Specificity of transglutaminase-catalyzed peptide synthesis

Article abstract of DOI:10.1016/j.molcatb.2015.11.009

Biocatalytic methods for peptide synthesis are of high value due to the rapidly increasing approval of peptide-based therapeutics and the need to develop new analogs. Guinea pig liver transglutaminase (gTG2) catalyzes the cross-linking of peptides and proteins via the formation of γ-glutamyl-ε-lysyl isopeptide bonds. In this study, we investigate gTG2-catalyzed peptide bond formation between various amino acid-derived donor and acceptor substrates. Using LC-MS analysis, we demonstrate that gTG2 forms Gly-Xaa and d-Ala-Gly dipeptide products, confirming that its natural transamidation activity can be co-opted for peptide synthesis. An aromatic ester of Gly was the most efficient acyl-donor substrate tested; aromatic esters of d-Ala and l-Ala showed 50-fold lower reactivity or no reactivity, respectively. A computational strategy combining computational protein design algorithms and molecular dynamics simulations was developed to model the binding modes of donor substrates in the gTG2 active site. We show that the inability of gTG2 to efficiently catalyze peptide synthesis from donors containing alanine results from the narrow substrate binding tunnel, which prevents bulkier donors from adopting a catalytically productive binding mode. Our observations pave the way to future protein engineering efforts to expand the substrate scope of gTG2 in peptide synthesis, which may lead to useful biocatalysts for the synthesis of desirable bioactive molecules.

Full text of DOI:10.1016/j.molcatb.2015.11.009

Accepted Manuscript
Title: Specificity of Transglutaminase-Catalyzed Peptide
Synthesis
Author: Antony D. St-Jacques Natalie M. Rachel Dan R.
Curry Steve M.F.G. Gillet Christopher M. Clouthier Jeffrey W.
Keillor Joelle N. Pelletier Roberto A. Chica
PII:
S1381-1177(15)30101-6
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.molcatb.2015.11.009
MOLCAB 3273
To appear in:
Journal of Molecular Catalysis B: Enzymatic
Received date:
Revised date:
Accepted date:
23-7-2015
23-10-2015
3-11-2015
Please cite this article as: Antony D.St-Jacques, Natalie M.Rachel, Dan R.Curry, Steve
M.F.G.Gillet, Christopher M.Clouthier, Jeffrey W.Keillor, Joelle N.Pelletier, Roberto
A.Chica, Specificity of Transglutaminase-Catalyzed Peptide Synthesis, Journal of
Molecular Catalysis B: Enzymatic http://dx.doi.org/10.1016/j.molcatb.2015.11.009
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Specificity of Transglutaminase-Catalyzed Peptide Synthesis
Antony D. St-Jacquesa,b, Natalie M. Rachelc,d,e, Dan R. Currya,b, Steve M.F.G. Gilletc,
Christopher M. Clouthiera,b, Jeffrey W. Keillora,b, Joelle N. Pelletierc,d,e, Roberto A. Chicaa,b,d*
rchica@uottawa.ca
aDepartment of Chemistry and Biomolecular Sciences, University of Ottawa, Ottawa, Ontario,
K1N 6N5, Canada
bCentre for Catalysis Research and Innovation, University of Ottawa, Ottawa, Ontario, K1N
6N5, Canada
cDépartement de Chimie, Université de Montréal, Montréal, Québec, H3T 1J4, Canada
dPROTEO Network, Université Laval, Québec, Québec, G1V 0A6, Canada
eCenter for Green Chemistry and Catalysis, Montréal, Québec, H3A 0B8, Canada
*Corresponding author at: Department of Chemistry and Biomolecular Sciences, University of
Ottawa, 10 Marie Curie, Ottawa, Ontario, Canada, K1N 6N5
1

Graphical Abstract
2

Highlights
1. Wild-type TG2 can catalyze synthesis of Cbz-protected dipeptides.
2. Donor substrate specificity is narrow.
3. Aromatic ester derivatives of Gly and D-Ala serve as donor substrates.
4. Hydrophobic cavity formed by Trp332 and Phe334 dictates substrate specificity.
5. Protein engineering based on our computational models may expand substrate scope.
3

Abstract
Biocatalytic methods for peptide synthesis are of high value due to the rapidly increasing
approval of peptide-based therapeutics and the need to develop new analogues. Guinea pig liver
transglutaminase (gTG2) catalyzes the cross-linking of peptides and proteins via the formation of
γ-glutamyl-ε-lysyl isopeptide bonds. In this study, we investigate gTG2-catalyzed peptide bond
formation between various amino acid-derived donor and acceptor substrates. Using LC-MS
analysis, we demonstrate that gTG2 forms Gly-Xaa and D-Ala-Gly dipeptide products,
confirming that its natural transamidation activity can be co-opted for peptide synthesis. An
aromatic ester of Gly was the most efficient acyl-donor substrate tested; aromatic esters of D-Ala
and L-Ala showed 50-fold lower reactivity or no reactivity, respectively. A computational
strategy combining computational protein design algorithms and molecular dynamics
simulations was developed to model the binding modes of donor substrates in the gTG2 active
site. We show that the inability of gTG2 to efficiently catalyze peptide synthesis from donors
containing alanine results from the narrow substrate binding tunnel, which prevents bulkier
donors from adopting a catalytically productive binding mode. Our observations pave the way to
future protein engineering efforts to expand the substrate scope of gTG2 in peptide synthesis,
which may lead to useful biocatalysts for the synthesis of desirable bioactive molecules.
Abbreviations
7HC: 7-hydroxycoumarin
Cbz: carbobenzyloxy
TG2: tissue transglutaminase
MD: molecular dynamics
4

GABA: γ-aminobutyric acid
gTG2: guinea pig liver transglutaminase
NTA: nitrilotriacetic acid
MOPS: 3-(N-morpholino)propanesulfonic acid
EDTA: ethylenediaminetetraacetic acid
DMF: N,N-dimethylformamide
HRMS: high-resolution mass spectrometry
ESI: electrospray ionization
FAB: fast atom bombardment
LC-MS: liquid chromatography-mass spectrometry
DMSO: dimethyl sulfoxide
HPLC: high-performance liquid chromatography
MOE: molecular operating environment
Boc: tert-butyloxycarbonyl
Keywords: TG2; enzyme kinetics; LC-MS; biocatalysis; molecular modeling
5

