact fluorescent light (CFL);
and the simplicity of the route, which produces high yields of
EP trans-ꢀ-lactams via a facile epimerization reaction, allowing
(12) Singh, G. S. Tetrahedron 2003, 59, 7631–7649.
(13) France, S.; Weatherwax, A.; Taggi, A. E.; Lectka, T. Acc. Chem. Res.
2004, 37, 592–600.
(14) Brandi, A.; Cicchi, S.; Cordero, F. M. Chem. ReV. 2008, 108, 3988–
4035.
(15) Gerstenberger, B. S.; Lin, J.; Mimieux, Y. S.; Brown, L. E.; Oliver,
A. G.; Konopelski, J. P. Org. Lett. 2008, 10, 369–372.
(16) Kirmse, W. Eur. J. Org. Chem. 2002, 2193–2256.
(17) Baldwin, J. E. J. Chem. Soc., Chem. Commun. 1976, 734–736.
(18) For another example of the 4-exo-dig route to ꢀ-lactams, see: Hachiya,
I.; Yoshitomi, T.; Yamaguchi, Y.; Shimizu, M. Org. Lett. 2009, 11,
3266–3268.
(22) Gassman, P. G.; Hodgson, P. K. G.; Balchunis, R. J. J. Am. Chem.
Soc. 1976, 98, 1275–1276.
(19) The possibility of this type of ketene addition reaction in the
biosynthesis of carbapenems has recently been discussed. See: Raber,
M. L.; Castillo, A.; Greer, A.; Townsend, C. A. ChemBioChem 2009,
10, 2904–2912.
(23) Hakimelahi, G. H.; Shiao, M.-J.; Hwu, J. R.; Davari, H. HelV. Chim.
Acta 1992, 75, 1840–1847.
(24) Brooks, D. W.; Lu, L. D.-L.; Masamune, S. Angew. Chem., Int. Ed.
Engl. 1979, 18, 72–74.
(20) Davies, J. R.; Kane, P. D.; Moody, C. J.; Slawin, A. M. Z. J. Org.
Chem. 2005, 70, 5840–5851, and references therein.
(21) Podlech, J.; Seebach, D. Liebigs Ann. Chem. 1995, 1217–1228.
(25) Mai, A.; Sbardella, G.; Artico, M.; Ragno, R.; Massa, S.; Novellino,
E.; Greco, G.; Lavecchia, A.; Musiu, C.; La Colla, M.; Murgioni, C.;
La Colla, P.; Loddo, R. J. Med. Chem. 2001, 44, 2544–2554.
9
11380 J. AM. CHEM. SOC. VOL. 132, NO. 32, 2010

Enantiomerically Pure trans-ꢀ-Lactams from R-Amino Acids
A R T I C L E S
Scheme 4. Synthesis and Reaction of R-Diazo-ꢀ-ketoamide 11
Figure 2. Solid-phase structure of diazoketone 14.
pounds are, in general, thermally unstable. Accordingly, on a
1 g scale, the reaction was complete in 3.5 h (vs 0.6 g, 7 h for
the batch process) and afforded an 81% isolated yield of a
roughly 3:1 separable mixture of 12 and 13, respectively. The
improved reaction time and ease of processing material on a
larger scale made the modified continuous-flow photochemical
process the preferred method, despite slightly diminished
yields.31
Nonetheless, the use of the MVL was not without concern
regarding heat and UV radiation for the experimentalist, and
we were drawn to consider alternative light sources by the
photochemist’s adage that the wavelength choice should be “as
long as possible and as short as necessary”. Therefore, a 100
W CFL (Figure S2 in the Supporting Information), which
promised far greater safety and distinctly lower cost than the
conventional light source, was investigated. To our knowledge,
there are no previous reports of CFL use for the photochemical
Wolff rearrangement, but we were delighted to find that while
the rate of the reaction was diminished relative to that for the
MVL, the product yield was increased and the distributions were
comparable in both the batch and continuous-flow modes
(Scheme 4). In addition, the CFL-promoted reaction required
no external cooling.
