Molecular insights into the enzyme promiscuity of an aromatic prenyltransferase

Article abstract of DOI:10.1038/nchembio.2263

Aromatic prenyltransferases (aPTases) transfer prenyl moieties from isoprenoid donors to various aromatic acceptors, some of which have the rare property of extreme enzymatic promiscuity toward both a variety of prenyl donors and a large diversity of acceptors. In this study, we discovered a new aPTase, AtaPT, from Aspergillus terreus that exhibits unprecedented promiscuity toward diverse aromatic acceptors and prenyl donors and also yields products with a range of prenylation patterns. Systematic crystallographic studies revealed various discrete conformations for ligand binding with donor-dependent acceptor specificity and multiple binding sites within a spacious hydrophobic substrate-binding pocket. Further structure-guided mutagenesis of active sites at the substrate-binding pocket is responsible for altering the specificity and promiscuity toward substrates and the diversity of product prenylations. Our study reveals the molecular mechanism underlying the promiscuity of AtaPT and suggests an efficient protein engineering strategy to generate new prenylated derivatives in drug discovery applications.

Full text of DOI:10.1038/nchembio.2263

article
puBLIsHed OnLIne: 19 deCeMBer 2016 | dOI: 10.1038/nCHeMBIO.2263
Molꢀcꢁlꢂꢃ iꢄꢅighꢆꢅ iꢄꢆo ꢆhꢀ ꢀꢄzymꢀ ꢇꢃomiꢅcꢁiꢆy
of ꢂꢄ ꢂꢃomꢂꢆic ꢇꢃꢀꢄylꢆꢃꢂꢄꢅfꢀꢃꢂꢅꢀ
1,7
2,3,7
1,6,7
1
4
4
1
riꢈꢂo Chꢀꢄ , Biꢄgqꢁꢂꢄ Gꢂo , Xiꢂo Liꢁ , Fꢀiyiꢄg rꢁꢂꢄ , Yoꢄg Zhꢂꢄg , Jizhoꢄg Loꢁ , Kꢀꢇiꢄg Fꢀꢄg ,
5
5
1
2,3
Cꢂꢃꢅꢆꢀꢄ Wꢁꢄꢅch , shꢁ-Miꢄg Li , Jꢁꢄgꢁi dꢂi * & Fꢀi sꢁꢄ *
Aromatic prenyltransferases (aPTases) transfer prenyl moieties from isoprenoid donors to various aromatic acceptors, some of
which have the rare property of extreme enzymatic promiscuity toward both a variety of prenyl donors and a large diversity of
acceptors. In this study, we discovered a new aPTase, AtaPT, from Aspergillus terreus that exhibits unprecedented promiscuity
toward diverse aromatic acceptors and prenyl donors and also yields products with a range of prenylation patterns. Systematic
crystallographic studies revealed various discrete conformations for ligand binding with donor-dependent acceptor specificity
and multiple binding sites within a spacious hydrophobic substrate-binding pocket. Further structure-guided mutagenesis of
active sites at the substrate-binding pocket is responsible for altering the specificity and promiscuity toward substrates and
the diversity of product prenylations. Our study reveals the molecular mechanism underlying the promiscuity of AtaPT and
suggests an efficient protein engineering strategy to generate new prenylated derivatives in drug discovery applications.
he transfer of prenyl moieties, as mediated by aPTases, is a particularly the lack of structures of these enzymes in complex
ubiquitous reaction in the biosynthesis of both primary and with diverse donors and/or acceptors, have hampered the efforts
1–4
T
secondary metabolites in plants, bacteria and fungi . The to characterize the molecular mechanism underlying the promiscu-
prenylation of aromatic compounds and a variety of subsequent ity of aPTases.
tailoring reactions, including oxidation, reduction and cyclization,
In the present work, we identified and functionally character-
give rise to a plethora of ‘hybrid’ natural products that exhibit diverse ized the aPTase gene AtaPT (A. terreus aromatic prenyltransferase)
biological properties. In particular, the introduction of prenyl from an A. terreus strain . Unlike any of the previously character-
21
moieties often increases the lipophilicity of these compounds, ized aPTases, AtaPT showed unprecedented substrate promiscu-
5
enhancing their interactions with biological membranes . ity, both for aromatic acceptors and for prenyl donors (C −C ),
5
20
Modification of drug-like natural products via one-step prenyla- including dimethylallyl, geranyl, farnesyl, geranylgeranyl and phytyl
tion is seen as a convenient approach for the synthesis of bioactive pyrophosphates (DMAPP, GPP, FPP, GGPP and PPP, respectively).
compounds in drug discovery programs. However, the present In addition, the prenylation catalyzed by AtaPT also had the
lack of a widely applicable method for the prenylation of diverse ability to form mono-, di- and/or triprenylated products with C–C
6
aromatic scaffolds remains a challenge .
and/or C–O bonds. Systematic structural and functional investiga-
The aPTases involved in secondary metabolism have attracted tions revealed the structural basis of AtaPT’s substrate promiscu-
considerable attention in recent years. Examples include the dis- ity. Moreover, structure-guided mutagenesis yielded new enzyme
covery of soluble ABBA-type and dimethylallyltryptophan syn- mutants with tuned catalytic specificities.
7,8
9–12
thase (DMATS)-type aPTases from bacteria and fungi as well
4,13
as the characterization of membrane-bound aPTases . These RESULTS
aPTases catalyze prenylations by forming C–C, C–O or C–N The unprecedented substrate tolerance of AtaPT
8
14
22
bonds via electrophilic alkylation . The ABBA-type aPTase NphB
We cloned the AtaPT gene from A. terreus A8-4 via RT-PCR
can catalyze the prenylation of several aromatic compounds (for (Supplementary Results, Supplementary Tables 1 and 2). The
example, hydroxynaphthalenes, flavonoids and polyketides) via a protein putatively encoded by AtaPT has a length of 424 amino
2
+
15
Mg -dependent mechanism. The DMATS-type aPTases, AnaPT , acids and a theoretical molecular weight of 47.5 kDa. AtaPT shares
16
17
18
BrePT , 7-DMATS and CdpNPT , can catalyze the prenylation high sequence identity with the EAU34068 protein from A. terreus
of tryptophan-containing cyclic dipeptides, hydroxynaphthalenes NIH2624 (Supplementary Table 1), differing at only three residues,
2+
or flavonoids via a Mg -independent mechanism. The aPTases including 290 (threonine in AtaPT versus lysine in EAU34068),
characterized to date have typically exhibited a strict specificity 292 (alanine versus glutamic acid) and 373 (asparagine versus
for their prenyl donors but have had relative diversity in their serine). Additionally, AtaPT shares 42% and 29% sequence identity
aromatic acceptors. However, BAE61387, TleC and MpnD can with AnaPT and FgaPT2 (ref. 23), respectively. Further sequence
10,19,20
accept various prenyl donors with differing chain lengths
Even though these enzymatic properties are chemically and biologi- and has close evolutionary relationships with aPTases from
.
analysis indicated that AtaPT belongs to the DMATS superfamily
cally intriguing, the limited number of available crystal structures, fungi (Supplementary Fig. 1).
