A labeled substrate approach to discovery of biocatalytic reactions: A proof of concept transformation with N-methylindole

Article abstract of DOI:10.1021/ja304767m

Biocatalysis has become an important method in the pharmaceutical industry for the incorporation of new functionality in small molecules. Currently this method is limited in the types of reactions that can be carried out and no strategy exists to systematically screen for new biocatalyzed reactions. This study involves the development of a medium throughput screen to identify and optimize new reactions using a series of marine-derived bacterial cell lines, which were screened against several ¹³C labeled organic substrates. The reactions were analyzed using ¹³C NMR as the primary screening tool. We describe the discovery of a bacterial catalyzed indole oxidation reaction in which complete conversion of ¹³C labeled N-methyl indole to 3-hydroxyindole was observed. In addition, the sensitivity of this reaction to dO₂ levels can be exploited to oxidize to either 3-hydroxyindole or 2-oxoindole. This new platform sets up an important tool for the discovery of new organic transformations using an extensive library of marine bacteria.

Full text of DOI:10.1021/ja304767m

Communication
pubs.acs.org/JACS
A Labeled Substrate Approach to Discovery of Biocatalytic Reactions:
A Proof of Concept Transformation with N-Methylindole
Jamie L. Rogers and John B. MacMillan*
Department of Biochemistry, The University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Boulevard, Dallas,
Texas 75390-9038, United States
S
* Supporting Information
ABSTRACT: Biocatalysis has become an important
method in the pharmaceutical industry for the incorpo-
ration of new functionality in small molecules. Currently
this method is limited in the types of reactions that can be
carried out and no strategy exists to systematically screen
for new biocatalyzed reactions. This study involves the
development of a medium throughput screen to identify
and optimize new reactions using a series of marine-
derived bacterial cell lines, which were screened against
several 13C labeled organic substrates. The reactions were
analyzed using 13C NMR as the primary screening tool.
We describe the discovery of a bacterial catalyzed indole
oxidation reaction in which complete conversion of 13C
labeled N-methyl indole to 3-hydroxyindole was observed.
In addition, the sensitivity of this reaction to dO2 levels can
be exploited to oxidize to either 3-hydroxyindole or 2-
oxoindole. This new platform sets up an important tool for
the discovery of new organic transformations using an
extensive library of marine bacteria.
Figure 1. Pharmaceuticals produced by biocatalysis.
number of commercially available synthetically useful enzymes
and limited development speed. Discovery methods that can
tap into the full potential of enzyme catalyzed reactions, such as
carbon bond-forming reactions (C−C, C−N, C−O, C−X) will
expand the toolbox for this growing field.
We have developed a library of marine-derived bacteria
(Actinomycetes, Bacilli, α-proteobacteria) that has been utilized
in a number of drug discovery efforts.6 We felt the library
presented a unique opportunity to develop a platform for the
exploration of new, whole-cell biocatalyzed organic trans-
formations. By screening various substrates in a range of
fermentation conditions we could search for interesting whole-
cell biotransformations. There are many examples of screening
microbial libraries for specific reactions, such as whole-cell
biocatalyzed Baeyer−Villiger reactions, but we are unaware of
unbiased approaches to biocatalysis discovery.7 We desired a
method to discover new or unexpected reactivity in an easy to
detect method, such as that developed by the Liu lab on DNA-
templated synthesis and in vitro selection to rapidly evaluate
combinations of substrates for bond-forming reactions.8
Additional approaches such as “accelerated serendipity” by
the lab of David MacMillan for the development of new metal
catalyzed reactions9 and an innovative MS screening approach
by Hartwig and co-workers have enhanced reaction discovery.10
The development of a robust, medium throughput method
to accomplish this goal posed some unique challenges. Because
we are using whole-cell fermentation with natural product
producing microorganisms, there is significant background
metabolite production, which interferes with analysis of
reactions.11 We desired a method to analyze a significant
number (>100) of transformations and look specifically for
changes to the substrate of choice. To this end, we selected
substrates 1−5 for initial evaluation of the biotransformation
iocatalysis has become an important component in the
Btoolbox of the pharmaceutical and fine chemical industries,
being used as reagents in multistep synthetic processes.1 The
biggest role for biocatalysis in the pharmaceutical sector lies in
the chemo-, regio-, and stereoselectivity properties that
biological systems can achieve.2 Additionally biocatalyzed