What is the origin of the facile Wolff rearrangement of
R-diazo-ꢀ-ketoamide 11? Optimized yields arise from photo-
chemical decomposition as opposed to the use of metal catalysts
and point to a very facile rearrangement reaction. Recent studies
of the Wolff rearrangement32 have suggested that ketene can
form from two different reaction pathways: an extremely rapid
rearrangement of the diazo excited state concomitant with
nitrogen loss and a slower rearrangement from the ketocarbene
after nitrogen loss. The rearrangement is facilitated by the
antiperiplanar geometry of the leaving and migrating groups
(N2 and the amino acid chiral center, respectively), and it is
this geometry that was found in the single-crystal X-ray study
of diazoketone 14 (Figure 2 and the Supporting Information),
which is derived from aspartic acid (see below). The structure
shows the molecule in the s-ZK,s-EW conformation (K for ketone,
W for Weinreb amide), with the dihedral angle defined by
N2-C-CCdO-CR (all marked with *) having a value of 172.6°.
Thus, the solid-state structure possesses the exact conformation
necessary for efficient Wolff rearrangement.
diazabicyclo[5.4.0]undec-7-ene (DBU) gave compound 7 in an
overall 80% yield for the two steps. Under standard photo-
chemical conditions, compound 7 afforded a 63% yield of a
separable mixture of the desired trans-ꢀ-lactam 8 and alkene
9. The structure of ꢀ-lactam 8 was confirmed by X-ray
crystallography; alkene 9 was assigned by 2D NMR analysis.
While the mechanistic pathway for the alkene product is not
entirely clear, the new amide functionality clearly suggests that
9 originates from Wolff rearrangement/ꢀ-lactam cascade fol-
lowed by further reaction.26
The modest yield of ꢀ-lactam 8 prompted the search for a
C3 amide functionality that would maintain or exceed the higher
product yields encountered in our initial studies while offering
facile functional group manipulation. After some consideration,
the N-methoxy-N-methyl (Weinreb) amide was chosen for study,
as Weinreb amides27 hold a reserved place as valuable and
versatile synthetic intermediates.28
Weinreb ꢀ-ketoamide 10 was prepared by a modified
literature procedure (Scheme 4).29 Accordingly, serine imida-
zolide 1 was acylated with the lithium enolate (LHMDS) of
N-methoxy-N-methylacetamide to give the desired ꢀ-ketoamide
10 in 86% yield on a 9 g scale. Treatment with diazo transfer
reagent gave the desired R-diazo-ꢀ-ketoamide 11 in 89% yield.
Under standard photolysis conditions, diazo 11 was irradiated
with a medium-pressure mercury vapor lamp (MVL), which
afforded an easily separable, roughly 2.5:1 mixture of ꢀ-lactams
12 and 13, respectively, in 90% isolated yield. The identities
of the two ꢀ-lactam isomers were confirmed by X-ray crystal-
lography. To improve upon our processing capabilities, a simple
continuous-flow photochemical reactor was constructed from
common laboratory equipment30 and consisted of the requisite
tubing and a medium-pressure liquid chromatography pump in
addition to standard flasks (see Figure S1 in the Supporting
Information). External cooling was employed, as diazo com-
To more fully understand the preferred solution conformation
of the diazo substrates, hybrid density functional theory calcula-
tions on compound 14 were employed. The calculations were
performed with no explicit solvent incorporation, since the
dielectric constant of toluene is very low (D ) 2.38). The crystal
structure coordinates of compound 14 were used as the starting
point, and four conformations were generated by setting the
dihedral angles related to the ꢀ-dicarbonyl system to either 180
(26) For alkene formation accompanying Wolff rearrangement, see: Stork,
G.; Szajewski, R. P. J. Am. Chem. Soc. 1974, 96, 5787–5791.
(27) Nahm, S.; Weinreb, S. M. Tetrahedron Lett. 1981, 22, 3815–3818.
(28) For a recent review, see: Balasubramaniam, S.; Aidhen, I. S. Synthesis
2008, 3707–3738.
(29) Kulesza, A.; Ebetino, F. H.; Mazur, A. W. Tetrahedron Lett. 2003,
44, 5511–5514.
(30) (a) For a recent review, see: Wiles, C.; Watts, P. Eur. J. Org. Chem.
