Reactivity of an Unusual Amidase May Explain Colibactin's DNA Cross-Linking Activity

Article abstract of DOI:10.1021/jacs.9b02453

Certain commensal and pathogenic bacteria produce colibactin, a small-molecule genotoxin that causes interstrand cross-links in host cell DNA. Although colibactin alkylates DNA, the molecular basis for cross-link formation is unclear. Here, we report that the colibactin biosynthetic enzyme ClbL is an amide bond-forming enzyme that links aminoketone and β-keto thioester substrates in vitro and in vivo. The substrate specificity of ClbL strongly supports a role for this enzyme in terminating the colibactin NRPS-PKS assembly line and incorporating two electrophilic cyclopropane warheads into the final natural product scaffold. This proposed transformation was supported by the detection of a colibactin-derived cross-linked DNA adduct. Overall, this work provides a biosynthetic explanation for colibactin's DNA cross-linking activity and paves the way for further study of its chemical structure and biological roles.

Full text of DOI:10.1021/jacs.9b02453

Subscriber access provided by UNIV AUTONOMA DE COAHUILA UADEC
Article
The reactivity of an unusual amidase may
explain colibactin’s DNA cross-linking activity
Yindi Jiang, Alessia Stornetta, Peter W. Villalta, Matthew R. Wilson,
Paul D. Boudreau, Li Zha, Silvia Balbo, and Emily P. Balskus
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.9b02453 • Publication Date (Web): 28 Jun 2019
Downloaded from http://pubs.acs.org on June 28, 2019
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 8
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
The reactivity of an unusual amidase may explain colibactin’s DNA
cross-linking activity
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Yindi Jiang,† Alessia Stornetta,‡ Peter W. Villalta,‡ Matthew R. Wilson,†,§ Paul D. Boudreau,† Li Zha,†,‖
Silvia Balbo,‡ and Emily P. Balskus*,†
†Department of Chemistry and Chemical Biology, Harvard University, 12 Oxford Street, Cambridge, MA 02138, United
States
‡Masonic Cancer Center, University of Minnesota, 2231 Sixth Street Southeast, Minneapolis, MN 55455, United States
ABSTRACT: Certain commensal and pathogenic bacteria produce colibactin, a small molecule genotoxin that causes interstrand
cross-links in host cell DNA. Though colibactin alkylates DNA, the molecular basis for cross-link formation is unclear. Here, we
report that the colibactin biosynthetic enzyme ClbL is an amide bond-forming enzyme that links aminoketone and β-keto thioester
substrates in vitro and in vivo. The substrate specificity of ClbL strongly supports a role for this enzyme in terminating the
colibactin NRPS-PKS assembly line and incorporating two electrophilic cyclopropane warheads into the final natural product
scaffold. This proposed transformation was supported by the detection of a colibactin-derived cross-linked DNA adduct. Overall,
this work provides a biosynthetic explanation for colibactin’s DNA cross-linking activity and paves the way for further study of
its chemical structure and biological roles.
characterized several stable candidate precolibactins (1-5) from
INTRODUCTION
pks+ E. coli strains lacking ClbP (Figure 1B).13-20 Notably,
metabolites 2-5 contain a cyclopropane ring, a structural feature
Colibactin is a genotoxin made by human gut commensal and
found in some DNA-alkylating agents.21,22 Formation of an ,-
extraintestinal pathogenic Escherichia coli strains and other
unsaturated imine after cleaving the prodrug motif in
Proteobacteria.1 Colibactin-producing E. coli (pks+ E. coli)
precolibactin likely enhances the electrophilicity of the
cause DNA double strand breaks,1-2 affect progression of colitis-
cyclopropane toward DNA bases, and the recent identification
associated colorectal cancer (CRC) in mouse models,3-4 and are
and structural characterization of colibactin-derived DNA
more frequently detected in patients with colorectal cancer.5-7
adducts in human cells and colonic epithelial cells provided
Recent studies have indicated colibactin’s genotoxicity likely
support for this functional group serving as an electrophilic
arises from a direct interaction with DNA, as we and others have
warhead.9,23 However, it was unclear how this structural motif
reported the accumulation of interstrand cross-links in human
could lead to generation of an interstrand cross-link.
cell lines incubated with pks+ E. coli.8,9 Cell lines with impaired
interstrand cross-link repair are also more sensitive to colibactin
To address this question, we focused on understanding
exposure.8 However, the underlying chemical and enzymatic
previously uncharacterized components of the colibactin
basis for colibactin’s DNA cross-linking activity remains
biosynthetic enzymes. Efforts to elucidate the biosynthetic
unclear.
origins of isolated candidate precolibactins have led to putative
functional assignments for all of the enzymes in the biosynthetic
pathway except for the putative amidase ClbL. This enzyme is
Colibactin is produced by a 54-kb gene cluster (the pks island)
encoding
a
nonribosomal peptide synthetase-polyketide
essential for the genotoxicity of pks+ E. coli1 but is not required
for biosynthesis of any isolated candidate precolibactins
reported to date, indicating that these metabolites were likely not
precursors of the final genotoxin(s). A recent finding that pks+
E. coli strains with inactive ClbL were incapable of cross-linking
DNA suggested that this enzyme participates in producing the
active cross-linking agent.24 However, the molecular basis for
ClbL’s contribution to colibactin’s cross-linking activity is
unknown.
