High-Throughput Screening of Reductive Amination Reactions Using Desorption Electrospray Ionization Mass Spectrometry

Article abstract of DOI:10.1021/acs.oprd.0c00230

This study describes the latest generation of a high-throughput screening system that is capable of screening thousands of organic reactions in a single day. This system combines a liquid handling robot with desorption electrospray ionization (DESI) mass spectrometry (MS) for a rapid reaction mixture preparation, accelerated synthesis, and automated MS analysis. A total of 3840 unique reductive amination reactions were screened to demonstrate the throughputs that are capable with the system. Products, byproducts, and intermediates were all monitored in full-scan mass spectra, generating a complete view of the reaction progress. Tandem mass spectrometry experiments were conducted to verify the identity of the products formed. The amine and electrophile reactivity trends represented in the data match what is expected from theory, indicating that the system accurately models the reaction performance. The DESI results correlated well with those generated using more traditional mass spectrometry techniques like liquid chromatography-mass spectrometry, validating the data generated by the system.

Full text of DOI:10.1021/acs.oprd.0c00230

Subscriber access provided by University of Wollongong Library
Full Paper
High Throughput Screening of Reductive Amination Reactions
using Desorption Electrospray Ionization (DESI) Mass Spectrometry
David Land Logsdon, Yangjie Li, Tiago Jose Paschoal Sobreira,
Christina R. Ferreira, David H Thompson, and R. Graham Cooks
Org. Process Res. Dev., Just Accepted Manuscript • DOI: 10.1021/acs.oprd.0c00230 • Publication Date (Web): 14 Aug 2020
Downloaded from pubs.acs.org on August 14, 2020
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 23
Organic Process Research & Development
1
2
3
4
5
6
High Throughput Screening of Reductive Amination Reactions using Desorption
Electrospray Ionization (DESI) Mass Spectrometry
7
8
9
David L. Logsdon1, Yangjie Li1, Tiago J. P. Sobreira2, Christina R. Ferreira2, David H.
Thompson1,2 and R. Graham Cooks*1,2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1 Department of Chemistry, Purdue University, West Lafayette, IN 47907, USA
2 Bindley Biosciences Center, Purdue University, West Lafayette, IN 47907, USA
*e-mail: cooks@purdue.edu
Abstract
This study describes the latest generation of a high-throughput screening system that is
capable of screening thousands of organic reactions in a single day. This system combines a liquid
handling robot with desorption electrospray ionization (DESI) mass spectrometry for rapid
reaction mixture preparation, accelerated synthesis, and automated MS analysis. A total of 3,840
unique reductive amination reactions were screened to demonstrate the throughputs that are
capable with the system. Products, byproducts, and intermediates were all monitored in full scan
mass spectra, generating a complete view of reaction progress. MS/MS experiments were
conducted to verify the identity of the products formed. The amine and electrophile reactivity
trends represented in the data match what is expected from theory indicating that the system
accurately models reaction performance. The DESI results correlated well with those generated
using more traditional mass spectrometry techniques like LC-MS, validating the data generated by
the system.
Keywords · high throughput experimentation · reductive amination · desorption electrospray
ionization · reaction screening · mass spectrometry
Introduction
High throughput experimentation (HTE) is a rapidly growing technology in which many
experiments are conducted in parallel, generating large datasets in a time efficient and cost-
1
effective manner. The applications of HTE are broad including evaluating gene expression ,
1
ACS Paragon Plus Environment

Organic Process Research & Development
Page 2 of 23
1
2
3
4
5
6
7
8
9
microbial strains 2, and protein-ligand interactions 3; however, HTE has found particular success
in reaction screening and discovery 4-7. HTE and automation are credited with revolutionizing the
pharmaceutical industry by significantly shortening the timeline and reducing the costs of the drug
development process 8, in part this been achieved through the adoption of flow chemistry which
has been impacted by HTE methodology 9, 10. One can readily imagine a system like this proving
useful in addressing a variety of synthetic and biosynthetic problems, to select just one recent case,
the biosynthesis of the classical lignans 11.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In an attempt to further develop technology which can expedite drug development, we
developed a high-throughput system that is capable of screening thousands of reactions per hour
9, 12. The system consists of a liquid handling robot for reaction mixture preparation, and a mass
spectrometer fitted with a desorption electrospray ionization (DESI) source, for reaction product
analysis. DESI is an ambient ionization technique in which a stream of charged droplets is directed
13, 14
at a surface, resulting in the desorption and ionization of analytes present on the surface
.
15
Accelerated reactions are known to occur in the droplets directed into the mass spectrometer .
With high special resolution (~500 µm) and an analysis rate of ~1 s per sample, DESI is excellently
suited for the analysis of high density reaction mixture arrays. Inspired by our work, other groups
16
have adopted the use of DESI for reaction screening . Our experiments, which rely of liquid
handling robots and DESI ionization, are compatible with most solvents which is not the case with
the fastest of the alternative methods, droplet injection technology. 7
The purpose of this work is to describe the latest generation of our reaction screening system
and demonstrate its performance. The objective of this study is to explore the value of DESI-MS
as a very rapid, ultra-small scale preliminary screening tool. In our previous work where we
scanned continuously across the DESI plate, it took about three hours to scan the whole plate no
matter how many samples spots there were on each plate. In the current system, the DESI-MS
scanning method has been improved using a spot-to-spot hopping method to collect data while
moving in a serpentine fashion to avoid delays associated with line returns. These two changes
have resulted in an analysis rate of ~1 s per spot. Two further consequences of these changes are
(i) when smaller numbers of spots are pinned the time to scan the plate is correspondingly reduced
and (ii) scanning is independent of the pinning pattern allowing much more repaid re-examination
of spots of interest, for example using tandem mass spectrometry. To demonstrate the realistic
throughputs that are capable using this system, we conducted an experiment in which 3,840 unique
2
ACS Paragon Plus Environment

