Directed evolution of sortase A mutants with altered substrate selectivity profiles

Article abstract of DOI:10.1021/ja205630g

The ligation of two polypeptides in a chemoselective manner by the bacterial transpeptidase sortase A has become a versatile tool for protein engineering approaches. When sortase-mediated ligation is used for protein semisynthesis, up to four mutations resulting from the strict requirement of the LPxTG sorting motif are introduced into the target protein. Here we report the directed evolution of a mutant sortase A that possesses broad substrate selectivity. A phage-display screen of a mutant sortase library that was randomized in the substrate recognition loop was used to isolate this mutant. The altered substrate selectivity represents a gain-of-function that was exploited for the traceless semisynthesis of histone H3. Our report is a decisive step toward a platform of engineered sortases with distinct ligation properties that will conceivably allow for more versatile assemblies of modified proteins in biotechnological approaches.

Full text of DOI:10.1021/ja205630g

COMMUNICATION
pubs.acs.org/JACS
Directed Evolution of Sortase A Mutants with Altered Substrate
Selectivity Profiles
Kirill Piotukh,† Bernhard Geltinger,‡ Nadja Heinrich,|| Fabian Gerth,† Michael Beyermann,*,||
Christian Freund,*,† and Dirk Schwarzer*,‡
Departments of †Protein Engineering, ‡Protein Chemistry, and Peptide Synthesis, Leibniz-Institut f€ur Molekulare Pharmakologie,
Robert-R€ossle-Strasse 10, 13125 Berlin, Germany
S Supporting Information
b
ABSTRACT: The ligation of two polypeptides in a che-
moselective manner by the bacterial transpeptidase sortase
A has become a versatile tool for protein engineering approaches.
When sortase-mediated ligation is used for protein semi-
synthesis, up to four mutations resulting from the strict
requirement of the LPxTG sorting motif are introduced into
the target protein. Here we report the directed evolution of a
mutant sortase A that possesses broad substrate selectivity.
A phage-display screen of a mutant sortase library that was
randomized in the substrate recognition loop was used to
isolate this mutant. The altered substrate selectivity repre-
sents a gain-of-function that was exploited for the traceless
semisynthesis of histone H3. Our report is a decisive step
toward a platform of engineered sortases with distinct ligation
properties that will conceivably allow for more versatile
assemblies ofmodified proteins in biotechnological approaches.
Figure 1. Directed evolution of sortase A. (a) Sortase-mediated ligation
of peptides and proteins. (b) The G5-sortase catalyzes self-ligation with
peptides containing the LPxTG sorting motif. (c) Product capture
strategy for selecting sortase mutants possessing an altered substrate
selectivity.
interest. Here we report the evolution of a sortase A mutant with
relaxed substrate selectivity by phage-display. Instead of the native
leucine, this mutant tolerates several other residues in the first
position of the sorting motif and recognizes a native sequence of
histone H3, which can be exploited for traceless semisynthesis of
this protein.
hemoselective ligation techniques have endowed scientists
C
with powerful tools for studying the structureÀfunction
relationship in proteins. These techniques enable site-specific
modification, semisynthesis, and even the total chemical synth-
esis of proteins.1 Complementary to chemical approaches like
native chemical ligation, enzyme-mediated ligation schemes
based on the bacterial transpeptidase sortase A have gained
considerable attention.2 Gram-positive bacteria use sortases to
anchor surface proteins to their peptidoglycan layers.3 The ligation
process is reminiscent of that of cysteine protease reactions: In
the first step, sortase A recognizes a highly conserved LPxTG
sorting motif in the target protein and cleaves this sequence
C-terminal to the Thr residue. The resulting thioester inter-
mediate—instead of being hydrolyzed—is ligated to the penta-
Gly unit of the bacterial peptidoglycan, leading to a covalently linked
cell wall protein. While the LPxTG sequence is a strict necessity
for sortase A recognition, a single glycine of the amino compo-
nent is sufficient for the ligation step, requiring only a minimal
modification of the N-terminal ligation fragment (Figure 1a).4b
However, when sortase-mediated ligation is used for protein
