Connecting Unexplored Protein Crystal Structures to Enzymatic Function

Full text of DOI:10.1002/cctc.201200544

CHEMCATCHEM
COMMUNICATIONS
DOI: 10.1002/cctc.201200544
Connecting Unexplored Protein Crystal Structures to
Enzymatic Function
Fabian Steffen-Munsberg,[a] Clare Vickers,[a] Ahmad Thontowi,[a] Sebastian Schꢀtzle,[a]
Tony Tumlirsch,[a] Maria Svedendahl Humble,[b] Henrik Land,[c] Per Berglund,[c]
Uwe T. Bornscheuer,*[a] and Matthias Hçhne*[a]
Biocatalysis has emerged as an important alternative to tradi-
tional chemical synthesis for the preparation of fine chemi-
cals,[1] as recently demonstrated by the biocatalytic manufac-
ture of the drug sitagliptin by using an (R)-amine transaminase
(ATA) created by intensive protein engineering.[2] This process
was superior with respect to optical purity, yield, and waste
generation to the already established transition-metal-cata-
lyzed production of sitagliptin.[3] Process development, identifi-
cation, and optimization of an appropriate enzyme represent
inportant requirements to obtain a successful and efficient
enzyme-catalyzed process. Nature has a rich reservoir from
which a suitable enzyme can be found as a starting point. In
contrast to classical approaches such as screening of strain col-
lections, modern developments in the area of metagenomics
offer enormous potential to screen for activity within noncul-
turable biodiversity.[4] A major advancement was the develop-
ment of next-generation sequencing techniques, which led to
a substantial increase in the genetic information deposited in
public databases;[5] currently >20 million protein sequences
are waiting to be explored. This development also led to the
accumulation of protein sequences with unknown (or wrongly
annotated) function, and thus a large part of this rich resource
cannot be used reliably. We recently took advantage of this in-
formation and developed an in silico enzyme discovery strat-
egy and identified 17 novel and unique (R)-selective ATAs in
>5000 sequences deposited in public databases,[6] although
no gene or protein sequence was described in the literature
ahead of our work. Furthermore, we could show that these (R)-
ATAs are synthetically useful, as demonstrated recently in the
asymmetric synthesis of a set of 12 chiral amines by using 7
out of the 17 enzymes.[7]
Modern protein engineering methods[8] rely on high-quality
structural information as obtained by X-ray crystallography,
which is then the basis for focused, directed evolution to im-
prove the properties of the enzyme. Similar to developments
in the genomics area, in the last years automated high-
throughput methods were developed for crystallization and
determination of crystal structures, which has resulted in
a rapid increase in the number of solved protein structures.[9]
Still, for many enzymes no structure is available, as for exam-
ple, (R)- and (S)-selective ATAs, which have become very popu-
lar in the last years.[10] Hence, structural information is highly
desired to understand experimental results and to guide pro-
tein engineering.[11] Interestingly, a trend similar to that pointed
out above for the discovery of sequences exists: there is
a growing number of crystal structures of enzymes that were
never characterized with respect to substrate scope, enantiose-
lectivity, or reaction specificity, and their physiological func-
tions are also often unknown. Similar to the lack of experimen-
tal validation of annotated proteins in sequence databases,
crystallized proteins with unknown function are only rarely
characterized.[12] Thus, many structures remain unexplored: For
example, from 104 crystal structures of different transaminases
found in the Brookhaven protein database (PDB), 46 structures
do not have a literature citation that provides a description of
the structure or the characterization data of the protein. Con-
necting the enzymatic function to unexplored proteins in the
PDB could provide information that would be of diverse inter-
est. These include protein engineering in the context of bio-
technology or biochemical studies to explore enzyme
mechanism.
