Rapid determination of enantiomeric excess using infrared thermography

Article abstract of DOI:10.1021/op010093k

Infrared thermography (IRT) is presented as a novel technique to screen a potentially large number of asymmetric catalysts or substrates in a high-throughput fashion. IRT was used as a simple, rapid, and practical approach for initial screening of the substrate specificity of Candida antarctica lipase. This was carried out using a 96-well microtitre plate format. Potential advantages and limitations of IRT for the enzymatic stereoselective acylation of primary and secondary alcohols of interest are discussed.

Full text of DOI:10.1021/op010093k

Organic Process Research & Development 2002, 6, 463−470
Rapid Determination of Enantiomeric Excess Using Infrared Thermography
Nicolas Millot, Phil Borman, Mike S. Anson, Ian B. Campbell, Simon J. F. Macdonald, and Mahmoud Mahmoudian*
GlaxoSmithKline Research and DeVelopment, Medicines Research Centre, Gunnels Wood Road,
SteVenage, Hertfordshire, SG1 2NY, England
Abstract:
detected by IRT as small temperature changes can be
measured in cellular metabolism, growth, and toxic re-
Infrared thermography (IRT) is presented as a novel technique
to screen a potentially large number of asymmetric catalysts
or substrates in a high-throughput fashion. IRT was used as a
simple, rapid, and practical approach for initial screening of
the substrate specificity of Candida antarctica lipase. This was
carried out using a 96-well microtitre plate format. Potential
advantages and limitations of IRT for the enzymatic stereose-
lective acylation of primary and secondary alcohols of interest
are discussed.
1
1
sponse. Acceleration of drug screening is one of the
potential applications of IRT in the pharmaceutical industry.
Given the recent reports from Reetz et al. on the use of
1
2
IRT for the selection of chiral catalysts, we decided to
extend these studies and to investigate its potential application
for the rapid quantification of ee in enzyme-catalysed
1
3,14
reactions.
Results and Discussion
We at GlaxoSmithKline frequently screen biocatalysts
isolated enzymes, recombinant microorganisms) and chiral
Introduction
(
Enzymes have been used extensively in chemical syn-
thesis to generate chiral synthons.1,2 Enzymatic kinetic
resolution is one of the most widely used approaches for
the preparation of chiral pharmaceutical intermediates as
enantiomers of racemic mixtures can often be efficiently
ligands for enantioselective transformations. For high-
throughput screening (HTS), the bottleneck is usually in the
speed of chiral analyses involving laborious development of
chiral HPLC and GC assays. Consequently, we are particu-
larly interested in developing new approaches for rapid
determination of ee. Recently Reetz et al. highlighted the
use of thermal imaging for the screening of enantioselective
3
resolved in this manner. Rapid determination of enantio-
meric excess (ee) is frequently hampered by the availability
of a suitable chiral assay. Chiral HPLC, GC, and chiral shift
NMR are the most preferred techniques, but these methods
can suffer from being too time-consuming to develop. Other
12
biocatalysts. Interestingly, they observed a differential heat
output during the thermal imaging of lipase-catalysed acy-
4,5
6
approaches used by several groups include fluorescence,
(7) Reetz, M. T.; Becker, M. H.; Klein, H.-W.; St o¨ ckigt, D. A Method for High-
Throughput Screening of Enantioselective Catalysts. Angew. Chem., Int.
Ed. 1999, 38, 1758.
8) Reetz, M. T.; K u¨ hling, K. M.; Deege, A.; Hinrichs, H.; Belder, D. Super-
High-Throughput Screening of Enantioselective Catalysts by Using Capillary
Array Electrophoresis Angew. Chem., Int. Ed. 2000, 39, 3891.
9) Ding, K.; Ishii, A.; Mikami, K. Super High-Throughput Screening (SHTS)
of Chiral Ligands and Activators: Asymmetric Activation of Chiral Diol-
Zinc Catalysts by Chiral Nitrogen Activators for the Enantioselective
Addition of Diethylzinc to Aldehydes Angew. Chem., Int. Ed. 1999, 38,
7
8
mass spectroscopy, capillary electrophoresis, and circular
9
dichroism techniques.
(
Among these methods, infrared thermography (IRT) has
attracted considerable interest in recent years because non-
invasive thermal imaging of chemical reactions can be
performed through detection of emitted infrared radiations.