Page 3 of 23
Organic Process Research & Development
1
2
3
4
5
6
reductive amination reactions were screened. Of particular note is that half of these reactions were
screened in a single day starting from commodity chemicals.
Reductive amination, as shown in Scheme 1, is among the most widely used reactions by
medicinal chemists in industry 17. It is a method of introducing nitrogen into a molecule that offers
superior control over alternative reactions. Amine alkylation using alkyl halides is typically
accompanied by over-alkylation 18. The reaction involves the nucleophilic attack of an amine on
an aldehyde or ketone, forming an imine after dehydration. The imine (or iminium if protonated)
is reduced by the reducing agent forming the product. Weak reducing agents such as sodium
cyanoborohydride (NaCNBH3) or sodium triacetoxyborohydride (NaBH(OAc)3) are often used
because they are strong enough to reduce the imine or iminium intermediate, but are usually not
strong enough to reduce the aldehyde or ketone starting material. Reductive amination reactions
can be carried out under a variety of conditions depending on what type of substrates are involved.
The choice of reducing agent, solvent, stoichiometry, acid catalyst, and reaction time can all have
a significant impact on the outcome of the reaction. This makes the reductive amination reaction
an excellent candidate to study using HTE.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
H+ (cat.) R3
R4
hydride
source
N+
-H2O
R1
R2
R3 R4
N
R3 R4
N
R3 R4
N
O
+
R1 R2
H
R1 R2
OH
R1 R2
H
R3
R4=H
-H2O
hydride, proton
source
N
R1, R2, R3, R4 = H, alkyl, aryl
R1
R2
Scheme 1. Reductive amination in which an amine and a carbonyl compound condense to afford
an imine or iminium ion that is reduced in situ to form an amine product
Results and Discussion
We evaluated the reactions of eight amines (Scheme 2) combined with 24 electrophiles
(Scheme 3, 12 aldehydes and 12 ketones) under a variety of reaction conditions, including different
reducing agents (NaCNBH3, NaBH(OAc)3, and sodium borohydride (NaBH4)), stoichiometries,
reaction times, and the presence or absence of an acid catalyst.
3
ACS Paragon Plus Environment

Organic Process Research & Development
Page 4 of 23
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 2. Amines used in the high throughput screening experiment.
Scheme 3. Electrophiles used in the high throughput screening experiment.
4
ACS Paragon Plus Environment

Page 5 of 23
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
The desired product from reductive amination is a single addition of the electrophile to the
amine nucleophile; however, intermediates and byproducts are also possible. Figure 1 contains
mass spectra with examples of the various outcomes that are possible for the reactions in this study.
These spectra highlight the advantage of acquiring full scan mass spectra. Any component of the
reaction mixture that can be ionized by electrospray ionization (starting material, product,
intermediate, or byproduct) can be detected, providing a complete picture of reaction progress,
while more detailed information on any component can be acquired by subsequently recording
MS/MS spectra. These representative full scan mass spectra are simple, showing single or double
amination products and, in Fig. 1 C, the unreduced intermediate imine, which has special stability
associated with the aniline substitution pattern.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. Example mass spectra from DESI-MS showing the various types of products,
byproducts, and intermediates monitored in this study. A) Reaction between A1 and B21
showing the final, reduced product. B) Reaction between A4 and B6 showing the product of two
additions of the amine to the dielectrophile as well as the reduced, single addition product. C)
Reaction between A3 and B3 showing the imine intermediate. D) Reaction between A6 and B11
showing two additions of the electrophile to the amine.
5
ACS Paragon Plus Environment

Organic Process Research & Development
Page 6 of 23
1
2
3
4
5
6
7
8
9
Full scan spectra provide an overview of reaction progress; however, more information
may be necessary to confirm the identity of individual peaks within a spectrum. We performed
MS/MS for all expected products. Figure 2 provides an example of how MS/MS was used to
confirm the identity of the detected products. MS/MS spectra could also be used to determine the
identity of unexpected byproducts and other unknown peaks in the mass spectra.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. MS/MS spectrum of the product of the reaction between A4 and B1. The precursor ion
selected was m/z 221, and a normalized collision energy of 30 was used.
Amine and Electrophile Reactivity Trends
We chose S/N>3 as representative of the presence of product signal; this a standard choice
when making qualitative measurements in analytical chemistry. It is important to note that the very
small sample sizes and high data acquisition speeds used put quantitative measurements out of
reach, even if one ignores the problems associated with differences in ionization efficiency.
Another practical question concerns the possibility of precipitation of product upon addition of
reagents in the sample well. This occurred in some cases, but the pin tool picks up solid as well
as solution and a successful reaction can therefore be recorded. Note that in such cases accelerated
reactions in the microdroplets are not needed to assess product formation. Samples were pinned in
quadruplicate to minimize the chance of pinning error or differences in volume. These replicates
agreed qualitatively but, as already noted, the small sample amounts (µg) restrict that experiments
to providing qualitative information.
In total, 3,840 unique reactions were analyzed in this study. Figure 3 shows an overall
summary of the DESI-MS results for these reactions broken down by amine and electrophile. The
aliphatic primary amines (cyclohexylamine, benzylamine, and butylamine) have the highest
6
ACS Paragon Plus Environment