1. Introduction
The amide bond is among the most versatile functional groups in synthetic organic
chemistry due to its high polarity, stability, and well-characterized conformational preference
[1]. In particular, facile peptide bond formation – whether between natural or unnatural amino
acids – is of extremely high value due to the rapidly increasing approval of peptide-based
therapeutics and the need to develop new analogues. Conventional chemical approaches to
peptide bond synthesis require chemical activation, protection, and deprotection steps for each
bond formed as well as orthogonal protection of reactive substituents. As a result, peptide bond
synthesis remains an important challenge in chemistry [2]. Enzymatic approaches have attempted
to alleviate these limitations. This is generally performed by running proteases “backward”,
toward bond synthesis rather than hydrolysis (recently reviewed in [3]). Despite engineering of
proteases and optimization of reaction conditions, hydrolysis of existing peptide bonds reduces
yield. Using an enzyme that has evolved to synthesize an amide bond, rather than hydrolyze it,
could prove advantageous in enzyme-catalyzed peptide bond synthesis.
One such enzyme is tissue transglutaminase (TG2), which catalyzes the Ca2+-dependent
cross-linking of peptides and proteins via the formation of γ-glutamyl-ε-lysyl isopeptide bonds
[4-6]. The catalytic reaction follows a modified ping-pong mechanism in which a glutamine-
containing protein or peptide, the acyl-donor substrate, reacts with the catalytic cysteine residue
to form a thioester bond. The resulting covalent acyl-enzyme intermediate then reacts with a
second substrate, the acyl-acceptor, to yield the isopeptide-containing product and free enzyme
in a transamidation reaction. In the absence of an amine acyl-acceptor, the acyl-enzyme
6

intermediate can be hydrolyzed, transforming the acyl-donor glutamine residue into glutamate
and regenerating the free enzyme [7].
TG2 enzymes exhibit broad specificity towards the acyl-acceptor substrate [8]. Although
the native acyl-acceptor substrate is generally a lysine-containing protein or peptide, many non-
natural primary amines, such as glycinamide [9, 10], and anilines, such as N,N-dimethyl-1,4-
phenylenediamine [11], can also react. However, amines containing free carboxylic acid groups,
such as free amino acids, do not act as substrates [10]. On the other hand, TG2 displays narrow
specificity for its acyl-donor substrates. The side chain of a protein or peptide-bound L-Gln
residue is the native substrate while the side chain of the similar amino acid L-Asn is not reactive
[9]. In addition to amides, γ-glutamyl aromatic ester derivatives of L-Glu, such as N-
carbobenzyloxy-L-glutamyl(γ-p-nitrophenyl ester)glycine (Fig. 1A), have also been shown to be
acyl-donor substrates of TG2 and are used to measure the enzyme’s activity [12]. However,
secondary amide derivatives of L-Gln, such as N-γ-methyl-L-glutamine or anilides, are not
substrates of TG2 [13]: the γ-carboxamide group of L-Gln is the only known amide that is an
acyl-donor substrate of TG2.
We and others previously demonstrated that TG2 could use a novel class of acyl-donor
substrates that are neither L-Gln nor L-Glu derivatives [14, 15].
Namely, 4-(N-
carbobenzyloxyglycylamino)-butyric acid-coumarin-7-yl ester (Cbz-Gly-GABA-7HC) and 4-(N-
carbobenzyloxyphenylalanylamino)-butyric acid-coumarin-7-yl ester (Cbz-Phe-GABA-7HC)
(Fig. 1B) can react with TG2 to release 7-hydroxycoumarin (7HC), resulting in a fluorescence
increase that makes these compounds useful for quantifying TG2 reaction rates. The scaffolds of
these substrates, based on known irreversible inhibitors of TG2 [16, 17], differ from L-Glu
aromatic ester acyl-donor substrates of TG2 in that the reactive ester function is located on the
7

main chain of the peptide analogue, rather than on the side chain. As a result, they give rise to
products that do not contain a γ-glutamyl-ε-lysyl isopeptide bond. An analogue in which the 7HC
leaving group is attached directly to the glycine residue carboxylate group, N-
carbobenzyloxyglycyl-coumarin-7-yl ester (Cbz-Gly-7HC, Fig. 1C), is also a donor substrate of
TG2 [18]. Significantly, the reaction of this substrate with an acceptor amine substrate would
result in the formation of a peptide-like α-amide bond (Scheme 1). These results illustrate that
specificity for acyl-donor substrates with aromatic ester functions is broader than had previously
been supposed and demonstrate that the enzyme can generate products with novel scaffolds.
In this study, we investigate guinea pig liver TG2 (gTG2)-catalyzed peptide bond
formation between the Cbz-Gly-7HC donor substrate in combination with various amino acid-
derived acceptors. Using LC-MS analysis of the reaction products, we demonstrate that the
enzyme is able to react directly with the α-carboxyl group of Cbz-Gly-7HC to form Gly-Xaa
dipeptide products, confirming that its natural transamidation activity can be co-opted for peptide
synthesis. Additionally, we explore the substrate specificity of the enzyme in peptide synthesis
by measuring its reactivity toward a variety of potential acyl-donor substrates having an aromatic
ester function on the α-carboxyl group of various amino acids. We observed that the aromatic
ester of Gly is an efficient acyl-donor substrate; the aromatic ester of D-Ala is also reactive
though to a lesser extent, and that of L-Ala showed no detectable reactivity.
To elucidate how the stereochemical configuration of the side-chain of alanine-containing
donor substrates affects gTG2 catalytic efficiency, we used a computational strategy combining
computational protein design and molecular dynamics simulations to model the binding modes
of donors in the gTG2 active site. We show that the inability of gTG2 to efficiently catalyze
peptide synthesis from donors other than Cbz-Gly-7HC results from the narrow substrate binding
8

tunnel, which prevents bulkier donors to adopt a catalytically productive binding mode. Our
observations pave the way to future protein engineering efforts to expand the substrate scope of
gTG2 in peptide synthesis, which may lead to useful biocatalysts for the synthesis of desirable
bioactive molecules.
2. Materials and Methods
2.1 Materials
All reagents used were of the highest available purity. Lysozyme, 7HC, N-acetyl-L-lysine
methyl ester hydrochloride (N-AcLysOMe), N-carbobenzyloxy-L-glutaminylglycine (Cbz-L-Gln-
Gly), glycinamide (GlyNH2) and L-leucine methyl ester (LeuOMe) hydrochlorides were
purchased from Sigma-Aldrich. L-Alaninamide hydrochloride (AlaNH2) was purchased from
Novabiochem (Mississauga, ON). Ni-NTA agarose resin was purchased from Qiagen
(Mississauga, ON). Restriction enzymes and DNA-modifying enzymes were from New England
Biolabs. All aqueous solutions were prepared using water purified with a Millipore BioCell
system.
2.2 Synthesis of donor substrates
2.2.1. Synthesis of Cbz-Gly-7HC and Cbz-L-Ala-7HC
The synthesis of Cbz-Gly-7HC was based on a previously reported protocol [14].
Namely, 0.2 g (1 mmol) of Cbz-Gly and 0.4 g (2.5 mmol) of 7HC were dissolved in 10 mL of
ethyl acetate. Then, 0.22 mL (0.2 g, 2 mmol) of N-methylmorpholine and 0.8 mL (0.63 g, 5
mmol) of N,N-diisopropylcarbodiimide were added with stirring at room temperature. Stirring
was continued until the complete disappearance of Cbz-Gly, as followed by thin layer
9