In Vitro and in ViVo biological processes have also been
(
10
497.
(
10) Taylor, S. J.; Morken, J. P. Thermographic Selection of Effective Catalysts
*
Author for correspondence. E-mail: mm6382@yahoo.co.uk.
1) Koeller, K. M.; Wong, C.-H. Enzymes for Chemical Synthesis. Nature 2001,
09, 232.
from an Encoded Polymer-Bound Library Science 1998, 280, 267.
(
(
(
(
(11) (a) Paulik M. A.,; Buckholz, R. G.; Lancaster, M.; E. Dallas, W. S.; Hull-
Ryde, E. A.; Weiel, J. E.; Lenhard, J. M. Development of Infrared Imaging
to Measure Thermogenesis in Cell Culture: Thermogenic Effects of
Uncoupling Protein-2, Troglitazone, and Adrenoceptor Agonists. Pharm.
Res. 1998, 15, 944. (b) Lenhard, J. M.; Paulik, M. A. Infrared Thermography
for Measuring Real-Time Thermogenesis in Organisms and Cells. PCT Int.
Appl. WO 9960630 A1 19991125, 1999.
(12) (a) Reetz, M. T.; Becker, M. H.; K u¨ hling, K. M.; Holzwarth, A. Time-
resolved IR-Thermographic Detection and Screening of Enantioselectivity
in Catalytic Reactions. Angew. Chem., Int. Ed. 1998, 37, 2647. (b) Reetz,
M. T.; Becker, M. H.; Liebl, M.; F u¨ rstner, A. IR-Thermographic Screening
of Thermoneutral or Endothermic Transformations: The Ring-Closing Olefin
Metathesis Reaction. Angew. Chem., Int. Ed. 2000, 39, 1236.
(13) High-throughput enzyme-inhibitor screening has been reported by an array
of thermistors: Connolly, R.; Sutherland, J. D. Catalyst Screening Using
an Array of Thermistors. Angew. Chem., Int. Ed. 2000, 39, 4268.
(14) Low-throughput calorimetric methods have been reported for monitoring
enzymes reactions: (a) Ramsden, D. K.; Webster, N. A.; Peron, Y.; Hughes,
J. Calorimetric Method for the Monitoring and Control of Bioreactions.
Biotechnol. Lett. 2000, 22, 1259. (b) Stefuca, V.; Gemeiner, P. Investigation
of Catalytic Properties of Immobilized Enzymes and Cells by Flow
Microcalorimetry. AdV. Biochem. Eng. Biotechnol. 1999, 64, 69 and
references therein.
4
2) Schmid, A.; Dordick, J. S.; Hauer, B.; Kiener, A.; Wubbolts, M.; Witholt,
B. Industrial Biocatalysis Today and Tomorrow. Nature 2001, 409, 258.
3) Keith, J. M.; Larrow, J. F.; Jacobsen, E. N. Practical Considerations in
Kinetic Resolution Reactions. AdV. Synth. Catal. 2001, 343, 5.
4) For excellent reviews see: (a) Reetz, M. T. Combinatorial and Evolution-
Based Methods in the Creation of Enantioselective Catalysts. Angew. Chem.,
Int. Ed. 2001, 40, 284. (b) Wahler, D.; Reymond, J.-L. Novel Methods for
Biocatalyst Screening. Curr. Opin. Chem. Biol. 2001, 5, 152.
5) (a) Le Bars, J.; H a¨ ussner, T.; Lang, J.; Pfaltz, A.; Blackmond, D. G. A
Scale-Transparent Reaction Calorimetric Assay for Rapid Catalyst Screening.
AdV. Synth. Catal. 2001, 343, 207; (b) Jandeleit, B.; Schaefer, D. J.; Powers,
T. S.; Turner, H. W.; Weinberg, W. H. Combinatorial Material Science
and Catalysis. Angew. Chem., Int. Ed. 1999, 38, 2494.