[a] F. Steffen-Munsberg, Dr. C. Vickers, A. Thontowi, Dr. S. Schꢀtzle, T. Tumlirsch,
Prof. Dr. U. T. Bornscheuer, Prof. Dr. M. Hçhne
Institute of Biochemistry
Herein we explore the crystal structures with unknown func-
tions in the cluster of “ornithine-aminotransferase (OAT)-like
proteins” (cd00610 of the NCBI conserved domain database)
deposited in the PDB database. OAT are pyridoxal-5’-phosphate
(PLP) dependent enzymes; they belong to PLP fold class I,[13]
which represent a very large and diverse superfamily. In the
OAT subfamily, eight different enzyme activities are known
(Table S1, Supporting Information). All 58 available 3D struc-
tures of this cluster show considerable similarity, but they are
different in important residues in the active site that are obvi-
ously involved in substrate recognition. This search identified
four structures (PDB codes: 3HMU, 3I5T, 3FCR, 3GJU), for which
we could not find any information associated with their struc-
tures or functions. Additionally, these structures differed in im-
Felix-Hausdorff-Str. 4, 17487 Greifswald (Germany)
Fax: (+49)3834-86-794367
E-mail: matthias.hoehne@uni-greifswald.de
uwe.bornscheuer@uni-greifswald.de
[b] Dr. M. Svedendahl Humble
Dept. of Organic Chemistry
Arrhenius Laboratories
Stockholm University, SE-106 91 Stockholm (Sweden)
[c] H. Land, Prof. Dr. P. Berglund
School of Biotechnology
KTH Royal Institute of Technology
AlbaNova University Center, 106 91 Stockholm (Sweden)
Supporting information for this article contains the full experimental de-
tails and is available on the WWW under http://dx.doi.org/10.1002/
cctc.201200544.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 150 – 153 150

CHEMCATCHEM
COMMUNICATIONS
www.chemcatchem.org
portant active-site residues involved in substrate coordination
relative to the enzymes with known structures in the OAT-like
cluster (Figure S1, Supporting Information). We therefore per-
formed extensive characterization of these proteins.
ure 1a; Table S2, Supporting Information). Additionally, we
could demonstrate that all four proteins exhibit (S)-enantiopre-
ference (Table 1).
1-Phenylethylamine (6), which is often used as a benchmark
substrate for many ATAs, was well accepted by 3HMU and 3I5T.
In contrast, 3FCR and especially 3GJU converted (S)-6 very
slowly. 3GJU preferred benzylamine over (S)-6 to a greater
extent than the other three enzymes, but displayed only low
specific activity. Surprisingly, 1-N-Boc-3-aminopyrrolidine (Boc=
tert-butoxycarbonyl), which is often not a preferred substrate
for ATAs,[6,19] was the best substrate for 3FCR and 3GJU.
After determination of the optimum pH value of approxi-
mately pH 9.5 (Figure S3, Supporting Information), we found
that various amino acceptors such as pyruvate, glyoxylate, and
succinate semialdehyde were accepted by the enzymes (Fig-
ure 1b). As a result of the very unusual optimum pH value (i.e.,
pH 9.5–10.5) and the observed dramatic decrease in activity at
neutral pH values, the ATA activities of 3FCR and 3GJU are vir-
tually absent at physiological pH (ꢀ7.5). Furthermore, other
substrates (ornithine, lysine, b-alanine, dimethylglycine, and 8-
amino-7-oxononanoic acid) that are accepted by enzymes that
are structurally similar to the OAT cluster (Scheme S1, Support-
ing Information) were converted only to a small extent. From
these observations, we conclude that the major activity of the
identified enzymes at physiological pH is the transamination of
small amino and a-keto acids, for example, alanine and succi-
nate semialdehyde (Table 1), and the catalytic performance for
these substrates was observed to be of the same order of
magnitude for all four enzymes. Additionally, we suggest that
ATA activity displays a type of substrate promiscuity, which is
more (for 3HMU, 3I5T) or less (for 3FCR, 3GJU) pronounced in
the four different enzymes.