2008, 1655–1671. (b) Lainchbury, M. D.; Medley, M. I.; Taylor, P. M.;
Hirst, P.; Dohle, W.; Booker-Milburn, K. I. J. Org. Chem. 2008, 73,
6497–6505. (c) Hook, B. D. A.; Dohle, W.; Hirst, P. R.; Pickworth,
M.; Berry, M. B.; Booker-Milburn, K. I. J. Org. Chem. 2005, 70,
7558–7564.
(31) See the Supporting Information for experimental details and Figure
S1 for a picture of the continuous-flow photochemical reactor.
(32) Burdzinski, G. T.; Wang, J.; Gustafson, T. L.; Platz, M. S. J. Am.
Chem. Soc. 2008, 130, 3746–3747.
9
J. AM. CHEM. SOC. VOL. 132, NO. 32, 2010 11381

A R T I C L E S
Vaske et al.
Table 1. Ketones from Weinreb Amide 12
Scheme 5. Hydrolysis and Curtius Rearrangement
entry
RLi or RBr; R′Li
ketone
yield (%)
1
2
3
4
MeLi
t-BuLi
4-bromoanisole; n-BuLi
2-bromopropene; t-BuLi
15
16
17
18
81
63
78
62 (79 brsm)
or 0° in recognition of the partial double bond character of these
bonds.16 These four structures were designated s-ZK,s-EW (the
conformation of 14), s-ZK,s-ZW, s-EK,s-EW, and s-EK,s-ZW, where
K refers to the conformation around the ketone-diazo carbon
bond (shown in blue in Figure 2) and W refers to the Weinreb
amide-diazo carbon bond (shown in red in Figure 2). Each of
these four conformers was minimized at the RI-PBE/TZVP(P)
level of theory.33 Four minimized structures, each residing in
the structural space of its starting conformation, were obtained
and are shown, with relative energies, in Figure S3 in the
Supporting Information. The calculated lowest-energy conformer
agrees closely with the X-ray conformer, with no other
minimized structure within 15 kJ of the s-ZK,s-EW conformation.
In addition, each structure has one of the trityl group aromatic
rings in close facial contact with the R-diazo-ꢀ-dicarbonyl
functionality. The origin of this close proximity is not understood
at the present time and is under investigation, but we are
unaware of other examples of this type of close interaction
between the π system of an aromatic ring and a diazo
functionality.
The Weinreb amide functionality at the C3 position afforded
the opportunity to explore various transformations to high-
light the versatility of this functional group within the present
structure. In particular, the carbapenem and monobactam classes
of antibiotics, exemplified by the structures in Figure 1, bear
the hydroxyethyl and amino functionalities, respectively, at this
position of the ring. The former group, a key element in
carbapenems having high biological activity, is usually devel-
oped via stereocontrolled reduction of the corresponding methyl
ketone. Thus, trans isomer 12 was treated with a series of
organolithium reagents, and the corresponding C3 ketones were
obtained in good to excellent yields (Table 1). Entries 1 and 2
proceeded directly from the commercially available organolith-
ium reagent. For entries 3 and 4, the organolithium agents were
prepared by lithium-halogen exchange using 4-bromoanisole34
and 2-bromopropene,35 respectively. Not unexpectedly, the
ꢀ-lactam carbonyl was fully protected by the large trityl group.
Hydrolysis of trans-ꢀ-lactam 12 was examined next (Scheme
5). Under microwave-assisted hydrolysis conditions,36 12 cleanly
afforded C3 carboxylic acid 19 in 83% isolated yield. For
comparison, standard ester hydrolysis of the trans isomer 8
afforded ent-19 in 90% yield after purification. The spectral data
for the two enantiomeric carboxylic acids were identical, and
their optical rotations were equal in magnitude and opposite in
direction. Continuing on the theme of versatility, treatment of
C3 carboxylic acid 19 with diphenylphosphoryl azide37 in the
presence of benzyl alcohol afforded benzyl carbamate 20, the
product of the expected Curtius rearrangement (Scheme 5).38
The remaining protection groups, those on the C4 amino acid
side chain and the ring nitrogen, were also easily manipulated
in an orthogonal manner with respect to the Weinreb amide.