6) (a) Korbel, G. A..; Lalic G.; Shair, M. D. Reaction Microarrays: A Method
for Rapidly Determining the Enantiomeric Excess of Thousands of Samples
J. Am. Chem. Soc. 2001, 123, 361. (b) Badalassi, F.; Wahler, D.; Klein, G.;
Crotti, P.; Reymond, J.-L. A Versatile Periodate-Coupled Fluorogenic Assay
for Hydrolytic Enzymes. Angew. Chem., Int. Ed. 2000, 39, 4067. (c) Reetz,
M. T.; Sostmann, S. 2,15-Dihydroxyhexahelicene (HELIXOL): Synthesis
and Use as an Enantioselective Fluorescent Sensor Tetrahedron 2001, 57,
(
(
2
515.
1
0.1021/op010093k CCC: $22.00 © 2002 American Chemical Society
Vol. 6, No. 4, 2002 / Organic Process Research & Development
•
463
Published on Web 03/05/2002

Scheme 1. HTS of enantioselective catalysts with infrared
IRT
tophenone 1) and R-2 (86% ee, prepared by CBS reduction
of 1) were added to the second row of the plate (Figure 1A).
Temperature changes were monitored periodically with an
infrared camera (thermal imaging), and reactions were started
by adding solutions of vinyl acetate in toluene to each well.
Before reactions were initiated, cold spots (blue colour)
were evident due to the endothermic process of solvent
evaporation (Figure 1A). Subsequently, we observed dif-
ferential heat outputs (yellow colour) corresponding to
various concentrations of the R-enantiomer (Figure 1B). This
compared well with previous reports (Reetz et al.1 ) that
temperature changes are affected by substrate concentrations.
Similarly, wells containing either Rac-2 or R-2 (prepared by
chemical reductions) showed elevated heat outputs (Figure
4
Scheme 2. Sodium borohydride (NaBH ) and CBS
reduction of 1 followed by enzymatic kinetic resolution of 2
catalysed by immobilised C. antarctica lipasea
2a
a
4
NaBH reduction of 1 affords Rac-2 (chiral HPLC) whereas, CBS reduction
affords R-2 (86% ee by chiral HPLC).
1B). C. antarctica lipase acylation of Rac-2 has been shown
to be R-enantioselective (i.e. the R-isomer is preferentially
acylated leaving behind the corresponding S-isomer, Scheme
1
2
2
). This suggested that an excellent R-enantioselectivity
1
5
was also obtained after CBS reduction of 1 (86% ee by
chiral HPLC), Figure 1. These results therefore, gave us
encouragement to further investigate the use of infrared
thermal imaging for rapid determination of ee.
To quantify temperature changes from enzymatic reac-
tions, heat output was measured as a function of time (Figure
2
). Initially, during the first 9 s of each reaction, cooling
was observed as a result of the addition of a solution of vinyl
1
6
acetate. Temperature changes were observed almost im-
17
mediately in each well after reactions had started. Further-
more, there was a noticeable differential heat output corre-
sponding to the various mixtures of optically pure R- and
S-2 (black curves, Figure 2). Indeed, the observed order of
heat output corresponded to different concentrations of the
R-enantiomer. We were also encouraged to observe that, at
given concentrations of alcohol, lipase, and vinyl acetate,
the heat output generated during reaction fitted perfectly on
the standard curves (black curves), data not shown. The
calculated ee for racemic alcohol 2 was found to be near
zero and 75-100% ee for enantiomerically enriched alcohol
18
R-2 (Figure 2). This corresponded well with the previously
observed values of zero and 86%, respectively, as analysed
by chiral HPLC. This therefore, demonstrated the proof-of-
principle that IRT was a viable approach for use in HTS of
asymmetric catalysts.
With these promising results in hand we decided to
investigate this further to see if the heat generated during
lipase-catalysed resolutions of 2 could be quantified more
Figure 1. Time-resolved thermal imaging of lipase-catalysed
enantioselective acylation of 1-phenylethanol, 2. Colours cor-
respond to the following temperatures. (A) At time zero; black
(12.5-13.5 °C), purple (15-16 °C), pink (17-18 °C), orange
(18.5-19.5 °C). (B) After 60 s; orange (18.5-19 °C), yellow
(19.5-20 °C), white (20.5 °C), grey (21.4-21.6 °C).
lation of chirally pure R- and S-alcohols.12a We therefore,
envisaged a high-throughput chiral screening approach based
on thermal detection of enzyme-catalysed kinetic resolutions
of racemic mixtures and proceeded to develop this further
(
15) Corey, E. J.; Helal, C. J. Reduction of Carbonyl Compounds with Chiral
Oxazaborolidine Catalysts: A New Paradigm for Enantioselective Catalysis
and a Powerful New Synthetic Method. Angew. Chem., Int. Ed. 1998, 37,
1986.