reactions can be carried out in H2O, without the use of
protecting groups and under mild temperatures. Examples of
processes that utilize biocatalysis in the manufacture of
pharmaceuticals include Crixivan, where Rhodococcus sp. is
used to perform a stereoselective epoxidation in the first step of
the synthesis.3 A second generation process route to Lyrica
includes the use of a lipolase catalyzed resolution to generate a
key (S)-mono acid enantiomer from a racemic cyanodiester.4
An efficient enzymatic desymmetrization route was recently
reported for the antidepression drug Paxil, where the key step
involves a protease catalyzed desymmetrization of a meso
diester, doubling the yield of the existing manufacturing process
(Figure 1).5
As the pharmaceutical industry continues to expand their
capabilities in biocatalysis, there is still a limitation in the
strategies to identify new reactions and enzymes. The majority
of biocatalyzed reactions focus on desymmetrizations and
resolutions. The more widespread application of large-scale
biocatalysis is currently restricted because of the limited
Received: May 16, 2012
Published: July 18, 2012
© 2012 American Chemical Society
12378
dx.doi.org/10.1021/ja304767m | J. Am. Chem. Soc. 2012, 134, 12378−12381

Journal of the American Chemical Society
Communication
Figure 2. (a) Approach to discovery of biotransformations with readout of 13C labeled substrates by 13C NMR. (b) Labeled substrates used in proof-
of-concept screen. (c) Matrix of bacteria and substrates used in initial screen. Boxes with gradient shading indicate bacteria/substrate combinations
that resulted in a new product.
discovery platform looking for substrates that fit the following
criteria: (1) substrates with motifs similar to those found in
natural products and/or pharmaceuticals, (2) substrates
wherein interesting biosynthetic reactions had been observed,
and (3) the ability to incorporate a 13C isotope label (Figure
2b).
Figure 3. Biocatalysis of 1 to products 6 and 7 under different oxygen
concentrations.
The strategy that we developed utilizes the incorporation of a
13C label on the exogenous substrate allowing for use of 13C
NMR spectroscopy as a primary screening tool (Figure 2a).
Compared to the low natural abundance of 13C, the use of
mediated oxidation in our screening approach. To verify the
conversion was not a result of air oxidation or from the
labeled substrates provides a sensitive technique to distinguish
fermentation media, we carried out a series of control
products of biotransformation. As demonstrated in Figure 1a,
the 13C NMR of N-[13C]methylindole only shows a single peak,
experiments with 1 in the absence of bacteria and saw no
conversion to 6. In addition to being a common structural
which upon fermentation with the actinomycete Salinispora
motif in natural products, oxidized indole substrates are
arenicola derives a new product with an upfield chemical shift.
common in pharmaceuticals.13 Typically the synthetic route
An initial screen was performed in which five 13C labeled
to an oxidized indole is through oxidation of a simpler substrate
substrates were added individually to seven unique strains of
followed by ring closing to get the indole. Direct oxidation on
bacteria (Figure 2c). The reactions were monitored over the
indole may allow for a more efficient synthesis of these
course of several days by removing a small aliquot of the culture
molecules.14
and extracting with ethyl acetate. The extracts were analyzed via
Using the initial screening conditions, a 125 mL Erlenmeyer
13C NMR and LC-MS. Shaded boxes in Figure 1c represent
flask containing bacteria was charged with 10 mg of 1, grown
substrate/bacteria combinations where we saw conversion to
for 10 days and subsequently purified via HPLC to reveal
complete conversion to N-[13C]methyl-3-hydroxyindole, with
one or more new products. We were able to find enzymatic
conversion with four of the five substrates that were tested in
this platform. Here, we will focus on the complete description
of one particular biotransformation.
no starting material or side products recovered. To follow up
on these results, the initial conditions were scaled up into a 2.8
L Fernbach flask. After observing complete conversion of 1, the
We were particularly drawn to one biotransformation; the
conversion of N-[13C]methylindole into an oxidized product by
the bacterium Salinispora tropica, strain SNB-003. Analysis of
the product revealed oxidation to N-[13C]methyl-3-hydrox-
yindole (6) (Figure 3). This is an unusual oxidation of N-
methylindole or indole, with the typical oxidation product
under both synthetic and biocatalysis conditions being 2-
oxindole.12 Of the few reports of direct indole oxidation, they
all suffer from poor yields and complex mixtures. Oxidation at
the 3-position of an indole is a common structural feature in
natural products; as such it is not surprising to find a bacterial
resulting product was isolated and determined to be an
alternative oxidation product, N-[13C]methyl-2-oxindole (7).
