Discovery of a Calcium-Dependent Enzymatic Cascade for the Selective Assembly of Hapalindole-Type Alkaloids: On the Biosynthetic Origin of Hapalindole U

Article abstract of DOI:10.1002/anie.201703932

Hapalindole U (4) is a validated biosynthetic precursor to ambiguine alkaloids (Angew. Chem. Int. Ed. 2016, 55, 5780), of which biogenetic origin remains unknown. The recent discovery of AmbU4 (or FamC1) protein encoded in the ambiguine biosynthetic pathway (J. Am. Chem. Soc. 2015, 137, 15366), an isomerocyclase that can rearrange and cyclize geranylated indolenine (2) to a previously unknown 12-epi-hapalindole U (3), raised the question whether 3 is a direct precursor to 4 or an artifact arising from the limited in vitro experiments. Here we report a systematic approach that led to the discovery of an unprecedented calcium-dependent AmbU1-AmbU4 enzymatic complex for the selective formation of 4. This discovery refuted the intermediacy of 3 and bridged the missing links in the early-stage biosynthesis of ambiguines. This work further established the isomerocyclases involved in the biogenesis of hapalindole-type alkaloids as a new family of calcium-dependent enzymes, where the metal ions are shown critical for their enzymatic activities and selectivities.

Full text of DOI:10.1002/anie.201703932

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Discovery of a calcium-dependent enzymatic cascade for
the selective assembly of hapalindole-type alkaloids: On the
biosynthetic origin of hapalindole U
Authors: Qin Zhu and Xinyu Liu
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.201703932
Angew. Chem. 10.1002/ange.201703932
Link to VoR: http://dx.doi.org/10.1002/anie.201703932
http://dx.doi.org/10.1002/ange.201703932

10.1002/anie.201703932
Angewandte Chemie International Edition
COMMUNICATION
Discovery of a calcium-dependent enzymatic cascade for the
selective assembly of hapalindole-type alkaloids: On the
biosynthetic origin of hapalindole U
Qin Zhu[a] and Xinyu Liu*[a]
Abstract: Hapalindole U (4) is a validated biosynthetic precursor to
ambiguine alkaloids (Angew. Chem. Int. Ed. 2016, 55, 5780), of
which biogenetic origin remains unknown. The recent discovery of
AmbU4 (or FamC1) protein encoded in the ambiguine biosynthetic
pathway (J. Am. Chem. Soc., 2015, 137, 15366), an isomerocyclase
that can rearrange and cyclize geranylated indolenine (2) to a
previously unknown 12-epi-hapalindole U (3), raised the question
whether 3 is a direct precursor to 4 or an artifact arising from the
limited in vitro experiments. Here we report a systematic approach
that led to the discovery of an unprecedented calcium-dependent
AmbU1-AmbU4 enzymatic complex for the selective formation of 4.
This discovery refuted the intermediacy of 3 and bridged the missing
links in the early-stage biosynthesis of ambiguines. This work further
established the isomerocyclases involved in the biogenesis of
hapalindole-type alkaloids as a new family of calcium-dependent
enzymes, where the metal ions are shown critical for their enzymatic
activities and selectivities.
incomplete. We recently showed that hapalindole U 4 is a
biosynthetic precursor to hapalindole and related
G
5
ambiguines in Fischerella ambigua UTEX1903 (Figures 1b and
S2) by characterizing the pathway-specific prenyltransferase
AmbP3 and halogenase AmbO5 (Figure 1c and S3).[6] However,
the latest discovery of a thermostable AmbU4 protein in the cell
lysate of F. ambigua UTEX1903 as a dedicated synthase for 12-
epi-hapalindole U 3 (Figures 1b), a previously structurally
unknown C12 epimer of 4,[10] added a twist to the ongoing
investigation of ambiguine biogenesis. The question immediately
raised by this unexpected observation is whether 3 is a
legitimate ambiguine pathway intermediate or it is an artifact due
to the limited experimental investigations. Besides AmbU4, there
are three AmbU homologs (U1-U3) encoded in the ambiguine
pathway of F. ambigua UTEX1903 (Figures 1c and S4), of which
functions remain unknown. Here we report the systematic in
vitro characterization of all AmbU isomerocyclases that allowed
us to conclusively establish the enzymatic basis for the selective
formation of 4 and refuted the intermediacy of 3 in the
biosynthesis of ambiguines. This effort further led to the
discovery of calcium ion as a critical element in modulating the
activities and selectivities of all isomerocyclases embedded in
the biosynthetic pathways of hapalindole-type alkaloids.