The first challenge was the identification of the reaction and
substrate specificity, as the natural function of these enzymes
was unclear. We hypothesized that these proteins might con-
vert amines, as the proteins are structurally similar to the
active site of an w-amino acid:pyruvate transaminase (PDB
code: 3A8U) from Pseudomonas putida.[14] This enzyme transa-
minates the terminal amino group of linear w-amino-n-alkyl
carboxylic acids of different chain lengths. Simple amines are
also converted,[15] but enantioselectivity towards chiral amines
was never analyzed.[16] Although many of the known ATAs
show broad substrate specificity,[17] there are also known en-
zymes that accept only a relatively narrow range of sub-
strates.[6] Hence, we needed to identify the amino donor while
keeping in mind that it could only be converted if the appro-
priate amino acceptor was present. Therefore, we designed
a screening set comprising eight structurally different amino
donors (aliphatic/arylaliphatic amines with different chain
lengths) and seven different amino acceptors (ketones, alde-
hydes, a-keto acids) in one pot (Scheme 1; Figure S2, Support-
ing Information).
As the pH profile of the enzymes is also unknown, we con-
ducted the reactions at two different pH values. This revealed
that all four proteins (crude extract from recombinant expres-
sion in E. coli) were able to convert a range of structurally dif-
ferent amines/ketones. From subsequent, detailed characteriza-
tion of the amino donor specificity, we could show that the
catalytic efficiency of the enzymes towards a panel of eight dif-
ferent amines differed by three orders of magnitude (Fig-
In the next step, we analyzed
the structures of the enzymes by
focusing on features that might
explain the observed substrate
profile. The four enzymes show
a sequence identity of approxi-
mately 35% with the exception
of 3FCR and 3GJU, which are
highly similar (75% sequence
identity). All structures align well
(Figure S4, Supporting Informa-
tion) and show a typical overall
structure of a protein of PLP fold
classI enzyme. 3GJU, 3FCR, and
3I5T crystallize in the apo form
containing PLP. 3HMU lacks the
PLP unit, but a sulfate ion is co-
ordinated in the position where
the phosphate group of PLP is
placed in other transaminases
(referred to as phosphate group
binding cup). Thus, we believe
that the structures represent
Scheme 1. Screening reaction for the initial detection of amine transaminase (ATA) activity. Various amines and
keto substrates were combined in a one-pot reaction. This increases the probability to identify the preferred (co)-
substrates of the enzymes
a proper active site conforma-
tion. High-resolution structures
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 150 – 153 151

CHEMCATCHEM
COMMUNICATIONS
www.chemcatchem.org
Figure 2. a) A view of the large binding pocket in the active site of 3HMU. A
docked quinonoid intermediate corresponding to the first planar quinonoid
in the reaction between PLP and 1-phenylethylamine is shown in magenta.
b) Superposition of the crystal structures. PLP and R417 are shown as stick
diagrams with different colors; 3HMU:A (green), 3HMU:B (green), 3FCR (ma-
genta), 3I5T (cyan).
Residues contributing to the active site pockets are high-
lighted in a multiple sequence alignment (Figure S5, Support-
ing Information). The large binding pocket provides a lot of
space, and this reflects the general trend that larger substrates
are converted with higher catalytic efficiency, for example, 2-
aminononane>2-aminohexane, or 1-methylnaphthylamine>
4-phenyl-2-aminobutane>1-phenylethylamine. This is in ac-
cordance with previously reported models of the active site.[20]
The preference of 3FCR and 3GJU for 1-N-Boc-3-aminopyrroli-
dine might be explained by the presence of Tyr59: its hydroxy
group forms a hydrogen bond to the carbonyl oxygen atom of
B3AP in docking simulations (Figure S6, Supporting Informa-
tion).
Figure 1. Characterization of the substrate spectrum of the 3HMU, 3I5T,
3FCR, and 3GJU proteins. a) Amino donor substrate spectrum for 1–8; see
Scheme 1 for the structures of 1–8. b) Amino acceptor substrate spectrum
for 9: pyruvate, 10: ethyl pyruvate, 11: glyoxylic acid, 12: succinate semial-
dehyde, 13: levulinic acid, 14: a-ketoglutarate, 15: hydroxypyruvate, 16:
mercaptopyruvate, 17: hexanal, 18: 2-hexanone.