chromatography (ethyl acetate). The reaction mixture was then washed once with 1 M NaOH,
three times with 0.1 M NaOH, 3 times with 0.1 M HCl, once with saturated NaHCO3, and once
with brine. The organic phase was then dried over MgSO4, filtered and evaporated under reduced
pressure. The resulting residue was purified by silica gel chromatography (ethyl acetate) to
remove traces of diisopropylurea, giving the desired ester in 70 % yield (0.25 g). Cbz-L-Ala-7HC
was synthesized according to a similar protocol.
1
Cbz-Gly-7HC. H NMR (300 MHz, CDCl3): δ 4.30 (2H, d), 5.18 (2H, s), 5.33 (1H, s),
13
6.44 (1H, d), 7.08 (1H, d), 7.10 (1H, s), 7.37 (5H, m), 7.51 (1H, d), 7.70 (1H, d). C NMR (75
MHz, CDCl3): δ 168.5, 160.6, 156.7, 154.9, 152.9, 143.1, 136.3, 129.0, 128.9, 128.6, 128.5,
118.4, 117.2, 116.6, 110.5, 67.7, 43.2. HRMS (FAB) calculated for C19H16NO6 ([M+H]+):
354.0972, found 354.0968.
Cbz-L-Ala-7HC. 1H NMR (300 MHz, CDCl3): δ 1.63 (3H, d), 4.62 (1H, m), 5.18 (2H, s),
5.33 (1H, d), 6.42 (1H, d), 7.07 (1H, d), 7.09 (1H, s), 7.39 (5H, m), 7.51 (1H, d), 7.70 (1H, d).
13C NMR (75 MHz, CDCl3): δ 171.5, 160.5, 156.0, 154.9, 153.1, 143.1, 136.4, 129.0, 128.9,
128.6, 128.5, 118.4, 117.2, 116.6, 110.5, 67.5, 50.2, 18.5. HRMS (FAB) calculated for
C20H18NO6 ([M+H]+): 368.1129, found 368.1118.
2.2.2. Synthesis of Cbz-D-Ala-7HC
The synthesis of Cbz-D-Ala-7HC followed the protocol employed for Cbz-L-Ala-7HC.
Equimolar amounts of Cbz-D-Ala (4 mmol, 0.89 g) and 7HC (4 mmol, 0.65 g) were dissolved in
15 mL of dichloromethane at room temperature. To the stirring solution, 0.38 mL (4.4 mmol) of
N-methylmorpholine and 0.82 mL (8 mmol) of N,N-diisopropylcarbodiimide were added. The
consumption of Cbz-D-Ala was monitored by thin-layer chromatography. Upon completion, the
10

reaction mixture was washed successively with 0.1 M NaOH, 0.1 M HCl, saturated sodium
bicarbonate, and brine. The organic phase was dried over MgSO4, filtered, and concentrated
under reduced pressure. The crude product was purified via flash column using
chloroform/methanol (9:1), affording the titular compound in 66 % yield (0.59 g).
Cbz-D-Ala-7HC. 1H NMR (CDCl3, 400 MHz): δ 1.41 (d, 3H), 4.25 (q, 1H), 5.10 (s, 2H),
13
6.63 (d, 1H), 7.37 (m, 5H), 7.55 (br s, 1H), 7.71 (d, 1H), 7.84 (d, 1H). C NMR (CDCl3, 100
MHz): δ 17.6, 53.5, 67.4, 110.1, 115.4, 116.1, 118.7, 127.9, 128.1, 129.7, 136.5, 146.3, 156.6,
156.9, 157.1, 161.8, 169.3. HRMS (ESI) Calculated for C20H17NO6: 367.1056. Found: 367.1060.
2.2.3. Synthesis of Cbz-GlyNH2
The synthesis of Cbz-GlyNH2 was adapted from a previously reported protocol [19].
Glycinamide hydrochloride (18 mmol, 2.00 g) was dissolved in water (60 mL) and acetone (8
mL), prior to the addition of Na2CO3 (54 mmol, 5.7 g) and NaHCO3 (18 mmol, 1.5 g). Benzyl
chloroformate (22 mmol, 3.20 mL) was added dropwise to the stirring solution over the course of
30 minutes. The resulting mixture was stirred for 3 hours at room temperature, after which the
products were isolated by washing with diethyl ether (50 mL). The protected product was
precipitated out of solution by the slow addition of 0.1 M HCl. The precipitate was filtered and
subsequently dried in vacuo to afford a white solid in 86% yield (3.24 g).
1H NMR (CDCl3, 400 MHz): δ 3.87 (s, 2H), 5.07 (s, 2H), 7.26 (m, 5H), 7.35 (s, 2H), 7.95
(s, 1H). 13C NMR (CDCl3, 100 MHz): δ 45.1, 67.3, 126.2, 126.9, 128.9, 136.3, 156.8, 170.1.
HRMS (ESI) calculated for C10H12N2O3: 208.0848. Found: 208.0851.
2.3 Overexpression and purification of gTG2
11

Recombinant gTG2 was overexpressed and purified from Escherichia coli according to a
protocol developed in our laboratory [20] with the following modifications. After Ni-NTA
purification, the eluant was transferred to a 15-mL Amicon Ultra tube (Millipore) with a
molecular weight cut-off of 30 kDa and the gTG2 solution was desalted by centrifugation with
25 mM Tris-acetate buffer (pH 7.0) containing 0.5 mM EDTA. The samples were aliquoted,
snap-frozen on dry ice and stored at -80 °C. Typical yields were 1.5-10 mg/L of approximately
85 % pure protein, as estimated from Coomassie Blue staining following SDS-PAGE, in
agreement with previous results [20].
2.4 Specific activity
The hydroxamate assay [9] was used to quantify gTG2 activity. Briefly, gTG2 was
incubated at 37 °C for 10 minutes with 30 mM Cbz-L-Gln-Gly and 100 mM hydroxylamine in
200 mM Tris-acetate buffer (pH 7.0) containing 5 mM CaCl2 and 1 mM EDTA. The reaction
was quenched with a solution containing 2.0 M ferric chloride hexahydrate, 0.3 M trichloroacetic
acid, and 0.8 M HCl. The mixture was vortexed and left at room temperature for 10 minutes
before measuring absorbance at 525 nm. One unit (U) of gTG2 produces 1 μmol of L-glutamic
acid γ-monohydroxamate per minute at 37 °C.
2.5 Kinetic assays
All assays were performed in triplicate. The following solutions were prepared: a
standard stock buffer solution (100 mM MOPS buffer pH 7.0, 5 mM CaCl2, and 0.05 mM
EDTA), a 2 mM (Cbz-Gly-7HC, Cbz-L-Ala-7HC) or 100 mM (Cbz-D-Ala-7HC) solution of
acyl-donor substrate in N,N-dimethylformamide (DMF), and a 200 mM solution of acyl-acceptor
12