The biosynthesis of hapalindole-type alkaloids, including
hapalindoles, fischerindoles, ambiguines and welwitindolinones,
has recently become a subject of intense investigations.[1-12]
These studies, collectively directed towards deciphering the
origin of structural complexities and diversities of hapalindole-
type alkaloids, have thus far unraveled several layers of puzzles
concerning their global biosynthetic diversifications (Figure S1).
These include the identification of indolenine 2 derived from the
enzymatic geranylation of cis-indolyl vinyl isonitrile 1 (Figure 1a),
as the cryptic common biosynthetic precursor to all hapalindole-
type alkaloids,[7] and establishing the enzymatic basis for
halogenations in these alkaloids that account for both complexity
and diversity generations.[4,6] More recently, a new family of
proteins embedded in the hapalindole-type alkaloid biosynthetic
pathways, originally named as U-proteins for their
bioinformatically unknown nature,[1-3] were shown able to
stereoselectively rearrange and cyclize 2 to three isomeric
hapalindole-type alkaloids, typified by 12-epi-fischerindole U
synthase WelU1, 12-epi-hapalindole C synthase WelU3 and 12-
epi-hapalindole U synthase AmbU4.[10.11] As the enzymatic
cyclizations by WelU1/WelU3/AmbU4 proteins are induced by
the geranyl group rearrangement in 2, these enzymes can be
a)
b)
R
H
NC
12
12
H
10
15
15
N
H
NC
NC
11
11
1
10
H
H
N
H
N
H
3
NC
4 hapalindole U (R=H)
12-epi-hapalindole U
5 hapalindole G (R=Cl)
2
N
ambiguine biosynthetic genes encoding diversification enzymes
c)
O5 O4 U4 U3 U2 U1 O3 O2 O1 P3
Figure 1. a) Structures of common biosynthetic precursors (1 and 2) to all
hapalindole-type alkaloids; b) structures of previously identified hapalindole
biosynthetic precursors (3 and 4) to ambiguines and hapalindole G (5) from
the ambiguine producer F. ambigua UTEX1903; c) organization of ambiguine
biosynthetic genes encoding diversification enzymes.
formally named as
a
family of isomerocyclases.[13] The
identification and characterization of WelU1 and WelU3
isomerocyclases effectively connected the missing links in the
early stage of the welwitindolinone biosynthetic pathway and
provided the enzymatic basis for the altered hapalindole-type
alkaloid structural diversities in two welwitindolinone
producers.[11,12]
N-terminal His7-tagged AmbU1, AmbU2, AmbU3 and
AmbU4 proteins (Figure S4) devoid of the signal peptide
sequences were procured by heterologous expression in E. coli.
An initial glimpse on the enzymatic activity of AmbU protein
complex was obtained by incubating stoichiometric amounts of
AmbU1, AmbU2, AmbU3, AmbU4 (10 µM each) with 1 (1.0 mM)
and GPP (1.5 mM) in the presence of MgCl2 (20 mM) at pH=6.0.
The enzymatic reaction was initiated by the addition of AmbP1
(10 µM) at 30 ºC and stopped by EtOAc extraction after 2 h.
HPLC analysis of the crude extract revealed three discernible
peaks eluted between 28 and 40 mins (Figure 2, HPLC trace a),
In contrast to the welwitindolinones, the enzymatic
pathways for the assembly of ambiguine precursors remain
[a]
Dr. Qin Zhu and Prof. Dr. Xinyu Liu
Department of Chemistry, University of Pittsburgh, 219 Parkman
Avenue, Pittsburgh, PA 15260. E-mail: xinyuliu@pitt.edu
This article is protected by copyright. All rights reserved.

10.1002/anie.201703932
Angewandte Chemie International Edition
COMMUNICATION
of which UV spectra reassemble hapalindoles (Figure 2 inset
and S5). Their structural identities were confirmed to be 4, 3 and
hapalindole H 6 by either 1H NMR analysis (Figures S6/7), or
HPLC co-elution with the known compound (for 3, vide infra).[11]
The detection of 4 as one of the enzymatic products from 1 and
GPP by the AmbU quartet and AmbP1 is distinctively different
from what was recently observed with the cell lysate of F.
ambigua UTEX1903,[10] or the purified AmbU4.[11] This suggests
the enzymatic formation of 4, an authentic pathway precursor to
all ambiguines, might have been overlooked (Figure S8) and
prompted us to conduct in-depth characterizations.