1
Sꢁaꢁꢃ Kꢃy laꢈꢅraꢁꢅry ꢅf bꢆꢅaꢄꢁꢆꢉꢃ Sꢂꢈsꢁaꢀꢄꢃ aꢀꢊ Fꢂꢀꢄꢁꢆꢅꢀ ꢅf naꢁꢂraꢇ Mꢃꢊꢆꢄꢆꢀꢃs, iꢀsꢁꢆꢁꢂꢁꢃ ꢅf Maꢁꢃrꢆa Mꢃꢊꢆꢄa, chꢆꢀꢃsꢃ Aꢄaꢊꢃmy ꢅf Mꢃꢊꢆꢄaꢇ Sꢄꢆꢃꢀꢄꢃs aꢀꢊ
2
pꢃkꢆꢀg uꢀꢆꢅꢀ Mꢃꢊꢆꢄaꢇ cꢅꢇꢇꢃgꢃ, bꢃꢆjꢆꢀg, chꢆꢀa. naꢁꢆꢅꢀaꢇ laꢈꢅraꢁꢅry ꢅf bꢆꢅmaꢄrꢅmꢅꢇꢃꢄꢂꢇꢃs, cAS cꢃꢀꢁꢃr fꢅr exꢄꢃꢇꢇꢃꢀꢄꢃ ꢆꢀ bꢆꢅmaꢄrꢅmꢅꢇꢃꢄꢂꢇꢃs, iꢀsꢁꢆꢁꢂꢁꢃ ꢅf
bꢆꢅꢋhysꢆꢄs, chꢆꢀꢃsꢃ Aꢄaꢊꢃmy ꢅf Sꢄꢆꢃꢀꢄꢃs, bꢃꢆjꢆꢀg, chꢆꢀa. uꢀꢆꢉꢃrsꢆꢁy ꢅf chꢆꢀꢃsꢃ Aꢄaꢊꢃmy ꢅf Sꢄꢆꢃꢀꢄꢃs, bꢃꢆjꢆꢀg, chꢆꢀa. Kꢃy laꢈꢅraꢁꢅry ꢅf RnA bꢆꢅꢇꢅgy,
iꢀsꢁꢆꢁꢂꢁꢃ ꢅf bꢆꢅꢋhysꢆꢄs, chꢆꢀꢃsꢃ Aꢄaꢊꢃmy ꢅf Sꢄꢆꢃꢀꢄꢃs, bꢃꢆjꢆꢀg, chꢆꢀa. iꢀsꢁꢆꢁꢂꢁ für pharmazꢃꢂꢁꢆsꢄhꢃ bꢆꢅꢇꢅgꢆꢃ ꢂꢀꢊ bꢆꢅꢁꢃꢄhꢀꢅꢇꢅgꢆꢃ, phꢆꢇꢆꢋꢋs-uꢀꢆꢉꢃrsꢆꢁäꢁ
Marꢈꢂrg, Marꢈꢂrg, Gꢃrmaꢀy. prꢃsꢃꢀꢁ aꢊꢊrꢃss: Mꢅꢊꢃrꢀ Rꢃsꢃarꢄh cꢃꢀꢁꢃr fꢅr traꢊꢆꢁꢆꢅꢀaꢇ chꢆꢀꢃsꢃ Mꢃꢊꢆꢄꢆꢀꢃ, bꢃꢆjꢆꢀg uꢀꢆꢉꢃrsꢆꢁy ꢅf chꢆꢀꢃsꢃ Mꢃꢊꢆꢄꢆꢀꢃ,
3
4
5
6
7
bꢃꢆjꢆꢀg, chꢆꢀa. thꢃsꢃ aꢂꢁhꢅrs ꢄꢅꢀꢁrꢆꢈꢂꢁꢃꢊ ꢃqꢂaꢇꢇy ꢁꢅ ꢁhꢆs wꢅrk. *ꢃ-maꢆꢇ: jgꢊaꢆ@ꢆmm.aꢄ.ꢄꢀ ꢅr fꢃꢆsꢂꢀ@sꢂꢀ5.ꢆꢈꢋ.aꢄ.ꢄꢀ
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.ꢀaꢁꢂrꢃ.ꢄꢅm/ꢀaꢁꢂrꢃꢄhꢃmꢆꢄaꢇꢈꢆꢅꢇꢅgy
1

NATURE ꢀꢁEꢂIꢀAL ꢃIOLOꢄꢅ dOI: 10.1038/nCHeMBIO.2263
article
H
N
R
2
R4
OH
O
O
H
3
CO
O
R1
R3
R2
HO
R2
R5
R4
HO
O
O
NH
4
N
O
O
O
HN
N
HN
OH
R
HO
InM
OH
R
1
R1
O
R3
O
OH
1
2
(4R)
(4S)
3 (R = InM)
4 (R = HPM)
1, 2, 3
= H; R4 = OH)
= H; R = OH)
(R = H; R = OH)
= H; R = OH)
11 (R = H; R = OH)
13 (R1, 2, 3, 4 = H; R5 = OH)
7
8
9
(R
(R
HO
1, 2, 4
1, 2
3
1, 3, 4, 5
1, 5
2
O
14 (R
= H; R = OH)
15 (R = OH; R
HPM
3, 4
2, 3, 4
1
(R = H; R2 = NH2)
(R = OH; R2 = H)
= H)
5
6
1, 3, 4
1, 4
2
1, 3, 5
= OH; R2, 4 = H)
3, 4, 5
1
0 (R
16 (R
1
2, 3
1, 2
OH
17 (R = OH; R
= H)
1
2, 3, 4
1, 2, 4
= OH; R3, 5 = H)
1
2 (R = OH; R
= H)
18 (R
OH
OH
R
4
R
R
3
2
O
R2
O
O
R
R2
OH
O
HO
O
O
R2
O
R
3
O
OH
OH
OGlu
R
1
R4
O
HO
R1
HO
O
O
R3
R1
O
OH
OH
3 (R1 = H; R2, 4 = OH)
4 (R = OH; R = H)
, 3
1, 2, 3, 4
1, 2, 4
HO
42 (R = H)
2
2
29 (R
30 (R
= H)
= H; R = OH)
3, 4
1
2, 3
OGal
R1
O
1
9 (R = H; R = OH)
1
, 2
3, 4
3
43 (R = CH3)
1
2
3
1, 2, 4
3
1, 2
35 (R1, 2
36 (R1, 2 = OH)
R1
2
0 (R = H; R =OH; R = OCH3)
25 (R
26 (R
27 (R
= OH; R = H)
31 (R = H; R = OH)
= H)
1
, 2
= OH; R3 = H; R = OCH3)
4
1, 4
2, 3
1 (R1, 2, 3 = OH)
2 (R1, 2 = OH; R3 = OGlu)
32 (R = H; R = OH)
OH
R2
2
2
1
, 2, 3, 4
33 (R = H; R = OH; R4 = OGal)
1
2, 3
= OH)
1, 4
8 (R = OH; R2 = OCH ; R3 = H)
34 (R1
, 3, 4
= OH; R2 = H)
2
3
1
2
4
4 (R = H; R = OH)
1
2
4
5 (R = OH; R = H)
OH
OH
OCH
OH
HO
O
3
OCH
3
OCH
3
H
HO
OH
O
H
3
CO
HO
O
O
HO
OH
HO
H
HO
O
O
H
OH
OH
O
O
O
O
37
38
39
40
41
46
Figure 1 | Structures of the aromatic compounds assayed as prenyl acceptors. iꢀM, 1H-ꢆꢀꢊꢅꢇ-3-yꢇmꢃꢁhyꢇ; HpM, 4-hyꢊrꢅxyꢋhꢃꢀyꢇmꢃꢁhyꢇ;
oGꢇꢂ, O-β-d-gꢇꢂꢄꢅꢋyraꢀꢅsyꢇꢅxy; oGaꢇ, O-β-d-gaꢇaꢄꢁꢅꢋyraꢀꢅsyꢇꢅxy. tꢃsꢁꢃꢊ arꢅmaꢁꢆꢄ ꢄꢅmꢋꢅꢂꢀꢊs wꢆꢁh ꢀꢅ ꢅr ꢇꢅw ꢄꢅꢀꢉꢃrsꢆꢅꢀ (<5%) arꢃ shꢅwꢀ
ꢆꢀ Supplementary Figure 2.
Recombinant AtaPT was produced in Escherichia coli and sub- It should be noted that our subsequent genetic studies showed
sequently purified for use in enzymatic assays. Initial incubation that AtaPT was not the aPTase gene involved in the biosynthesis
24