substrate N-AcLysOMe in water. Prior to performing the assays, a “fluorescence coefficient”
was determined daily by measuring the arbitrary fluorescence intensities corresponding to five
concentrations of 7HC at concentrations ranging from 0.05 to 12.5 μM in 5 % DMF in the stock
buffer solution at 25 °C. The value of this “fluorescence coefficient” varied only slightly (<5 %)
each day. For the hydrolysis reaction, activity was measured by adding 10 mU of purified gTG2
to each well of a TCT Luminescence 96-well microtiter plate (Thermo Electron) containing a
0.5-100 μM solution of the acyl-donor substrate in stock buffer. For the transamidation reaction,
the same amount of purified enzyme was added to 0.5-80 mM of the acyl-acceptor substrate in
stock buffer solution containing 100 μM of acyl-donor substrate Cbz-Gly-7HC. The acyl-
acceptor substrate was replaced by water in the blank. DMF was present at 5 % in the final
reaction mixtures. The increase in fluorescence due to the release of 7HC was followed at 25 °C
against a blank at λex 340 nm and λem 465 nm in a FluoStar Optima microtiter plate reader (BMG
Labtech). Linear slopes of fluorescence versus time were measured over the first <10 %
conversion of substrate to product and were converted into initial rates using the fluorescence
coefficient.
2.6 LC-MS
Reaction mixtures containing 150 µM of ester donor substrate (Cbz-Gly-7HC, Cbz-L-
Ala-7HC, or Cbz-D-Ala-7HC) or 20 mM of amide donor substrate (Cbz-GlyNH2) and 50 mM of
acceptor substrate (GlyNH2, AlaNH2, or LeuOMe) were prepared in a buffer composed of 100
mM MOPS pH 7.0, 5 mM CaCl2, 0.05 mM EDTA, and 5 % DMF. The pH of each substrate
mixture was verified with indicator paper prior to the addition of enzyme. The reaction was
initiated upon the addition of 0.1 U/mL of gTG2 (or an equivalent volume of buffer for reactions
13

run in the absence of enzyme) in a final volume of 2 mL. Reactions were incubated at 37 °C for
up to 20 min. Control reactions without gTG2 or without amine acceptor were run for each
combination of substrate mixtures. Experiments and controls were performed in triplicate.
Disappearance of substrates and appearance of dipeptide products were monitored by ESI
LC-MS. Aliquots of reaction mixture were taken immediately after the addition of enzyme (0
min) and after 2 min, 10 min, and 20 min of reaction time. Formic acid (98 %, 10 μL) was added
to each aliquot (480 μL) and the mixture was vortexed to quench the reaction. The change in pH
from 7.0 to < 2 was verified with indicator paper. An internal standard solution (10 μL of 33.1
mM 4-methoxybenzamide in neat DMSO) was added to the quenched reaction, which was then
filtered using 0.2 µm polytetrafluoroethylene filters (Corning) to remove particulates. The
filtered sample (20 µL, or 10 µL for the reactions containing Cbz-GlyNH2) was injected onto a
Synergi 4 μm, polar reverse phase, 80 Å, 50 × 2 mm liquid chromatography column
(Phenomenex) on a Waters 2545 HPLC apparatus. Elution was achieved with a 5-70 %
MeOH/H2O gradient. Masses were detected under positive ionization mode with a Waters 3100
single quadrupole mass detector.
2.7 Homology modeling
A homology model of gTG2 was prepared as described previously [21]. Briefly, an
alignment of the human and guinea pig liver TG2 sequences (83% identity) was performed using
ClustalW [22] with default parameters. Atomic coordinates for human TG2 in complex with a
covalent inhibitor were retrieved from the Protein Data Bank (PDB ID: 2Q3Z [23]). Using the
sequence alignment and the crystal structure with all non-protein atoms removed, ten models
were generated by Modeller 9.15 [24] with default parameters. All models had regions with
14

unfavorable residue interactions and had to be further refined using the following procedure.
Following the addition of hydrogens, the Protonate 3D utility [25], available in the Molecular
Operating Environment (MOE) software package [26], was used to solvate the ten models with
water in rectangular boxes under periodic boundary conditions with a box cut-off of 6 Å, and to
add counter-ions (Na+ and Cl-). Then, each structure was energy minimized by conjugate
gradient minimization to a root-mean-square gradient below 0.01 kcal mol-1 Å-1 using the
AMBER99 force field [27] with a combined explicit and implicit reaction field solvent model set
up using MOE. Following analysis of all-atom contacts and geometry using MolProbity [28], the
best homology model was selected for further experiments.
2.8 Construction of acyl-enzyme intermediates
Using MOE, the catalytic Cys residue (Cys277) on the gTG2 homology model was
acetylated. The carbonyl moiety of the acetyl group was then manually oriented via dihedral
angle rotation to form a hydrogen bond with the indole nitrogen of Trp241, which has been
shown to be essential for catalytic activity, presumably by stabilizing the transition state [29].
Following energy minimization as described above (root-mean-square gradient below 0.1 kcal
mol-1 Å-1) to optimize H-bonding between the acetyl group and Trp241, the acetylated Cys277
residue was extracted from the gTG2 structure and used as a template to build a Cys residue
acylated with the Cbz-glycyl moiety in MOE. The added Cbz and Gly atoms were then energy
minimized in vacuo by conjugate gradient minimization to a root-mean-square gradient below
0.1 kcal mol-1 Å-1 in order to refine bond lengths and angles. The resulting minimized acylated
Cys residue was then used to generate rotamers via the introduction of the following dihedrals:
O1-C1-C2-N1, 90 ± 20° and -90 ± 20°; C1-C2-N1-C3, 180 ± 20°; C2-N1-C3-O3, 180°; N1-C3-
15

