1
Modification and de novo design of non-ribosomal peptide synthetases (NRPS)
using specific assembly points within condensation domains
Kenan A. J. BozhĆ¼yĆ¼k
1
, Annabell Linck
1
, Andreas Tietze
1
, Frank Wesche
1
, Sarah
Nowak
1
, Florian Fleischhacker
1
, Helge B. Bode
1,2
*
1
Fachbereich Biowissenschaften, Merck Stiftungsprofessur fĆ¼r Molekulare
Biotechnologie, Goethe-UniversitƤt Frankfurt, Frankfurt am Main 60438, Germany.
2
Buchmann Institute for Molecular Life Sciences, Goethe-UniversitƤt Frankfurt,
Frankfurt am Main 60438, Germany.
K.A.J.B. and A.L. contributed equally to this work.
*e-mail: h.bode@bio.uni-frankfurt.de
Abstract
Many important natural products are produced by non-ribosomal peptide synthetases
(NRPSs)
1
.These giant enzyme machines activate amino acids in an assembly line
fashion in which a set of catalytically active domains is responsible for the section,
activation, covalent binding and connection of a specific amino acid to the growing
peptide chain
1,2
. Since NRPS are not restricted to the incorporation of the 20
proteinogenic amino acids, their efficient manipulation would give access to a diverse
range of peptides available biotechnologically. Here we describe a new fusion point
inside condensation (C) domains of NRPSs that enables the efficient production of
peptides, even containing non-natural amino acids, in yields higher than 280 mg/L.
The technology called eXchange Unit 2.0 (XU
2.0
) also allows the generation of
targeted peptide libraries and therefore might be suitable for the future identification
of bioactive peptide derivatives for pharmaceutical and other applications.
not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which wasthis version posted June 24, 2018. ; https://doi.org/10.1101/354670doi: bioRxiv preprint
2
Introduction
Secondary metabolite derived drugs have become essential agents to cure infectious
diseases during the last almost 70 years
3,4
. Yet, infectious diseases are still the
second major cause of death worldwide and furthermore, the world is facing a global
public-health crisis as there is a growing risk of re-entering a pre-antibiotic era, since
more and more infections are caused by multi-drug-resistant bacteria
5
.
One source of new antibacterial agents are non-ribosomally made peptides (NRPs).
Their high structural diversity imparts to them many properties of biological relevance
and peptides have been identified with antibiotic, antiviral, anti-cancer, anti-
inflammatory, immunosuppressant and surfactant qualities
6, 7, 8
. However, natural
products often need to be modified to improve clinical properties and/or bypass
resistance mechanisms
9,10
. To date, most clinically used NP derivatives are created
by means of semi-synthesis
9,11
. A promising alternative strategy is the use of
engineering approaches to modify NRP producing non-ribosomal peptide synthetase
(NRPS) directly in order to produce optimized or non-natural natural products
12
.
However, to date most attempts to achieve this have yielded impaired or non-
functional biosynthetic machineries
7,13
.
NRPSs are large multienzyme complexes (megasynthases)
14
that form peptides not
limited to the twenty proteinogenic amino acids (AA)
15
. Furthermore, these NRPS can
generate linear or cyclic peptides containing D-AA, N-methylated AA, N-terminal
attached fatty acids (FA) or heterocycles
1,2,14,15
. NRPS do this by exhibiting a strict
modular architecture in which a module is defined as the catalytic unit responsible for
the incorporation of one specific building block (e.g. AA) into the growing peptide
chain (N ā C) and associated functional group modifications
16
. Modules are
composed of domains that catalyze the single reaction steps like activation, covalent
binding, optional modification of the building blocks, and condensation with the amino
not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which wasthis version posted June 24, 2018. ; https://doi.org/10.1101/354670doi: bioRxiv preprint
3
acyl or peptidyl group on the neighboring module
17
. At least three domains or
essential enzymatic activities, respectively, are necessary for the non-ribosomal
production of peptides (Fig. 1). They reside in the adenylation (A) for AA activation,
thiolation (T) for AA tethering, and condensation (C) domains for peptide bond
formation. Finally, most NRPS termination modules harbor a TE domain that releases
the peptide, often in a cyclized form. These standard domains are additionally joined
by tailoring domains that can catalyze epimerization (E), methylation (MT), cyclization
(CY) or other modifications of the building blocks or the growing peptide chain
1
.
Due to the modular character of the NRPS scientists strived to reprogram these
systems via (I) the substitution of the A or paired A-T domain activating an alternative
substrate, (II) the targeted alteration of just the substrate binding pocket of the A
domain or (III) substitutions that treat C-A or C-A-T domain units as inseparable pairs
7
. These strategies are complemented by recombination studies which have sought
to re-engineer NRPS by T
18
, T-C-A
19
, communication domain
20
and A-T-C
swapping
21
. However, with exception of the latter and recently published strategy,
denoted as the concept of eXchange Units (XU)
22
, it has been difficult to develop
clearly defined, reproducible and validated guidelines for engineering modified
NRPS.