AmbU1 with AmbU2 or AmbU3 did not alter their respective
inactive or unselective nature (Figure 3, traces i-j), however its
AmbP1 and AmbU
1 + GPP
C21H24N2
pH=6.0, Mg2+
H
NC
H
3
8
9 6
7
8
U4
U3
a
N
H
2/4
b
c
3
2
8
9 6
7
2
U1+2
U2
i
2/4
AmbP1 and AmbU
U1
d
e
U1+3
U2+3
j
1 + GPP
C
21H24N2
pH=6.0, Mg2+
4
U1+2+4
4
k
3
Boiled
U1+U2+
U3+U4
12
H
15
2
NC
11
10
7
H
b
U2+3+4
f
l
U1+4
U3+4
U2+4
3
U1+U2+
U3+U4
6
N
H
4
a
U1+2+3
U1+3+4
g
m
6
hapalindole H
28.0
30.0
32.0
34.0
36.0
38.0
40.0
Time (min)
h
n
60.0
40.0
60.0
%
%
222.0
28.0
32.0
Time (min)
36.0
40.0
28.0
32.0
Time (min)
36.0
40.0
2 UV spectrum
40.0
20.0
4 UV spectrum
7
Figure 3. Enzymatic characterizations of individual AmbU protein, their 1-to-1
stoichiometric trio and duo complexes. Assays were conducted with 1 (1.0
mM), GPP (1.5 mM) in the presence of MgCl2 (20 mM) at pH=6.0 and AmbU
proteins (10 µM each). The enzymatic reaction was initiated by the addition of
AmbP1 (10 µM) at 30 ºC and stopped by EtOAc extraction after 2 h.
20.0
280.0
nm
250 300 350 400
nm
-10.0
190
-10.0
190
250 300 350 400
Figure 2. Initial characterizations of AmbU1-U2-U3-U4 protein mixture.
combination with AmbU4 yielded a small amount of 4, ca. 3.5%
to 3 (Figure 3, trace l). Interestingly, the addition of AmbU2 or
AmbU3 to AmbU4 did not appear eroding the fidelity of AmbU4
(Figure 3, traces m-n), suggesting the detection of 4 from the
AmbU1+AmbU4 assay is unlikely an artifact. In addition, as
perceived from the AmbU trio assay, the combination of AmbU2
and AmbU3 robustly generated 6 with the concomitant formation
of 4 (ca. 17% to 6) (Figure 3, trace k).
As AmbU4 was shown highly thermostable,[10] we first boiled
the AmbU quartet and re-assessed their enzymatic activity with
AmbP1. HPLC analysis (Figure 2, HPLC trace b) showed 3
became the sole detectable hapalindole along with the AmbP1
product 2 and its rearranged byproduct 7, suggesting the
formation of 4 requires the remaining AmbU protein(s). Under
the HPLC condition we employed, 2 and 4 nearly co-elute,
however they can be easily distinguished by their characteristic
UV absorption spectra (Figure 2 insets). To trace the enzymatic
origin of 4, we first examined each individual AmbU protein. This
analysis showed only AmbU3 and AmbU4 exhibited enzymatic
activities (Figure 3, traces a-d). In contrast to AmbU4, AmbU3 is
unselective and generated four isomeric hapalindole-type
products (4, 8, 9 and 6) that were readily identifiable by UV and
HRMS (Figure S9). While compound 9 was not obtainable in a
pure form as it co-eluted with 6, compound 8 was isolated and
determined to be 12-epi-hapalindole C (Figure S10). This
indicates AmbU3 is an active isomerocyclase but incapable of
controlling the selectivity of all three elemental steps proposed
for this enzymatic cascade (Figure S11). Next we examined the
AmbU trio complex (Figure 3, traces e-h) by systematically
omitting individual AmbU protein from the initially tested AmbU
quartet (Figure 2). These assays showed the co-presence of
AmbU2 and AmbU3 appeared strongly promoting the formation
of 6 (Figure 3, traces f-g) and a small amount of 4 was
detectable in all AmbU trio mixtures. Building on these
observations, we proceeded further to examine all possible
AmbU duo complexes (Figure 3, traces i-n). The combination of
1 + GPP +AmbP1
pH=6.0, Mg2+
AmbU1
AmbU4
AmbU2
AmbU3
4
3
10xU1+U4
5xU1+U4
g
f
6
4
5xU2+U3
c
2xU2+U3
2xU1+U4
e
d
b
a
U2+U3
U1+U4
29.0 31.0
33.0
35.0
37.0
28.0
30.0
Time (min)
32.0
34.0
Time (min)
Figure 4. Effects of AmbU2 and AmbU1 stoichiometries in their respective
AmbU2/AmbU3 and AmbU1/AmbU4 enzymatic duos for the selective
formation of 6 and 4.