O3-C4, 120 ± 20°, 180 ± 20°, and -120 ± 20°; C3-O3-C4-C5, 60 ± 20°, 180 ± 20°, and -60 ± 20°;
O3-C4-C5-C6A, 0 ± 20°, 60 ± 20°, and 120 ± 20° (see Fig. S1 for atom names). The internal
energy of the resulting 13,122 rotamers was evaluated in vacuo using MOE with the previously
described force field, and only rotamers whose energy was within 10 kcal/mol from the lowest
energy rotamer were included in the final rotamer library. A similar procedure was utilized to
prepare a rotamer library for Cys277 acylated with the Cbz-D-alanyl moiety.
To build the gTG2 acyl-enzyme intermediates, computational protein design was
performed using the fast and accurate side-chain topology and energy refinement (FASTER)
algorithm as implemented in PHOENIX [30-32]. Rotamers for acylated Cys277 and surrounding
residues (positions Gln169, Trp241, Asn243, Tyr245, Met252, Gln276, Trp278, Phe316,
Arg317, Met330, Trp332, Asn333, Phe334, His335, and Cys336) were optimized on the fixed
backbone of the gTG2 homology model. The backbone independent Dunbrack rotamer library
with expansions of ± 1 standard deviation around χ1 and χ2 [33] was used to model side-chain
conformations. A four-term potential energy function consisting of a van der Waals term from
the Dreiding II force field with atomic radii scaled by 0.9 [34], a direction sensitive hydrogen-
bond term with well depth at 8.0 kcal/mol [35], an electrostatic energy term modelled using
Coulomb’s law with a distance dependent dielectric of 10, and an occlusion-based solvation
penalty term [31] were used to evaluate rotamer combinations. The lowest energy acyl-enzyme
intermediate structure obtained from each donor was retained for further analysis.
2.9 Molecular dynamics
For generation of molecular dynamics (MD) trajectories, structures of the gTG2 acyl-
enzyme intermediates prepared as described in Section 2.8 were used as templates. The thioester
16

bond between Cys277 and the acyl groups was hydrolyzed in silico, resulting in noncovalent
complexes (gTG2 bound with Cbz-Gly or Cbz-D-Ala) that were energy minimized to alleviate
steric clashes following the procedure described in Section 2.7, with the exception that
minimization was conducted until a root-mean-square gradient below 0.1 kcal mol-1 Å-1 was
achieved. For the complex with Cbz-L-Ala, a methyl group was added with MOE to the Cα of
Gly prior to minimization. The minimized and solvated noncovalent complexes were used as
input to NPT (constant number, pressure, and temperature) MD simulation at 300 K. MD
trajectories were heated over 500 picoseconds and equilibrated for an additional nanosecond.
This was followed by a 1.5-nanosecond production run sampled at 10-picosecond increments.
All MD simulations were performed using the AMBER99 and extended Hückel theory [36]
force fields in NAMD [37].
3. Results
3.1 Wild-type gTG2 can form peptide bonds
Previously, we showed that recombinant gTG2 can react with Cbz-Gly-7HC as a donor
substrate in conjunction with a variety of amino acid derivatives as acceptor substrates [18].
These observations were based on the increased rate of 7HC release during the enzymatic
reaction in the presence of acceptor substrates, relative to the rate of 7HC release in their
absence. While we had not monitored the appearance of the final reaction products, our
observations were consistent with gTG2 having an intrinsic peptide synthase activity (Scheme
1). In the current work, we applied a sensitive LC-MS assay to monitor dipeptide product
formation directly, and thus confirm that gTG2 catalyzes peptide bond synthesis.
17

In the first step in the development of this LC-MS assay, we synthesized Cbz-Gly-7HC
and determined the kinetic parameters of its gTG2-mediated hydrolysis using
spectrofluorometric analysis. Michaelis-Menten kinetics demonstrated that Cbz-Gly-7HC is an
acyl-donor substrate of wild-type gTG2, having an apparent KM of 15 ± 2 µM and an apparent
kcat of 0.128 ± 0.007 s-1 (Table 1). This KM is similar to that measured for the gTG2-catalyzed
hydrolysis of Cbz-Gly-GABA-7HC (9 ± 2 µM) while the kcat is approximately 10-fold lower
(1.25 ± 0.08 s-1) [14]. The lower kcat results from the absence of the γ-aminobutyric acid linker in
Cbz-Gly-7HC relative to Cbz-Gly-GABA-7HC (Fig. 1), the lack of which may decrease
accessibility of the substrate’s reactive carbonyl group for nucleophilic attack by the catalytic
thiol.
We also determined the kinetic parameters for the gTG2-catalyzed transamidation
reaction of acyl-donor substrate Cbz-Gly-7HC with the widely used acyl-acceptor substrate N-
AcLysOMe. As previously observed with other amino acid derivatives [18], we confirmed that
the rate of release of 7HC from Cbz-Gly-7HC increased in the presence of N-AcLysOMe relative
to the rate of hydrolysis. This result suggests that gTG2 catalyzes the formation of a covalent
bond between the α-carboxyl group of Gly in Cbz-Gly-7HC and the ε-amino group of Lys in N-
AcLysOMe, whose apparent KM and kcat values were determined to be 4 ± 1 mM and 0.20 ± 0.02
s-1, respectively.
Next, we confirmed gTG2-mediated peptide synthesis by LC-MS analysis to identify the
reaction products. We assayed derivatives of three different amino acids previously shown to act
as acyl-acceptor substrates of gTG2 [18] in which the negatively-charged carboxylate is
neutralized under the form of a primary amide or a methyl ester. Chromatograms of the reaction
time-course of Cbz-Gly-7HC with acceptors GlyNH2 (Fig. 2A), AlaNH2 (Fig. 2B), and LeuOMe
18