Interestingly, the position of the side chain of an arginine
residue near the active site, R417, varies greatly in the four
structures (Figure 2b), which is indicative of the high flexibility
of this residue. In 3HMU, two different positions of R417 are
observed in the two monomers of the enzyme, and in 3FCR,
the final model actually contains two different positions of this
arginine side chain in each monomer. A mutagenesis experi-
ment and MD-simulation studies show that this flexible R417
residue allows the enzymes to accommodate different types of
substrates. This is known as dual substrate specificity and is
described in more detail elsewhere.[21]
Table 1. Specific activities for the transamination of the donor/acceptor
pairs 1-phenylethylamine (6)/pyruvate and succinate semialdehyde (12)/
alanine at two pH values.
Enzyme
pH 7.5
12[a]
pH 9.5
(S)-6[b]
[c]
(S)-6[b]
12[a]
(R)-6[b]
Eapp
3HMU
3I5T
3FCR
3GJU
1.50
0.46
1.05
1.63
1.25
0.34
0.008
0.007
11.4
0.37
1.73
1.96
14.9
4.95
0.17
0.092
0.014
0.029
0.004
0.002
1070
169
44
Finally, all four enzymes were tested in the asymmetric syn-
thesis of a panel of amines (Table 2) by using l-alanine as the
amino donor. The co-product pyruvate was removed by lactate
dehydrogenase to shift the equilibrium of the reaction. Al-
though 3FCR and 3GJU turned out to be inactive for most of
the substrates, 1-N-Boc-3-pyrrolidinone was converted into
product 4 in a nearly quantitative manner.
In conclusion, owing to the advancements in high-through-
put crystal structure determination, the protein database is
worth exploring in more detail, as it contains a rich source of
structures of many proteins that are not characterized in suffi-
cient detail or that have unknown function. The structures of
53
[a] Activities (UmgproteinÀ1) were measured by spectrophotometric deter-
mination of pyruvate with lactate dehydrogenase (LDH assay). [b] Activi-
ties were measured by using the acetophenone assay.[18] [c] Eapp =v(S)/v(R)
.
[1.6 (for 3GJU), 1.8 (for 3FCR), 2.0 (for 3I5T), and 2.1 ꢂ (for
3HMU)] allowed a detailed view of the active site of these en-
zymes. From earlier characterization studies,[20] it was hypothe-
sized that the active site of an ATA should consist of a large
and a small binding pocket, which should accommodate the
large and small substituents of an amine/ketone (Figure 2a).
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 150 – 153 152

CHEMCATCHEM
COMMUNICATIONS
www.chemcatchem.org
[4] P. Lorenz, J. Eck, Nat. Rev. Microbiol. 2005, 3, 510–516.
Table 2. Conversion and enantiomeric excess values of the amines
[5] a) S. Yooseph, G. Sutton, D. B. Rusch, A. L. Halpern, S. J. Williamson, K.
Remington, J. A. Eisen, K. B. Heidelberg, G. Manning, W. Li, L. Jaroszew-
ski, P. Cieplak, C. S. Miller, H. Li, S. T. Mashiyama, M. P. Joachimiak, C. van
Belle, J. M. Chandonia, D. A. Soergel, Y. Zhai, K. Natarajan, S. Lee, B. J. Ra-
phael, V. Bafna, R. Friedman, S. E. Brenner, A. Godzik, D. Eisenberg, J. E.
Dixon, S. S. Taylor, R. L. Strausberg, M. Frazier, J. C. Venter, PLoS Biol.
2007, 5, e16; b) J. C. Venter, K. Remington, J. F. Heidelberg, A. L. Halpern,
D. Rusch, J. A. Eisen, D. Wu, I. Paulsen, K. E. Nelson, W. Nelson, D. E.
Fouts, S. Levy, A. H. Knap, M. W. Lomas, K. Nealson, O. White, J. Peter-
son, J. Hoffman, R. Parsons, H. Baden-Tillson, C. Pfannkoch, Y. H. Rogers,
H. O. Smith, Science 2004, 304, 66–74.
[6] M. Hçhne, S. Schꢀtzle, H. Jochens, K. Robins, U. T. Bornscheuer, Nat.