The limitation of the XU-concept is that the natural downstream C domain specificity
must be obeyed clearly restricting its applicability and the C-domain specificities have
to be met - at the donor as well as at the acceptor site. This disadvantage can be
accepted if a large number of XUs with different downstream C domains are
available. Due to these limitations also at least two XUs have to be exchanged to
produce a new peptide derivative that differs in one AA position from the primary
sequence of the wild type (WT) peptide
22
. However, a more flexible system reducing
the limitations of C-domain specificities would drastically reduce the amount of NRPS
not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which wasthis version posted June 24, 2018. ; https://doi.org/10.1101/354670doi: bioRxiv preprint
4
building blocks necessary to produce or alter particular peptides and would enable
the creation of artificial natural product libraries with hundreds or thousands of
entities for large scale bioactivity screenings.
Results and Discussion
C-domains have acceptor site substrate specificity
To verify the influence of the C-domains acceptor site (C
Asub
) proof reading activity,
the GameXPeptide producing NRPS GxpS of Photorhabdus luminescens TT01
(Supplementary Figure 1 and 2) was chosen as a model system
23,24
. A recombinant
GxpS was constructed, not complying with the C-domain specificity rules of the XU
concept
22
. Here, XU2 of GxpS (Fig. 1b, NRPS-1) was exchanged against XU2 of the
bicornutin producing NRPS (BicA, Fig. 1c)
25
. Although both XUs are Leu specific,
they are differentiated by their C
Asub
specificities - Phe for XU2 of GxpS and Arg for
XU2 of BicA. Therefore, no peptide production was observed as expected. This
experiment confirmed previously published scientific results from in vitro experiments
26ā29
, and illustrates that C domains indeed are highly substrate specific at their C
Asub
.
From the available structural data of C domains it is clear that they show a pseudo-
dimer configuration
28,30ā32
with their catalytic center, including the HHXXXDG motif,
having two binding sites - one for the electrophilic donor substrate and one for the
nucleophilic acceptor substrate
29
(Fig. 1a and Supplementary Figure 3). Therefore
we concluded that the four AA long conformationally flexible loop/linker between both
subdomains might be the ideal target to reconfigure C domain specificities via the
engineering of C domain hybrids (Fig. 1a). For this purpose the Arg specific C
Asub
of
the GxpS-BicA hybrid NRPS (Fig. 1b, NRPS-1) was re-exchanged to the Leu specific
C
Asub
of GxpS, restoring the functionality of the hybrid NRPS (NRPS-2) and leading
to the production of GameXPeptide A-D (1-5) in 217% (107 mg/L) yield compared to
not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which wasthis version posted June 24, 2018. ; https://doi.org/10.1101/354670doi: bioRxiv preprint
5
the WT GxpS (Fig. 1b) as confirmed by MS/MS analysis and comparison of the
retention times with a synthetic standard (Supplementary Figure 4).
The eXchange Unit 2.0 concept
From these results in conjunction with bioinformatics analysis, we concluded that C-
domains acceptor and donor site (C
Dsub
) mark a self-contained catalytically active unit
C
Asub
-A-T-C
Dsub
(XU
2.0
) without interfering major domain-domain
interfaces/interactions during the NRPS catalytic cycle
33
. In order to validate the
proposed XU
2.0
building block (Fig. 1c) and to compare the production titers with a
natural NRPS, we reconstructed GxpS (Fig. 1b) in two variants (Fig. 2a, NRPS-3 and
-4). Each from five XU
2.0
building blocks from four different NRPSs (XtpS, AmbS,
GxpS, GarS, HCTA) (Supplementary Figure 5):
NRPS-3 showed a mixed C/E
Dsub
-C
Asub
-domain
between XU
2.0
3 and XU
2.0
4 (Fig. 2a),
to reveal if C and C/E domains can be combined. In NRPS-4 XU
2.0
3 from HCTA
instead of GarS was used in order to prevent any incompatibilities (Fig. 2a).
Whereas NRPS-3 (Fig. 2a) showed no detectable production of any peptide, NRPS-4
(Fig. 2) resulted in the production of 1 and 3 in 66 and 6 % yield, respectively,
compared to the natural GxpS, as confirmed by MS/MS analysis and comparison of
the retention times of synthetic standards (Fig. 2a, Supplementary Figure 6). In line
with expectations from domain sequences, phylogenetics, as well as structural
idiosyncrasies of C/E- and C-domains
29
, it may be deduced from these results that
C/E and C-domains cannot be combined with each other. Although NRPS-4 (Fig. 2a)
showed moderately reduced production titers, most likely due to the non-natural
C
Dsub
-C
Asub
pseudo-dimer interface, the formal exchange of the promiscous XU
2.0
1
from GxpS (for Val/Leu) against the Val-specific XU
2.0
1 from XtpS led to exclusive
production of 1 and 3 (Fig. 2a) without production of 2 and 4 observed in the original
not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which wasthis version posted June 24, 2018. ; https://doi.org/10.1101/354670doi: bioRxiv preprint