synthase (NRPS-PKS) hybrid assembly line,1 and its
biosynthesis involves a self-protection mechanism in which an
N-acyl-D-asparagine scaffold (prodrug motif) is initially
assembled and elaborated to produce an inactive metabolite
(precolibactin) (Figure 1A).10-12 Precolibactin is cleaved by the
periplasmic peptidase ClbP to release the active colibactin
genotoxin. This active species has not yet been identified,
isolated, or structurally characterized, presumably due to its
instability. We and others have isolated and structurally
1
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. The pks enzymatic assembly line and the biosynthesis of characterized candidate precolibactins. (A) Hypothetical
biosynthetic pathway for candidate precolibactins 1-3 and 5. C, condensation; A, adenylation; T, thiolation; E, epimerization; KS,
ketosynthase; AT, acyltransferase; KR, ketoreductase; DH, dehydratase; ER, enoyl reductase; AT*, atypical acyltransferase; Cy, cyclization;
Ox, oxidase. Cyclopropane rings are highlighted in orange. Thiazoles are highlighted in blue. (B) Chemical structures of isolated and
characterized candidate precolibactins 1-5.
strains failed to cause cell cycle arrest in HeLa cells (Figure S1),
suggesting that these three residues are crucial for the function
of ClbL and confirming that ClbL’s catalytic activity is critical
for genotoxicity.
Here, we elucidate the function of ClbL in colibactin
biosynthesis.
By
identifying,
isolating,
structurally
characterizing a ClbL-dependent candidate precolibactin from
pks+ E. coli lacking ClbP and reconstituting its biosynthesis in
vitro, we show that ClbL is an amide bond-forming enzyme.
ClbL couples β-keto thioesters and aminoketones, structural
motifs known to be generated by the colibactin assembly line.
Based on this observation, we hypothesize that ClbL assembles
a pseudo-dimeric precolibactin precursor possessing two
electrophilic cyclopropane warheads which upon ClbP
activation can cross-link DNA. This proposal is supported by the
detection of two new colibactin-derived DNA adducts in pks+ E.
coli-treated DNA samples, including a putative cross-link.
Overall, this work reveals the activity of ClbL and provides a
molecular explanation for colibactin’s DNA cross-linking
activity.
Figure 2. Discovery of
a
ClbL-dependent candidate
precolibactin. (A) Extracted ion chromatograms (EICs) of
metabolite 6 (m/z 729.4334) in the extracts of E. coli expressing
wild-type or mutant ClbL. (B) LC-MS analysis of feeding of 1-13C-
L-methionine to ClbL expressing E. coli strains.
Results
ClbL belongs to the amidase signature (AS) family of enzymes,
which contain a Ser-cisSer-Lys catalytic triad (Figure S1).25
These enzymes catalyze a wide variety of hydrolytic reactions,
including the breakdown of the neurotransmitter anandamide by
fatty acid amide hydrolase (FAAH) in humans and hydrolysis of
glutamine by Glu-tRNAGln amidotransferase subunit A (GatA)
in bacteria.26-27 Phylogenetic analysis of representative amidases
indicated that ClbL is part of a distinct clade, potentially
implying functional differences (Figure S2). Alignment of ClbL
with structurally characterized amidases revealed the three
conserved, essential catalytic residues (K80, S155, and S179)
(Figure S1). To test whether these active site amino acids are
essential for ClbL function and genotoxicity, plasmids
expressing versions of ClbL with these residues individually
mutated to alanine were constructed and used to complement a
strain of pks+ E. coli lacking clbL (BACpksΔclbL). These mutant
To identify candidate precolibactins linked to ClbL, we
overexpressed wild-type ClbL or the ClbL active site mutants in
E. coli DH10B BACpksΔclbP/ΔclbL. Methanol extracts from
these cultured strains were compared to identify mass features
affected by the activity of ClbL. As amidases are typically amide
bond-hydrolyzing enzymes, we expected the substrate for ClbL
to accumulate in the mutant strains. Surprisingly, we were
unable to detect any up-regulated metabolites in these strains.
Consistent with the previous findings,16 ClbL mutant strains
produced detectable levels of 2 and other known candidate
precolibactins (Figure S3), confirming that these metabolites
are not ClbL-dependent. We did observe a novel metabolite (6,
m/z 729.4334) accumulate only in strains expressing wild-type
ClbL (Figure 2A, Table S4). To confirm metabolite 6 was pks-
2
ACS Paragon Plus Environment

Page 3 of 8
Journal of the American Chemical Society
associated, we fed wild-type ClbL-expressing strains L-[1-
pathway A, the intermediate generated by ClbJ-NPRS117,18
could be intercepted and offloaded via nucleophilic attack by the
known E. coli metabolite indole.29 In pathway B, 2-amino-1-
(1H-indol-3-yl)ethan-1-one (7) could act as a nucleophile to
offload the β-keto thioester intermediate generated by the
upstream biosynthetic enzyme ClbI.30
13C]Met, the amino acid precursor of the cyclopropane ring.15,16
Observation of a +1 mass shift demonstrated that methionine is
1
2
3
4
5
6
7
8
incorporated into
6
(Figure 2B). Moreover, MS/MS
fragmentation analysis indicated that 6 contains the prodrug
scaffold (m/z 324.2437)18 and
a colibactin-characteristic
28
fragment ion (m/z 231.1119)17, (Figure S4). These results
suggested that metabolite 6 is a novel ClbL-dependent candidate
precolibactin.