(Scheme 1).
(
(
(
16) Vinyl acetate is a highly volatile compound (bp ) 77 °C); consequently,
an endothermic process is observed by IRT due to evaporation of this
reagent.
17) We believe the heat observed during the first stages of the lipase-catalysed
reactions is partly due to the mixing enthalpy of the highly concentrated
solutions (exothermic effect in this case).
18) The use of emmisivity-corrected IRT did not improve the accuracy of the
ee reading. The method consists of subtracting the thermal background
before the start of reactions to data gathered during the catalytic reactions.
For a leading reference, see: Holzwarth, A.; Schmidt, H.-W.; Maier, W. F.
Detection of Catalytic Activity in Combinatorial Libraries of Heterogeneous
Catalysts by IR Thermography. Angew. Chem., Int. Ed. 1998, 37, 2644.
We chose to study the lipase-catalysed acylation of
1-phenylethanol, 2, with Candida antarctica as a model
system (Scheme 2). Reactions were carried out in a 96-well
microtitre plate containing solutions of 2 and the immobilised
lipase. Standard solutions of various mixtures of optically
pure R- and S-2 (0-100% R or S-isomer) were used for
determining ee values (first row, Figure 1). Solutions of
Rac-2 (prepared by sodium borohydride reduction of ace-
464
•
Vol. 6, No. 4, 2002 / Organic Process Research & Development

Figure 2. Time-resolved temperature profile for C. antarctica acylation of 1-phenylethanol, 2. Black curves represent first rows
(
100% S - 100% R) of each plate (standard curves) as shown in Figure 1. Red and green curves represent reaction wells (second
row) containing Rac-2 and R-2 respectively.
Scheme 3. Enzymatic kinetic resolutions of 5 and 8
catalysed by immobilised C. antarctica lipase
Figure 3. Correlation between temperature areas and enan-
tiomeric excess for lipase-catalysed resolution of 1-phenyletha-
nol, 2. Positive values of ee were conventionally matched with
the R-configuration, and negative values, with the S-configu-
ration.
2
with the observed ee values (R ) 0.9953), data not shown.
Table 1. Determination of enantiomeric excess (ee) by IRTa
Similarly, standard curves for the determination of ee by IRT,
alcohols
Rac-2
R-2
using solutions of 8 and ethyl methoxyacetate as an acylating
20
agent, were obtained. Temperature areas (s‚K), representing
solutions of 8 of varying optical purity (100% S - 100%
R), were determined in each row (A-G) of a microtitre plate
heat areas (s.K)
IRT calculated ee (%)
chiral HPLC ee (%)
368.32
8
1
508.93
94
86
(
Table 2). These gave a good linear correlation with the
a
Temperature areas are expressed as second‚Kelvin (s‚K)
2
observed ee values (R ) 0.9835) (data not shown). The data
were analysed using the statistical software package Minitab.
This experiment was deliberately designed to allow seven
different concentrations of optically pure 8 to be analysed
simultaneously and was replicated four times (rows A-G)
to test for “lack of fit” (Table 2). Lack of fit is observed
when a pattern is discernible in the residual plot (residuals
are the distances of the experimental points from the fitted
regression line, measured in a direction parallel to the
response axis). By replicating measurements we can define
whether an apparent pattern is significant. This gives
information about the inherent variability of the response
measurements (called the “pure error”). Unexpectedly,
although data from the replicate on row A (at the very edge
of the plate) demonstrated a relationship between ee and
temperature area, this relationship differed slightly from the
other three replicates (Figure 5). This difference in temper-
ature area for row A could be due to its being at the edge of
the plate where dissipation of heat may be different from
that at the centre of the plate.
accurately. We calculated areas under each temperature curve
from the start of the reactions (9 s) to the point where the
maximum temperature rise was reached (117 s).19 A linear
correlation could be obtained between heat areas and ee
values (Figure 3). Plots of areas were fitted to a straight line
2
using the nonweighted linear least-squares method (R )
0.988). This linear equation could be used to calculate of ee
values of Rac-2 (-8% ee) and optically enriched R-2 (94%
ee). These corresponded very well with ee values previously
measured by IRT and chiral HPLC (Table 1).