Deprotection of the serine side chain was accomplished by
standard hydrogenolysis conditions, affording alcohol 21 from
12 in quantitative yield. Deprotection of the ring nitrogen was
initially investigated by employing the only literature procedure
available for a ꢀ-lactam N-Tr functionality.23 However, this
methodology (neat TFA, cat. KClO4) proved to be capricious
in our hands and was abandoned in favor of the reagents
employed for the deprotection of 3 (Scheme 1). Under these
conditions [TFA/H2O/(CH2SH)2], a reproducible 85% yield of
22 was obtained (Scheme 5).
Structural Studies. After X-ray diffraction data for several
of the N-trityl ꢀ-lactams prepared in this study were obtained,
the details of the solid-state structures of these compounds were
examined. A ꢀ-lactam differs from a normal amide because of
the intrinsically strained four-membered ring, which confers
decreased amide resonance.39 For this reason, a monocyclic
ꢀ-lactam structure exhibits a longer C2-N(amide) bond length
(typically 1.35-1.38 Å vs 1.33 Å for a normal amide) and
shorter CdO bond length (1.21-1.23 Å vs 1.24 Å for a standard
amide).
The data from our structural studies of the N-trityl
ꢀ-lactams are summarized in Figure 3 together with data for
ꢀ-lactams from the Cambridge Structural Database (CSD,
identified by CSD refcode), including both monocyclic
ꢀ-lactams (CEJROZ, ALASAH, CPIPLA) and carbapenems
(CEKQUE, BALSOW).40 Selected interatomic distances of
interest (Å) are highlighted in red; pyramidalization at the
carbonyl carbon (∆, Å) is given in blue. Compounds 12a
and 12b are two different molecules found in the asymmetric
unit cell of the X-ray diffraction data.
(33) Neese, F. ORCA: An ab Initio Density Functional and Semiempirical
Program Package, version 2.6.35; University of Bonn: Bonn,
Germany, 2008.
(34) Jones, R. G.; Gilman, H. Chem. ReV. 1954, 54, 835–890.
(35) (a) Fuller, N. O.; Morken, J. P. Org. Lett. 2005, 7, 4867–4869. (b)
Ebel, H.; Zeitler, K.; Steglich, W. Synthesis 2003, 101–106.
(36) Jaipuri, F. A.; Jofre, M. F.; Schwarz, K. A.; Pohl, N. L. Tetrahedron
Lett. 2004, 45, 4149–4152.
(37) Shioiri, T.; Ninomiya, K.; Yamada, S.-i. J. Am. Chem. Soc. 1972, 94,
6203–6205.
(38) Go´mez-Sa´nchez, E.; Marco-Contelles, J. Tetrahedron 2005, 61, 1207–
1219.
(39) Page, M. I. The Chemistry of ꢀ-Lactams; Blackie Academic &
Professional: Glasgow, 1992; Chapter 2.
(40) Allen, F. H. Acta Crystallogr. 2002, B58, 380–388.
9
11382 J. AM. CHEM. SOC. VOL. 132, NO. 32, 2010

Enantiomerically Pure trans-ꢀ-Lactams from R-Amino Acids
A R T I C L E S
Figure 3. Bond lengths and interatomic distances (red) and pyramidalization (∆, blue) of N-trityl ꢀ-lactams from this study (left) and selected ꢀ-lactams
from the CSD (right). All distances are given in Å, and the definition of ∆ is shown in the inset at the right.
Scheme 6
Inspection of the structural data revealed that the N-trityl
ꢀ-lactam CdO bonds were within the expected range for a
monocyclic ꢀ-lactam, as were the C2-N(amide) and C2-C3
bond lengths (1.52-1.55 Å). The N-C4 bond lengths were
found to be consistent within the structures defined by this study
(1.49-1.51 Å). However, on average, these values were longer
than the corresponding N-C4 values of the CSD ꢀ-lactams
(1.45-1.49 Å), likely as a result of the steric interaction of the
trityl group with the C4 substituent. Finally, a consistent close
contact between the C3 proton and the oxygen atom of
the Weinreb -OMe group was noted in the structures of 12
and 13.
