Synthetic Zippers as an Enabling Tool for Engineering of Non-Ribosomal Peptide Synthetases

TLDR: This work describes a strategy to functionally combine NRPS fragments of Gram-negative and -positive origin, synthesising novel peptides at titres up to 290 mg l-1, and inserts synthetic zippers to split single protein NRPSs into up to three independently expressed and translated polypeptide chains.

Abstract: Non-ribosomal peptide synthetases (NRPSs) are the origin of a wide range of natural products, including many clinically used drugs. Engineering of these often giant biosynthetic machineries to produce novel non-ribosomal peptides (NRPs) at high titre is an ongoing challenge. Here we describe a strategy to functionally combine NRPS fragments of Gram-negative and -positive origin, synthesising novel peptides at titres up to 290 mg l-1. Extending from the recently introduced definition of eXchange Units (XUs), we inserted synthetic zippers (SZs) to split single protein NRPSs into up to three independently expressed and translated polypeptide chains. These synthetic type of NRPS (type S) enables easier access to engineering, overcomes cloning limitations, and provides a simple and rapid approach to building peptide libraries via the combination of different NRPS subunits. One Sentence Summary Divide and Conquer: A molecular tool kit to reprogram the biosynthesis of non-ribosomal peptides.

Figures

Figure 1. Introduction to SZ mediated in trans protein-protein interaction. (a) 417 Schematic overview of type A and type S NRPSs. Natural DDs (in trans type A NRPS) 418 are shown in green and artificial SZs in red (in trans type S NRPS). For domain 419 assignment the following symbols are used: A, adenylation domain, large circles; T, 420 thiolation domain, rectangle; C, condensation domain, triangle; C/E, dual 421 condensation/epimerization domain, diamond; TE, thioesterase domain, small circle. 422 (b) Top: excised C-A di-domain and linker region (ribbon representation) from the SrfA-423 C termination module (PDB-ID: 2VSQ). Removed area of the C-A linker region to 424 introduce SZs is highlighted red. Bottom: modelled 41 AAs comprising SZ pair. (c) Top: 425 antiparallel (left) and parallel (right) interacting hetero-specific SZs. Bottom: SZ17:18 426 and SZ1:2 are forming an orthogonal interaction network. (d) SZ interactions. SZ17:18 427 are predicted to be electrostatic complementary at adjacent interfacial e and g 428 positions. 429

Figure 4. (a) Type S XtpS using parallel interacting SZ18:19, (b) (GS)x-elongated C-A 449 linker sequences (NRPS-21 – NRPS-23), (c) using different split positions in between 450 the T-C (NRPS-24) and A-T domains (NRPS-25). See Fig. 1 for assignment of the 451 domain symbols. 452

Figure 5. Schematic representation of three subunit type S XtpS using the SZ1:2 and 455 SZ17:18 pairs splitting in between the C-A (NRPS-26), T-C (NRPS-27) and A-T linker 456 region (NRPS-28). See Fig. 1 for assignment of the domain symbols. 457

Figure 2. (a) Type S NRPS-1 – 4, as well as corresponding peptide yields obtained 433 from triplicate experiments. (b) Type S NRPS-5 and -6, where GxpS and RtpS are split 434 in two subunits. (c) Structures of 1–6 produced from NRPS-1 to 435 NRPS-6 expressed in E. coli. See Fig. 1 for assignment of the domain symbols; further 436 symbols: CY, heterocyclization domain; E, epimerization domain. Boxed are the colour 437 coded NRPSs used as building blocks and the used SZ pairs. 438

Full text

1
Synthetic Zippers as an Enabling Tool for Engineering of Non-Ribosomal
1
Peptide Synthetases
2
Kenan A. J. Bozhueyuek
1,
†, Jonas Watzel
1,
†, Nadya Abbood
1,
†, Helge B. Bode
1,2,3,
*
3
4
1 Molecular Biotechnology, Department of Biosciences, Goethe University Frankfurt,
5
60438, Frankfurt am Main, Germany.
6
2 Buchmann Institute for Molecular Life Sciences (BMLS), Goethe University Frankfurt,
7
60438, Frankfurt am Main, Germany.
8
3 Senckenberg Gesellschaft für Naturforschung, 60325, Frankfurt am Main, Germany
9
† equal contribution
10
11
* Corresponding author
12
13
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted May 6, 2020. ; https://doi.org/10.1101/2020.05.06.080655doi: bioRxiv preprint