Page 7 of 23
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
number of successful reactions (S/N > 3) and in general higher signal to noise values, indicating
greater product formation. The aliphatic secondary amines (piperidine and N-benzylmethylamine)
also have relatively high signal to noise values, but there are fewer successful reactions overall.
The aromatic amines (aniline and 4-methoxyaniline) have few successful reactions and lower
signal to noise values. Benzimidazole, an aromatic heterocycle, is the least reactive amine having
the fewest number of successful reactions. In reductive amination reactions, aliphatic primary
amines are expected to be the most reactive nucleophiles followed by aliphatic secondary amines
with aromatic amines being the least reactive. These expected trends are well represented in the
data, indicating that the results generated by the system accurately model reaction performance.
When examining the electrophile trends in Figure 3, one can immediately see that the
aldehydes (B1-B12) produce many more successful reactions than the ketones (B13-B24). This is
expected as aldehydes are much more electrophilic than ketones. Among the aldehydes, reagents
with strong electron donating groups, such a 4-(dimethylamino)benzaldehyde (B8), have lower
signal to noise ratios because the increased electron density at the aldehyde carbonyl discourages
nucleophilic attack by the amine. By contrast, aldehydes with strong electron withdrawing groups,
such 4-nitrobenzaldehyde (B1), which favor nucleophilic attack, have high signal to noise ratios.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
7
ACS Paragon Plus Environment

Organic Process Research & Development
Page 8 of 23
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. The DESI-MS signal to noise ratio of the final, reduced product as a function of amine
and electrophile. Each subplot represents a different amine, and the electrophiles (B1-B24) are
represented on the x-axis. The dashed line represents a signal to noise of 3:1.
Effect of Reducing Agent, Reaction Time, Acid and Stoichiometry
Reductive amination reactions have several adjustable variables that can affect the reaction
outcome. The variables explored in this study include the choice of reducing agent (NaBH(OAc)3,
NaBH4, or NaCNBH3), the reaction time (0 or 42 hours), the presence or absence of a
8
ACS Paragon Plus Environment

Page 9 of 23
Organic Process Research & Development
1
2
3
4
5
6
stoichiometric amount of acetic acid, and the stoichiometry of the amine. Time 0 means that there
is no delay in time between reagent mixing, pinning and DESI-MS screening.
Figure 4 shows how these variables impacted the number of successful reactions (S/N > 3)
for the final, reduced product. NaCNBH3 proved to be the least effective reducing agent for the
aldehyde reactions but the most effective for the ketone reactions while NaBH4 and NaBH(OAc)3
showed similar efficacy for both aldehydes and ketones. NaCNBH3 and NaBH(OAc)3 are
commonly used in reductive amination reactions; however, NaBH4 is not commonly used because
it is strong enough to reduce the aldehyde or ketone starting material. This makes the fact that
NaBH4 and NaBH(OAc)3 showed similar efficacy surprising. This result highlights an advantage
of high-throughput experimentation. Because the time and cost of running each individual reaction
is reduced, unusual conditions, such as using NaBH4 for reductive amination reactions, can be
explored readily.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. Effect of reducing agent, reaction time, addition of acetic acid, and stoichiometry on
the number of successful reactions. A successful reaction is defined as having a signal to noise
ration greater than 3:1 for the final, reduced product.
Increasing the reaction time to 42 hours resulted in a minor increase in the number of
successful reactions for the ketones but made little difference in the aldehyde reactions. The
aldehyde reactions are apparently fast enough to produce detectable amounts of product with no
9
ACS Paragon Plus Environment

Organic Process Research & Development
Page 10 of 23
1
2
3
4
5
6
7
8
9
incubation period; however, one would expect that the increased reaction time would increase the
amount of product formed in the slower ketone reactions. Figure 5 shows that imine formation did
increase significantly over time, indicating that the increased reaction time did encourage imine
formation.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. Effect of reaction time on the number of successful intermediate reactions. A
successful intermediate reaction is defined as having a signal to noise ratio greater than 3:1 for
the imine intermediate.
The presence or absence of acetic acid had little effect on the outcome of the reactions. The
results of the NaCNBH3 reactions are not factored into these numbers because there was no “With
Acid” condition. Acid is not typically added to reductive amination reactions involving aldehydes
because it could potentially accelerate the reduction of the aldehyde. Acetic acid is commonly
added in a stoichiometric amount to reactions involving ketones when NaBH(OAc)3 is used. The
acid is added to accelerate the imine formation, which is slower with ketones. One of the
limitations in maintaining the high throughput of DESI-MS screening, in its present state of
development, is that only a single reaction step can be performed. Reductive amination involves
two separate steps and the reducing agent is normally added after the imine formation. Our
imperative to perform the screening as fast as possible means that reducing agent was present in
the original reaction mixture. This factor too might have contributed to errors in prediction of bulk
reactivity.
Using a stoichiometry of 3:1 (amine:electrophile) vs. 1:1 resulted in a slight increase in the
number of successful aldehyde reactions and a fairly significant increase in the number of
10
ACS Paragon Plus Environment