9
We then isolated and structurally characterized this new
metabolite, obtaining approximately 0.7 mg of purified 6 from
140 L of an E. coli DH10B BACpksΔclbP strain overexpressing
ClbL. The high-resolution mass of this compound ([M+H]+ =
729.4334) yielded a molecular formula of C41H56N6O6. One-
dimensional (1D) and two-dimensional (2D) NMR along with
MS/MS fragmentation analysis allowed us to elucidate its
chemical structure (Figure 3A, Figure S5-S8, Table S5).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. Investigating the formation of candidate precolibactin
6. (A) Two proposed pathways could generate 6. (B) LC-MS
analysis of extracts from double mutants of E. coli DH10B BACpks.
(C) LC-MS analysis of in vitro reconstitution assays. (D) LC-MS
analysis of cell pellet extracts from E. coli DH10B BACpksΔclbP +
pTrc-clbL supplemented with 7. Panels B-D show the EICs of 6 (m/z
729.4334).
Figure 3. Structure determination of candidate precolibactin 6.
(A) Chemical structure of metabolite 6. (B) Key COSY and HMBC
correlations that support this assignment. (C) EICs of metabolite 6
(m/z 729.4334) from the extracts of bacterial cultures, synthetic
standard, and co-injection of synthetic standard and bacterial culture
extracts.
To test the involvement of pathway A, we made multiple double
mutants in E. coli DH10B BACpksΔclbP (Figure 4B).
Detection of 6 in the absence of ClbJ and ClbP indicated that
ClbJ was not required for its biosynthesis. To confirm this result,
we fed [1,2-13C2]Gly, the building block used by ClbJ-NPRS1,
to a glycine auxotrophic BACpksΔclbP strain. We observed a +2
mass shift in the known glycine-derived precolibactin 3, but not
in 6 (Figure S13), confirming that pathway A is not the source
of 6.
Similarities between the NMR spectra of known precolibactins
and metabolite 6 enabled us to assign fragments I and II easily
(Figure 3B). Unexpectedly, COSY and HMBC correlations
indicated the presence of an indole ring (Figure 3B). We were
unable to assign the precise locations of a methylene and
carbonyl within 6 using NMR, but MS/MS fragmentation
analysis revealed a loss of indole (m/z 117.0607) from the parent
ion, suggesting that the ketone is directly attached to the indole
ring (Figure S9). To further support this structural assignment,
we chemically synthesized a standard of metabolite 6. This
standard possessed the same exact mass, MS/MS fragmentation
We next explored the involvement of pathway
B by
reconstituting the biosynthesis of metabolite 6 in vitro. We
cloned, overexpressed, and purified C-terminal His6-tagged
ClbL (Figure S14). An assay mixture containing Sfp, ClbN,
ClbB, ClbC, ClbH, ClbI, ClbL, myristoyl-CoA, malonyl-CoA,
L-Asn, L-Ala, NADPH, ATP, SAM, and 7 generated 6 (Figure
4C, trace iv). To further confirm the involvement of 7 in this
process, 7 was fed to an E. coli DH10B BACpksΔclbP strain
overexpressing ClbL. An approximately 20-fold increase in the
production of 6 was observed (ion count of 4 x 104 vs. 7 x 105,
Figure 4D), consistent with pathway B being operative in vivo.
Notably, these results suggested that ClbL catalyzes amide bond
formation rather than amide bond hydrolysis.
1
pattern, retention time, H spectrum, and UV spectrum as 6
isolated from bacterial strains (Figure 3C, Figure S10-S12),
confirming our structural assignment.
The chemical structure of 6 guided subsequent efforts to
elucidate the function of ClbL. Metabolite 6 does not contain
either a free amine or a carboxylic acid, which are the functional
groups liberated by the activity of characterized amidases. Based
on the known steps of colibactin biosynthesis, we envisioned
two possible pathways for the formation of 6 (Figure 4A). In
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. In vitro characterization of ClbL reveals its role in amide formation. (A) Reaction scheme of an in vitro assay to form 9. (B)
HPLC time course for the formation of 9 by ClbL (monitored at 250 nm). (C) Thioesters and -aminoketones accepted by ClbL. (D) The
pks assembly line may produce -aminoketone nucleophiles. (E) EICs of 12 (m/z 690.4225) formed in an in vitro reconstitution assay. (F)
ClbL catalyzes the formation of 16 in vitro. The two peaks likely correspond to two tautomers of the -ketoamide product.
product arising from 11 (Figure S17). This suggests ClbL
To further characterize the activity of ClbL in vitro, we
preferentially recognizes the ClbI-bound intermediate.