with a range of routinely tested candidate substrates (for example, of antitumor metabolite (+)-butyrolactone I (47a) in A. terreus
2
5
tryptophan-containing cyclic dipeptides, hydroxynaphthalenes and (Supplementary Fig. 6), a finding consistent with a recent report .
flavonoids as acceptors) and with DMAPP as the prenyl donor in
the absence of divalent metal ions revealed the exceptionally relaxed AtaPT is capable of promiscuous prenylation
substrate specificity of AtaPT.
In addition to substrate promiscuity, we also investigated the
We next used a panel of 106 drug-like aromatics as potential promiscuous prenylation capacity of AtaPT. On the basis of the
acceptors and used DMAPP, GPP and FPP as prenyl donors to results of the initial exploratory assays, we selected nine drug-
extensively investigate the substrate promiscuity of AtaPT (Fig. 1 like scaffolds with complex structures, including 1, (−)-cyclo-
and Supplementary Fig. 2). HPLC–diode array detection (DAD)– L-Trp-L-Trp (3), 9-aminocamptothecin (5), norathyriol (11),
MS analyses of reaction products revealed substantial substrate 4,4′-dihydroxybenzophenone (13), genistein (21), sophoricoside
promiscuity of AtaPT for both the acceptors and donors. AtaPT (22), resveratrol(41)and7-hydroxycoumarin(42), touseinprepara-
can catalyze the prenylations of 72 aromatic compounds with all tive-scalereactionsandtomeasureenzymekinetics(Supplementary
three different prenyl donors. Included among the accepted com- Table 3). In total, we obtained and determined the structures of 42
pounds were many drug-like natural products, including lignanoids mono-, di- and/or triprenylated products, with a diversity of both
(
1 and 2), tryptophan-containing cyclic dipeptides (3 and 4), quino- C- and/or O-prenylations at differing positions, using high-reso-
line alkaloids (5 and 6), xanthones (7–12), benzophenones (13–18), lution MS and NMR analyses (Fig. 2b; Supplementary Fig. 7 and
flavonoid aglycons and glycosides (19–37, 40), a phenylethyl Supplementary Note). The prenylated products included 35 new
chromone (38), a curcuminoid (39), a stilbene (41), coumarins (42 compounds and seven previously reported bioactive compounds,
and 43), hydroxynaphthalenes (44 and 45) and 4-hydroxybenzalde- 47a, 50a, 50b, 51a, 51b, 58a and 59a. We performed in vitro bio-
hyde (46). Furthermore, we found that the initial monoprenylated activity assays with all of these compounds. For instance, we found
products were subsequently accepted by AtaPT as new substrates that prenylated products with dimethylallyl or geranyl substitutions
for further reactions that formed di- and/or triprenylated products (50a–50c and 51a–51d) of the anticancer agent 21 (refs. 26,27)
with both C-prenylation and O-prenylation, thus providing further exhibited stronger antiproliferation effects against three human
evidence of the extreme degree of substrate promiscuity of AtaPT cancer cell lines (HCT-8, Bel7402 and BGC823) and had reduced
(Fig. 2, Supplementary Fig. 3 and Supplementary Note).