Chem. Biol. 2010, 6, 807–813.
[7] S. Schꢀtzle, F. Steffen-Munsberg, M. Thotowi, M. Hçhne, K. Robins, U. T.
Bornscheuer, Adv. Synth. Catal. 2011, 353, 2439–2445.
[8] a) U. T. Bornscheuer, G. W. Huisman, R. J. Kazlauskas, S. Lutz, K. Robins,
J. C. Moore, Nature 2012, 485, 185–194; b) R. J. Kazlauskas, U. T. Born-
scheuer, Nat. Chem. Biol. 2009, 5, 526–529; c) Protein Engineering Hand-
book (Eds.: S. Lutz, U. T. Bornscheuer), Wiley-VCH, Weinheim, 2009;
d) M. T. Reetz, Angew. Chem. 2011, 123, 144–182; Angew. Chem. Int. Ed.
2011, 50, 138–174.
[9] a) S. Khurshid, L. F. Haire, N. E. Chayen, J. Appl. Crystallogr. 2010, 43,
752–756; b) B. A. Manjasetty, A. P. Turnbull, S. Panjikar, K. Bussow, M. R.
Chance, Proteomics 2008, 8, 612–625.
obtained by asymmetric synthesis from the corresponding ketones.
Product[a]
Enzyme
Conversion[b]
[%]
ee[c]
[%]
3HMU
3I5T
3HMU
3I5T
3HMU
3I5T
3HMU
3I5T
93
9
26
<5
>99
>99
92
87
2
6
3
8
96
>99
>99
98
88
99
>99
97
90
95
90
9
3HMU
3I5T
3FCR[d]
3GJU[d]
>99
99
>99
97
4
[a] Structures of the products are given in Scheme 1. [b] Measured after
24 h by HPLC or by capillary electrodes (CE). [c] After derivatization with
trifluoroacetic anhydride measured by chiral GC. [d] Using 3FCR and
3GJU, no significant conversion (<5%) was observed for the other ke-
tones, and thus, optical purity could not be determined.
[10] M. Hçhne, U. T. Bornscheuer, ChemCatChem 2009, 1, 42–51.
[11] In parallel to the characterization of the four transaminases of this
paper, we solved the crystal structure of the amine transaminase of
Chromobacterium violaceum; see M. Svedendahl-Humble, K. E. Cassim-
jee, M. Hakansson, Y. R. Kimbung, B. Walse, V. Abedi, H. J. Federsel, P.
Berglund, D. T. Logan, FEBS J. 2012, 279, 779–792.
[12] a) C. Kalyanaraman, H. J. Imker, A. A. Fedorov, E. V. Fedorov, M. E. Glas-
ner, P. C. Babbitt, S. C. Almo, J. A. Gerlt, M. P. Jacobson, Structure 2008,
16, 1668–1677; b) D. Dunaway-Mariano, Structure 2008, 16, 1599–1600;
c) N. de Souza, Nat. Methods 2007, 4, 771–771; d) L. Song, C. Kalyanara-
man, A. A. Fedorov, E. V. Fedorov, M. E. Glasner, S. Brown, H. J. Imker,
P. C. Babbitt, S. C. Almo, M. P. Jacobson, J. A. Gerlt, Nat. Chem. Biol.
2007, 3, 486–491; e) J. C. Hermann, R. Marti-Arbona, A. A. Fedorov, E.
Fedorov, S. C. Almo, B. K. Shoichet, F. M. Raushel, Nature 2007, 448,
775–779; f) D. F. Xiang, P. Kolb, A. A. Fedorov, C. Xu, E. V. Fedorov, T.
