4008
M. M. Kayser et al. / Tetrahedron: Asymmetry 16 (2005) 4004–4009
tor differently; instead, the substrate likely occupies mir-
ror image orientations. As noted above (Fig. 1), this
would lead to antipodal products, consistent with the
experimental observations. The loop A sizes and confor-
mations in models of Yjr096w and Ydl124w are similar,
consistent with both mediating Si-face hydride transfer
to 1. It is also interesting to compare the loop A portions
of the stereoselective reductases with that of a non-selec-
tive reductase such as Ydr368w. In this case, the confor-
mation of the loop A region of Ydr368w resembles that
of Ybr149w; however, its slightly shorter length makes it
tempting to speculate the fact that it is insufficient to
enforce a single binding mode for 1.
ing possibility that more stereoselective reductions
might be obtainable by further alterations in this region
remains to be tested experimentally.
4. Experimental
4.1. General
a-Keto-b-lactam 1 was prepared as described previ-
ously.15 Authentic samples of 2, ent-2, and 3 were pre-
pared by literature methods.15 Yeast GST-fusion
proteins were isolated as described previously13 and
stored at À20 ꢂC in buffer containing 50% glycerol.
Screening reactions were sampled by mixing 100 lL of
the reaction with an equal quantity of EtOAc containing
methyl benzoate (internal standard), followed by GC
analysis with a DB-1301 column (15 m · 0.53 mm).
The enantiomeric compositions of reduction reactions
were determined by chiral-phase HPLC on a Chiracel
OD-H column (4.6 · 250 mm) operated isocratically
(90:10 hexanes/i-PrOH) at a flow rate of 1.0 mL/min.
UV detection (254 nm) was used and peaks were identi-
fied by co-injection with authentic standards of 2, ent-2,
and 3.
Although the length and spatial position of A loops
appear to play an important role in the selectivity of
aldose reductases, their amino acid composition is
clearly also a factor. For example, A loops of Ycr107W
and Ynl331C are identical except for a single amino acid
change: phenylalanine in the selective Ycr107W to
valine in the only marginally selective Ynl331C.
3. Conclusion
This study examined the behaviour of a series of yeast
reductases toward a probe a-keto-b-lactam 1. The
results differed from those observed in our previous
studies of this library in several respects.11–14 First, all
of the reductases examined accepted 1. This was unusual
since even for small substrates, it was generally found
that a given substrate would be reduced by only a subset
of the library. The prior exception we had observed was
ethyl 4-chloroacetoacetate, whose carbonyl was acti-
vated by inductive effects of the neighboring chlorine
atom.12 Likewise, the ketone carbonyl of 1 is also highly
activated both by geometric strain and Coulombic
repulsion, and it is likely that these effects enhance its
ability to act as a substrate for a broad array of
enzymes.
4.2. Small-scale reductions with bakersÕ yeast reductases
Screening was carried out on 0.5 mL scales at 30 ꢂC in
100 mM KPi (pH 7.0 or 8.0). Reaction mixtures con-
tained 5 mM 1 (25 lL of a 100 mM stock in EtOH),
5 mM hydroxypropyl-b-cyclodextrin (125 lL of
a
20 mM stock in KPi buffer), 0.10 mM NADP+ (1.5 lL
of a 34 mM stock in KPi buffer), 7.5 mM glucose-6-
phosphate (2.6 lL of a 1.5 M solution in KPi buffer),
0.5 U of glucose-6-phosphate dehydrogenase (2.0 lL of
a 0.25 U/lL stock in KPi buffer) plus 344 lL of KPi buf-
fer. After vortex mixing, 5 lg of the appropriate yeast
GST-fusion protein was added. The reactions were then
rotated gently and sampled periodically for GC and
HPLC analysis.
The relative lack of stereoselectivity in enzymatic reduc-
tions of 1 was also somewhat surprising. Initial experi-
ments on the prochiral a- and b-ketoesters suggested
that when a substrate is accepted by a reductase, the
reaction tends to proceed with a high stereoselectiv-
ity.11,12 Clearly, the reductions of 1 did not follow this
trend. Instead, even the most highly selective reactions
afforded alcohols with ca. 90% ee. It is important, how-
ever, to note the similarity between reductions of 1 and
those of a-ketoesters investigated earlier.12 Three a-
ketoesters, ethyl 2-oxobutyrate, ethyl 2-oxopentanoate
and ethyl 2-oxo-3-phenylbutyrate, accepted by seven
(out of eight) aldose reductases listed in Table 1, were
reduced with lower than anticipated enantiomeric excess.
Furthermore, a clear change in selectivity was parallel to
that observed for a-keto-b-lactam 1; Yjr096w and
Ydl124w catalyzed Si-face and Ybr149w and Ycr107w
Re-face hydride transfer.12
Acknowledgments
We thank NSERC (RGP 20940), the National Science
Foundation (CHE-0130315), and the University of
New Brunswick for generous financial support of this
project.
References
1. Ojima, I.; Habus, I.; Zhao, M.; Georg, G. I.; Jayasinghe,
L. R. J. Org. Chem. 1991, 56, 1681–1683.
2. Georg, G. I.; Cheruvallath, Z. S.; Himes, R. H.; Mejillano,
M. R.; Burke, C. T. J. Med. Chem. 1992, 35, 4230–4237.
3. Thaisrivongs, S.; Pals, D. T.; Kroll, L. T.; Turner, S. R.;
Han, F. S. J. Med. Chem. 1987, 30, 976–982.
4. Huff, J. R. J. Med. Chem. 1991, 34, 2305–2314.
5. Kearns, J.; Kayser, M. M. Tetrahedron Lett. 1994, 35,
2845–2848.
6. Kayser, M. M.; Mihovilovic, M. D.; Kearns, J.; Feicht,
A.; Stewart, J. D. J. Org. Chem. 1999, 64, 6603–6608.
Differences in loop A conformations provide one possi-
ble reason for the divergent stereoselectivities observed,
at least within the family of aldose reductases. The entic-