The characterizations of individual AmbU protein as well as
their 1-to-1 stoichiometric duos and trios indicate the formation
of 4, as observed in the initial examination of AmbU quartet,
originates primarily from AmbU2/AmbU3 and AmbU1/AmbU4
This article is protected by copyright. All rights reserved.

10.1002/anie.201703932
Angewandte Chemie International Edition
COMMUNICATION
duos. Although neither duo afforded 4 as a major product, these
experiments clearly showed the isomerocyclase selectivity of
AmbU4 or AmbU3 were altered when mixed with only one
equivalent of AmbU1 or AmbU2. This suggested to us that
varying the stoichiometry of AmbU proteins in each duo might
lead to an altered production of 4. To this end, we increased the
ratio of AmbU2 and AmbU1 in their respective duos with AmbU3
and AmbU4 from 1:1 to 2:1 and 5:1 and analyzed their
enzymatic products by HPLC (Figure 4). While increasing the
AmbU2 stoichiometry to AmbU3 led to only marginal selectivity
increase in the formation of 6 versus 4 (Figure 4, traces a-c), a
nearly 14-fold selectivity enhancement for the formation of 4
versus 3 was observed when the ratio of AmbU1 to AmbU4 was
increased to 5:1 (Figure 4, traces d-f). When this ratio was
further increased to 10:1, 4 became the major enzymatic product
(~2.5 times more than 3, Figure 4, trace g), which translated to a
>70-fold selectivity enhancement from the 1-to-1 stoichiometric
AmbU1/AmbU4 duo.
enzymatic activity of 10xAmbU1-AmbU4 duo was directly
assessed using 2 as its substrate at pH=6.0 (Figure 5c). This
analysis convincingly showed that Ca2+, but not Mg2+, is critical
for the selective generation of 4. These results further indicate
that the enzymatic reaction by 10xAmbU1-AmbU4 duo that
converts 2 to 4 is kinetically more rapid than that of AmbP1 that
generates 2 (Figure S12). These kinetic differences effectively
diminished the nonenzymatic rearrangement from 2 to 7 (Figure
S12), when AmbP1 is coupled to 10xAmbU1-AmbU4 duo with
Ca2+ at pH=6.0, mirroring what was recently observed for
WelU1.[11] In addition, we examined the calcium effect on the
enzymatic selectivity of standalone AmbU4 and AmbU1 as well
as 1xAmbU1-AmbU4, 2xAmbU1-AmbU4 and 5xAmbU1-AmbU4
duos (Figure S13). While the exogenous addition of Ca2+ did not
alter the selectivity profile of AmbU4 or AmbU1, it enhanced the
selectivity of AmbU1-AmbU4, 2xAmbU1-AmbU4 and 5xAmbU1-
AmbU4 duos towards 4, albeit the near-exclusive formation of 4
as shown by 10xAmbU1-AmbU4 was not achieved. These
results collectively indicate that both the stoichiometry of
AmbU1/AmbU4 and the presence of Ca2+ are critical for the
selective formation of 4.
1 + GPP +AmbP1 (pH=6.0)
a)
b)
10xAmbU1+AmbU4
varied metal ion
2+
c)
1 + GPP +AmbP1+Mg
10xAmbU1+AmbU4
4
3
varied pH
b)
a)
1 + GPP +AmbP1
Ca2+
1 + GPP +AmbP1 (pH=6.0)
pH=6.0,
WelU1 (treated)
4
3
AmbU2 or AmbU3 or
AmbU2-AmbU3
4
3
2+
2+
pH=9.0
Ca
d
c
d
c
2+
10
2 +
(pH=6.0)
Ca
WelU1
10xAmbU1+AmbU4
dialyzed against
c
b
a
AmbU3
pH=8.0
pH=7.0
pH=6.0
Zn
2+7
EDTA, then Ca
9
8
7
7
2/4
6
c
2+
2 +
(pH=6.0)
CN
H
Mg
c
10xAmbU1+AmbU4
WelU1
no additive
dialyzed against
6
b
b
a
2+
EDTA, then Mg
H
AmbU2
2
b
b
2+
2 std
Mg
a
N
6
WelU1
dialyzed against
EDTA
H
10
12-epi-fischerindole U
AmbU2-
AmbU3
28.0
30.0
Time (min)
32.0
34.0
28.0
30.0
Time (min)
32.0
34.0
30.0
35.0
Time (min)
40.0
a
a
30.0
35.0
Time (min)
40.0
Figure 5. Effects of pH and divalent metal ions on 10xAmbU1-AmbU4 duo for
the selective formation of 4: the critical role of Ca2+ revealed
30.0
35.0
Time (min)
40.0
Figure 6. The calcium effect on the selectivity of AmbU2-AmbU3 duo and the
activity of AmbU2 and WelU1.