(Fig. 2C) unequivocally demonstrate the time-dependent increase in concentration of the
corresponding dipeptide product. Since the Cbz-Gly-7HC donor substrate is an activated ester, it
is highly reactive with nucleophilic amines such as the amino acid derivatives tested herein. This
is illustrated by the fact that the LC-MS chromatograms also show significant amounts of
dipeptide product being formed in the absence of gTG2 (2.4- to 4-fold catalyzed/uncatalyzed
product ratio at the 20 min time-point). However, dipeptide products are formed more rapidly in
the presence of enzyme, confirming the intrinsic peptide synthase activity of wild-type gTG2.
This is particularly clear at the earlier time points, where the catalyzed/uncatalyzed product ratio
is 4- to 8-fold after 10 min and greater than 10-fold for GlyNH2 at the 2 min time point.
3.2 Donor substrate specificity of gTG2-catalyzed peptide synthesis
Having confirmed that gTG2 catalyzes the synthesis of Cbz-Gly-Xaa dipeptides, we next
investigated whether alternate donor substrates could be utilized. Because the natural substrate of
gTG2 is the -carboxamide group of an L-Gln residue, we first tested the amide analog of Cbz-
Gly-7HC, Cbz-GlyNH2, using GlyNH2 as the acceptor substrate. Cbz-GlyNH2 does not react
with gTG2 at concentrations up to 50 mM (Fig. S2), its solubility limit. It has previously been
observed that aromatic ester acyl-donor substrates of gTG2 have a lower KM value than the
corresponding amide: Cbz-L-Gln-Gly has an apparent KM of 4.1 mM in the hydrolysis reaction
[38] whereas its aromatic ester analog N-carbobenzyloxy-L-glutamyl(γ-p-nitrophenyl
ester)glycine has an apparent KM of 0.02 mM [12]. The two orders of magnitude lower KM of the
aromatic ester could be due to improved binding conferred by the p-nitrophenol aromatic leaving
group. This improved binding could also occur in Cbz-Gly-7HC relative to Cbz-GlyNH2 through
19

beneficial π-stacking interactions between the aromatic leaving group and the aromatic side
chains of the tunnel-wall residues of gTG2 (Trp241 and Trp332) [23, 39].
To investigate whether the donor-substrate specificity of gTG2 includes compounds with
a substituted α-carbon, we verified whether activated ester donors containing an amino acid other
than Gly would react with gTG2. Thus, we synthesized coumarin-7-yl esters of Cbz-protected L-
Ala and D-Ala, which contain small methyl-group side chains. Cbz-L-Ala-7HC did not react with
gTG2 (Fig. 2D) suggesting that the methyl side chain of the L-alanine residue may hinder
productive binding at the active site. However, a clearly detectable activity was observed with
Cbz-D-Ala-7HC (Fig. 2E). We thus measured the kinetic parameters for this donor substrate with
gTG2 using a fluorometric assay. Although we could not saturate the enzyme with this
compound at its solubility limit (50 µM in 5 % DMF), we were able to measure its kcat/KM, which
is approximately 50-fold lower than that of Cbz-Gly-7HC (Table 1). These results support
observations that substituents, even small ones such as methyl groups, located in close proximity
to the reactive carbonyl group of the donor cannot be accommodated readily in the gTG2 active
site and are detrimental to activity [40, 41].
3.3 Structural basis for donor substrate specificity of gTG2
To elucidate the structural basis for the observed acyl-donor specificity of gTG2, we
generated models of the acyl-enzyme intermediates formed during the gTG2-catalyzed
hydrolysis of the Cbz-Gly-7HC and Cbz-D-Ala-7HC substrates. We did not generate an acyl-
enzyme intermediate structure for hydrolysis of Cbz-L-Ala-7HC as this compound is not a gTG2
substrate. In the acyl-enzyme intermediates, the catalytic Cys277 residue of gTG2 is covalently
bound to the Cbz-glycyl or Cbz-D-alanyl moiety through a thioester bond. In the acyl-enzyme
20

intermediate for gTG2-catalyzed hydrolysis of Cbz-Gly-7HC, the glycyl group fits, with no
steric clashes, into a tunnel formed by residues Trp241, Gln276, Trp278, Trp332, and Phe334
(Fig. S3), while the Cbz phenyl ring is positioned outside the tunnel and lies in a cleft on the
surface of the enzyme. Closer inspection of this acyl-enzyme intermediate suggests that the
presence of a methyl side chain on the Cα atom resulting in an L or D configuration would be
detrimental to binding as it would clash with either residue Phe334 or Trp332, respectively (Fig.
3A). This is indeed what is observed in the acyl-enzyme intermediate structure of gTG2 with a
Cbz-D-alanyl moiety (Fig. 3B). In this model structure, the side chain of Trp332 adopts an
alternate conformation, presumably to alleviate unfavorable steric interactions with the methyl
side chain of D-Ala.
Based on these observations, we hypothesized that the methyl side chain of alanine is
detrimental to activity because it decreases productive binding for Cbz-D-Ala-7HC and abolishes
binding for Cbz-L-Ala-7HC. To test these hypotheses, we generated noncovalent complexes of
gTG2 bound to the hydrolysis products Cbz-Gly, Cbz-L-Ala, and Cbz-D-Ala from the acyl-
enzyme structures, and used these complexes as input structures for molecular dynamics
simulations. The goal of these simulations was to evaluate the binding modes of products in the
gTG2 active site. Carboxylic acid products of hydrolysis were selected as ligands because gTG2
should be able to bind these compounds due to microscopic reversibility and because we could
not unambiguously specify where the 7HC group would bind. To evaluate the efficiency with
which these compounds are bound in the gTG2 active site, we measured the distance between the
nucleophilic sulfur atom of the catalytic residue Cys277 and the electrophilic carbonyl carbon of
the Cbz-Gly, Cbz-D-Ala, or Cbz-L-Ala products during the course of a 1.5-nanosecond MD
simulation. As shown in Figure 4A, sulfur-carbon distances for Cbz-Gly and Cbz-D-Ala are
21

much lower than those obtained for Cbz-L-Ala. Specifically, the sulfur-carbon distance for Cbz-
Gly is centered at approximately 5 Å for the duration of the MD simulation while for Cbz-D-Ala,
this distance increases to approximately 7 Å after 0.5 nanosecond, suggesting a second distinct
binding mode (Fig. 4B). On the other hand, the sulfur-carbon distance for Cbz-L-Ala remains
centered at approximately 15 Å throughout the simulation. This significantly higher distance
results from the fact that the Cbz-L-Ala molecule exits rapidly the gTG2 active site, suggesting
that it cannot be bound by the enzyme, in agreement with our kinetic data.
In light of our results, we propose that Cbz-D-Ala-7HC is a poor substrate and that Cbz-
L-Ala-7HC is not a substrate of gTG2 because their methyl side chain clashes with tunnel wall
residue Trp332 or Phe334, respectively. We postulate that the clash between the methyl side
chain of D-Ala and Trp332 can be more easily accommodated in the active site because Trp332
is located on a loop formed by residues Asn318-Asn333. This loop is highly flexible, as
illustrated by the fact that no electronic density is present for most of the residues comprising it
in the human TG2 crystal structure [23]. The higher mobility of this loop would enable Trp332
to move out of the way from the methyl side chain of D-Ala, allowing retention of catalytic
activity, albeit at a lower level. On the other hand, Phe334 is part of a β-sheet formed by residues
Thr295-Phe301 and Phe334-Trp341, and its phenyl side chain is stacked against the backbone of
residues Gln169 and Gly170. These interactions would make Phe334 more rigid, preventing it
from moving away from the methyl side chain of L-Ala. These observations raise the possibility
that the acyl-donor substrate scope of gTG2 may be expanded by mutating Phe334 and Trp332
in order to increase the space available for substrates containing alternate side chains.
4. Discussion
22