half-maximal inhibitory concentration (IC ) values compared to
50
28
To investigate the prenyl donor promiscuity of AtaPT, we 21 when assessed using an MTT assay method (Supplementary
incubated (+)-butyrolactone II (1) with GGPP or PPP. Notably, Table 4). This result suggested that these prenylated products might
HPLC–MS analyses revealed that AtaPT can use these substrates be promising anticancer leads in the future.
to produce a series of geranylgeranylated and phytylated products
It is typically thought that prenylation reactions catalyzed by the
(
Supplementary Figs. 4 and 5). To the best of our knowledge, AtaPT aPTases have a regioselective preference at the ortho position to the
is the first aPTase shown to be able to catalyze prenylations of lig- electron-donating hydroxyl group of an aromatic ring, a scenario
nanoids, quinoline alkaloids, benzophenones, flavonoid glycosides that follows the Friedel–Crafts rule. However, for the prenylations
and curcuminioids while accepting various types of prenyl donors. catalyzed by AtaPT, we observed unusual prenylation patterns
2
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.ꢀaꢁꢂrꢃ.ꢄꢅm/ꢀaꢁꢂrꢃꢄhꢃmꢆꢄaꢇꢈꢆꢅꢇꢅgy

NATURE ꢀꢁEꢂIꢀAL ꢃIOLOꢄꢅ dOI: 10.1038/nCHeMBIO.2263
article
a
100
*
80
60
40
20
0
*
*
*
*
*
*
*
*
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
R2
8
b
OH
O
'
H3CO
O
H3CO
O
R2
PPi
4
HO
O
O
PPi
AtaPT
HO
O
R3
3
R1
3' R3
3
'
''
4''
R1
AtaPT
DMAPP
O
6
O
R4
OH
OH
R4
HO
O
OH
DMAPP
OH
O
OH
4'
O
O
O
Genistein (21)
50a (R1, 3 = dimethylallyl; R2, 4 = H)
47a (R1
, 2, 4
, 3, 4
, 3
= H; R3 = dimethylallyl)*
(
+) -Butyrolactone II (1)
1
2
4
4
4
4
7b (R
= H; R = dimethylallyl)*
1, 4
2, 3
5
0b (R = H; R = dimethylallyl)
1
2, 4
7c (R = dimethylallyl; R = H)*
1
2, 3, 4
5
0c (R = H; R
= dimethylallyl)
1
, 4
2, 3
7d (R = H; R = dimethylallyl)*
1
, 3, 4
2
R2
7e (R
= dimethylallyl; R = H)*
PPi
HO
O
O
PPi
H3CO
O
R2
R3
R3
O
AtaPT
GPP
R1
R1
AtaPT
GPP
R4
O
OH
O
OH
OH
51a (R1, 2 = H; R3 = geranyl)**
O
8a (R1
8b (R = geranyl; R
, 2, 4
= H; R3 = geranyl)
1, 3
2
4
4
4
4
4
4
51b (R = H; R = geranyl)***
1
2, 3, 4
1, 3
2
= H)
51c (R = geranyl; R = H)
1
, 3, 4
2
1
2, 3
8c (R
= H; R = geranyl)
51d (R = H; R = geranyl)
1
, 3
2, 4
8d (R = geranyl; R = H)
R3
1
, 4
2, 3
8e (R = geranyl; R = H)
1
, 4
2, 3
PPi
R2
O
O
O
8f (R = H; R = geranyl)
R4
O
AtaPT
FPP
R1
O
H3CO
O
PPi
AtaPT
R
R5
OH
52a (R1, 4 = farnesyl; R2, 3, 5 = H)
2b (R1, 2, 5 = H; R3, 4 = farnesyl)
2c (R1, 3, 5 = H; R2, 4 = farnesyl)
2d (R1, 2, 4 = H; R3, 5 = farnesyl)
O
OH
FPP
HO
5
5
5
O
4
9 (R = farnesyl)
Dimethylallyl
Geranyl
Farnesyl
Figure 2 | Prenylation of aromatics by AtaPT. (a) pꢃrꢄꢃꢀꢁ ꢄꢅꢀꢉꢃrsꢆꢅꢀ ꢅf ꢃaꢄh arꢅmaꢁꢆꢄ whꢃꢀ ꢂsꢆꢀg Gpp as ꢋrꢃꢀyꢇ ꢊꢅꢀꢅr. thꢃ ꢄꢅꢇꢅrs ꢆꢀ ꢁhꢃ ꢈar graꢋhs
rꢃꢋrꢃsꢃꢀꢁ ꢊꢆffꢃrꢃꢀꢁ raꢁꢆꢅs ꢅf ꢉarꢆꢅꢂs ꢋrꢃꢀyꢇaꢁꢃꢊ ꢋrꢅꢊꢂꢄꢁs ꢆꢀ ꢁhꢃ ꢁꢅꢁaꢇ ꢋrꢅꢊꢂꢄꢁ yꢆꢃꢇꢊ ꢅf ꢃaꢄh arꢅmaꢁꢆꢄ, aꢀꢊ ꢁhꢃ ꢁraꢄꢃ ꢋrꢅꢊꢂꢄꢁs (yꢆꢃꢇꢊ < 5%) arꢃ ꢀꢅꢁ ꢄꢅꢂꢀꢁꢃꢊ.