2
Abstract
14
Non-ribosomal peptide synthetases (NRPSs) are the origin of a wide range of natural
15
products, including many clinically used drugs. Engineering of these often giant
16
biosynthetic machineries to produce novel non-ribosomal peptides (NRPs) at high titre
17
is an ongoing challenge. Here we describe a strategy to functionally combine NRPS
18
fragments of Gram-negative and -positive origin, synthesising novel peptides at titres
19
up to 290 mg l
-1
. Extending from the recently introduced definition of eXchange Units
20
(XUs), we inserted synthetic zippers (SZs) to split single protein NRPSs into up to three
21
independently expressed and translated polypeptide chains. These synthetic type of
22
NRPS (type S) enables easier access to engineering, overcomes cloning limitations,
23
and provides a simple and rapid approach to building peptide libraries via the
24
combination of different NRPS subunits.
25
26
27
One Sentence Summary:
28
Divide and Conquer: A molecular tool kit to reprogram the biosynthesis of non-
29
ribosomal peptides.
30
31
32
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted May 6, 2020. ; https://doi.org/10.1101/2020.05.06.080655doi: bioRxiv preprint

3
Introduction
33
Non-ribosomal peptide synthetases (NRPSs) are multifunctional enzymes, producing
34
a broad range of structural diverse and valuable compounds with diverse applications
35
in medicine and agriculture (1) making them key targets for bioengineering. The
36
structural diversity of non-ribosomal peptides (NRPs) arises from the assembly line
37
architecture of their biosynthesis. According to their biosynthetic logic, known NRPS
38
systems are classified into three groups, linear (type A), iterative (type B), and
39
nonlinear NRPSs (type C) (2). Type A NRPSs are composed of sequential catalytically
40
active domains organised in modules, each responsible for the incorporation and
41
modification of one specific amino acid (AA). The catalytic activity of a canonical
42
module is based upon the orchestrated interplay of an adenylation (A) domain for AA
43
selection and activation, a condensation (C) domain to catalyse peptide bond
44
formation, and a thiolation/ peptidyl-carrier protein (T) onto which the AAs or
45
intermediates are covalently tethered (3). In addition, tailoring domains, including
46
epimerization (E), methylation, and oxidation domains can be part of a module, or a
47
heterocyclization (Cy) domain instead of a C-domain can be present. Finally, most
48
NRPS termination modules harbour a TE-domain, usually responsible for the release
49
of linear, cyclic or branched cyclic peptides (4).
50
Type A NRPSs (Fig. 1a) follow the collinearity rule, i.e. the number of NRPS modules
51
corresponds directly to the number of monomers incorporated into the associated
52
product, and the arrangement of the modules directly follows the peptides primary
53
sequence (5). Whereas in in cis type A NRPSs all modules are arranged on a single
54
polypeptide chain (e.g. ACV-synthetase (6)), in trans assembly-lines comprise a
55
number of individual proteins (Daptomycin-synthetase (7)). Mutual protein-protein
56
interactions of the latter are mediated by specialized C- (donor) and N-terminally
57
(acceptor) attached ~30 AAs long a-helical structural elements, so called
58
communication mediating (COM) or docking domains (DDs) (8). DDs typically are
59
located in between two modules and only interact with weak affinities (4-25 µM) (9–
60
13), but are crucial to ensure biosynthesis of the desired product(s) (8, 11, 14). Despite
61
recent progress on applying DD substitutions to program new assembly lines, in most
62
cases structural information is lacking to effectively apply DDs for general engineering
63
purposes (11, 15, 16).
64
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted May 6, 2020. ; https://doi.org/10.1101/2020.05.06.080655doi: bioRxiv preprint