To extend the scope of this method further, we applied
IRT determination of ee to other substrates such as 2-hexanol,
5, and 1-phenylethylamine, 8, (Scheme 3). The time-resolved
IRT temperature profile for C. antarctica at various mixtures
of optically pure 5 is shown in Figure 4. The initial rates
under each temperature curve gave a good linear correlation
(
19) The time for which maximum temperature rise is reached (top of temperature
curve) is believed to be close to the time for which the reaction is over. For
more information, see ref 12a and Simultaneous Assessment of Thermo-
dynamic and Kinetic Behaviour of Chemical Reactions: Theory and
Experiment. Davies, G. C.; Hutton, R. S.; Millot, N.; Anson, M. S.;
Macdonald, S. J. F.; Campbell, I. B. Manuscript in preparation.
(20) Ditrich, F.; Hauer, K.; Ladner, B. W. Optisch aktive amine durch lipase-
katalysierte methoxyacetylierung Backenhohl. J. Prakt. Chem. 1997, 339,
381.
Vol. 6, No. 4, 2002 / Organic Process Research & Development
•
465

Figure 4. Time-resolved temperature profile for C. antarctica resolution at various mixtures of optically pure 2-hexanol, 5.
Table 2. Lipase-catalysed resolution of 8a
ee of 8 in each row
100% (S)
75% (S)
50% (S)
0
50% (R)
75% (R)
100% (R)
rows of plate
a
a
a
a
a
a
a
A
C
E
G
369.129
395.121
401.959
401.715
464.148
435.323
423.204
432.426
501.250
449.861
458.833
457.215
603.263
549.652
513.634
526.046
646.656
598.866
609.919
602.104
644.218
615.796
608.994
634.191
691.162
628.882
636.040
654.406
a
Temperature areas (s‚K), representing various mixtures of optically pure 8 [100% (S) to 100% (R)], were determined in each row (A-G) of a microtitre plate
Figure 6. Linear relationship with 95% confidence intervals
between ee and temperature area using row C, E, and G data
as stated in Table 2.
along with confidence intervals at the 95% level were plotted
(
Figure 6). This can be used to predict an unknown ee from
samples with known temperature area to approximately
5%. There is no significant lack of fit associated with the
(
above model; in other words, the residuals plotted against
ee resembled a random sample from a normal distribution
with zero mean. A normal probability plot of the residuals
also suggested that there was no evidence against normality
(The Anderson-Darling normality test produced a prob-
ability value of 0.798). The 95% confidence intervals (Figure
6), fitted around the above first order polynomial model,
demonstrate very high confidence in the mean temperature
area response. To use these intervals as a predictive tool in
analysing a sample of unknown ee, at least six replicate
experiments should be carried order to achieve a good
approximation of the mean temperature area. If, however,
Figure 5. Relationship between rows, ee, and temperature area
as stated in Table 2.
In the following analysis, to illustrate how the ee can be
predicted from temperature area, the data from row A was
excluded. It should be noted that, if the cause of the
variability of the row A data cannot be determined, then the
estimated precision for predicted ee in the following analysis
will be an overestimate. A first order polynomial was fitted
to the data (rows C, E, and G). The y-intercept (the constant)
and the linear (x term) predictors were found to be highly
significant (Probability < 0.0005) to the model. This model
466
•
Vol. 6, No. 4, 2002 / Organic Process Research & Development

Table 3. Ninety-six-well microtitre plate showing C. antarctica-catalysed acylation of 96 racemic alcohols using IRTa
a
Numbers (T; %) are related to the temperature rise (T) measured by IRT and conversions (%) measured by GC after 90 min. A-H and 1-12 represent rows of
microtitre plate.
only a rough estimate of ee is required rather than a
statistically significant value, a duplicate measurement is
probably all that is needed.
volume, 0.2 mL). Toluene was chosen as the reaction solvent
because of its low volatility. In some cases dioxane was
preferred for solubility reasons (alcohols A2, A3, H2, H7,
Table 3). Analysis of thermal imaging pictures at regular
time intervals showed some surprising results (Figure 7). In
particular, the strongest heat output was immediately ob-
served for alcohol G9 having a tertiary amino group in the
To extend the application of IRT to biocatalysis, we turned
our attention to the screening of alcohols for acylations
catalysed by C. antarctica lipase. It is important to note that
an initial objective of a HTS campaign is its speed. In other
words, the process must rapidly indicate which substrates
are worthy of further study. Although desirable, it is
unrealistic to expect an HTS process to deliver completely
reliable and reproducible data for every substrate.