The pyramidalization (∆) at the carbonyl carbon of the
ꢀ-lactam is the perpendicular distance of this atom from the
plane defined by the three substituents (Figure 3 inset).39,41
A
striking feature of these data is that the majority of the N-trityl
ꢀ-lactams had substantial pyramidalization at the carbonyl
carbon (∆ ) 0.018-0.024 Å), in sharp contrast to most
monocyclic ꢀ-lactams, which are essentially planar at this center.
This trend is highlighted by the CSD structures in Figure 3,
which display large pyramidalizations for the two carbapenem
structures (CEKQUE and BALSOW) and smaller nonplanarity
measurements for the monocyclic ꢀ-lactams. For further com-
parison, a larger subset of CSD ꢀ-lactams (listed by CSD refcode
in Table 1 of Appendix A in the Supporting Information) were
examined, and a pyramidalization database was compiled. A
histogram of these data revealed that a large percentage of the
CSD ꢀ-lactams had pyramidalizations of ∼0.01 Å. The smaller
set of ꢀ-lactam structures that accounted for the larger pyra-
midalizations (∆ g 0.02 Å) is given in Tables 2 and 3 of
Appendix A in the Supporting Information.
C3 Stereochemistry and Epimerization. The overall yield of
ꢀ-lactam product remained consistently high (∼90%) throughout
the many instances of photolysis of R-diazo-ꢀ-ketoamide 11.
However, the trans/cis (12/13) diastereomeric ratio (dr) varied
on a case-by-case basis, always favoring a higher yield of the
more thermodynamically stable and highly crystalline trans
isomer. Thus, 12 could be directly isolated from the concentrated
crude reaction mixture (toluene) and recrystallized to purity
(ethyl acetate/hexanes), thus avoiding the need for column
chromatography. However, the isolated yield of the minor
product, cis-ꢀ-lactam 13, varied significantly. Furthermore,
despite rigorous column chromatography using various eluting
systems, cis isomer 13 was often accompanied by cocontami-
nation with trans isomer 12, making isolation of 13 as a single
pure compound a challenge. Moreover, this crystalline sample
of 13, from which the single crystal for the X-ray structure
analysis was taken, epimerized on standing at room temperature
in a desiccator from 100% cis to 62% cis over the course of 4
months. The X-ray crystal structure showed no signs of included
solvent, which could have facilitated the epimerization.
Mechanistically, the Wolff rearrangement route to ꢀ-lactams
provides the strained ring system and a new chiral center (C3
of the ꢀ-lactam ring) through proton transfer. As first shown in
Scheme 1 for the synthesis of 3 and elaborated in Scheme 6 for
a synthesis of 12, Wolff rearrangement affords a ketene such
as A (Scheme 1). Theoretical work42 has indicated that addition
of amines to ketenes occurs via a cyclic transition state (depicted
as TS-1) that forms the corresponding enol amide B as a high-
energy intermediate. Compound B undergoes intramolecular
tautomerization (possibly through TS-2) to afford the final amide
product with a new chiral center at C3. The intramolecular
tautomerization, which appears reasonable in the present case
since the photolysis reaction takes place in pure nonprotic
solvent (toluene), is nonetheless likely circumvented by impuri-
ties in the system that facilitate the 1,3-proton shift.
(41) Lwowski, W. In ComprehensiVe Heterocyclic Chemistry; Pergamon
Press: Oxford, U.K., 1964; Vol. 7, Part 5, pp 300-301.
(42) Cannizzaro, C. E.; Houk, K. N. J. Am. Chem. Soc. 2004, 126, 10992–
11008, and references therein.
9
J. AM. CHEM. SOC. VOL. 132, NO. 32, 2010 11383

A R T I C L E S
Vaske et al.
yielded an approximately 94:6 mixture of trans isomer 15 and
the compound assigned as the cis structure 23 on the basis of
LC-MS data.44 Clearly, more experiments must be performed
before suggesting a mechanism for the purely thermal epimer-
ization of 13 to 12. One possibility would be an intramolecular,
nonbasic epimerization reaction mediated by the Weinreb amide
functionality, a previously unknown reaction. However, regard-
less of the mechanism of the reaction, a facile route to high
yields of EP trans-ꢀ-lactams and compounds directly related
to these systems, such as ꢀ2,3-amino acids, is available.