4
Although early engineering attempts, including the exchange of DDs, the targeted
65
modification of the A-domains substrate specificity conferring AA residues, and the
66
substitution of domains as well as whole modules, gave mixed results, several notable
67
advances have been published recently (16–18). To give but one example, we
68
comprehensively analysed structural data as well as inter-domain linkers in NRPSs to
69
define novel fusion sites and to provide guidelines for exchanging A-T-C units, denoted
70
as eXchange Units (XUs), as opposed to canonical modules (C-A-T) (19). By
71
combining XUs from 15 NRPSs in cis, it was possible to reconstitute naturally available
72
peptides, peptide derivatives, and to generate new-to-nature peptides de novo in high
73
yields.
74
Herein, starting from our recently published XU concept, we explored the ability of
75
synthetic zippers (SZs (20)) to manipulate collinear type A NRPSs by introducing
76
artificial in trans regulation. SZs interact with high affinity (KD<10 nM) via a coiled-coil
77
structural motif, enabling the specific association of two proteins. Such a strategy not
78
only would allow creating a synthetic type of in trans regulated mega-synthetases
79
(type S), by combining NRPSs with high-affinity SZs (20) (Fig. 1a), but to overcome
80
cloning and protein size limitations associated with heterologous NRP production.
81
Results
82
In depth structural analysis of the crystallised termination module SrfA-C (PDB-ID:
83
2VSQ) suggested splitting NRPSs in between consecutive XUs at the previously
84
defined W]-[NATE motif of the conformationally flexible C-A linker (21–23) region. As
85
already known (21, 22), this splicing position bears several advantages. Of particular
86
importance is that it keeps intact the short (~ 10 AAs) a-helical structure at the C-
87
terminus of the resulting truncated protein (subunit 1) – as in wild type (WT) NRPSs
88
this helical structure not only regulates the C-A distance throughout the catalytic cycle
89
(21), but also associates with the A-domains hydrophobic protein surface (23).
90
Attempts of in silico creating NRPS domains connected via SZs, composed of ~40
91
amino acids (AAs), were unsuccessful. Nevertheless, careful revision of available
92
structural data indicated that ~10 AAs from the unstructured N-terminus of subunits 2
93
must be removed to meet the distance-criteria set out by the WT C-A inter-domain
94
linker to ensure correct C-A di-domain contacts before SZs N-terminally could be
95
introduced (Fig. 1b). After perusing characterized SZ pairs (20), to begin with we chose
96
the SZ pair 17 & 18 (Fig. 1c & d).
97
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted May 6, 2020. ; https://doi.org/10.1101/2020.05.06.080655doi: bioRxiv preprint

5
Proof of Concept
98
To assess the general suitability of SZ-pairs to in trans connect two NRPS proteins
99
and mediate biosynthetically functional protein-protein interface interactions, we
100
targeted the xenotetrapeptide (1) producing NRPS (XtpS; Fig. S1) from the Gram-
101
negative entomopathogenic bacterium Xenorhabdus nematophila HGB081 (24). We
102
decided to split XtpS into two subunits in between XUs 2 and 3 and four artificial two
103
component type S NRPS (Fig. 2a) were constructed and heterologously produced in
104
E. coli DH10B::mtaA (25) – either with SZs fused to both subunits (NRPS-1: subunit 1-
105
SZ17, SZ18-subunit 2); only fused to subunit 1 (NRPS-2: subunit 1-SZ17, subunit 2)
106
or subunit 2 (NRPS-3: subunit 1, SZ18-subunit 2), and without SZs (NRPS-4: subunit
107
1, subunit 2).
108
NRPS-2 and NRPS-4 showed no detectable peptide production, whereas NRPS-1
109
lead to the production of 1 with ~30% (28 mg l
-1
) yield compared to WT XtpS (Fig. 2a,
110
Fig. S2), confirming that SZs indeed can be used to functionally mediate new-to-nature
111
in trans regulation of NRP biosynthesis. Interestingly, NRPS-3 with SZ18 fused to
112
subunit 2, but lacking SZ17 on subunit 1, showed moderate yields of 1. Despite lacking
113
SZ17, the C-terminus of XtpS subunit 1 is forming a Leucine rich a-helical structure
114
(PDB-ID: 2VSQ) that might be able to interact with SZ18 of subunit 2 and mediate an
115
impaired but catalytically active C-A interface (21–23, 26).
116
Additionally, SZ17:18 were used to split the GameXPeptide A-D (2-5) producing NRPS
117
(GxpS (27, 28)) and the recombinant thiazole-peptide (6) producing NRPS (RtpS (29)).
118
Whereas GxpS originates from the Gram-negative bacterium Photorhabdus
119
luminescens TTO1, RtpS was constructed previously (29) from building blocks (BBs)
120
of Gram-positive origin (using NRPSs for the production of bacitracin (30) and surfactin
121
(31)). Both resulting type S NRPSs (Fig. 2b) showed good to very good titres of desired
122
peptides. NRPS-5 produced 2 (Fig. S3) with yields of ~64 % (4.9 mg l
-1
) compared to
123
WT GxpS and NRPS-6 produced 6 (Fig. S4) at WT RtpS level (~20 mg l
-1
).
124
All product structures and yields were confirmed by tandem mass spectrometry
125
(MS/MS) analysis and comparison of the retention times with synthetic standards.
126
Creating Synonymous Chimeras
127
To explore the recombination potential of chimeric type S NRPSs, initially we co-
128
expressed non-cognate subunits from NRPS-1 and -5 (Fig. 3a). Both, NRPS-1
129
.CC-BY-NC-ND 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted May 6, 2020. ; https://doi.org/10.1101/2020.05.06.080655doi: bioRxiv preprint