exꢋꢃrꢆmꢃꢀꢁs wꢃrꢃ ꢋꢃrfꢅrmꢃꢊ ꢆꢀ ꢁrꢆꢋꢇꢆꢄaꢁꢃ, aꢀꢊ ꢁhꢃ s.ꢊ. ꢆs shꢅwꢀ. thꢃ ꢋrꢃꢀyꢇaꢁꢃꢊ ꢋrꢅꢊꢂꢄꢁs markꢃꢊ wꢆꢁh a sꢆꢀgꢇꢃ asꢁꢃrꢆsk wꢃrꢃ ꢆsꢅꢇaꢁꢃꢊ aꢀꢊ fꢂrꢁhꢃr
ꢄharaꢄꢁꢃrꢆzꢃꢊ ꢈy hꢆgh-rꢃsꢅꢇꢂꢁꢆꢅꢀ MS aꢀꢊ nMR aꢀaꢇysꢃs. (b) iꢊꢃꢀꢁꢆfꢆꢃꢊ ꢋrꢅꢊꢂꢄꢁs frꢅm ꢃꢀzymaꢁꢆꢄ rꢃaꢄꢁꢆꢅꢀs ꢈy Aꢁapt wꢆꢁh (+)-ꢈꢂꢁyrꢅꢇaꢄꢁꢅꢀꢃ ii (1) aꢀꢊ
gꢃꢀꢆsꢁꢃꢆꢀ (21). *thꢃ ꢋrꢅꢊꢂꢄꢁs frꢅm ꢁhꢃ ꢋrꢃꢀyꢇaꢁꢆꢅꢀs ꢅf 1 ꢈy ꢁhꢃ mꢂꢁaꢀꢁ G326M. **thꢃ ꢋrꢅꢊꢂꢄꢁ frꢅm ꢁhꢃ ꢋrꢃꢀyꢇaꢁꢆꢅꢀs ꢅf 21 ꢈy ꢁhꢃ mꢂꢁaꢀꢁ W397A.
*
**thꢃ ꢋrꢅꢊꢂꢄꢁ frꢅm ꢁhꢃ ꢋrꢃꢀyꢇaꢁꢆꢅꢀs ꢅf 21 ꢈy ꢁhꢃ mꢂꢁaꢀꢁ e91A.
at various electron-negative, atypical positions of aromatic rings residues at the N terminus, which had been computationally pre-
(
Supplementary Fig. 7). When using 5 as the substrate and GPP as dicted to be disordered, were truncated, yielding a new recombi-
11–424
theprenyldonor, theprenylgroupwasintroducedtotheortho and/or nant protein, AtaPT
, that had the same catalytic activities as
para position relative to an amino moiety to produce 10-geranyl-9- the full-length protein. Hereafter, without explicit statement indi-
11–424
aminocamptothecin (54a), 12-geranyl-9-aminocamptothecin (54b) cating otherwise, AtaPT
is referred to as ‘A taPT ’. The crystal
and 10,12-digeranyl-9-aminocamptothecin (54c). Additionally, structure of AtaPT was solved to 1.90-Å resolution via molecu-
when using 42 and GPP, the prenylations occurred at both the ortho lar replacement using the structure of FgaPT2 (ref. 23; Protein
position to the carbonyl group (C3) and the ortho position to the Data Bank (PDB) code 3I4Z) as the initial model (Supplementary
phenolic hydroxyl moiety (C6), yielding the diprenylated product Table 5). All of the residues, excepting those from K309 to A320
3
,6-digeranyl-7-hydroxycoumarin (59b). To the best of our knowl- (inclusive), were clearly traced and assigned according to clear
edge, this is the first report on prenylation at the para position to electron density information.
an amino moiety and the ortho position to a carbonyl group by an
As a member of the DMATS superfamily, AtaPT shares a com-
aPTase. These highly diverse catalytic outcomes are representative mon aPTase fold comprising five consecutive αββα units with a
of the multitudinous enzymatic promiscuity of AtaPT.
catalytic chamber at the center (Fig. 3a). Comparing the structure
of AtaPT with the previously reported structures of DMATS-type
29
Structural characterization of AtaPT
aPTases, AnaPT (PDB code 4LD7), FgaPT2, FtmPT1 (ref. 30; PDB
18
To explore the molecular mechanism (or mechanisms) underly- code 3O24) and CdpNPT (PDB code 4E0T), the corresponding
ing the promiscuity of AtaPT, we solved its 3D molecular structure structural r.m.s. deviation of C atoms are 1.08 Å, 1.63 Å, 1.57 Å
α
using crystallography. To obtain high-quality crystals, the first ten and 1.72 Å, respectively.
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.ꢀaꢁꢂrꢃ.ꢄꢅm/ꢀaꢁꢂrꢃꢄhꢃmꢆꢄaꢇꢈꢆꢅꢇꢅgy
3

NATURE ꢀꢁEꢂIꢀAL ꢃIOLOꢄꢅ dOI: 10.1038/nCHeMBIO.2263
article
a
N terminus
C terminus
c
R264
K95
1
80°
Q342
K266
K188
N411
R104
N328
Y268
Y190
L252
Y413
W397
Y344
b
L249
L204
A270
C175
P324
E91
R396
1
80°
S192
L194
L323
F272
Q108
Figure 3 | ꢀrystal structure of AtaPT and its catalytic chamber. (a) thꢃ ꢅꢉꢃraꢇꢇ sꢁrꢂꢄꢁꢂrꢃ ꢅf Aꢁapt ꢆs shꢅwꢀ ꢆꢀ a ꢄarꢁꢅꢅꢀ rꢃꢋrꢃsꢃꢀꢁaꢁꢆꢅꢀ ꢅꢉꢃrꢇaꢆꢊ wꢆꢁh
a ꢁraꢀsꢋarꢃꢀꢁ sꢅꢇꢉꢃꢀꢁ-aꢄꢄꢃssꢆꢈꢇꢃ sꢂrfaꢄꢃ. (b) lꢃfꢁ, ꢁꢅꢋ ꢉꢆꢃw ꢅf ꢁhꢃ ꢃꢇꢃꢄꢁrꢅsꢁaꢁꢆꢄ sꢂrfaꢄꢃ ꢅf Aꢁapt shꢅwꢆꢀg ꢁhꢃ ꢋꢅsꢆꢁꢆꢉꢃꢇy ꢄhargꢃꢊ sꢂrfaꢄꢃ ꢅf ꢁhꢃ ꢄaꢁaꢇyꢁꢆꢄ
ꢄhamꢈꢃr fꢅr ꢋrꢃꢀyꢇ ꢊꢅꢀꢅr ꢈꢆꢀꢊꢆꢀg. bꢇꢂꢃ rꢃꢋrꢃsꢃꢀꢁs ꢋꢅsꢆꢁꢆꢉꢃ ꢄhargꢃs, aꢀꢊ rꢃꢊ rꢃꢋrꢃsꢃꢀꢁs ꢀꢃgaꢁꢆꢉꢃ ꢄhargꢃs. Rꢆghꢁ, ꢈꢅꢁꢁꢅm ꢉꢆꢃw ꢅf ꢁhꢃ ꢃꢇꢃꢄꢁrꢅsꢁaꢁꢆꢄ sꢂrfaꢄꢃ
ꢅf Aꢁapt shꢅwꢆꢀg ꢁhꢃ hyꢊrꢅꢋhꢅꢈꢆꢄ ꢃꢀꢉꢆrꢅꢀmꢃꢀꢁ ꢅf ꢁhꢃ ꢄaꢁaꢇyꢁꢆꢄ ꢄhamꢈꢃr fꢅr arꢅmaꢁꢆꢄ sꢂꢈsꢁraꢁꢃ ꢈꢆꢀꢊꢆꢀg. (c) thꢃ ꢄaꢁaꢇyꢁꢆꢄ ꢄhamꢈꢃr ꢅf Aꢁapt ꢄꢅꢅrꢊꢆꢀaꢁꢃꢊ
wꢆꢁh kꢃy rꢃsꢆꢊꢂꢃs, whꢆꢄh arꢃ ꢊꢆꢉꢆꢊꢃꢊ ꢆꢀꢁꢅ ꢁhrꢃꢃ rꢃgꢆꢅꢀs: ꢁhꢃ ꢋrꢃꢀyꢇ ꢊꢅꢀꢅr ꢈꢆꢀꢊꢆꢀg ꢋꢅꢄkꢃꢁ (ꢁꢅꢋ), ꢁhꢃ ꢁyrꢅsꢆꢀꢃ shꢆꢃꢇꢊ (mꢆꢊꢊꢇꢃ) aꢀꢊ ꢁhꢃ arꢅmaꢁꢆꢄ aꢄꢄꢃꢋꢁꢅr
ꢈꢆꢀꢊꢆꢀg ꢋꢅꢄkꢃꢁ (ꢈꢅꢁꢁꢅm).
Using information from previous structural studies7 , we structure of AtaPT (Supplementary Figs. 8c,d and 9). The prenyl