Page 11 of 23
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
successful ketone reactions. Reductive amination reactions are often carried out with an excess of
the amine to favor the formation of the imine intermediate and to reduce byproduct formation (two
additions of the electrophile to the amine). It appears that the slower ketone reactions did benefit
from the favored imine formation provided by an excess of the amine.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Validation of DESI-MS Results using LC-MS and Flow Injection
Each of the ten 384 well plates was analyzed by DESI-MS, LC-MS and flow injection MS.
The LC-MS and flow injection experiments were carried out in order to validate the results of the
DESI-MS experiments. LC-MS is a technique that is widely used throughout industry to evaluate
reaction mixtures. Flow injection is a simpler MS experiment where no chromatography is
involved, so all of the reaction components are ionized together and appear in the same spectra,
much like the DESI-MS spectra. Since the use of DESI-MS for reaction screening is a new concept,
a comparison with more traditional techniques like LC-MS and flow injection lends validity to the
use of DESI-MS as tool for reaction screening.
Heat maps showing results from each of the three analytical techniques for one of the ten
384 well plates are show in Figure 6. The corresponding heat maps for the other nine well plates
are show in Figures S3-S11. One can see that there is significant agreement among the experiments.
All three techniques reveal that the ketones are much less reactive than the aldehydes. Aniline (A2),
4-methoxyaniline (A3), and benzimidazole (A8) are identified as being the least reactive amines,
and 4-(dimethylamino)benzaldehyde (B8) is shown to be the least reactive aldehyde in all three
experiments.
A description of the true positive and false positive and the false negative hits rates
achieved by DESI is made by explicit comparisons to LCMS data that are presented in Figure 7.
with more detailed discussions in SI (Sec. 3). Briefly, for 273 (71%) reductive amination reactions
DESI-MS and LC-MS provided the same outcome (YES/YES or NO/NO). For 58 reactions (15%)
the expected product m/z was observed by DESI-MS above the ion count threshold, but it was not
observed by LC-MS (YES/NO outcome). For 53 reactions (14%) the expected reaction product
was not observed by DESI-MS, but it was present at the LC-MS data (NO/YES). The DESI false
positives (15% of the reactions studied) may be due to droplet acceleration which is not possible
in the LC-MS experiment. At least some of DESI-MS false negatives (14% of the reactions studied)
may be related to the relatively low sensitivity of the mass spectrometer used for the DESI-MS
11
ACS Paragon Plus Environment

Organic Process Research & Development
Page 12 of 23
1
2
3
4
5
6
analysis (linear ion trap mass spectrometer) compared to the high sensitivity instrument used for
LC-MS analysis (triple quadrupole).
The major difference among the three experiments is that LC-MS shows higher peak
intensities for the products than DESI-MS or flow injection. This is expected as each component
of the reaction is separated on the column prior to ionization in LC-MS. This results in less
suppression of the product ionization by other components in the reaction mixture. DESI-MS and
flow injection do not have any separation prior to ionization, so the product is subject to ion
suppression.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
12
ACS Paragon Plus Environment

Page 13 of 23
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 6. Heat maps comparing the peak height of the final, reduced product analyzed by DESI-
MS (Top), flow injection (Middle), and LC-MS (Bottom). Dark red indicates the areas of highest
intensity, and dark blue indicates the areas of lowest intensity. A1-A8 are the amines, and B1-
B24 are the electrophiles. 3:1 and 1:1 represent the amine:electrophile stoichiometry. This data
was obtained from the NaBH4, 0 h, without acid 384 well plate.
13
ACS Paragon Plus Environment

Organic Process Research & Development
Page 14 of 23
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 7. Plot showing the log ion counts for DESI-MS (x axis) and for LC-MS (y axis) for 384
reductive aminations obtained from the NaBH4, 0 h, without acid 384 well plate. Using ion
count thresholds for each of the methods, four quadrants have been used to test correlations.
Complete agreement between DESI-MS and LC-MS (YES/YES or NO/NO) is seen in 71%
(273/384) of the cases while DESI-MS false positives occurred in 15% and false negatives in
14%.
Advantages of DESI-MS as a Reaction Screening Tool
The most obvious advantage of using DESI-MS for reaction screening is its extremely
short analysis time of 1 s per sample. Table 1 compares the time it takes to screen one 384 well
plate of reactions mixtures using DESI-MS, flow injection, and LC-MS. A 384 well plate can be
analyzed in 7 minutes using DESI-MS while the same plate takes 33 hours using LC-MS. This is
based on an LC-MS cycle time of 5.1 minutes, which is relatively short compared to many
gradients which often take 10 or 20 minutes. LC-MS cycle times can be shortened by using shorter
columns with smaller particle sizes and higher flowrates, but even with a cycle time of 1.4 minutes,
the same as that for flow injection, a 384 well plate still takes 9 hours to analyze.
14
ACS Paragon Plus Environment