Frequently asked questions

Q1. What is the role of a thiolation domain in a nrp?

The catalytic activity of a canonical 42 module is based upon the orchestrated interplay of an adenylation (A) domain for AA 43 selection and activation, a condensation (C) domain to catalyse peptide bond 44 formation, and a thiolation/ peptidyl-carrier protein (T) onto which the AAs or 45 intermediates are covalently tethered (3). 

Q2. What is the name of the tool kit?

DDs typically are 59 located in between two modules and only interact with weak affinities (4-25 µM) (9–60 13), but are crucial to ensure biosynthesis of the desired product(s) (8, 11, 14). 

Q3. What is the name of the NRPS?

Type A NRPSs are composed of sequential catalytically 40 active domains organised in modules, each responsible for the incorporation and 41 modification of one specific amino acid (AA). 

Q4. What are the three types of NRPSs?

According to their biosynthetic logic, known NRPS 38 systems are classified into three groups, linear (type A), iterative (type B), and 39 nonlinear NRPSs (type C) (2). 

Q5. What is the name of the book?

Whereas in in cis type A NRPSs all modules are arranged on a single 54 polypeptide chain (e.g. ACV-synthetase (6)), in trans assembly-lines comprise a 55 number of individual proteins (Daptomycin-synthetase (7)). 

Q6. What is the significance of the a-helical structure?

Of particular 86importance is that it keeps intact the short (~ 10 AAs) a-helical structure at the C-87 terminus of the resulting truncated protein (subunit 1) – as in wild type (WT) NRPSs 88 this helical structure not only regulates the C-A distance throughout the catalytic cycle 89 (21), but also associates with the A-domains hydrophobic protein surface (23). 

Q7. What is the role of a TE domain in NRPSs?

most 48 NRPS termination modules harbour a TE-domain, usually responsible for the release 49 of linear, cyclic or branched cyclic peptides (4). 

Q8. How do SZs interact with each other?

SZs interact with high affinity (KD<10 nM) via a coiled-coil 77 structural motif, enabling the specific association of two proteins. 

Q9. What is the structure of the NRPSs that the authors have studied?

In depth structural analysis of the crystallised termination module SrfA-C (PDB-ID: 83 2VSQ) suggested splitting NRPSs in between consecutive XUs at the previously 84 defined W]-[NATE motif of the conformationally flexible C-A linker (21–23) region. 

Q10. What is the way to combine NRPSs with SZs?

careful revision of available 92 structural data indicated that ~10 AAs from the unstructured N-terminus of subunits 2 93 must be removed to meet the distance-criteria set out by the WT C-A inter-domain 94 linker to ensure correct C-A di-domain contacts before SZs N-terminally could be 95 introduced (Fig. 1b). 

Q11. How can a synthetic zipper be created?

Such a strategy not 78 only would allow creating a synthetic type of in trans regulated mega-synthetases 79 (type S), by combining NRPSs with high-affinity SZs (20) (Fig. 1a), but to overcome 80 cloning and protein size limitations associated with heterologous NRP production.