21
γ-position (Figure 7). Significant temperature rises were
(
21) As a control experiment a reaction in the same conditions has also been
run for alcohol G9 without any enzyme catalyst. It was found that no reaction
occurred after a prolonged time, suggesting that the heat output observed
by IRT was not due to intramolecular catalysis from tertiary amino group
but rather by enzyme catalysis.
Ninety-six racemic mixtures of secondary alcohols of
interest (Table 3) were screened in a microtitre plate (reaction
Vol. 6, No. 4, 2002 / Organic Process Research & Development
•
467

Figure 7. Time-resolved thermal imaging for screen of lipase-catalysed acylations of 96 alcohols as stated in Table 3. Colours
correspond to the following temperatures: black (14-15 °C), purple-blue (16-17 °C), pink-red (19-20 °C), red-orange (20-21
°
C), orange-yellow (21.5-22.5 °C), yellow (23-24.5 °C), white (25-26 °C), grey (26-27 °C).
Table 4. Enantiomeric ratio (E) and temperature rises as
observed by IRT after lipase-catalysed reactions of selected
alcohols (GC analysis)
also seen after 10 s for substrates with a tertiary amino group
at the R-position (E9 and F9). For comparison, heat outputs
from reactions of similar alcohols not containing any basic
functionality (e.g. A4, B4, D4, B12, and G11) were detected
much later (Figure 7). This may suggest that the kinetics of
acylation of alcohols E9-G9 could be substantially different
than that for reactions with alcohols having no basic
functionality.
wells
conversion (%)
ee (%)
E
IRT (K)
A8
C2
D2
D7
G8
G10
H10
67
72
41
20
30
40
40
0
33
24
72
7
30
52
1.0
4.8
1.9
7.3
1.2
2.2
4.4
3.2
3.1
3.9
3.1
2.1
3.2
3.1
Attempts to obtain a more accurate readout for each well
proved to be more tricky as quantification of the generated
heat needs to be considered. In fact, we tentatively applied
the emissivity-corrected IRT method as previously reported
The findings on the substrate specificity of C. antarctica
1
8
by Maier et al. by subtracting the maximum reaction
temperatures measured for each well from initial tempera-
tures measured before addition of the final reactant. It was
apparent that a small temperature rise of at least 1.7 °C was
lipase for acylations of alcohols are consistent with previ-
23
ously reported data on its active site. Finally, we have
highlighted a potential drawback of IRT in detection of the
heat output in a 96-well format, and the apparent lack of
correlation between the observed heat output and enantiose-
lectivity. However, we believe that IRT can provide a crude
but practical initial method, when screening large numbers
of catalysts and/or substrates, to rapidly identify a particular
structural class that can be further accurately analysed by
conventional methods. For example the data generated by
IRT for the whole plate (96 substrates) took only 15 min,
several orders of magnitude faster than analysis by GC or
HPLC.
22
detected in each well. To investigate this further, the extents
of reactions were determined by GC (Table 3). Interestingly,
despite the appearance of small temperature rises, no reaction
at all was detected in 11 wells (B1, H3, F6, A7, B7, C9,
D9, B11, C11, E11, D12) as judged by GC highlighting a
potential limitation of IRT. For example, the highest tem-
perature rise was 3.1°K for a well with no reaction. This
was therefore used as the background reading; 46 wells gave
a temperature rise under 3.1 K, whereas the remaining 50
wells showed a temperature rise above 3.1 K. This detection
rate is typical for HTS where the emphasis is on rapid
detection of interesting wells for further study rather than
an exhaustive assessment of each substrate. Furthermore,
although chiral analyses (chiral GC) of selected wells showed
that in some cases acylations were indeed enantioselective
Experimental Section
General Methods. All chemicals, reagents and com-
pounds used were of highest purity. All IRT experiments
were performed in polypropylene microtitre plates having
9
0
6 U-shaped wells to allow smooth stirring (well volume,
.25 mL). Reactions were stirred at 1100 rpm with 5 × 2
(Table 4), this however, did not correspond well with the
generated heat outputs.