Page 15 of 23
Organic Process Research & Development
1
2
3
4
5
Table 1. Comparison of the analysis time, solvent consumption, and sample consumption for
each of the analysis techniques used in this study.
6
7
8
9
Analysis Time
(per 384 well plate)
7 minutes
Solvent Consumption
Sample Consumption
(per 384 well plate)
DESI-MS
Flow Injection
LC-MS
20 uL
54 mL
2 L
50 nL
2 uL
2 uL
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
9 hours
33 hours
Table 1 also highlights the low solvent consumption of DESI-MS. Only 20 L of DESI
spray solvent is required to screen one 384 well plate. LC-MS requires 2 L of mobile phase (1
mL/min flowrate, 5.1 minute cycle time) to analyze the same plate, thus DESI-MS solvent
consumption is five orders of magnitude lower than LC-MS, thereby reducing costs and producing
less chemical waste.
DESI-MS also has an advantage over LC-MS and flow injection in the amount of sample
required. The reagents used in reaction screening are often very expensive and difficult to obtain
in large amounts, so making reaction mixtures in small volumes lowers the costs associated with
reaction screening. Since DESI-MS requires only 50 nL of sample (ca. 1 µg at 150 mM), reaction
mixtures can be made in extremely small volumes (<10 µL) while still enabling repeated sampling.
Experimental
All chemicals and reagents were purchased from Sigma-Aldrich (St Louis, Missouri) and
used without any purification.
Reaction Mixture Preparation
All reaction mixtures were prepared in 384 well microtiter plates (Analytical Sales &
Services, max volume: 120 µL) using a Biomek i7 liquid handling robot (Beckman Coulter). Stock
solutions of each amine (450 mM), electrophile (150 mM), NaBH4 (150 mM), NaCNBH3 (150
mM), NaBH(OAc)3 (225 mM), and acetic acid (500 mM) were prepared in methanol. The stock
solutions were added by the Biomek i7 to the 384 well plate using the pattern shown in Figure S2.
For the 3:1 (amine:electrophile) stoichiometry reactions, 30 µL of each amine stock solution was
added to the appropriate wells. For the 1:1 stoichiometry reactions, 20 µL of methanol followed
15
ACS Paragon Plus Environment

Organic Process Research & Development
Page 16 of 23
1
2
3
4
5
6
7
8
9
by 10 µL of each amine stock solution and was added the wells. Next, 30 µL of each electrophile
stock solution was added to the plate. Using the 384 multichannel head of the Biomek i7, 30 µL
of the reducing agent stock solution was added to all wells of the plate at once. Each reducing
agent stock solution was prepared immediately before the addition step to reduce degradation. The
plate was then mixed by aspirating and dispensing within the well plate several times, and then 45
µL of each well was transferred to a new 384 well plate. Using the 384 multichannel head of the
Biomek i7, 5 µL of the acetic acid stock solution was added to this new plate.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
After the plate preparation procedure described above, which takes a total of 30 min, two
384 well plates are generated. Each plate contains 384 unique reaction mixtures with a final
concentration of 50 mM of each electrophile, 50 mM (1:1 stoichiometry) or 150 mM (3:1
stoichiometry) of each amine, and 50 mM (NaBH4 and NaCNBH3) or 75 mM (NaBH(OAc)3) for
the reducing agent. The difference between the two plates is that one plate also contains one
equivalent (50 mM) of acetic acid. This procedure was performed for each reducing agent
generating a total of five 384 well plates (1,920 unique reactions). Only five well plates were
generated because the addition of acetic acid was not tested for the NaCNBH3 reactions for fear of
releasing hydrogen cyanide.
After the pinning procedure described below, each of the five 384 well plates was either
analyzed immediately or sealed with a foil seal (AlumaSeal II, EXCEL Scientific) and allowed to
sit at room temperature for 42 hours. This was done to evaluate if additional reaction time would
promote product formation for the slower reactions, particularly the ones involving the ketones.
Counting reaction time as an additional variable, a total of ten unique 384 well plates (3,840
reactions) were evaluated in this study (Figure S1).
DESI Slide Preparation
To enable analysis by DESI, the reaction mixtures were transferred from the 384 well plate
to a flat surface. This surface, called a DESI slide, consists of a PTFE membrane (Zitex G-115,
Saint-Gobain Performance Plastics) glued via spray adhesive to a glass support. The glass support
is custom fabricated (Abrisa Technologies) to have the same footprint (5.030 in long, 3.365 in
wide, 0.08 in thick) as a standard microtiter plate so that it can fit correctly on the deck of the
Biomek i7.
16
ACS Paragon Plus Environment