Furthermore, reexamination of extracts from ClbL-
overexpressing strains did not reveal products of amide bond
synthesized ethyl thioester 8 as a surrogate for the ClbI-tethered
biosynthetic intermediate (Figure 1A and 5A). Incubating ClbL
with 7 and 8 (Figure 5A) revealed a single new product peak
formation between 7 and ClbC- or ClbJ-tethered thioester
accumulating during an HPLC time course experiment (Figure
intermediates (Figure S17). Together, these data indicate that
5B). We confirmed this product was amide 9 by comparison to
a synthetic standard (Figure S15).
ClbL prefers the ClbI-bound thioester as an electrophile for
amide bond formation.
To evaluate ClbL’s preference for potential thioester substrates,
We next investigated the relevance of indole-containing
we designed and synthesized ethyl thioester mimics of assembly
nucleophile 7 to colibactin biosynthesis. 7 has only previously
line-tethered colibactin biosynthetic intermediates (Figure S16).
been found in termitomycamide B, which was isolated from the
ClbL and 7 were incubated with these mimics individually. In
addition to 8, we found that thioesters 10 (a mimic for the ClbC-
giant mushroom Termitomyces titanicus (Figure S18).31
Intriguingly, 7 contains an -aminoketone, a structural motif
bound intermediate) and 11 (a mimic for the ClbJ-bound
that could originate from decarboxylation of the essential
intermediate) could be recognized by ClbL (Figure 5C, Figure
S17). To assess ClbL’s ability to discriminate among these
substrates, we conducted a competition experiment. Mixing 8
colibactin building block aminomalonyl-ACP (Figure 5D). To
examine the involvement of aminomalonyl-ACP biosynthetic
enzymes in the production of 7, we used an E. coli DH10B
and 11, which are predicted to have very similar
BACpksΔclbPΔclbG strain, in which the transfer of
hydrophobicities, in equal molar amounts with ClbL and 7
aminomalonate to assembly line enzymes is abolished. This
produced more of the product arising from 8 compared to the
4
ACS Paragon Plus Environment

Page 5 of 8
Journal of the American Chemical Society
strain still produced 6, the candidate precolibactin derived from
diasteromeric colibactin-derived DNA adducts (18) that likely
arise from degradation of a larger lesion.9 Depurination of the
proposed colibactin-DNA cross-link 19 would give adduct 20,
which could undergo oxidative cleavage to give monoadducts
21 and 18 (Figure 7A). A similar cleavage has been shown to
occur readily in model substrates and a synthetic colibactin
mimic.33,34
1
2
3
4
5
6
7
8
7, indicating that aminomalonyl-ACP is not required (Figure
4B). We further showed that 7 does not originate from any of the
pks island enzymes as it was detected in E. coli DH10B (Figure
S19). 7 was also absent from uninoculated growth medium,
indicating it is produced in E. coli endogenously. Therefore, we
questioned whether the incorporation of 7 was relevant for the
biosynthesis of the active colibactin genotoxin. Supplementing
7 to pks+ E. coli during infection of host cells did not boost
genotoxicity (Figure S20), suggesting that 7 is likely not a
relevant substrate for ClbL in colibactin biosynthesis. We
therefore hypothesize that metabolite 6 arises from the
promiscuous activity of ClbL and is not an on-pathway
intermediate.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Next, we considered the possibility that an alternative
nucleophile may be generated by colibactin biosynthetic
enzymes. Assays with various amine substrates showed that the
-aminoketone of 7 is crucial for recognition by ClbL. No
products were observed when tryptamine replaced 7 in the in
vitro reconstitution of the biosynthesis of 6. In contrast, we
identified a new product 12 (m/z 690.4225) in assay mixtures
containing 2-aminoacetophenone in place of 7 (Figure 5E). This
result suggested the -aminoketone and not the indole ring is
important for substrate recognition by ClbL.
To identify the true nucleophilic coupling partner for ClbL, we
examined its reactivity toward mimics of potential -
aminoketone substrates derived from the colibactin assembly
line. We previously identified four PKS assembly line enzymes
(ClbC, ClbI, ClbK, and ClbO) that accept aminomalonate in
vitro.18 Isolation of candidate precolibactin 5 also showed that
the PKS module of ClbK accepts this building block in vivo,
generating an -aminoketone that is further oxidized to an -
iminoketone.19 Once offloaded from PKS modules, the
immediate products of extension with aminomalonyl-ACP
should readily undergo decarboxylation to give rise to -
aminoketones resembling 7 (Figure 5D). To examine whether
ClbL can use assembly line derived -aminoketones for amide
bond formation, we synthesized mimics of these potential
substrates (13-15) and incubated them with ClbL and preferred
thioester 8 (Figure S21). While ClbL can use all three mimics,
it preferentially uses 15, which mimics the -aminoketone
derived from ClbO, the final module of the colibactin assembly
line (Figure 5C and 5F, Figure S22). Together, these results
suggest a potential role for ClbL in colibactin biosynthesis:
forming an amide bond between a ClbI-bound thioester and the
-aminoketone derived from ClbO (Figure 6).