Encouraged by this observation, we sought to identify
additional factor(s) for the AmbU1/AmbU4 enzymatic duo to
achieve the selective formation of 4. Initial examination of the pH
effect revealed the selectivity of 10-to-1 stoichiometric
AmbU1/AmbU4 (10xAmbU1-AmbU4) duo was optimal at pH=6.0
and deteriorated gradually to afford more 3 with the increased
pH (Figure 5a). This observation mirrors our recent finding that
the 12-epi-fischerindole U synthase WelU1 is most active at
pH=6.0.[11] We next examined the influence of divalent ions as
they have been found to influence the activity of AmbP1, the
aromatic prenyltransferase that generates substrate 2 for the
AmbU proteins (Figure 5b).[7] While the absence or presence of
Mg2+ (20 mM) did not alter the selectivity of 10xAmbU1-AmbU4
duo (Figure 5b, traces a-b), the addition of Ca2+ (20 mM) led to a
near-exclusive formation of 4 (>98% vs <2% for 3) (Figure 5b,
trace d). As a negative control, the addition of Zn2+ (20 mM),
known to inhibit the activity of AmbP1,[7] completely abolished
the formation of 4 and 3 (Figure 5b, trace c). The dramatic
selectivity enhancement of 10xAmbU1-AmbU4 duo by the
addition of Ca2+ was unexpected, in part because we have
previously shown Ca2+ partially impeded the activity of AmbP1 at
pH=8.0.[7] To ascertain the observed calcium effect, the
Analogous to what was observed for the 10xAmbU1-AmbU4
duo, the addition of Ca2+ (20 mM) to the AmbU2-AmbU3 duo
also enhanced its diastereoselectivity for the generation of 6
relative to 4 from a ratio of ~5:1 to >98:2 (Figure 4 trace a vs.
Figure 6a trace a). Furthermore, in the presence of Ca2+, AmbU2
alone gained activity and was able to generate 6 selectively with
the concomitant formation of 7 (Figure 3 trace c vs. Figure 6a
trace b), whereas the selectivity profile of AmbU3 in generating
hapalindoles was unaltered (Figure 3 trace b vs. Figure 6a trace
c). The critical role of Ca2+ in modulating the enzymatic
selectivity of 10xAmbU1-AmbU4 and AmbU2-AmbU3 duos as
well as the activity of AmbU2 prompted us to trace its
biochemical basis. Routine BLAST-P search (ref 14) showed the
U-type isomerocyclases encoded in the hapalindole-type
alkaloid biosynthetic pathways had no homology to any
functionally characterized proteins (Figure S14). However,
bioinformatics analysis using HHpred (ref 15) for remote protein
homology detection and structure prediction revealed they are
weakly related to a broad family of polysaccharide hydrolases,
which are known to naturally bound to calcium (Figure S15).
This article is protected by copyright. All rights reserved.

10.1002/anie.201703932
Angewandte Chemie International Edition
COMMUNICATION
Sequence alignment of all characterized AmbU and WelU
proteins revealed that the calcium binding residues as observed
in the related polysaccharide hydrolases are well conserved
(Figure S16). These bioinformatical data along with the calcium
effect observed for the 10xAmbU1-AmbU4, AmbU2-AmbU3
duos and standalone AmbU2 led us to hypothesize that the
entire family of U-type isomerocyclases encoded in the
hapalindole-type alkaloid biosynthetic pathways are calcium-
dependent.
When the 10xAmbU1-U4 and AmbU2-AmbU3 duos are mixed in
a
1:1 ratio and coupled with AmbP1, we observed the
preferential formation of 4 to 6 (Figure 7a, trace a). When this
enzymatically generated mixture of 4 and 6 was further treated
with AmbP3 and DMAPP, only 4 was processed to ambiguine H
11 (Figure 7a, traces b-c, for the structure of 11, see Figure S2).