The peptide synthase activity of gTG2 described here results from its transferase activity:
the enzyme can transfer a Gly or a D-Ala moiety onto the α-amino group of various amino acid
derivatives, thus forming peptide bonds. The transamidase activity of gTG2 relies on its capacity
to exclude water from the active site [42]. If water had free access to the thioester bond of the
covalent acyl-enzyme intermediate, amine acyl-acceptor substrates would not be able to compete
with it for the acyl transfer reaction, as water is much more abundant. Thus, the intermediate
thioester must be sequestered in the active site long enough for amines to enter and act as acyl-
acceptor substrates. The exclusion of water may result from the hydrophobicity of residues that
form the tunnel leading to the catalytic residues, namely Trp241, Trp332, and Phe334. This
ability to exclude water from the active site differentiates TG2 from the cysteine proteases, such
as papain, that share a similar segment of α-helix and β-sheet containing the catalytic triad [43].
Protease-catalyzed peptide synthesis is the topic of much current research [3]. Serine and
cysteine proteases can catalyze peptide synthesis through a kinetically controlled process in
which the protease (hydrolase) acts as a transamidase [44, 45]. This process requires a protease
that can form a covalent acyl-enzyme intermediate, as is the case with gTG2. Competition
between hydrolysis and aminolysis is always present during the degradation of this acyl-enzyme
intermediate, resulting in lower yields for the synthesis of peptides, since proteases are not
efficient at excluding water from their active site. Furthermore, proteases hydrolyze the peptide
products, further lowering the overall yield of peptide synthesis.
Proteases used in kinetically controlled peptide synthesis have reported
transamidation/hydrolysis ratios in the range of 102-104 [46, 47] whereas gTG2 has a similar 102
increase in rate of transamidation relative to hydrolysis when the acceptor substrate is
hydroxylamine [12]. Further, the catalytic efficiency of wild-type gTG2 for the synthesis of
23

various Cbz-Gly-L-Xaa dipeptides ranges from 12 to 141 M-1s-1 [18]. Papain, a cysteine protease
that has been used in peptide synthesis, has catalytic efficiencies of 5 and 49 M-1s-1 for the
synthesis of the Boc-Gly-L-Phe-N2H2Ph dipeptide and the Boc-L-Tyr(Bzl)-Gly-Gly-L-Phe-L-
Leu-N2H2Ph pentapeptide, respectively [48]. These comparisons suggest that gTG2 could also be
used as a catalyst for the synthesis of peptide bonds.
An advantage of gTG2-catalyzed peptide synthesis is that it requires no organic co-
solvent. Indeed, the 5 % DMF used in the transamidation assay of Cbz-Gly-7HC and amino acid
derivatives by gTG2 is required only to help solubilize the acyl-donor substrate. This is not the
case with papain, with which the synthesis of peptides must be carried out in a mixture
containing 40 % ethanol [48, 49] in order to decrease the activity of water. A further advantage
of gTG2 for the synthesis of peptide bonds is that the enzyme cannot recognize secondary
amides as acyl-donor substrates, thus limiting hydrolysis of the dipeptide product and potentially
increasing yields. However, gTG2 suffers from its apparent need of an aromatic leaving group in
acyl-donor substrates, a limitation for peptide synthesis. In addition, the narrow specificity for
the amino acid residue found at the C-terminus of acyl-donors hinders the general applicability
of gTG2 for peptide synthesis. Nevertheless, it may be possible to expand the specificity of
gTG2 for additional donor substrates by mutating active site residues Trp332 and Phe334 that
form part of the substrate binding tunnel.
5. Conclusion
Herein, we confirmed the peptide synthase activity of wild-type gTG2 using LC-MS.
This enzyme can form peptide bonds between Cbz-protected Gly or D-Ala, and a variety of polar
or hydrophobic amino acid derivatives with a catalytic efficiency similar to the cysteine protease
24

papain. Although the specificity of gTG2 for peptide bond formation is limited, future
engineering efforts based on our computational models to expand its donor substrate specificity
may lead to the development of a new tool for the enzymatic synthesis of peptides and
complement the known specificities of other proteases.
Acknowledgements
R. A. C., J. N. P and J. W. K. acknowledge grants from the Natural Sciences and Engineering
Research Council of Canada (NSERC), and the Canada Foundation for Innovation. R. A. C. also
acknowledges a grant from the Ontario Research Fund; J. N. P and J. W. K. acknowledge a grant
from FRQNT. A. D. S. is the recipient of an Ontario Graduate Scholarship.
25