Page 17 of 23
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
The reaction mixtures were transferred from the 384 well plate to the DESI slide using a pin
tool, which consists of an array of 384 stainless steel, slotted pins (FP3NS50, V & P Scientific)
that are machined to retain 50 nL of liquid after being dipped into a solution. The pins ride in a
fixture (AFIX384FP3, V & P Scientific) that attaches to the 384 multichannel head of the Biomek
i7. The pins were dipped into the 384 well plate containing the reaction mixtures, which “aspirates”
50 nL of each reaction mixture, and then the pins are touched to the surface of the DESI slide
transferring 50 nL of each reaction mixture to the surface. One transfer using the pin tool creates
384 reaction mixture “spots” on the DESI slide, each of which is approximately 1 mm in diameter.
Additional transfers can be made to the same DESI slide by slightly offsetting the location of the
pins relative to the DESI slide during each transfer. In this study, each 384 well plate was pinned
four times onto the DESI slide generating four replicate spots of each reaction mixture. An
additional copy of each DESI slide was also made for MS/MS analysis.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
DESI-MS Conditions
DESI-MS data was collected using a Prosolia DESI 2D stage connected to a Thermo LTQ
XL mass spectrometer. The DESI solvent (0.1 % formic acid in methanol) was supplied at a
flowrate of 2.75 µL/min, and nitrogen was used as the DESI nebulizing gas at a pressure of 150
psi. Full scan mass spectra were collected over the m/z range of 50-700 in positive ion mode with
a spray voltage of 5 kV. The automatic gain control (AGC) of the LTQ XL was turned on with a
maximum injection time of 250 ms.
The DESI stage and the mass spectrometer were controlled using a custom software suite
called CHRIS (CHemical Reaction Integrated Screening). This software will be discussed in
greater detail in a future publication, but briefly, it allows spot-to-spot acquisition of the data rather
than a traditional DESI imaging approach which acquires data from the entire surface of the DESI
slide. With an analysis time of ~1 sec per spot, 384 reaction mixtures can be analyzed in under
seven minutes. In this study, data was acquired from each of the four replicates of the 384 reaction
mixtures and from 384 “blank” spots, which were areas of the DESI slide that did not contain any
reaction mixtures. These blank spots were used during the data analysis to calculate signal to noise
ratios. In total, 1,920 spots were analyzed per DESI slide resulting in an analysis time of 32 min
per slide.
17
ACS Paragon Plus Environment

Organic Process Research & Development
Page 18 of 23
1
2
3
4
Dilution of the 384 well plates for LC-MS and Flow Injection
5
6
7
8
After the pinning procedure, each 384 well plate was diluted 1000x in preparation for the
LC-MS and flow injection experiments described below. This dilution was performed using the
Biomek i7. In the first step of the dilution, 2 µL of each reaction mixture was transferred to a
second 384 well plate (Analytical Sales & Services, max volume: 225 µL) containing 178 µL of
methanol. After mixing, 5 µL from the second plate was transferred to a final plate containing 50
µL of water. After mixing, the final plate was sealed with a pre-scored, silicone cap mat to prevent
evaporation. Note that the amounts of analyte used in LC-MS and flow injection experiments were
the same as those used in DESI-MS.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
LC-MS Conditions
LC-MS data was collected using a Dionex UltiMate 3000 UHPLC system (WPS-3000
autosampler, HPG-3200SD binary pump, TCC-3000 column oven, and SRD-3200 degasser)
connected to a Thermo TSQ Quantum Access Max mass spectrometer. The column used was an
XBridge BEH C18 with a particle size of 2.5 µm, and the column was held a 45 °C throughout.
Moblie phase A was 0.1% formic acid in water, and mobile phase B was acetonitrile. A flowrate
of 1 mL/min was used throughout, and the gradient was as follows: 5% B at 0.0 min; linear
transition to 95% B until 2.0 min; held at 95% B until 2.5 min; linear transition back to 5% B until
3.0 min; held at 5% B until 3.75 min for equilibration. The needle wash solution was
methanol:water (1:1). An injection volume of 5 µL was used, and the injection to injection cycle
time for each sample was 5.1 minutes.
Full scan mass spectra were collected over the m/z range of 50-700 in positive ion mode with
a scan time of 0.200 s. The source conditions were as follows: spray voltage: 5 kV; vaporizer
temperature: 500 °C; sheath gas pressure: 60 (arbitrary units); ion sweep gas pressure: 1.5
(arbitrary units); aux gas pressure: 20 (arbitrary units); capillary temperature: 250 °C.
Flow Injection Conditions
The flow injections were performed using a Dionex UltiMate WPS-3000 autosampler and
ISO-3100SD isocratic pump connected to a Thermo TSQ Quantum Access Max mass
spectrometer. A flowrate of 100 µL/min of methanol was used to carry each injection (5 µL)
18
ACS Paragon Plus Environment