were able to identify the highly positively charged end of the AtaPT donor binding pocket and the tyrosine shield in all of these aPTases
catalytic chamber for prenyl donor binding, the hydrophobic end are highly conserved; the aromatic acceptor binding pocket is less
of the chamber for aromatic acceptor binding (Fig. 3b) and addi- conserved. Comparison of the aromatic acceptor-binding pocket
tional active sites of the reaction chamber (Fig. 3c). The reaction of AtaPT with those of the other DMATS-type aPTases revealed
chamber can be divided into three regions: the top prenyl donor that AtaPT has a uniquely spacious hydrophobic substrate-binding
binding pocket, which is composed of basic residues (K95, R104, pocket (Supplementary Fig. 10a–e), having both the largest solvent
K188, R264 and K266) with positive charges and three polar resi- accessible surface area and the greatest overall volume among all of
dues (N328, Q342 and N411), interacts with the β-phosphate of the proteins examined (Supplementary Table 6).
,23,30
2
3,30
the prenyl donor by negating the negative charges ; the middle
The aromatic acceptor-binding pocket of AtaPT is surrounded
tyrosine shield, which is composed of four tyrosines (Y190, Y268, mainly by hydrophobic residues with relatively short side chains
Y344 and Y413), interacts with the α-phosphate of the prenyl compared with the other DMATS-type aPTases. For instance, the
donor and is responsible for stabilizing the nascent carbocation residues forming the pocket of AnaPT include F103, Q192, F212
intermediate during catalysis ; and the bottom aromatic accep- and F265 (Supplementary Fig. 10b), whereas the corresponding
tor binding pocket, which is mainly composed of hydrophobic residues at the equivalent positions of AtaPT are L83, N172, S192
residues, including L252, L249, L323, F272, W397, A270, L204 and and G251 (Supplementary Fig. 10a). In another example, the resi-
23,30
L194 (Fig. 3c).
dues forming the pocket of FgaPT2 include T102, Y191 and M328
We also noted that there are two AtaPT molecules in the asym- and the positively charged resides K174 and R244 (Supplementary
metric unit of the crystal, and the area of the buried interface Fig. 10c), whereas the corresponding residues at the equivalent posi-
2
between them is ~4,500 Å . The overall solvent accessible surface tions of AtaPT are all small and hydrophobic (G106, S192, G326,
2
area of one monomer is ~15,400 Å (Supplementary Fig. 8a), C175 and G251) (Supplementary Fig. 10a). Similar observations
suggesting a strong interaction between two monomers. Further were also noted when comparing the structures of FtmPT1 and
31
quantitative analysis using PISA suggested a tetramer form of CdpNPT with that of AtaPT (Supplementary Fig. 10d,e). We also
AtaPT (Supplementary Fig. 8b), a supposition consistent with examined the electron density map of the high-resolution structure
our previous data, which were collected using size-exclusion chro- of AtaPT and observed only two water molecules inside its aromatic
matography with multi-angle light scattering (SEC–MALS) in the acceptor-binding pocket, reflecting its highly hydrophobic nature.
22
aqueous phase . The interfaces of AtaPT within the tetramer are
The spacious substrate-binding pocket of AtaPT provides addi-
mainly composed of charged residues (Supplementary Fig. 8a). tional flexibility to accommodate various donors and acceptors,
Structural analysis suggested that the oligomerization of AtaPT and the hydrophobic nature of the pocket can probably increase the
does not restrict the substrate access to the binding pocket stability of prenyl carbocation intermediates, thereby increasing the
(
Supplementary Fig. 8b) and might not affect the enzymatic possibility of prenylation at multiple positions.
activity of AtaPT. However, a potential allosteric and synergetic
effect might exist within the tetramer.