References
[1] V.R. Pattabiraman, J.W. Bode, Nature, 480 (2011) 471-479.
[2] D.J.C. Constable, P.J. Dunn, J.D. Hayler, G.R. Humphrey, J.L. Leazer, R.J. Linderman, K. Lorenz, J.
Manley, B.A. Pearlman, A. Wells, A. Zaks, T.Y. Zhang, Green Chem, 9 (2007) 411-420.
[3] K. Yazawa, K. Numata, Molecules, 19 (2014) 13755.
[4] L. Fesus, M. Piacentini, Trends in biochemical sciences, 27 (2002) 534-539.
[5] M. Griffin, R. Casadio, C.M. Bergamini, Biochem J, 368 (2002) 377-396.
[6] L. Lorand, R.M. Graham, Nature reviews. Molecular cell biology, 4 (2003) 140-156.
[7] J.W. Keillor, C.M. Clouthier, K.Y. Apperley, A. Akbar, A. Mulani, Bioorganic chemistry, 57 (2014) 186-
197.
[8] D. Aeschlimann, M. Paulsson, Thrombosis and haemostasis, 71 (1994) 402-415.
[9] J.E. Folk, S.I. Chung, Methods Enzymol, 113 (1985) 358-375.
[10] D.D. Clarke, M.J. Mycek, A. Neidle, H. Waelsch, Archives of biochemistry and biophysics, 79 (1959)
338-354.
[11] P. de Macedo, C. Marrano, J.W. Keillor, Analytical Biochemistry, 285 (2000) 16-20.
[12] A. Leblanc, C. Gravel, J. Labelle, J.W. Keillor, Biochemistry, 40 (2001) 8335-8342.
[13] D. Halim, K. Caron, J.W. Keillor, Bioorganic & Medicinal Chemistry Letters, 17 (2007) 305-308.
[14] S.M. Gillet, J.N. Pelletier, J.W. Keillor, Anal Biochem, 347 (2005) 221-226.
[15] C. Klock, X. Jin, K. Choi, C. Khosla, P.B. Madrid, A. Spencer, B.C. Raimundo, P. Boardman, G. Lanza,
J.H. Griffin, Bioorg Med Chem Lett, 21 (2011) 2692-2696.
[16] K. Choi, M. Siegel, J.L. Piper, L. Yuan, E. Cho, P. Strnad, B. Omary, K.M. Rich, C. Khosla, Chemistry &
Biology, 12 (2005) 469-475.
[17] C. Pardin, S.M.F.G. Gillet, J.W. Keillor, Bioorgan Med Chem, 14 (2006) 8379-8385.
[18] J.W. Keillor, R.A. Chica, N. Chabot, V. Vinci, C. Pardin, E. Fortin, S.M.F.G. Gillet, Y. Nakano, M.T.
Kaartinen, J.N. Pelletier, W.D. Lubell, Can J Chem, 86 (2008) 271-276.
[19] A.D. Pehere, A.D. Abell, Tetrahedron Lett, 52 (2011) 1493-1494.
[20] S.M. Gillet, R.A. Chica, J.W. Keillor, J.N. Pelletier, Protein Expr Purif, 33 (2004) 256-264.
[21] C. Pardin, I. Roy, R.A. Chica, E. Bonneil, P. Thibault, W.D. Lubell, J.N. Pelletier, J.W. Keillor,
Biochemistry, 48 (2009) 3346-3353.
[22] M.A. Larkin, G. Blackshields, N.P. Brown, R. Chenna, P.A. McGettigan, H. McWilliam, F. Valentin, I.M.
Wallace, A. Wilm, R. Lopez, J.D. Thompson, T.J. Gibson, D.G. Higgins, Bioinformatics, 23 (2007) 2947-
2948.
[23] D.M. Pinkas, P. Strop, A.T. Brunger, C. Khosla, PLoS Biol, 5 (2007) e327.
[24] N. Eswar, B. Webb, M.A. Marti-Renom, M.S. Madhusudhan, D. Eramian, M.Y. Shen, U. Pieper, A.
Sali, Current protocols in bioinformatics / editoral board, Andreas D. Baxevanis ... [et al.], Chapter 5
(2006) Unit 5 6.
[25] P. Labute, Proteins-Structure Function and Bioinformatics, 75 (2009) 187-205.
[26] Chemical Computing Group Inc., in, Chemical Computing Group Inc., 1010 Sherbooke St. West, Suite
#910, Montreal, QC, Canada, H3A 2R7, 2012.
[27] J. Wang, P. Cieplak, P.A. Kollman, Journal of Computational Chemistry, 21 (2000) 1049-1074.
[28] I.W. Davis, A. Leaver-Fay, V.B. Chen, J.N. Block, G.J. Kapral, X. Wang, L.W. Murray, W.B. Arendall,
3rd, J. Snoeyink, J.S. Richardson, D.C. Richardson, Nucleic Acids Res, 35 (2007) W375-383.
[29] S.N. Murthy, S. Iismaa, G. Begg, D.M. Freymann, R.M. Graham, L. Lorand, Proc Natl Acad Sci U S A,
99 (2002) 2738-2742.
[30] B.D. Allen, A. Nisthal, S.L. Mayo, Proc Natl Acad Sci U S A, 107 (2010) 19838-19843.
[31] R.A. Chica, M.M. Moore, B.D. Allen, S.L. Mayo, Proc Natl Acad Sci U S A, 107 (2010) 20257-20262.
26

[32] H.K. Privett, G. Kiss, T.M. Lee, R. Blomberg, R.A. Chica, L.M. Thomas, D. Hilvert, K.N. Houk, S.L.
Mayo, Proc Natl Acad Sci U S A, 109 (2012) 3790-3795.
[33] R.L. Dunbrack, F.E. Cohen, Protein Science, 6 (1997) 1661-1681.
[34] S.L. Mayo, B.D. Olafson, W.A. Goddard, J. Phys. Chem., 94 (1990) 8897-8909.
[35] B.I. Dahiyat, S.L. Mayo, Proc Natl Acad Sci U S A, 94 (1997) 10172-10177.
[36] P. Gerber, K. Müller, Journal of computer-aided molecular design, 9 (1995) 251-268.
[37] J.C. Phillips, R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid, E. Villa, C. Chipot, R.D. Skeel, L. Kalé, K.
Schulten, Journal of Computational Chemistry, 26 (2005) 1781-1802.
[38] N. Day, J.W. Keillor, Anal Biochem, 274 (1999) 141-144.
[39] R.A. Chica, P. Gagnon, J.W. Keillor, J.N. Pelletier, Protein Sci, 13 (2004) 979-991.
[40] M. Gross, J.E. Folk, J Biol Chem, 248 (1973) 1301-1306.
[41] M. Gross, J.E. Folk, J Biol Chem, 248 (1973) 6534-6542.
[42] Z. Nemes, T. Csosz, G. Petrovski, L. Fesus, Prog Exp Tumor Res, 38 (2005) 19-36.
[43] T. Kashiwagi, K. Yokoyama, K. Ishikawa, K. Ono, D. Ejima, H. Matsui, E. Suzuki, Journal of Biological
Chemistry, 277 (2002) 44252-44260.
[44] C. Lombard, J. Saulnier, J.M. Wallach, Protein Peptide Lett, 12 (2005) 621-629.
[45] J. Mitsuhashi, T. Nakayama, A. Narai-Kanayama, Enzyme and Microbial Technology, 75–76 (2015)
10-17.
[46] D. Kumar, T.C. Bhalla, Applied microbiology and biotechnology, 68 (2005) 726-736.
[47] L. Riechmann, V. Kasche, Biochimica Et Biophysica Acta, 830 (1985) 164-172.
[48] W. Kullmann, Biochemical Journal, 220 (1984) 405-416.
[49] W. Kullmann, Journal of Biological Chemistry, 255 (1980) 8234-8238.
27

Figure Captions
Scheme 1. Peptide synthesis reaction catalyzed by gTG2.
28

Figure 1. Aromatic ester donor substrates of gTG2. The reactive carbonyl group of these
compounds is indicated by an arrow. (A) N-carbobenzyloxy-L-glutamyl(γ-p-nitrophenyl
ester)glycine; (B) Cbz-Gly-GABA-7HC and Cbz-Phe-GABA-7HC; (C) Cbz-Gly-7HC.
29