Figure 6. A model for ClbL-catalyzed formation of the active
colibactin genotoxin). Key structural features colored as in Figures
1, 4, and 5.
To test this proposal, we used targeted selection ion monitoring
(SIM), untargeted adductomics,9 and MS2 analyses to identify
and support the structures of these proposed colibactin-derived
DNA adducts in linearized plasmid DNA treated with pks+ E.
coli. We identified a large DNA adduct consistent with cross-
link 20 (m/z 537.1719, z = 2) (Figure 7B). We also used
untargeted DNA adductomics9 to find a cross-link adduct
lacking one of the adenine moieties 22 (m/z 938.2821, z = 1; m/z
469.6447, z = 2) (Figure S23). To confirm 20 and 22 are pks-
associated, we supplemented assay mixtures with either L-[1-
13C]Met or L-[1-13C]Cys. The observation of +2 mass shifts in
20 and 22 suggested that they derive from two methionines and
two cysteines (Figure 7C, Figure S23-24), supporting our
proposal that colibactin contains two electrophilic warheads and
two thiazoles. We also detected the proposed decomposition
product 21 (m/z 568.1721) (Figure 7D-E, Figure S25-27).
Further characterization of colibactin-derived adducts will be
necessary to clarify the identity of the active cross-linking agent.
This proposed assembly line logic has important implications for
colibactin’s biological activity. Amide bond formation by ClbL
could link the precursors of two electrophilic cyclopropane
warheads together. Cleavage of the two prodrug motifs by ClbP
would generate a species that would contain two electrophilic
warheads and could be capable of generating interstrand cross-
links (17) (Figure 6). This metabolite could be the active
colibactin genotoxin. This scenario is consistent with previous
observations that pks+ E. coli promote interstrand cross-link
formation in linearized plasmid DNA and human cell lines,8,9 as
well as a recent finding that a synthetic ‘colibactin mimic’
containing two cyclopropane rings cross-links DNA in vitro,
while a monomer cannot.32 This proposal is also consistent with
our current understanding of colibactin-mediated DNA damage.
Conclusions
ClbL belongs to the AS enzyme family, which participates in
numerous biological processes, including the biosynthesis of the
natural product actinonin in Streptomyces sp. ATCC 14903.35 In
addition to catalyzing amide bond hydrolysis, some amidases
exhibit acyltransferase activity. For example, the amidase from
Rhodococcus sp. R312 transfers the acyl group of an amide to
We previously identified and characterized
a pair of
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 8
hydroxylamine at a 33-fold faster rate than the hydrolysis of the
we were unable to detect a ClbO-derived -aminoketone in vivo.
Attempts to generate this substrate through in vitro
reconstitution were also unsuccessful.
1
2
3
4
5
6
7
8
same amide, an activity that could be explained by the stronger
nucleophilicity of hydroxylamine toward the acyl-enzyme
intermediate compared to water.36 Our discovery of ClbL’s
activity expands the known roles of amidases to include
interfacing with enzymatic assembly lines. It is intriguing that
ClbL preferentially recognizes -aminoketones over the
corresponding primary amines. Future crystallographic studies
may shed further light on this specificity.
Nonetheless, ClbL’s in vitro activity supports the hypothesis that
this enzyme links two intermediates generated by the colibactin
NRPS-PKS assembly line, a ClbI-bound thioester and a ClbO-
derived -aminoketone (Figure 6). This transformation may
play a critical role in generating a DNA cross-linking agent that
could represent the active colibactin genotoxin. This proposal is
supported by ClbL’s essential role in genotoxicity and cross-link
formation, as well as our detection of adduct 20 in linearized
plasmid DNA treated with pks+ E. coli. We cannot exclude the
possibility that ClbL accepts additional substrates. For example,
we were unable to test whether ClbL could use a mimic of the
putative ClbO-bound aminomalonate thioester intermediate as
an electrophile due to its intrinsic instability. However,
precedented colibactin biosynthetic logic and its DNA
interstrand cross-linking activity allow us to establish two
minimal criteria for the structure of colibactin: 1) it contains two
electrophilic cyclopropane rings and 2) its construction requires
all of the essential colibactin biosynthetic enzymes. Application
of these conditions greatly narrows the potential structures for
the active colibactin genotoxin(s) (Figure S28). These criteria
also help in assessing the biological relevance of other reported
candidate precolibactins. For example, Qian and Zhang recently
disclosed a candidate precolibactin they proposed is the
precursor to colibactin (Figure S29).20 Though the
corresponding ClbP cleavage product appeared to damage DNA
in vitro and in cells in a Cu2+-dependent manner, it did not form
cross-links. It is also unclear whether ClbL is required for its
biosynthesis. Together, these observations suggest this
metabolite may not be a major contributor to genotoxicity.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
While this work was in revision, the Herzon and Crawford
groups independently identified colibactin-derived cross-link
adduct 20 providing further support for our model.34 They also
chemically synthesized diketone 17 and showed that it cross-
links linearized plasmid DNA at a concentration of 10 µM. This
relatively low cross-linking reactivity suggests the active
genotoxin(s) may contain an -iminoketone motif. As our
results suggest ClbL may be able to construct additional
candidate genotoxins (Figure S28), we feel that the final
structure(s) of colibactin still requires further investigation.