The only variable in this reaction was the size of the flask and
the volume of media, which can change the amount of
dissolved oxygen (dO2) in the media (2.8 L Fernbach flasks are
designed for aeration of cultures). This leads to the possibility
that there are two enzymatic pathways at work, operating under
different oxygen levels.
To probe this hypothesis we carried out a series of whole-cell
biotransformations at various dO2 levels. By altering the media
volume in standard 500 mL Erlenmeyer flasks we were able to
12379
dx.doi.org/10.1021/ja304767m | J. Am. Chem. Soc. 2012, 134, 12378−12381

Journal of the American Chemical Society
Communication
vary the dO2 concentration in each culture and cover a wide
range of relative O2 levels from 21 to 91% (Air = 100%)
(Tables 1 and S1). We measured dO2 concentration using an
Table 1. Role of % dO2 on Conversion of 1 to 6 or 7 under
Whole-Cell Biotransformation Conditions
entry
%6
%7
% dO2 relative to air
1
0
0
100
100
100
0
91
89
81
77
73
67
61
61
58
58
50
46
44
39
21
Figure 4. Cell-free extract conversion.
2
3
0
of P450s and monooxygenases; however oxidation results in
the formation of multiple products and is driven toward
formation of indigo or indirubin.17 In the case of the whole-cell
biotransformation of 1 with S. arenicola, the oxidation can be
controlled to give 6, making this unique compared to previous
examples. We are carrying out further studies to determine the
oxidase responsible for the biotransformation.
4
100
63
100
70
80
75
100
60
5
27
0
6
7
30
0
8
9
0
10
11
12
13
14
15
0
We probed the substrate scope of the biotransformation
using a variety of substituted N-methylindole and indole
substituents (Table 1). The substrate scope is somewhat
limited at this time, with more electron-rich substrates being
readily oxidized with complete conversion to the 3-oxindole
product (entries a−d). However, the electron-poor substrates
we tested were completely unreactive and only starting material
was recovered (entries h and i). Entries f and g represent
substrates found in a number of natural products. 1,2,3,4-
Tetrahydrocarbazole (f), lacking the N-methyl, was oxidized at
the 3-position, which induced a rearrangement to the known
spirocycle 20. The product 20 has been reported from the
oxidation of tetrahydrocarbazole as early as 1951 by Witkop
and co-workers.18 A radical mechanism is proposed for this
reaction in which homolytic cleavage of the NH bond initiates
the oxidation. However, with N-methyl-1,2,3,4-tetrahydrocar-
bazole no reaction was observed and starting material was
quantitatively recovered (entry g). We believe this is related to
the fact that 21 is the least soluble substrate in Table 2 and is
unlikely to be incorporated into the cells.
The development of additional “reagents” for biocatalysis is
an important avenue of research as the pharmaceutical industry
continues to build their capabilities for manufacturing scale
biological approaches to chemistry. We have been able to use a
biodiscovery platform to develop an enzyme catalyzed,
regiospecific oxidation of a common aromatic moiety that is
present in a number of biologically active natural products.
Further efforts to isolate the enzyme responsible for the
transformation would allow directed evolution approaches to
optimize for a given substrate. Further efforts to look for
stereospecific incorporation of oxygen in 3-alkylindoles would
provide additional synthetic utility. However, this reaction
establishes a robust discovery platform for whole-cell
biocatalysis.
0
100
40
0
0
40
0
30
0
Oakton submersible dissolved oxygen probe. At dO2 levels
lower than 45% there was little conversion of 1 to either of the
oxidized indole products (Table 1, entries 13−15). In the dO2
range of 45%−80% there was nearly quantitative conversion to
the desired oxidized product 6 (Table 1, entries 4−12). There
is a dramatic switch at 80% relative dO2 where production of 6
is completely eliminated and uniform conversion to 7 is
obtained (Table 1, entries 1−3). We can conclude that to
obtain the 3-oxindole product most efficiently the oxygen level
in the fermentation flask should be maintained at a relative dO2
of 60%.
With these optimized dO2 conditions we obtained a 35%
overall isolated yield of 6. Interestingly, no other product was
observed in this reaction and no starting material was recovered
to account for the low yield. This could be due to challenges in
extraction of 6 from media, loss of the 13C label, or
incorporation of the indole into the organism. To overcome
this problem we looked at the possibility of moving the reaction
from whole-cell biotransformation to an enzymatic preparation
from the bacteria.