Page 19 of 23
Organic Process Research & Development
1
2
3
4
5
6
directly to the mass spectrometer. The needle wash solution was methanol:water (1:1), and the
injection to injection cycle time for each sample was 1.4 minutes.
Full scan mass spectra were collected over the m/z range of 50-700 in positive ion mode with
a scan time of 0.500 s. The source conditions were as follows: spray voltage: 5 kV; vaporizer
temperature: 150 °C; sheath gas pressure: 20 (arbitrary units); ion sweep gas pressure: 1.0
(arbitrary units); aux gas pressure: 5 (arbitrary units); capillary temperature: 250 °C.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
DESI-MS/MS Data Acquisition
DESI-MS/MS data was acquired for products which had a signal to noise ratio of at least
3:1. This data was acquired from an identical copy of each DESI slide using the CHRIS software
suite. During the full scan data acquisition, the CHRIS software records the XY coordinates of
each spot on the DESI slide. This allows the user to return to that same location on a copy of the
original slide to acquire MS/MS data. A copy was used rather than returning to the original DESI
slide position in these experiments because reaction products are removed by the DESI spray
during the original acquisition leaving less material for MS/MS analysis. CHRIS software
automatically loads the appropriate instrument method as it transitions from spot to spot based on
a user provided a list of m/z values and their corresponding XY coordinates.
The MS/MS acquisition took approximately 30 s per spot because product ion scans were
acquired at three different relative collision energies (20 first, then 10, and finally 30) in three
consecutive 10 s long segments to get a broad view of the fragmentation behavior of each
molecule. Except for the scan type, the DESI and mass spectrometric conditions used for the
MS/MS acquisition were the same as those used for the full scan acquisition.
CHRIS Data Analysis
The output of each full scan DESI experiment was a single RAW file containing the data for
all 1,920 spots (384 x 4 reaction spots and 384 blank spots). In order to correlate the mass spectra
in each RAW file with the corresponding reaction information, the CHRIS data analysis software
was used. This software takes two inputs: the raw data and a plate layout file. The plate layout file
is an Excel file provided by the user which contains the 384 well plate layout (Figure S1), all m/z
values of interest (starting materials, products, intermediates, byproducts, etc.) and the positions
on the DESI slide where the 384 well plate was pinned.
19
ACS Paragon Plus Environment

Organic Process Research & Development
Page 20 of 23
1
2
3
4
5
6
7
8
9
During the full scan data acquisition, the CHRIS software records the start and stop times
for the acquisition of each spot. This allows CHRIS to correlate the mass spectra in the RAW file
with each spot on the DESI slide. Using the information in the plate layout file, CHRIS then
correlates each spot with the corresponding reaction conditions and m/z values. The output of the
CHRIS data analysis software is a set of csv files which contain the intensity for each m/z value
for each replicate spot of each unique reaction condition.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Conclusions
In this study, a total of 3,840 unique reductive amination reactions were screened using a
DESI-MS based high throughput reaction screening system. Using this system, 1,920 unique
reactions were screened in a single day starting from commodity chemicals. Unusual reaction
conditions were explored, such as using NaBH4 as a reducing agent, which proved to be effective
in producing product. HTE enables the evaluation of these high-risk reaction conditions because
the costs associated with performing each individual reaction are significantly reduced.
In addition to collecting DESI-MS data, all reactions were analyzed using traditional MS
techniques such as LC-MS and flow injection. The results of the DESI-MS correlated well with
the results from these more traditional techniques validating the effectiveness DESI-MS for
reaction screening. MS/MS spectra also were collected to confirm the identity of the peaks detected
in the full scan spectra.
A custom-built software suite was developed to control both the DESI stage and mass
spectrometer allowing for the acquisition of both full scan and MS/MS data. This software also
processes the raw data from each experiment and generates csv files from which the user can
access the reaction data.
This system represents a breakthrough in organic reaction screening. With an analysis time
of only 1 second per sample, a solvent consumption 20 µL per 384 well plate, and a sample
consumption of only 50 nL, this system allows for the cost-effective screening of thousands of
organic reactions per day.
Acknowledgements
The authors gratefully acknowledge the support of DARPA Award #W911NF-16-2-0020.
Yangjie Li acknowledges support by the Henry Bohn Hass Memorial Fellowship.
20
ACS Paragon Plus Environment