Donor-dependent specificity with conformational changes
To further reveal the molecular mechanism underlying the pro-
miscuity of AtaPT, we determined the crystal structures of AtaPT
The spacious hydrophobic substrate-binding pocket
To explore the structural basis of the enzymatic promiscuity of bound with various donors and/or acceptors. To obtain a stable
AtaPT, we mapped the sequence conservation of other DMATS-type complex with substrates, we used the nonhydrolyzable prenyl donors
aPTases, including AnaPT, FgaPT2, FtmPT1 and CdpNPT, onto the DMSPP (dimethylallyl S-thiolodiphosphate) and GSPP (geranyl
4
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.ꢀaꢁꢂrꢃ.ꢄꢅm/ꢀaꢁꢂrꢃꢄhꢃmꢆꢄaꢇꢈꢆꢅꢇꢅgy

NATURE ꢀꢁEꢂIꢀAL ꢃIOLOꢄꢅ dOI: 10.1038/nCHeMBIO.2263
article
S-thiolodiphosphate) instead of DMAPP and GPP. Eventually, we
obtained the structures of binary complexes bound with DMSPP
and GSPP as well as the structures of ternary complexes bound with
GSPP–1 (DBu), DMSPP–2 (LBu) and DMSPP–21 (Gen) (Fig. 1;
Supplementary Fig. 11 and Supplementary Table 5).
a
R396
W397
During the cocrystallization experiments, we found that, with-
out donors, we were not able to obtain binary complexes between
AtaPT and acceptors, suggesting that there is a sequential substrate
binding mechanism for AtaPT. We selected GSPP–DBu as the
substrate to explore in detail in our attempt to characterize the
donor-dependent acceptor binding of AtaPT. In the complex struc-
ture of AtaPT–GSPP–DBu, both GSPP and DBu are clearly resolved
in our 2.3-Å electron density map (Fig. 4a). DBu is specifically
coordinated via hydrogen bonds with E91, H88, R396 and W397.
We noted a strong hydrophobic interaction between the prenyl
tail of GSPP and DBu, suggesting the contribution of GSPP in the
binding of DBu with AtaPT.
E91
GSPP
H88
DBu
To further understand the role of GSPP in the interaction
between DBu and AtaPT, we sought to measure the binding affin-
ity of DBu to AtaPT by using an isothermal titration calorimetry
b
(
ITC) approach. Notably, we found that without GSPP, the titration
of DBu into the AtaPT protein solution did not produce substantial
heat, suggesting that there is no interaction between DBu and AtaPT
(Supplementary Fig. 12a). However, in the presence of GSPP, sig-
nificant heat was produced during the titration, indicating a strong
interaction between DBu and AtaPT, with a measured binding
affinity of 0.96 ± 0.05 μM (Supplementary Fig. 12b,c). Thus, both
the crystallographic results and the ITC results indicated a sequen-
tial manner for the binding of AtaPT to the donor GSPP and the
acceptor DBu. Furthermore, in the presence of DMAPP, AtaPT
hardly catalyzed the prenylation of DBu, (Supplementary Fig. 3a);
however, with GPP and FPP, the catalytic efficiency was increased
substantially (Fig. 2a and Supplementary Fig. 3b). The K of DBu
M
bound to AtaPT in the presence of DMAPP was determined to be
3
13 μM, which is significantly greater than the value determined
when GPP was the prenyl donor (89 μM; Supplementary Table 3).
Together, the crystallographic, ITC and enzymatic assay results sup-
port the conclusion that the acceptance of DBu as a substrate by
AtaPT is dependent on the structure of the prenyl donor.
We superimposed the structure of AtaPT–GSPP–DBu with the
apo structure and found that, following binding with DBu, AtaPT
underwent a marked conformational change at the loop 1 region
located within the first repetitive αββα motif (Fig. 4b). Residue
H88 shifted 6.1 Å to interact with DBu, forming a hydrogen
bond. Such substantial conformational changes have not been
previously reported.
R396
W397
Loop 1
E91
H88
Loop 1’
H88’
GSPP
DBu
Various conformations and multiple acceptor-binding sites
In addition to observing donor-dependent acceptor binding, we
further compared the apo structure with the complex structures at
the substrate-binding pocket. First, we observed diverse conforma-
tional fluctuations of the tyrosine shield region, especially residues
Y190 and Y268, at different substrate binding states (Fig. 5). We
observed two conformations each for Y190 and Y268; we defined
one such state as the ‘open state’ and the other as the ‘closed state’.
In the apo structure, both Y190 and Y268 were in the closed state
Figure 4 | Donor-dependent acceptor binding by AtaPT. (a) Sꢁrꢂꢄꢁꢂrꢃ ꢅf
hꢃ ꢁꢃrꢀary ꢄꢅmꢋꢇꢃx ꢅf Aꢁapt–GSpp–dbꢂ. thꢃ ꢋrꢃꢀyꢇ ꢊꢅꢀꢅr GSpp aꢀꢊ ꢁhꢃ
arꢅmaꢁꢆꢄ aꢄꢄꢃꢋꢁꢅr dbꢂ (1) arꢃ shꢅwꢀ ꢆꢀ sꢁꢆꢄk fꢅrm aꢀꢊ arꢃ ꢅꢉꢃrꢇaꢆꢊ wꢆꢁh a
F –F ꢃꢇꢃꢄꢁrꢅꢀ ꢊꢃꢀsꢆꢁy maꢋ ꢄꢅꢀꢁꢅꢂrꢃꢊ aꢁ 1.5 σ. thꢃ rꢃsꢆꢊꢂꢃs ꢆꢀꢉꢅꢇꢉꢃꢊ ꢆꢀ ꢁhꢃ
ꢁ
2
ꢅ
ꢄ
ꢄꢅꢅrꢊꢆꢀaꢁꢆꢅꢀ ꢅf dbꢂ arꢃ aꢇsꢅ shꢅwꢀ ꢆꢀ sꢁꢆꢄk fꢅrm aꢀꢊ ꢇaꢈꢃꢇꢃꢊ aꢄꢄꢅrꢊꢆꢀgꢇy.
(b) Sꢂꢋꢃrꢋꢅsꢆꢁꢆꢅꢀ ꢅf ꢁhꢃ ꢄꢅmꢋꢇꢃx sꢁrꢂꢄꢁꢂrꢃ ꢅf Aꢁapt–GSpp–dbꢂ (grꢃꢃꢀ)
ꢅꢀꢁꢅ ꢁhꢃ aꢋꢅ sꢁrꢂꢄꢁꢂrꢃ ꢅf Aꢁapt (ꢄyaꢀ). thꢃ ꢊyꢀamꢆꢄ ꢇꢅꢅꢋ 1 ꢆs ꢆꢀꢊꢆꢄaꢁꢃꢊ
wꢆꢁh ꢁhꢃ ꢇargꢃ ꢊꢆsꢋꢇaꢄꢃmꢃꢀꢁ ꢅf H88.