In addition, when 6 was treated with the pathway-specific
halogenase AmbO5,[6] no enzymatic conversion was observed
(Figure S18). These results thus conclusively demonstrate 6 is
irrelevant to the biogenesis of ambiguines (Figure 7b).
During the previous characterization of 12-epi-fischerindole
U synthase WelU1,[11] we have shown that the addition of Ca2+
or EDTA to the WelU1 assay mixture immediately prior to
initiating the enzymatic reaction had little effect on its activity.
We reasoned this conundrum was due to the presence of
prebound Ca2+ in the heterologously purified WelU1 that might
not be readily sequestered by EDTA. ICP-MS analysis of
purified WelU1 confirmed the presence of Ca2+. Furthermore,
when we pre-dialyzed the WelU1 protein against EDTA (20 mM)
for 1 h, its enzymatic activity was completely abolished (Figure
6b, trace a). The diminished activity was readily rescued by
further dialyzing the EDTA-treated WelU1 against Ca2+ (20 mM)
for 1 h (Figure 6b, trace c), but not against Mg2+ (20 mM) (Figure
6b, trace b), confirming the enzymatic activity of WelU1 is
calcium-dependent. Analogously, the enzymatic activities of the
In summary, intrigued by the biosynthetic origin of 4 and the
conundrum presented by the recent discovery of its C-12 epimer
(3) synthase AmbU4, we systematically reconstituted all AmbU-
type isomerocyclases encoded in the ambiguine pathway. This
endeavor led to the discovery of two calcium-dependent
enzymatic cascades for the selective formation of
4 by
10xAmbU1-U4 and 6 by AmbU2-U3 or AmbU2 (Figure 6b). The
mechanism for their formations from 2 can be formulated as
proceeding through a Cope rearrangement via a boat transition
state, followed by a stereodivergent aza-Prins cyclization and a
regioselective carbocation deposition to the C4 indole (Figure
S19). Notably, in the presence of Ca2+, the 10xAmbU1
completely altered the initial Cope rearrangement selectivity of
AmbU4, which otherwise is a dedicated synthase for 3 in vitro
(Figure 6b, right panel). AmbU2, on the other hand, is a
dedicated synthase for 6 in the presence of Ca2+, and its activity
can be further enhanced by AmbU3, an unselective
isomerocyclase that generates four isomeric hapalindoles (4, 6,
8, 9).
standalone 12-epi-hapalindole
C synthase WelU3, 12-epi-
hapalindole U synthase AmbU4 as well as 10xAmbU1-U4 and
AmbU2-AmbU3 duos were shown completely abolished upon
dialyzing against EDTA but rescuable by re-dialyzing against
Ca2+ (Figure S17). Collectively, these results validate that the
activities and selectivities of U-type isomerocyclases depend on
the presence of calcium.
The unprecedented calcium-dependency of the 10xAmbU1-
U4, AmbU2-U3 duos and standalone AmbU2 led us to re-
evaluate and re-define the isomerocyclase proteins encoded in
the hapalindole-type biosynthetic pathways as a new type of
calcium-dependent enzymes. To the best of our knowledge,
these U-type isomerocyclases represent the only known
calcium-dependent isomerase or cyclase involved in the
biogenesis of microbial secondary metabolites. While the
elucidation of the mechanism underlying how calcium modulates
the activities and selectivities of these isomerocyclases and their
complexes remain underway, the studies presented here
effectively bridged the missing links and refuted the intermediacy
of 3 in the early stage biosynthesis of ambiguines (Figure 6b).
The identification of the 10xAmbU1-U4 enzymatic complex as
the dedicated hapalindole U synthase also allowed to establish 4
as the bona fide biosynthetic intermediate for the assembly of
hapalindole A, the founding member of hapalindole-type
alkaloids,[18] based on the co-occurrence of AmbU1-AmbU4
coding genes in its biosynthetic gene cluster sequenced from
Fischerella muscicola UTEX LB1829 (Figure S20).[19,20] This
corroborates with our recent observation that an enzyme-
mediated tertiary carbon epimerization at the C-10 center of 4
and 5 is operant for the stereodivergent biogenesis of the
hapalindole-type alkaloids in the same organism. [21]
a)
b) minor pathway irrelevant
major pathway relevant
to ambiguine biogenesis
to ambiguine biogenesis
Ambiguine H (11) std
11
AmbU3 alone
mixture of
3, 4, 6, 9
AmbU4 alone
3 only
c
AmbU2 alone
w/ Ca2+ added
AmbU1 alone
no activity
1 + GPP +AmbP1
10xAmbU1-AmbU4
AmbU2-AmbU3
6
4
2+
pH=6.0, Ca
AmbU2-AmbU3
10AmbU1-AmbU4
AmbP3, DMAPP
4 + 3
6 +
4
2
6
b
a
w/o Ca2+ added
w/ Ca2+ added
w/o Ca2+ added
w/ Ca2+ added
~ 3
:
1
~ 5
:
1
1 + GPP +AmbP1
>98
:
<2
>98
:
<2
10xAmbU1-AmbU4
AmbU2-AmbU3
11
5
2+
AmbP3
pH=6.0, Ca
AmbP3
6
4
AmbO5
AmbO5
30.0
35.0 40.0 45.0
Time (min)
Figure 7. a) Contrast and compare the biosynthetic relevancy of 4 and 6 with
regards to the assembly of ambiguine H (11), b) Summary of the enzymatic
basis for the early stage structural diversifications in the biogenesis of
ambiguines. All U-type proteins are most active at pH=6.0. The activity and
selectivity profiles are based AmbU proteins purified heterologously from E.