Figure 7. Detection of two new colibactin-derived DNA adducts.
(A) Predicted structures of colibactin-derived DNA adduct 20 and
its decomposition products 21 and 18. (B) EICs of 20 (m/z 537.1719,
z = 2) in pks+ E. coli treated DNA samples. (C) Adduct 20
incorporates two alanines, two methionines, and two cysteines. (D)
EICs of 21 (m/z 568.1721) in DNA samples treated with pks+ E. coli.
The two peaks likely arise from multiple diastereomers. (E) Adduct
21 incorporates one alanine, one methionine, and one cysteine.
In summary, these studies provide a biosynthetic rationale for
the formation of interstrand DNA cross-links by the gut bacterial
genotoxin colibactin. As such cross-links represent the most
toxic form of DNA damage in human cells, it is possible this
activity could explain the effects of pks+ E. coli on tumor
development. Future efforts to characterize additional
precolibactins, colibactins, and colibactin-DNA adducts will
provide us with the additional molecular information needed to
decipher how colibactin influences CRC development in
patients.
We uncovered the activity of ClbL through studying the
biosynthesis of metabolite 6, an off-pathway candidate
precolibactin that incorporates an endogenously produced -
aminoketone (7). A competition assay with 7 and 13-15 revealed
that 7 is a preferred substrate for ClbL (Figure S22), explaining
the presence of 6 in E. coli strains expressing the entire pks
assembly line. However, we cannot conclude whether ClbL is
likely to preferentially recognize 7 in vivo because we cannot
examine its reactivity toward complex assembly line derived -
aminoketones. These proposed substrates are not readily
accessed using chemical synthesis or in vitro reconstitution
because of the instability of the -aminoketone located between
the two thiazole heterocycles.33,34 Consistent with its instability,
ASSOCIATED CONTENT
Supporting Information
The Supporting Information is available free of charge on the ACS
Publications website.
6
ACS Paragon Plus Environment

Page 7 of 8
Journal of the American Chemical Society
generates DNA interstrand cross-links in infected cells. mBio 2018, 9,
e02393–17.
Materials and Methods for enzyme purification, synthetic
procedures, and additional experimental data (PDF)
1
2
3
4
5
6
7
8
(9) Wilson, M. R.; Jiang, Y.; Villalta, P. W.; Stornetta, A.; Boudreau,
P. D.; Carrá, A.; Brennan, C. A.; Chun, E.; Ngo, L.; Samson, L. D.;
Engelward, B. P.; Garrett, W. S.; Balbo, S.; Balskus, E. P. The human
gut bacterial genotoxin colibactin alkylates DNA. Science 2019, 363,
eaar7785.
(10) Dubois, D.; Baron, O.; Cougnoux, A.; Delmas, J.; Pradel, N.;
Boury, M.; Bouchon, B.; Bringer, M. A.; Nougayrède, J.-P.; Oswald,
E.; Bonnet, R. ClbP is a prototype of a peptidase subgroup involved in
biosynthesis of nonribosomal peptides. J. Biol. Chem. 2011, 286,
35562–35570.
AUTHOR INFORMATION
Corresponding Author
* balskus@chemistry.harvard.edu
Present Addresses
§Vertex Pharmaceuticals, 50 Northern Ave, Boston, MA 02210,
United States.
‖ Boston Children’s Hospital, 300 Longwood Avenue, Boston, MA
02115, United States.
9
(11) Brotherton, C. A.; Balskus, E. P. A prodrug resistance
mechanism is involved in colibactin biosynthesis and cytotoxicity. J.
Am. Chem. Soc. 2013, 135, 3359–3362.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(12) Bian, X.; Fu, J.; Plaza, A.; Herrmann, J.; Pistorius, D.; Stewart,
A. F.; Zhang, Y.; Müller, R. In vivo evidence for a prodrug activation
mechanism during colibactin maturation. ChemBioChem 2013, 14,
1194–1197.
(13) Vizcaino, M. I.; Engel, P.; Trautman, E.; Crawford, J. M.
Comparative metabolomics and structural characterizations illuminate
colibactin pathway-dependent small molecules. J. Am. Chem. Soc. 2014,
136, 9244–9247.
(14) Brotherton, C. A.; Wilson, M.; Byrd, G.; Balskus, E. P. Isolation
of a metabolite from the pks island provides insights into colibactin
biosynthesis and activity. Org. Lett. 2015, 17, 1545–1548.
(15) Vizcaino, M. I.; Crawford, J. M. The colibactin warhead
crosslinks DNA. Nat. Chem. 2015, 7, 411–417.
(16) Bian, X. Y.; Plaza, A.; Zhang, Y. M.; Müller, R. Two more
pieces of the colibactin genotoxin puzzle from Escherichia coli show
incorporation of an unusual 1-aminocyclopropanecarboxylic acid
moiety. Chem. Sci. 2015, 6, 3154–3160.