An enzyme preparation was performed with S. arenicola
SNB003 in which the cells were grown in A1 media for 10 days
and then lysed via repeated sonication. The large, insoluble
particles were separated from the lysate via centrifugation at
5200 rpm (15 min, 4 °C), and the presence of proteins was
established via Western blot. 1 was added to the flask
maintained at 4 °C for four days. Multiple minor products
were observed by LC-MS and 13C NMR; however the two
major products, following purification, were identified as the
desired 3-hydroxyindole 6 and the higher molecular weight
indole trimer 8 (Figure 4). Compounds 6 and 8 accounted for
the 75% yield of this reaction in a 1:4 ratio of products. 8 is a
known compound and is also structurally related to a previously
reported natural product, which is lacking the N-methyl
groups.15 The trimer has been postulated to arise from a 3-
hydroxyindole intermediate.16 When the cell-free extract
reaction is allowed to proceed for seven days, only 8 is
observed, indicating that 6 is a substrate for further reactivity.
Oxidation of indoles is a common transformation by a variety
There are two technical challenges to be considered in using
the approach: maintenance of the bacterial screening collection
and detection. Advancements in microscale fermentation (12-
well to 96-well plates)19 make it simple to generate bacterial
screening plates, and relatively little microbiology expertise is
needed to carry out a screen, through the use of cryogenically
stored screening plates. The only instrumentation needed for
these studies is a small shaker table and well-plate compatible
centrifuge. One challenge that we have encountered that has
limited throughput is the differential growth rates of the
bacteria in our collection. In order to utilize well plates
efficiently it has been important to optimize growth conditions
12380
dx.doi.org/10.1021/ja304767m | J. Am. Chem. Soc. 2012, 134, 12378−12381

Journal of the American Chemical Society
Communication
Texas Higher Education Coordinating Norman Hackermann
ARP Grant 010019-0047 and Welch Foundation Grant I-1689.
J.B.M. is a Chilton/Bell Foundation Endowed Scholar.
Table 2. Substrate Scope of Whole-Cell Indole Oxidation
REFERENCES
■
(1) Schmid, A.; Dordick, J. S.; Hauer, B.; Kiener, A.; Wubbolts, M.;
Witholt, B. Nature 2001, 409, 258. (b) Straathof, A. J.; Panke, S.;
Schmid, A. Curr. Opin. Biotechnol. 2002, 13, 548−. (c) Tao, J.; Xu, J. H.
Curr. Opin. Chem. Biol. 2009, 13, 43.
(2) Pollard, D. J.; Woodley, J. M. Trends Biotechnol. 2006, 25, 66.
(3) Buckland, B. C.; Drew, S. C.; Connors, N. C.; Chartrain, M. M.;
Lee, C.; Salmon, P. M.; Gbewonyo, K.; Zhou, W.; Gailliot, P.; Singhvi,
R.; Olewinski, R. C.; Sun, W. J.; Reddy, J.; Zhang, J.; Jackey, B. A.;
Taylor, C.; Goklen, K. E.; Junker, B.; Greasham, R. L. Metab. Eng.
1999, 1, 63.
(4) Martinez, C. A.; Hu, S.; Dumond, Y.; Tao, J.; Kelleher, P.; Tully,
L. Org. Process Res. Dev. 2008, 12, 392.
(5) Glaser, V. Gene Eng. Biotechnol. News 2008, 28, 30.
(6) Hu, Y.; Legako, A. G.; Espindola, A.-P. D. M.; MacMillan, J. B. J.
Org. Chem. 2012, 77, 3401. (b) Pan, E.; Jamison, M.; Yousufuddin, M.;
MacMillan, J. B. Org. Lett. 2012, 14, 2390. (c) Hu, Y.; MacMillan, J. B.
Org. Lett. 2011, 13, 6580. (d) Hu, Y.; Espindola, A.-P. D. M.; Steward,
N. A.; Wei, S.; Posner, B. A.; MacMillan, J. B. Bioorg. Med. Chem. 2011,
19, 5183.
(7) Alphand, V.; Furstoss, R. J. Mol. Catal. B: Enzym. 2000, 9, 209.