semisynthesis, where the native amino acid sequence of the target
protein is ideally restored, the high selectivity of the LPxTG motif
becomes a downside. We reasoned that a broader selectivity
profile of sortase A would be advantageous, because fewer or no
amino acid changes need to be introduced into the protein of
In the first step of this process, we developed a product capture
strategy to establish a covalent bond between the enzyme and its
substrate. Such a strategy prevents the accumulation of sortase
mutants that hydrolyze the sorting motif, and it was successfully
applied to the directed evolution of subtiligase, an engineered
protease that catalyzes ligation reactions under conditions of
kinetically controlledreverseproteolysis.5 To this end, we elongated
the N-terminus of Staphylococcus aureus sortase A by a penta-Gly-
containing flexible tail (G5-sortase) (Figure 1b). G5-sortase
efficiently captured a fluorescein-conjugated substrate peptide
(Fl-Ahx-LPKTGGRR-NH2, where Fl = 5(6)-carboxyfluorescein
and Ahx = 6-aminohexanoic acid) by intramolecular ligation
(Supporting Information, Figure S1). To ensure that the ob-
served reaction is indeed an intramolecular product capture and
not an intermolecular reaction, we additionally performed experi-
ments under enforced intermolecular ligation conditions. There-
fore, we created a G5-sortase C184A mutant which is catalytically
inactive but still serves as a ligation partner for the Fl-Ahx-
LPKTGGRR-NH2 peptide when wild-type (wt) sortase without
the N-terminal extension is added. When competing free penta-
Gly peptide was added to either G5-sortase or G5-sortase-C184A/
Received: June 17, 2011
Published: October 06, 2011
r
2011 American Chemical Society
17536
dx.doi.org/10.1021/ja205630g J. Am. Chem. Soc. 2011, 133, 17536–17539
|

Journal of the American Chemical Society
COMMUNICATION
Figure 3. Comparison of the ligation of preferred substrates by wt- and
F40-sortase (equimolar concentrations). The left graph highlights the
difference in reaction kinetics (time scale in minutes), indicating the
reduced ligation efficiency of the F40 mutant. The right graph includes a
time point at 24 h showing that equilibrium for ligation of the preferred
substrates is reached. After 24 h, slow hydrolysis of the LPxTG motif by
wt-sortase A is observed under these conditions, resulting in decreased
ligation yields for the last data point.
Figure 2. Characterizing the substrate preferences of the first residue in
sorting motif (X-PKTG) of evolved sortase. Ligation reactions with a
sortase motif library were analyzed for product formation by LC-MS.
Product formation yields of identified ligation products were determined
using individual substrate peptides. At ∼55% the ligation reactions have
reached equilibrium. Reaction profiles of (a) wt-sortase, (b) F21-sortase,
and (c) F40-sortase are shown. (d) Amino acid sequence of the randomized
β6Àβ7 loop region of selected sortase mutants. n.t. = not tested; F* =
Dns-FAKTGGRR-NH2.
Phe as well as other amino acids as the first residue of the sorting
motif. Two mutants (F21 and F40) were further characterized in
quantitative terms for the individual peptides that were identified
as substrates in the library screen. The F21 mutant displayed a
2.5-fold higher ligation efficiency with respect to the FPxTG-
motif than wt-sortase, but the native LPxTG remained the favored
substrate of this mutant (Figure 2b). In contrast, the F40 mutant
preferred ligation of the FPxTG motif over the native LPxTG
sequence, but the overall ligation efficiency was low for both
motifs. In general, the F40-sortase possessed a remarkable broad
selectivity and accepted aromatic amino acids as well as residues
with small side chains, including Ala, Asp, Ser, Pro, and Gly, at
position 1 of the sorting motif (Figure 2c and Figure S7). Based
on the measured reaction profile, Ala is among the most
preferred amino acids in the first position of the F40-sorting
motif. To assess the ligation yield that can be obtained with the
F40-sortase, we monitored the ligation reaction with a Dns-
APKTGGRR-NH2 peptide (where Dns = dansyl) over a time
course of 24 h (Figure 3). A final equilibrium level of ∼55% was
reached after 24 h. For the wt-sortase, in contrast, we detected no
significant ligation of the APxTG peptide under comparable
conditions, while the equilibrium level for the ligation of a LPxTG
peptide catalyzed by wt-sortase is reached within 20 min (Figure 3).