Application. Thienamycin, shown in Figure 1, is the ancestor
of all carbapenem antibiotics45 and has a storied place in the
annals of synthetic organic chemistry through the efforts of the
Merck process group.46 Synthetic work on key intermediates
leading toward the carbapenem ring system has occupied many
research groups47 and offered an ideal platform for extending
this methodology toward a recognizable target. As envisioned,
a concise approach to a carbapenem synthetic intermediate
would abandon the serine derivatives that dominated our initial
development work in favor of aspartic acid. In addition, there
was interest in demonstrating a process directly from the amino
acid, which would allow either R or S stereochemistry to be
incorporated into the synthetic target with equal ease and the
lowest possible cost.
Figure 4. Thermal epimerization of 13 to 12. Data were obtained by HPLC
analysis during the course of the reaction.
Scheme 7. Epimerization Studies
Our target became ꢀ-lactam 30, the enantiomer of which has
been prepared previously48,49 and employed in the synthesis of
thienamycin and other carbapenems.48,50 Thus, L-aspartic acid
was subjected to the Rapoport protocol51 to provide H-
Asp(OMe)-OH in 72% yield without a purification step, identical
to the literature (Scheme 8). Installation of the trityl protection
group and imidazolide formation was inspired by the protocols
of Hoffman52 and performed without purification, providing
highly crystalline 24 in 93% yield from H-Asp(OMe)-OH. Thus,
the route from aspartic acid to 24 was accomplished without
chromatography in 67% overall yield.
Following the methods developed for the production of 11
(Scheme 4), the anion of N-methoxy-N-methylacetamide was
reacted with 24 to provide ꢀ-ketoamide 25 in 72% yield. Diazo
transfer to afford 14 was accomplished in 94% yield through
the use of MsN3. Photochemical decomposition of the diazo
functionality afforded the expected mixture of trans isomer 26
and cis isomer 27, which was not isolated. Rather, direct
treatment of the reaction mixture with DBU at room temperature
afforded exclusively the desired isomer 26 as a crystalline solid
in 81-90% yield (depending on the light source; the highest
yield was obtained using CFL continuous flow). Thus, highly
functionalized crystalline ꢀ-lactam 26 was produced in 43%
yield from L-Asp with only one chromatographic purification.
Introduction of the methyl ketone functionality to afford 27
was far from trivial, since the inherent reactivity of the methyl
Alternatively, this crucial proton transfer could be mediated
by a functionality within the molecule, with the serine side chain
-OBn group and the Weinreb amide oxygen atoms key
possibilities. The solid-state epimerization data seemed to
support this view, and the proximity of the Weinreb -OMe
oxygen to the key C3 proton in the X-ray structures of 12 and
13 reinforced this idea. Finally, and most importantly, it was
discovered that simply heating a 90:10 13/12 mixture in toluene
at 90 °C for several hours delivered a 90:10 12/13 mixture
(Figure 4). No base was added for the transformation. These
results are in sharp contrast to the harsh conditions needed to
effect epimerization of amide 3 (Scheme 2).
To investigate this transformation more thoroughly, cis amide
isomer 4 and trans ketone isomer 15 were subjected to toluene
at 90 °C (Scheme 7); no change in dr was recorded in either
reaction. However, heating pure trans isomer 12 at 90 °C for
20 h afforded a 97:3 mixture of 12 and 13. As was seen in
other equilibrium experiments on ꢀ-lactam systems, the time
to equilibrium was much longer when starting from the more
stable trans isomer than from the cis isomer.43 Likewise,
treatment of ketone 15 with DBU at room temperature for 18 h
(44) Only trans isomer 15 was obtained upon attempted isolation of the
new compound assigned as structure 23.
(45) Bassetti, M.; Nicolini, L.; Esposito, S.; Righi, E.; Viscoli, C. Curr.
Med. Chem. 2009, 16, 564–575.
(46) Nicolaou, K. C.; Sorensen, E. J. Classics in Total Synthesis; Wiley-
VCH: Weinheim, Germany, 1996; p 249.
(47) Berks, A. H. Tetrahedron 1996, 52, 331–375.
(48) Shih, D. H.; Baker, F.; Cama, L.; Christensen, B. G. Heterocycles
1984, 21, 29–40.