(
23) Anderson, E. M.; Larsson, K. M.; Kirk, O. One BiocatalystsMany
Applications: The Use of Candida antarctica B-Lipase in Organic Synthesis.
Biocatal. Biotransform. 1998, 16, 181.
(22) When an empty plate is placed under the IRT camera, a temperature rise of
approximately 2 °C is observed over 1 h. For more information see ref 19.
468
•
Vol. 6, No. 4, 2002 / Organic Process Research & Development

mm magnetic bars. Empty wells were interspersed between
each reaction well to prevent any heat cross-contamination.
IRT measurements were carried out under a fume hood
without taking any other precautions to enhance sensitivity
of the system.
Reaction temperatures were measured with a Ther-
maCAM SC3000 (Flir Systems AB, Sweden) equipped with
a GaAs detector (320 × 240 pixels FPA QWIP resolution)
and placed at 35 cm above the top of each microtitre plate
ethanol (60 mL). This was stirred for 1 h at 0 °C and then
warmed to room temperature for a further 1 h. After the
reaction mixture was carefully quenched at 0 °C with an
4
aqueous saturated solution of NH Cl (100 mL), the resulting
biphasic mixture was extracted with ether (3 × 100 mL).
The combined organic layers were dried with brine and
magnesium sulphate. After filtration the solution was evapo-
rated under vacuo. The resulting crude oil was purified by
flash chromatography on silica gel (cyclohexane:ether, 4:1)
to yield 2.3 g of a colourless oil (37%, 0% ee).
(top-viewing camera reading). The detection system was
1
5
sensitive to IR radiation in the wavelength range 8-9 µm
and to temperature changes of 5 mK. 16 frames were taken
for every thermal picture. Well Plate Thermal Profiler was
used as a computer interface to manipulate data from thermal
imaging and temperature changes as a function of time.
Where applicable, thermal data were transferred to Microsoft
Excel for further manipulations.
Immobilized C. antarctica lipase (Novozyme-435) was
purchased from Novozymes (Denmark). Chiral HPLC analy-
ses of 1-phenylethanol were carried out with Chiralcel OB-H
column (flow rate, 1.0 mL/min; heptane/2-propanol, 90/10;
column temperature, 30 °C). Retention times for the S- and
R-isomer were 5.6 and 7.0 min, respectively. Chiral GC
analyses were carried out with a Chiraldex â-cyclodextrin
permethylated hydroxypropyl column (Astec, 20 m × 0.25
mm) with a temperature gradient of 60-200 °C at 5 °C/min
and a 2-min hold at 200 °C. Achiral GC analyses were
carried out with a HP-1 cross-linked methyl siloxane column
Synthesis of R-2 by CBS Reduction. To a stirred 1 M
toluene solution of S-2-methyl-CBS-oxazaborolidine catalyst
(
5 mL, 5 mmol, 10 mol %) at room temperature was added
a 1 M solution of BH THF (5 mL, 5 mmol) under a flow
of nitrogen. After cooling to 0 °C, a 3.3 M THF solution of
(5.8 mL, 50 mmol) and a 1 M solution of BH THF (25
3
1
3
mL, 25 mmol) were added simultaneously dropwise (addition
rate 1 drop/sec). The resulting mixture was stirred at room
temperature for a further 30 min, and 10 mL of methanol
was slowly added at 0 °C. A 2 M aqueous HCl solution was
finally added before extraction with ether (3 × 100 mL).
The combined organic layers were dried with brine and
magnesium sulphate. After filtration the solution was evapo-
rated under vacuo. The resulting crude oil was purified by
flash chromatography on silica gel (cyclohexane:ether, 4:1)
to yield 2.2 g of a colourless oil (36%, 86%ee).
Method for Calculating Heat Areas from Temperature
Curves. Heat areas were calculated from the corresponding
temperature curves. Initial rates were calculated between the
start (9 s) and end of each reaction (117 s). Areas (A) were
approximately obtained by addition of all small areas
calculated between two successive plots i and i + 1: A ) Σ
(
Agilent Technologies, 10 m × 0.1 mm, 0.4 µm film
thickness).