(17) Li, Z. R.; Li, Y.; Lai, J. Y.; Tang, J.; Wang, B.; Lu, L.; Zhu, G.;
Wu, X.; Xu, Y.; Qian, P. Y. Critical intermediates reveal new
biosynthetic events in the enigmatic colibactin pathway. ChemBioChem
2015, 16, 1715–1719.
(18) Zha, L.; Wilson, M. R.; Brotherton, C. A.; Balskus, E. P.
Characterization of polyketide synthase machinery from the pks island
facilitates isolation of a candidate precolibactin. ACS Chem. Biol. 2016,
11, 1287–1295.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENT
We thank Alex Sieg for help with cloning, Lihan Zhang for help
with NMR experiments, and Matthew Volpe for comments on the
manuscript. We acknowledge financial support from National
Institutes of Health (R01CA208834-02) (E.P.B). Salary support for
P.W.V. was provided by the U.S. National Institutes of Health and
National Cancer Institute [Grant R50-CA211256]. Mass
spectrometry for DNA adduct analysis was carried out in the
Analytical Biochemistry Shared Resource of the Masonic Cancer
Center, supported in part by the U.S. National Institutes of Health
and National Cancer Institute [Cancer Center Support Grant CA-
77598]. M.R.W. acknowledges support from the American Cancer
Society-New England Division Postdoctoral Fellowship PF-16-
122-01-CDD.
REFERENCES
(1) Nougayrède, J.-P.; Homburg, S.; Taieb, F.; Boury,
Brzuszkiewicz, E.; Gottschalk, G.; Buchrieser, C.; Hacker, J.; Dobrindt,
U.; Oswald, E. Escherichia coli induces DNA double-strand breaks in
eukaryotic cells. Science 2006, 313, 848–851.
(2) Cuevas-Ramos, G.; Petit, C. R.; Marcq, I.; Boury, M.; Oswald,
E.; Nougayrède, J.-P. Escherichia coli induces DNA damage in vivo and
triggers genomic instability in mammalian cells. Proc. Natl. Acad. Sci.
U.S.A. 2010, 107, 11537–11542.
(3) Arthur, J. C.; Perez-Chanona, E.; Mühlbauer, M.; Tomkovich, S.;
Uronis, J. M.; Fan, T.-J.; Campbell, B. J.; Abujamel, T.; Dogan, B.;
Rogers, A. B.; Rhodes, J. M.; Stintzi, A.; Simpson, K. W.; Hansen, J.
J.; Keku, T. O.; Fodor, A. A.; Jobin, C. Intestinal inflammation targets
cancer-inducing activity of the microbiota. Science 2012, 338, 120–123.
(4) Arthur, J. C.; Gharaibeh, R. Z.; Mühlbauer, M.; Perez-Chanona,
E.; Uronis, J. M.; McCafferty, J.; Fodor, A. A.; Jobin, C. Microbial
genomic analysis reveals the essential role of inflammation in bacteria-
induced colorectal cancer. Nat. Commun. 2014, 5, 4724.
(5) Buc, E.; Dubois, D.; Sauvanet, P.; Raisch, J.; Delmas, J.;
Darfeuille-Michaud, A.; Pezet, D.; Bonnet, R. High prevalence of
mucosa-associated E. coli producing cyclomodulin and genotoxin in
colon cancer. PLoS ONE 2013, 8, e56964.
(6) Eklof, V.; Löfgren-Burström, A.; Zingmark, C.; Edin, S.; Larsson,
P.; Karling, P.; Alexeyev, O.; Rutegård, J.; Wikberg, M. L.; Palmqvist,
R. Cancer-associated fecal microbial markers in colorectal cancer
detection. Int. J. Cancer. 2017, 141, 2528–2536.
(7) Dejea, C. M.; Fathi, P.; Craig, J. M.; Boleij, A.; Taddese, R.; Geis,
A. L.; Wu, X.; DeStefano Shields, C. E.; Hechenbleikner, E. M.; Huso,
D. L.; Anders, R. A.; Giardiello, F. M.; Wick, E. C.; Wang, H.; Wu, S.;
Pardoll, D. M.; Housseau, F.; Sears, C. L. Patients with familial
adenomatous polyposis harbor colonic biofilms containing tumorigenic
bacteria. Science 2018, 359, 592–597.
(19) Li, Z. R.; Li, J.; Gu, J. P.; Lai, J. Y. H.; Duggan, B. M.; Zhang,
W. P.; Li, Z. L.; Li, Y. X.; Tong, R. B.; Xu, Y.; Lin, D. H.; Moore, B.
S.; Qian, P. Y. Divergent biosynthesis yields
a cytotoxic
aminomalonate-containing precolibactin. Nat. Chem. Biol. 2016, 12,
773–775.
(20) Li, Z. R.; Li, J.; Cai, W.; Lai, J.; McKinnie, S.; Zhang, W. P.;
Moore, B.; Zhang, W.; Qian, P. Y. Macrocyclic colibactin induces DNA
double-strand breaks via copper-mediated oxidative cleavage. bioRxiv
2019, doi: 10.1101/530204.