(49) Evans, D. A.; Sjogren, E. B. Tetrahedron Lett. 1986, 27, 4961–4964.
(50) Andrus, A.; Christensen, B. G.; Heck, J. V. Tetrahedron Lett. 1984,
25, 595–598.
(51) Gmeiner, P.; Feldman, P. L.; Chu-Moyer, M. Y.; Rapoport, H. J. Org.
Chem. 1990, 55, 3068–3074.
(43) Bose, A. K.; Narayanan, C. S.; Manhas, M. S. Chem. Commun. 1970,
(52) Hoffman, R. V.; Maslouh, N.; Cervantes-Lee, F. J. Org. Chem. 2002,
67, 1045–1056.
975–976.
9
11384 J. AM. CHEM. SOC. VOL. 132, NO. 32, 2010

Enantiomerically Pure trans-ꢀ-Lactams from R-Amino Acids
A R T I C L E S
tions53 and afforded the desired material 30 in 24% isolated
yield for the final two steps; the product was identical with the
enantiomer described in the literature on the basis of 1H NMR
and melting point data.48,49
Scheme 8
Conclusion
The methodology outlined herein delineates a route to
differentially substituted EP ꢀ-lactams. The power of the
technology arises from the unique one-step formation of an EP
N-trityl ꢀ-lactam via a stereospecific intramolecular Wolff
rearrangement directly from the corresponding R-amino acid.
The use of a Weinreb amide substituent permits a control
element on the C3 position, allowing ready transformation to
other key functionalities. In addition, the N-trityl group, which
is rarely encountered on the ꢀ-lactam nucleus, offers superb
protection to the strained heterocycle during the course of
treatment with reactive nucleophiles such as organolithium
reagents and strong bases. Both CFL and MVL continuous-
flow reactors were utilized in the production of ꢀ-lactam
structures. Incorporation of different amino acid functionalities
at the C4 position of the ꢀ-lactam nucleus and examination of
diastereomer control in the formation of the ꢀ-lactam ring are
ongoing efforts in our laboratory and will be reported in due
course.
ester proved greater than that of the Weinreb amide toward
standard methyl nucleophiles. We theorized, however, that
complexation of lithium ion to the Weinreb amide would
activate the functionality toward nucleophilic attack. In the
event, addition of MeMgBr to a solution of 26 and anhydrous
LiCl in THF at 0 °C, with warming to room temperature,
provided the desired ketone 27 together with unreacted 26 as a
roughly 1:1.4 separable mixture in 87% yield based on recovered
starting material (brsm).
Reduction of the methyl ketone functionally of 27 under
previously described conditions53 led to a 76% yield (88% brsm)
of 28 as an inseparable 6:1 mixture of stereoisomers at the newly
installed secondary alcohol favoring the stereoisomer shown.
The identity of this major isomer was confirmed by the
production of known compound 30. Removal of the trityl
protection group through the use of TFA and Et3SiH was
followed by silyl ether formation under the published condi-
Acknowledgment. We thank NIH IMSD (Grant R25GM58903)
and NSF AGEP for the generous financial support (Y.S.M.V.). The
purchase of the 600 MHz NMR spectrometer used in these studies
was supported by funds from the National Institutes of Health
(S10RR019918) and the National Science Foundation (CHE-
0342912). The single-crystal X-ray diffraction data for compounds
8, 12, 13, 14, and 20 were recorded on an instrument supported by
the National Science Foundation Major Research Instrumentation
(MRI) Program under Grant CHE-0521569. We thank Prof. Roger
Linington for help with the spectroscopic assignment of alkene 9
and Dr. Charles J. Vaske for computational support for the
pyramidalization studies.
Supporting Information Available: Synthetic procedures,
1
complete spectroscopic data, H and 13C NMR spectra for all
new compounds, and CIF files for all new X-ray structures. This
material is available free of charge via the Internet at http://
pubs.acs.org.
(53) Wieber, G. M.; Hegedus, L. S.; Åkermark, B.; Michalson, E. T. J.
Org. Chem. 1989, 54, 4649–4653.
JA1050023
9
J. AM. CHEM. SOC. VOL. 132, NO. 32, 2010 11385