coli without EDTA treatment. The activity and selectivity profiles of standalone
AmbU1, AmbU4 and AmbU4 were unaltered with the exogenous addition of
Ca2+, whereas AmbU2 was inactive without exogenous Ca2+
.
Having revealed the enzymatic basis for the selective
formation of 4 and 6, we sought to further examine and compare
their biosynthetic relevancy to the ambiguine assemblies.
Although 6 has been isolated from two hapalindole-type alkaloid
In addition, the disclosure on the in vitro enzymatic assembly
of 4, an on-pathway biogenetic intermediate to ambiguines and
related hapalindole alkaloids, that requires a 10-to-1 mixture of
AmbU1 and AmbU4 in the presence of calcium contrasts a
producers including the ambiguine producer F. ambigua
[16,17]
UTEX1903 (Figure S2),
its chlorinated or prenylated
derivatives are not known, indicating it is a pathway outlier.
This article is protected by copyright. All rights reserved.

10.1002/anie.201703932
Angewandte Chemie International Edition
COMMUNICATION
[12] Q. Zhu, X. Liu, Beilstein J. Org. Chem. 2017, 13, 1168–1173.
doi:10.3762/bjoc.13.115
recent claim, emerged during the preparation of this
manuscript,[22] that states the structural diversifications from 2 in
the biosynthesis of hapalindole-type alkaloids is controlled by
the homo- or hetero-dimeric complex of the U-type
isomerocyclases. We reasoned this over-interpretation is likely
due to the failure to appreciate that 6 is a pathway outlier in the
biogenesis of ambiguines and related hapalindole alkaloids. As
the in vitro enzymatic reconstitution remains the only viable
method to dissect the biochemical basis for the biosynthesis of
hapalindole-type alkaloids, we caution for the pitfalls in the
interpretation of these in vitro data without thoroughly correlating
them with the observed structural diversity in the known
hapalindole-type alkaloid producer.
[13] For related enzyme nomenclatures: a) oxidocyclase (an enzyme
catalyzing an oxidation induced cyclization) in cyclopiazonic acid
biosynthesis, J. C. Schabort, D. C. Wilkens, C. W. Holzapfel, D. J. Potgieter,
A. W. Neitz, Biochim Biophys Acta 1971, 250, 311-328, b) isomeroreductase
(an enzyme catalyzing the sequential isomerization and reduction) in valine
and isoleucine biosynthesis, S. M. Arfin, H. E. Umbarger, J Biol Chem 1969,
244, 1118-1127.
[14] S. F. Altschul, W. Gish, W. Miller, E. W. Myers, D. J. Lipman, J Mol Biol
1990, 215, 403-410.
[15] J. Soding, A. Biegert, A. N. Lupas, Nucleic Acids Res 2005, 33, W244-
248.
[16] T. A. Smitka, R. Bonjouklian, L. Doolin, N. D. Jones, J. B. Deeter, W. Y.
Yoshida, M. R. Prinsep, R. E. Moore, G. M. L. Patterson, J Org Chem 1992,
57, 857-861.
[17] R. E. Moore, C. Cheuk, X. Q. G. Yang, G. M. L. Patterson, R. Bonjouklian,
T. A. Smitka, J. S. Mynderse, R. S. Foster, N. D. Jones, J Org Chem 1987, 52,
1036-1043.
Acknowledgements
We thank University of Pittsburgh and National Institute of
Health for supporting this work.
[18] R. E. Moore, C. Cheuk, G. M. L. Patterson, J Am Chem Soc 1984, 106,
6456-6457.