(21) Tanasova, M.; Sturla, S. J. Chemistry and biology of
acylfulvenes: sesquiterpene-derived antitumor agents. Chem. Rev. 2012,
112, 3578-3610.
(22) Ghosh, N.; Sheldrake, H. M.; Searcey, M.; Pors, K. Chemical
and biological explorations of the family of CC-1065 and the
duocarmycin natural products. Curr. Top. Med. Chem. 2009, 9, 1494-
1524.
(23) Xue, M.; Shine, E.; Wang, W.; Crawford, J. M.; Herzon, S. B.
Characterization of natural colibactin-nucleobase adducts by tandem
mass spectrometry and isotopic labeling. Support for DNA alkylation
by cyclopropane ring opening. Biochemistry 2018, 57, 6391–6394.
(24) Shine, E. E.; Xue, M.; Patel, J. R.; Healy, A. R.; Surovtseva, Y.
V.; Herzon, S. B.; Crawford, J. M. Model colibactins exhibit human cell
genotoxicity in the absence of host bacteria. ACS Chem. Biol. 2018, 13,
3286–3293.
(25) Shin, S.; Lee, T. H.; Ha, N. C.; Koo, H. M.; Kim, S. Y.; Lee, H.
S.; Kim, Y. S.; Oh, B. H. Structure of malonamidase E2 reveals a novel
Ser-cisSer-Lys catalytic triad in a new serine hydrolase fold that is
prevalent in nature. EMBO J. 2002, 21, 2509–2516.
(26) Cravatt, B. F.; Giang, D. K.; Mayfield, S. P.; Boger, D. L.;
Lerner, R. A.; Gilula, N. B. Molecular characterization of an enzyme
that degrades neuromodulatory fatty-acid amides. Nature 1996, 384,
83–87.
(8) Bossuet-Greif, N.; Vignard, J.; Taieb, F.; Mirey, G.; Dubois, D.;
Petit, C.; Oswald, E.; Nougayrède J.-P. The colibactin genotoxin
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 8
(27) Curnow, A. W.; Hong, K. W.; Yuan, R.; Kim, S. I.; Martins, O.;
isolated from the mushroom Termitomyces titanicus, suppress
endoplasmic reticulum stress. Org. Lett. 2010, 12, 5012–5015.
(32) Healy, A. R.; Nikolayevskiy, H.; Patel, J. R.; Crawford, J. M.;
Herzon, S. B. A mechanistic model for colibactin-induced genotoxicity.
J. Am. Chem. Soc. 2016, 138, 15563–15570.
Winkler, W.; Henkin, T. M.; Söll, D. Glu-tRNAGln amidotransferase: a
novel heterotrimeric enzyme required for correct decoding of glutamine
codons during translation. Proc. Natl. Acad. Sci. U.S.A. 1997, 94,
11819–11826.
1
2
3
4
5
6
7
8
9
(28) Trautman, E. P.; Healy, A. R.; Shine, E. E.; Herzon, S. B.;
Crawford, J. M. Domain-targeted metabolomics delineates the
heterocycle assembly steps of colibactin biosynthesis. J. Am. Chem.
Soc. 2017, 139, 4195–4201.
(29) Li, G.; Young, K. D. Indole production by the tryptophanase
TnaA in Escherichia coli is determined by the amount of exogenous
tryptophan. Microbiology 2013, 159, 402-410.
(30) Zha, L.; Jiang, Y.; Henke, M. T.; Wilson, M. R.; Wang, J. X.;
Kelleher, N. L.; Balskus, E. P. Colibactin assembly line enzymes use S-
adenosylmethionine to build a cyclopropane ring. Nat. Chem. Biol.
2017, 13, 1063–1065.
(31) Choi, J. H.; Maeda, K.; Nagai, K.; Harada, E.; Kawade, M.;
Hirai, H.; Kawagishi, H. Termitomycamides A to E, fatty acid amides
(33) Healy, A. R.; Wernke, K. M.; Kim, C. S.; Lees, N. R.; Crawford,
J. M.; Herzon, S. B. Synthesis and reactivity of precolibactin 886.
ChemRxiv 2019, doi: 10.26434/chemrxiv.7849151.v1.
(34) Xue, M.; Kim, C. S.; Healy, A. R.; Wernke, K. M.; Wang, Z.;
Frischling, M. C.; Shine, E. E.; Wang, W.; Herzon, S. B.; Crawford, J.
M. Structure elucidation of colibactin. bioRxiv 2019, doi:
10.1101/574053.
(35) Wolf, F.; Leipoldt, F.; Kulik, A.; Wibberg, D.; Kalinowski, J.;
Kaysser, L. Characterization of the actinonin biosynthetic gene cluster.
ChemBioChem 2018, 19, 1189–1195.
(36) Thiéry, A.; Maestracci, M.; Arnaud, A.; Galzy, P.
Acyltransferase activity of the wide spectrum amidase of
Brevibacterium sp. R312. J. Gen. Microbiol. 1986, 132, 2205–2208.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
TOC graphic:
8
ACS Paragon Plus Environment