(8) Kanan, M. W.; Rozenman, M. M.; Sakurai, K.; Syder, T. M.; Liu,
D. R. Nature 2004, 431, 545.
(9) McNally, A.; Prier, C. K.; MacMillan, D. W. C. Science 2011, 334,
1114.
(10) Robbins, D. W.; Hartwig, J. H. Science 2011, 333, 1423.
(11) Although LC-MS profiling is an option, many low molecular
weight compounds do not ionize well and have weak UV
chromophores making detection of new molecules difficult.
(12) Medici, R.; Garaycoechea, J. I.; Dettorre, L. A.; Iribarren, A. M.;
Lewkowicz, E. S. Biotechnol. Lett. 2011, 33, 1999. (b) Trost, B. M.; Kie,
J.; Sieber, J. D. J. Am. Chem. Soc. 2011, 133, 20611. (c) Ling, K.; Li, W.;
Sayre, L. M. J. Am. Chem. Soc. 2008, 130, 933.
(13) Jimenez, J. L.; Huber, U.; Moore, R. E.; Patterson, G. M. L. J.
Nat. Prod. 1999, 62, 569. (b) Birch, A. L.; Wright, I. I. J. Chem. Soc.,
Chem Commun. 1969, 12, 644−. (c) Liu, J. R.; Jiang, Z. Y.; Wang, R.
R.; Zheng, Y. T.; Chen, J. L.; Zhang, X. M.; Ma, Y. B. Org. Lett. 2007,
9, 4127.
(14) Onyeibor, O.; Croft, S. L.; Dodson, H. I.; Feiz-Haddad, M.;
Kendrick, H.; Millington, N. J.; Parapini, S.; Phillips, R. M.; Seville, S.;
Shnyder, S. D. J. Med. Chem. 2005, 48, 2701. (b) Gaywood, A. P.;
McNab, H. Synthesis 2010, 8, 1361. (c) Azadi-Ardakani, M.; Alkhader,
M. A.; Lipplatt, J. H.; Patel, D. I.; Smalley, R. R.; Higson, S. J. Chem.
Soc., Perkin Trans. 1 1986, 6, 1107.
of each bacteria strain independently. From the analytical side,
although 13C NMR is not typically viewed as high-throughput,
NMR systems with autosamplers and 13C optimized probes
allow for hundreds of samples to be run in a 12-h period. To
enhance the throughput, we have begun looking at the
possibility of adding multiple substrates with orthogonal 13C
labeling. The continual identification of ever more interesting
enzymes in natural products biosynthesis justifies an unbiased
approach to harness microbial capabilities for synthetically
useful biotransformations.
(15) Kiraz, C. I. A.; Enge, T. J.; Jimenez, L. S. J. Org. Chem. 2004, 69,
2200.
(16) Ganachaud, C.; Garfagnoli, V.; Tron, T.; Lacazio, G.
Tetrahedron Lett. 2008, 49, 2476.
ASSOCIATED CONTENT
* Supporting Information
General procedures, chemical derivatization, compound char-
acterization, and data table. This material is available free of
charge via the Internet at http://pubs.acs.org.
■
S
(17) Ensley, B. D.; Ratzkin, B. J.; Osslund, T. D.; Simon, M. J.;
Wackett, L. P.; Gibson, D. T. Science 1983, 222, 167. (b) Murdock, D.;
Ensley, B. D.; Serdar, C.; Thalen, M. Bio/Technology 1993, 11, 381.
(c) Mermod, N.; Harayama, S.; Tinnis, K. N. Bio/Technology 1986, 4,
321. (d) Meyer, A.; Wursten, M.; Schmidt, A.; Kohler, H-P. E.;
Witholt, B. J. Biol. Chem. 2002, 277, 34161.
(18) Wiktop, B.; Patrick, J. B. J. Am. Chem. Soc. 1951, 73, 2188.
(19) (a) Ferreira-Torres, C.; Michelleti, M.; Lye, G. J. Biosyst. Eng.
2005, 28, 83. (b) Minas, W.; Bailey, J. E.; Duetz, W. Antonie
Leeuwenhoek 2000, 78, 297.
AUTHOR INFORMATION
Corresponding Author
john.macmillan@utsouthwestern.edu
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Elizabeth Orth for help in developing an approach to
screening varying dO2 concentrations for Table 1. We also
thank Joseph Ready for helpful discussions about the approach.
We acknowledge the following grants for funding this project:
12381
dx.doi.org/10.1021/ja304767m | J. Am. Chem. Soc. 2012, 134, 12378−12381