Narindoshivili, H. J. Williams, B. K. Shoichet, S. C. Almo, F. M. Raushel,
Biochemistry 2012, 51, 1762–1773.
the amine transaminases used in this study were deposited
three years ago and remained unrecognized, although struc-
tural information of these popular biocatalysts is highly de-
sired. With a growing number of solved crystal structures (in-
cluding those of uncharacterized proteins), the PDB continues
to develop into a great resource that can be considered a gold
mine for various areas. We hope that this work will convince
other researchers to connect structures of uncharacterized pro-
teins with their function. This will help to uncover and apply
these resources in the field of protein engineering or the
discovery of novel biocatalysts.
Acknowledgements
[13] a) R. Percudani, A. Peracchi, BMC Bioinf. 2009, 10, 273; b) J. N. Jansonius,
Curr. Opin. Struct. Biol. 1998, 8, 759–769.
[14] N. Watanabe, K. Sakabe, N. Sakabe, T. Higashi, K. Sasaki, S. Aibara, Y.
Morita, K. Yonaha, S. Toyama, H. Fukutani, J. Biochem. 1989, 105, 1–3.
[15] K. Yonaha, S. Toyama, M. Yasuda, K. Soda, Agric. Biol. Chem. 1977, 41,
1701–1706.
We thank Prof. Karl Hult and Dr. Karim Engelmark Cassimjee for
fruitful discussions. U.T.B thanks the European Union (KBBE-2011-
5, grant no. 289350) for financial support.
[16] Our attempts to characterize the substrate scope and enantioselectivity
towards different chiral amines failed, as this Pseudomonas putida w-
amino acid transaminase could not be expressed in E. coli in a soluble
form (a synthetic, codon-optimized gene was used and various condi-
tions, e.g, temperature, chaperones, were investigated).
Keywords: amines · enantioselectivity · enzyme catalysis ·
protein structures · asymmetric synthesis
[17] D. Koszelewski, K. Tauber, K. Faber, W. Kroutil, Trends Biotechnol. 2010,
28, 324–332.
[18] S. Schꢀtzle, M. Hçhne, E. Redestad, K. Robins, U. T. Bornscheuer, Anal.
Chem. 2009, 81, 8244–8248.
[19] M. Hçhne, K. Robins, U. T. Bornscheuer, Adv. Synth. Catal. 2008, 350,
807–812.
[20] J.-S. Shin, B.-G. Kim, J. Org. Chem. 2002, 67, 2848–2853.
[21] F. Steffen-Munsberg, C. Vickers, A. Thontowi, S. Schꢀtzle, T. Meinhardt,
M. Svedendahl Humble, H. Land, P. Berglund, U. T. Bornscheuer, M.
Hçhne, ChemCatchem 2012, DOI: 10.1002/cctc.201200545.
[1] a) Enzymes in Organic Synthesis Vol. 1–3, 3rd ed. (Eds.: O. May, H. Grçger,
K. Drauz), VCH, Weinheim, 2012; b) K. Buchholz, V. Kasche, U. T. Born-
scheuer, Biocatalysts and Enzyme Technology, 2nd ed., Wiley-VCH, Wein-
heim, 2012; c) Biocatalysis in the Pharmaceutical and Biotechnology In-
dustries (Ed.: R. N. Patel), CRC Press, London, 2006; d) Industrial Biotrans-
formations, 2nd ed. (Eds.: A. Liese, K. Seelbach, C. Wandrey), Wiley-VCH,
Weinheim, 2006; e) M. Breuer, K. Ditrich, T. Habicher, B. Hauer, M. Keßel-
er, R. Stꢃrmer, T. Zelinski, Angew. Chem. 2004, 116, 806–843; Angew.
Chem. Int. Ed. 2004, 43, 788–824.
[2] C. K. Savile, J. M. Janey, E. C. Mundorff, J. C. Moore, S. Tam, W. R. Jarvis,
J. C. Colbeck, A. Krebber, F. J. Fleitz, J. Brands, P. N. Devine, G. W. Huis-
man, G. J. Hughes, Science 2010, 329, 305–309.
Received: August 8, 2012
[3] A. A. Desai, Angew. Chem. 2011, 123, 2018–2020; Angew. Chem. Int. Ed.
2011, 50, 1974–1976.
Published online on November 8, 2012
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2013, 5, 150 – 153 153