These observations showed that the F40-sortase has lost activity
compared to the wt enzyme but is able to produce similar ligation
yields over an extended period of time.
The mutations selected in the F40-sortases varied largely from
the wt sequence of the β6Àβ7 loop (Figure 2d). A simple
explanation for the enhanced promiscuity is not evident from the
current sortase A structures, but we noticed an E171A mutation
in the F40-sortase. The E171 residue is involved in calcium binding,
and we therefore analyzed whether a loss of Ca2+ coordination
caused the promiscuity of the F40 mutant. Such loss of Ca2+
binding conceivably increases the flexibility of the long β6Àβ7
loop and thereby allows for the recognition of a wider variety of
substrates.8 As expected, in the absence of calcium we did not
observe any ligation activity of the wt-sortase with a Dns-
APKTGGRR-NH2 peptide, which is preferentially ligated by
the F40-sortase mutant (Figure S8a,b). More importantly, we
observed that the F40 mutant maintained Ca2+ dependency
when ligating the Dns-APKTGGRR-NH2 peptide, probably
because the neighboring E172 mutation restores the Ca2+ binding
site (Figure S8c). This observation indicates that the selected
wt-sortase, we observed efficient self-ligation of the G5-sortase
but noligationin trans with G5-sortase-C184A/wt-sortase (Figure S1).
These experiments showed that the reaction in cis is clearly
favored over the ligation in trans. In the following we establish
monovalent display of the G5-sortase as pIII fusion protein on
the envelop of the M13 phage (Figure 1c). Western blot analysis
of the ligation reaction with a biotinylated substrate peptide
(biotin-GLPKTGGRR-NH2) confirmed functional presentation
of the G5-sortase (Figure S2).
Having established a phage-display system for selecting evolved
sortases, we set out to construct a library of sortase mutants
(Figures S3 and S4). Initially, we intended to screen for sortases
which recognize an FPxTG sorting motif instead of the native
LPxTG sequence. The reasons for choosing this variation were
two-fold: on the one hand bioinformatics approaches indicated
that FPxTG sorting sequences exist in nature (although the
corresponding sortases remain unknown), and on the other hand
we observed marginal ligation of this motif catalyzed by wt-
sortase (Figure 2a).4a Library design was based on the crystal
structure of a C184A mutant of sortase A in complex with a
natural substrate.6 Six amino acids in the β6Àβ7 loop, represent-
ing solvent-exposed positions in spatial proximity to the leucine
of the sorting motif, were randomized by using NNK incorpora-
tion into synthetic oligonucleotides (Figure S4b).7 A library of a
complexity of ∼108 was subsequently screened for the ligation of
the biotin-GFPKTGGRR-NH2 peptide to the phage-displayed
G5-sortase (Figure 1c). After three rounds of panning, individual
clones were sequenced and analyzed by Western blot for ligation
of the biotinylated peptide (Figure S5).
Four G5-sortases showing enhanced ligation activity toward
the FPxTG motif were subcloned and expressed without the
N-terminal G5 extension and C-terminal pIII fusion. The sub-
strate selectivity of these mutants was studied with an HPLC/
MS-based ligation assay and a library of substrate peptides, which
covered 23 variations of the sorting motif, mostly at position 1,
and an alanine scan of the remaining three invariant residues
(Figure S6 and Table S2). All tested sortase mutants tolerated
17537
dx.doi.org/10.1021/ja205630g |J. Am. Chem. Soc. 2011, 133, 17536–17539

Journal of the American Chemical Society
COMMUNICATION
A by exchanging the β6Àβ7 loop with that of sortase B, which
recognizes a different sorting motif, resulted in mutants that
possessed poor activity and hydrolyzed their substrates.11 Further-
more, the engineered F40-sortase was able to catalyze traceless
ligations of histone H3. This gain of function represents a very
useful feature for chromatin biochemistry, enabling simple access
to semisynthetic histones with defined modification patterns. In a
very recent publication, it was further shown that sortase A can be
evolved for improved catalytic activity.12 Together these inves-
tigations demonstrate amenability of sortase A to engineering
approaches, inspiring future manipulation experiments focused
on the substrate selectivity and catalytic properties of this class of
enzymes.