Determination of ee by IRT. Procedure A: 1-Phenyl-
ethanol (2) and 2-Hexanol (5). Optically active solutions
of 2 or 5, used as standard curves for ee determinations, were
prepared by mixing the appropriate amounts of commercially
pure R- and S-alcohols. Immobilised C. antarctica lipase (25
mg/well) and 2 M toluene solutions of alcohol (0.1 mL, 0.2
mmol) were dispensed into each well. The temperature
reading was started with an IR camera, and reactions were
initiated by the addition of 4 M toluene solution of vinyl
acetate (0.1 mL, 0.4 mmol) with a multichannel pipet robot.
Addition was completed within 10 s. The experiment was
run for 5 min and temperature reading was taken every 3 s
with 2. The experiment was run for 10 min, and temperature
readings were taken every 5 s with 5.
A
t
i
where A
i+1 are time coordinates for plots i and i + 1, T
are temperature coordinates for plots i and i + 1, and A
i+1 are heat areas calculated between plots i and i + 1.
Screen of 96 Alcohols With C. antarctica Lipase Using
i
) (ti+1 - t
i
)[(T
i
+ Ti+1)/2 - 16], where t
i
and
i
and Ti+1
i
and
A
IRT. The immobilised lipase (25 mg/well) and 4 M toluene
solutions of 96 racemic mixtures of various alcohols (0.1
mL, 0.4 mmol), containing n-tetradecane as an internal
standard (10 µL/ well), were dispensed into 96 wells of a
microtitre plate. For solubility reasons, dioxane was used as
a solvent in some cases. An IR camera was used to record
temperature readings, and reactions were started by the
addition of 8 M toluene solution of vinyl acetate (0.1 mL,
Procedure B: 1-Phenylethylamine (8). Similarly, opti-
cally active solutions of 8, used as standard curves for ee
determinations, were prepared by mixing the appropriate
amounts of commercially pure R- and S-8. Immobilised C.
antarctica lipase (25 mg/well) and 4 M toluene solutions of
amine (0.1 mL, 0.4 mmol) were dispensed into each well.
Temperature reading was started with an IR camera, and
reactions were initiated by the addition of 4 M toluene
solution of vinyl acetate (0.1 mL, 0.4 mmol). Addition was
completed within 10 s. The experiment was run for 10 min,
and temperature reading was taken every 5 s.
0
.8 mmol). Additions were completed in less than 15 s using
a multichannel pipet robot. The experiment was run for 15
min, and the temperature readings were taken every 10 s.
Upon completion, reaction mixtures were quenched with
methanol and analysed by GC.
Conclusions
We report here on the evaluation and development of IRT
as a means to screen a potentially large number of asym-
metric catalysts or substrates in a high-throughput fashion.
We have shown that a differential heat output, corresponding
to various mixtures of R- and S-isomers, can be achieved
and have demonstrated a good linear correlation with the
observed ee values with selected alcohols and amines.
Synthesis of Rac-2 by Borohydride Reduction. Sodium
borohydride (2.27 g, 60 mmol) was added portionwise at 0
°
C to a solution of acetophenone 1 (5.8 mL, 50 mmol) in
Vol. 6, No. 4, 2002 / Organic Process Research & Development
•
469

However, a potential limitation and drawback of this
approach was highlighted when screening alcohols in a 96-
well format. It was evident that there were limitations in
detection of the heat output from reactions, and a good
correlation between the observed heat output and enantiose-
lectivity could not be obtained.
whereas using IRT would only take a few hours (if not
minutes) to crudely identify a particular structural class that
can be further accurately analysed giving valuable informa-
tion on structure-activity relationships for lead optimisation
programmes.
It was suggested that although IRT is less accurate than
other methods, and despite its limitations, it can however
provide a crude but practical initial method to rapidly
visualise and narrow down a selection of substrates for
further evaluation. This is acceptable in an industrial scenario,
when dealing with screening campaigns involving huge
number of catalysts and substrates, where the speed of
analysis becomes an issue. Conventionally such analyses may
take several days to complete using HPLC or GC techniques,
Acknowledgment
N.M. thanks the European Union for a Marie Curie
Industry Host fellowship under the Improving Human
Potential Programme.
Received for review October 8, 2001.
OP010093K
470
•
Vol. 6, No. 4, 2002 / Organic Process Research & Development