Page 21 of 23
Organic Process Research & Development
1
2
3
4
Supporting Information
5
6
7
8
Section S1
Section S2
Section S3
Experimental Design and Layout of the 384 Well Plates
Additional Heat Maps Comparing DESI, Flow Injection, and LC-MS
Correlation of LC-MS and DESI for Reductive Amination
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
References
1.
Fan, J.-B.; Yeakley, J. M.; Bibikova, M.; Chudin, E.; Wickham, E.; Chen, J.; Doucet, D.;
Rigault, P.; Zhang, B.; Shen, R.; McBride, C.; Li, H.-R.; Fu, X.-D.; Oliphant, A.; Barker,
D. L.; Chee, M. S. A Versatile Assay for High-Throughput Gene Expression Profiling on
Universal Array Matrices. Genome Res. 2004, 14, 878-885.
2.
3.
4.
Saleski, T. E.; Kerner, A. R.; Chung, M. T.; Jackman, C. M.; Khasbaatar, A.; Kurabayashi,
K.; Lin, X. N. Syntrophic co-culture amplification of production phenotype for
highthroughput screening of microbial strain libraries. Metab. Eng. 2019, 54, 232-243.
D’Agostino, V. G.; Sighel, D.; Zucal, C.; Bonomo, I.; Micaelli, M.; Lolli, G.; Provenzani,
A.; Quattrone, A.; Adami, V. Screening Approaches for Targeting Ribonucleoprotein
Complexes: A New Dimension for Drug Discovery. SLAS Discovery 2019, 24(3), 314-331.
Santanilla, A. B.; Regalado, E. L.; Pereira, T.; Shevlin, M.; Bateman, K.; Campeau, L.-C.;
Schneeweis, J.; Berritt, S.; Shi, Z.-C.; Nantermet, P.; Liu, Y.; Helmy, R.; Welch, C. J.;
Vachal, P.; Davies, I. W.; Cernak, T.; Dreher, S. D. Nanomole-scale high-throughput
chemistry for the synthesis of complex molecules. Science 2015, 347, 49-53.
Troshin, K.; Hartwig, J. F. Snap deconvolution: An informatics approach to high-
throughput discovery of catalytic reactions. Science, 2017, 357, 175-181.
5.
6.
Perera, D.; Tucker, J. W.; Brahmbhatt, S.; Helal, C. J.; Chong, A.; Farrell, W.; Richardson,
P.; Sach, N. W. A platform for automated nanomolescale reaction screening and
micromole-scale synthesis in flow. Science, 2018, 359, 429-434.
7.
8.
Wang, Y.; Shaabani, S.; Ahmadianmoghaddam, M.; Gao, L.; Xu, R.; Kurpiewska, K.;
Kalinowska-Tluscik, J.; Olechno, J.; Ellson, R.; Kossenjans, M.; Helan, V.; Groves, M.;
Dömling, A. Acoustic Droplet Ejection Enabled Automated Reaction Scouting. ACS Cent.
Sci. 2019, 5, 451-457.
Mennen, S. M.; Alhambra, C.; Allen, C. L.; Barberis, M.; Berritt, S.; Brandt, T. A.;
Campbell, A. D.; Castañón, J.; Cherney, A. H.; Christensen, M.; Damon, D. B.; Eugenio
de Diego, J.; García-Cerrada, S.; García-Losada, P.; Haro, R.; Janey, J.; Leitch, D. C.; Li,
L.; Liu, F.; Lobben, P. C.; MacMillan, D. W. C.; Magano, J.; McInturff, E.; Monfette, S.;
Post, R. J.; Schultz, D.; Sitter, B. J.; Stevens, J. M.; Strambeanu, I. I.; Twilton, J.; Wang,
K.; Zajac, M. A. The Evolution of High-Throughput Experimentation in Pharmaceutical
Development and Perspectives on the Future. Org. Process Res. Dev. 2019, 23, 1213-1242.
Jaman, Z.; Mufti, A.; Sah, S.; Avramova, L.; Thompson, D. H. High Throughput
Experimentation and Continuous Flow Validation of Suzuki–Miyaura Cross-Coupling
Reactions. Chem. Eur. J. 2018, 24, 9546-9554.
9.
10.
Szeto, J.; Vu, V.-A.; Malerich, J. P.; Collins, N. Multi-step continuous flow synthesis of
fluconazole. J. Flow Chem. 2019, 9, 35–42.
21
ACS Paragon Plus Environment

Organic Process Research & Development
Page 22 of 23
1
2
3
4
5
6
7
8
9
11.
12.
Alfonzo, E.; Beeler, A. B. A sterically encumbered photoredox catalyst enables the unified
synthesis of the classical lignan family of natural products. Chem. Sci. 2019, 10, 7746–
7754.
Wleklinski, M.; Loren, B. P.; Ferreira, C. R.; Jaman, Z.; Avramova, L.; Sobreira, T. J. P.;
Thompson, D. H.; Cooks, R. G. High throughput reaction screening using desorption
electrospray ionization mass spectrometry. Chem. Sci. 2018, 9, 1647-1653.
Takáts, Z.; Wiseman, J. M.; Gologan, B.; Cooks, R. G. Mass Spectrometry Sampling Under
Ambient Conditions with Desorption Electrospray Ionization. Science 2004, 306, 471-473.
Cooks, R. G.; Ouyang, Z.; Takats, Z.; Wiseman, J. M. Ambient Mass Spectrometry.
Science 2006, 311, 1566-1570.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
13.
14.
15.
Müller, T.; Badu-Tawiah, A.; Cooks, R. G. Accelerated carbon-carbon bond-forming
reactions in preparative electrospray. Angew. Chem. Int. Ed. 2012, 51, 11832–11835;
Angew. Chem. 2012, 124, 12002–12005.
16.
Sawicki, J. W.; Bogdan, A. R.; Searle, P. A.; Talaty, N.; Djuric, S. W. Rapid analytical
characterization of highthroughput chemistry screens utilizing desorption electrospray
ionization mass spectrometry. React. Chem. Eng. 2019, 4, 1589-1594.
Roughley, S. D.; Jordan, A. M. The Medicinal Chemist’s Toolbox: An Analysis of
Reactions Used in the Pursuit of Drug Candidates. J. Med. Chem. 2011, 54, 3451-3479.
Abdel-Magid, A. F.; Mehrman, S. J. A Review on the Use of Sodium
Triacetoxyborohydride in the Reductive Amination of Ketones and Aldehydes. Org.
Process Res. Dev. 2006, 10, 971-1031.
17.
18.
22
ACS Paragon Plus Environment

Page 23 of 23
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table-of-Content Figure
26x15mm (600 x 600 DPI)
ACS Paragon Plus Environment