Figure 4. Semisynthesis of histone H3. (a) Scheme of histone H3
semisynthesis catalyzed by the F40-sortase. The native sequence of H3 is
restored at the ligation site (marked in red). (b) Fl-conjugated peptides
containing either the LPxTG or the native H3 APATG sequence were
ligated to recombinant histone H3 (33À135). The unligated H3
(33À135), 1, and the ligation products, 2, were separated on SDS-
PAGE and analyzed by Coomassie Brilliant Blue staining (bottom
panel) and UV irradiation (top panel). L = LPxTG peptide (Fl-Ahx-
LPKTGGRR-NH2); A = APxTG peptide (Fl-Ahx-SAPATGGK-NH2).
’ ASSOCIATED CONTENT
S
Supporting Information. Experimental procedures for
b
solid-phase peptide synthesis, construction and expression of
sortase constructs, phage-display, library design, ligation assays,
and additional experimental data. This material is available free of
charge via the Internet at http://pubs.acs.org.
amino acid sequence of the β6Àβ7 loop and not the loss of Ca2+
binding causes the promiscuity of the F40 mutant.
We next analyzed the human proteome for traceless ligation
sites of the F40-sortase mutant. The most interesting sequence
was the SAPATGG site in histone H3 covering residues 28À34.
Histones package DNA into chromatin and regulate gene activity
through a multitude of posttranslational modifications on their
N-terminal tail regions.9 Several ligation approaches have already
been tried in order to introduce site-specific modifications into
the tail region of histones for biological investigations.10 The
F40-sortase mutant appeared ideally suited for such an under-
taking because the APATG motif is located at the interface
between the tail and the globular domain of H3, and the Ala
residue in position 1 of the sorting motif is preferred by the F40
mutant. We subcloned and expressed a truncated version of
X. laevis H3 covering residues 33À135. Two Fl-conjugated peptides
containing either the sorting motif (Fl-Ahx-LPKTGGRR-NH2)
or the native 28À34 sequence of histone H3 (Fl-Ahx-SA-
PATGGK-NH2) were studied in ligation assays (Figure 4a).
Wt-sortase efficiently ligated the LPxTG but not the APxTG
peptide to H3 (33À135). In contrast, F40-sortase catalyzed the
ligation of the LPxTG only with marginal ligation efficiency, but
was able to ligate the APxTG to H3 (33À135), and thereby
restored the native amino acid sequence of H3 at the ligation site
(Figure 4b). Encouraged by these findings, we performed a
semisynthesis reaction to generate full-length histone H3. A
peptide covering residues 1À33 of histone H3 was synthesized
and subsequently ligated to H3 (33À135) by the F40-sortase.
Western blot analysis of the ligation reaction confirmed that F40-
sortase catalyzed the semisynthesis of full-length histone H3
(Figure S9).
In summary, we have reported the evolution of sortase A
mutants with modified substrate selectivities. Not surprisingly,
the F40-sortase mutant lost activity compared to the wt enzyme,
since our screen did not select for enhanced ligation efficiency.
However, the good solubility and stability of the engineered
sortase allowed for efficient ligation within 24 h of reaction time.
Therefore, the F40-sortase mutant can be used as a versatile tool
for in vitro applications for which low enzymatic efficacy is not
limiting. To the best of our knowledge, this investigation repre-
sents the first example of an engineered sortase A that maintained
ligation activity when the substrate recognition site was manipu-
lated. Earlier attempts to alter the substrate recognition of sortase
’ AUTHOR INFORMATION
Corresponding Author
beyerman@fmp-berlin.de; freund@fmp-berlin.de; schwarzer@
fmp-berlin.de
’ ACKNOWLEDGMENT
This work was supported by the Deutsche Forschungsge-
meinschaft to D.S. (SCHW 1163/3-1) and C.F. (SFB958,
project A07). We thank Rebecca Klingberg, Till Teschke, and
Bernhard Schmikale for support with solid-phase peptide synthesis.
’ REFERENCES
(1) (a) Hackenberger, C. P. R.; Schwarzer, D. Angew. Chem., Int. Ed.
2008, 47, 10030–10074. (b) Kent, S. B. H. Chem. Soc. Rev. 2009,
38, 338–351. (c) Muralidharan, V.; Muir, T. W. Nat. Methods 2006,
3, 429–438. (d) Sletten, E. M.; Bertozzi, C. R. Angew. Chem., Int. Ed.