[19] A. Park, R. E. Moore, G. M. L. Patterson, Tetrahedron Lett 1992, 33,
3257-3260.
Keywords: alkaloid • biosynthesis • calcium • enzyme • cascade
[20] KF819511, https://www.ncbi.nlm.nih.gov/nuccore/583831257, deposited
on Nov 06, 2013
[1] M. L. Hillwig, Q. Zhu, X. Liu, ACS Chem Biol 2014, 9, 372-377.
[2] M. L. Hillwig, H. A. Fuhrman, K. Ittiamornkul, T. J. Sevco, D. H. Kwak, X.
Liu, ChemBioChem 2014, 15, 665-669.
[21] X. Liu, M. L. Hillwig, Q. Zhu, H. A. Fuhrman, K. Ittiamornkul, T. J. Sevco,
manuscript in preparation
[22] S. Li, A. N. Lowell, S. A. Newmister, F. Yu, R. M. Williams, D. H. Sherman,
Nat Chem Biol 2017, doi:10.1038/nchembio.2327. The authors of this paper
also reported that both AmbU3 and AmbU2 had no observable activity but
were active when combined and gave the exclusive formation of 6 at pH=7.8
[3] M. L. Micallef, D. Sharma, B. M. Bunn, L. Gerwick, R. Viswanathan, M. C.
Moffitt, BMC Microbiol 2014, 14, 213.
[4] M. L. Hillwig, X. Liu, Nat Chem Biol 2014, 10, 921-923.
[5] K. Ittiamornkul, Q. Zhu, D. S. Gkotsi, D. R. M. Smith, M. L. Hillwig, N.
Nightingale, R. J. M. Goss, X. Liu, Chemi Sci 2015, 6, 6836-6840
[6] M. L. Hillwig, Q. Zhu, K. Ittiamornkul, X. Liu, Angew. Chem. 2016, 128,
5874-5878; Angew. Chem. Int. Ed. 2016, 55, 5780-5784.
[7] X. Liu, M. L. Hillwig, L. M. Koharudin, A. M. Gronenborn, Chem Commun
2016, 52, 1737-1740.
in the absence of Ca2+
.
[8] A. J. Mitchell, Q. Zhu, A. O. Maggiolo, N. R. Ananth, M. L. Hillwig, X. Liu, A.
K. Boal, Nat Chem Biol 2016, 12, 636-640.
[9] Q. Zhu, M. L. Hillwig, Y. Doi, X. Liu, ChemBioChem 2016, 17, 466-470
[10] S. Li, A. N. Lowell, F. Yu, A. Raveh, S. A. Newmister, N. Bair, J. M.
Schaub, R. M. Williams, D. H. Sherman, J Am Chem Soc 2015, 137, 15366-
15369.
[11] Q. Zhu, X. Liu, Chem Commun 2017, 53, 2826-2829.
This article is protected by copyright. All rights reserved.

10.1002/anie.201703932
Angewandte Chemie International Edition
COMMUNICATION
COMMUNICATION
minor pathway irrelevant
to ambiguine biogenesis
major pathway relevant
to ambiguine biogenesis
Qin Zhu and Xinyu Liu*
Calcium is the key! The pursuit of the
enzymatic basis for hapalindole U (4),
a validated biogenetic precursor for the
ambiguine alkaloids, led to the
AmbU2, pH=6.0
w/ Ca2+ added
6
10AmbU1-AmbU4
Page No. – Page No.
NC
pH=6.0
4 +
3
1
AmbU2-AmbU3
pH=6.0
w/o Ca2+ added
w/ Ca2+ added
~ 3 :
N
Discovery of a calcium-dependent
enzymatic cascade for the selective
assembly of hapalindole-type
alkaloids: On the biosynthetic
origin of hapalindole U
6
~ 5
+
:
4
1
2
>98
:
<2
w/o Ca2+ added
w/ Ca2+ added
discovery
of
an
unprecedented
isomerocyclase
>98
:
<2
calcium-dependent
validated ambiguine
biogenetic precursor
complex for its selective assembly.
This discovery allowed to establish the
isomerocyclases involved in the
hapalindole-type alkaloid biogenesis
as a new family of calcium-dependent
enzymes.
12
12
12
H
15
H
15
H
15
NC
11
NC
11
NC
11
10
10
H
10
H
H
6
4
3
N
H
N
H
N
H
hapalindole H
hapalindole U
12-epi-hapalindole U
This article is protected by copyright. All rights reserved.