2009, 48, 6974–6998.
(2) (a) Clancy, K. W.; Melvin, J. A.; McCafferty, D. G. Biopolymers
2010, 94, 385–396. (b) Popp, M. W.; Antos, J. M.; Grotenbreg, G. M.;
Spooner, E.; Ploegh, H. L. Nat. Chem. Biol. 2007, 3, 707–708. (c) Popp,
M. W.; Ploegh, H. L. Angew. Chem., Int. Ed. 2011, 50, 5024–5032.
(d) Pritz, S.; Kraetke, O.; Klose, A.; Klose, J.; Rothemund, S.; Fechner, K.;
Bienert, M.; Beyermann, M. Angew. Chem., Int. Ed. 2008, 47, 3642–3645.
(e) Tsukiji, S.; Nagamune, T. Chembiochem 2009, 10, 787–798.
(3) (a) Ton-That, H.; Marraffini, L. A.; Schneewind, O. Biochim.
Biophys. Acta 2004, 1694, 269–278. (b) Ton-That, H.; Mazmanian,
S. K.; Alksne, L.; Schneewind, O. J. Biol. Chem. 2002, 277, 7447–7452.
(4) (a) Kruger, R. G.; Otvos, B.; Frankel, B. A.; Bentley, M.; Dostal,
P.; McCafferty, D. G. Biochemistry 2004, 43, 1541–1551. (b) Mao, H. Y.;
Hart, S. A.; Schink, A.; Pollok, B. A. J. Am. Chem. Soc. 2004, 126, 2670–
2671.
(5) Atwell, S.; Wells, J. A. Proc. Natl. Acad. Sci. U.S.A. 1999, 96,
9497–9502.
(6) Zong, Y.; Bice, T. W.; Ton-That, H.; Schneewind, O.; Narayana,
S. V. J. Biol. Chem. 2004, 279, 31383–31389.
(7) Bentley, M. L.; Lamb, E. C.; McCafferty, D. G. J. Biol. Chem.
2008, 283, 14762–14771.
(8) (a) Ilangovan, U.; Iwahara, J.; Ton-That, H.; Schneewind, O.;
Clubb, R. T. J. Biomol. NMR 2001, 19, 379–380. (b) Ilangovan, U.;
17538
dx.doi.org/10.1021/ja205630g |J. Am. Chem. Soc. 2011, 133, 17536–17539

Journal of the American Chemical Society
COMMUNICATION
Ton-That, H.; Iwahara, J.; Schneewind, O.; Clubb, R. T. Proc. Natl. Acad.
Sci. U.S.A. 2001, 98, 6056–6061.
(9) (a) Jenuwein, T.; Allis, C. D. Science 2001, 293, 1074–1080.
(b) Kouzarides, T. Cell 2007, 128, 693–705.
(10) (a) Chiang, K. P.; Jensen, M. S.; McGinty, R. K.; Muir, T. W.
Chembiochem 2009, 10, 2182–2187. (b) Chatterjee, C.; Muir, T. W.
J. Biol. Chem. 2010, 285, 11045–11050. (c) Schwarzer, D. J. Pept. Sci.
2010, 16, 530–537. (d) Shogren-Knaak, M.; Ishii, H.; Sun, J. M.; Pazin,
M. J.; Davie, J. R.; Peterson, C. L Science 2006, 311, 844–847. (e) He, S.;
Bauman, D.; Davis, J. S.; Loyola, A.; Nishioka, K.; Gronlund, J. L.;
Reinberg, D.; Meng, F.; Kelleher, N.; McCafferty, D. G. Proc. Natl. Acad.
Sci. U.S.A. 2003, 100, 12033–12038.
(11) Bentley, M. L.; Gaweska, H.; Kielec, J. M.; McCafferty, D. G.
J. Biol. Chem. 2007, 282, 657–681.
(12) Chen, I.; Dorr, B. M.; Liu, D. R. Proc. Natl. Acad. Sci. U.S.A.
2011, 108, 11399–11404.
17539
dx.doi.org/10.1021/ja205630g |J. Am. Chem. Soc. 2011, 133, 17536–17539