(
Fig. 5a). After binding with GSPP, the states of these two tyrosines
did not change (Fig. 5b); however, following further binding with
DBu, both Y190 and Y268 were in the open state (Fig. 5c). These
We next performed molecular dynamics simulations
conformational changes for Y190 and Y268 from the closed to the (Supplementary Fig. 13a) to investigate whether the multiple
open state allow enough additional space to accommodate the bind- conformations of the tyrosine shield were induced by substrate
ing of DBu. When using DMSPP as the prenyl donor, however, we binding or by spontaneous structural fluctuations of AtaPT
observed that Y190 and Y268 each showed both the closed and the itself. We observed that two regions, G68–H88 and K309–P324,
open states in the same crystal with equal occupancies (Fig. 5d). showed substantial structural fluctuations during the simulations
This type of conformational flexibility was reduced during binding (Supplementary Fig. 13b). This observation was consistent with
with the acceptors LBu (Fig. 5e) and Gen (Fig. 5f).
the facts that the electron density from K309 to A320 could not be
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.ꢀaꢁꢂrꢃ.ꢄꢅm/ꢀaꢁꢂrꢃꢄhꢃmꢆꢄaꢇꢈꢆꢅꢇꢅgy
5

NATURE ꢀꢁEꢂIꢀAL ꢃIOLOꢄꢅ dOI: 10.1038/nCHeMBIO.2263
article
a
b
c
d
e
f
Y413
Y190
Y413
Y190
(CS)
Y413
Y413
Y190
CS)
Y190
CS)
(CS/OS)
(
Y190
(OS)
(
Y190
Y413
Y268
(CS)
(
CS/OS)
Y268
(CS)
Y413
Y268
CS)
Y268
CS)
Y268
OS)
(
(
Y268
(
(
CS/OS)
AtaPT11–424–Apo
AtaPT11–424 –GSPP
AtaPT11–424 –GSPP–DBu
AtaPT11–424 –DMSPP
AtaPT11–424 –DMSPP–LBu
AtaPT11–424 –DMSPP–Gen
Figure 5 | ꢀonformational dynamics of the tyrosine shield and various acceptor binding sites of AtaPT. (a–f) thꢃ sꢂꢈsꢁraꢁꢃ-ꢈꢆꢀꢊꢆꢀg ꢋꢅꢄkꢃꢁs, ꢁyrꢅsꢆꢀꢃ shꢆꢃꢇꢊs,
ꢋrꢃꢀyꢇ ꢊꢅꢀꢅrs aꢀꢊ arꢅmaꢁꢆꢄ aꢄꢄꢃꢋꢁꢅrs frꢅm ꢁhꢃ sꢁrꢂꢄꢁꢂrꢃs ꢅf aꢋꢅ Aꢁapt (a), Aꢁapt–GSpp (b), Aꢁapt–GSpp–dbꢂ (c), Aꢁapt–dMSpp (d), Aꢁapt–dMSpp–lbꢂ
(
e) aꢀꢊ Aꢁapt–dMSpp–Gꢃꢀ (f), shꢅwꢀ ꢆꢀ sꢂrfaꢄꢃs aꢀꢊ sꢁꢆꢄks. tyr190 aꢀꢊ tyr268 shꢅw ꢁwꢅ ꢄꢅꢀfꢅrmaꢁꢆꢅꢀs wꢆꢁh ꢁhꢃ ꢄꢇꢅsꢃꢊ sꢁaꢁꢃ (cS) aꢀꢊ ꢁhꢃ aꢇꢁꢃrꢀaꢁꢆꢉꢃ ꢅꢋꢃꢀ
37
sꢁaꢁꢃ (oS). thꢃ sꢂꢈsꢁraꢁꢃ-ꢈꢆꢀꢊꢆꢀg ꢋꢅꢄkꢃꢁs wꢃrꢃ ꢄꢅmꢋꢂꢁꢃꢊ ꢂsꢆꢀg HolloW aꢀꢊ gꢃꢀꢃraꢁꢃꢊ ꢆꢀ pyMol (hꢁꢁꢋ://www.ꢋymꢅꢇ.ꢅrg).
solved owing to its high mobility and that, in the crystal structure of many short hydrophobic residues, and this unique pocket is
of AtaPT–GSPP–DBu, loop 1 (from R71 to G87) became flexible most likely responsible for AtaPT’s flexibility in using multiple
and could not be modeled owing to a lack of electron density. We prenyl donors. To study the key sites controlling donor selectivity,
further investigated the fluctuations of the four tyrosine residues we compared the complex structures of AtaPT–GSPP with other
of the tyrosine shield and found that Y190, Y268 and Y344 exhib- aPTase structures and found that residue G326 in AtaPT contributed
ited substantial flexibility (Supplementary Fig. 13c) and had two to the enlarged substrate binding pocket (Supplementary Fig. 15a),
stable positions, whereas Y413 was less flexible and was stabilized whereas the equivalent residue in FgaPT2 (M328), AnaPT (V340),
in one position (Supplementary Fig. 13d). This observation was in FtmPT1 (M364) and CdpNPT (M349) protruded into the pocket
accordance with the above findings, although the conformational and narrowed the cavity, thereby preventing the tolerance for an
fluctuation of Y344 was not observed from the current crystallo- extended carbon tail of a prenyl donor compound (Supplementary
graphic data. We further performed simulations with the complex Figs. 9 and 15b). It is possible that the presence of residues
structures AtaPT–DMSPP and AtaPT–GSPP (Supplementary with shorter hydrophobic side chains in the pocket of AtaPT elimi-
Fig. 13e–h). Comparing these with the simulations of the apo state, nates some or all of the steric barriers that are present in homologs
we observed a reduced degree of conformational fluctuation follow- from other species. Doing so may prevent the binding of longer
ing substrate binding.
prenyl donors, resulting in AtaPT’s broad tolerance for a diversity
We next compared the binding sites of acceptors from the struc- of prenyl donors.
tures of AtaPT–GSPP–DBu, AtaPT–DMSPP–LBu and AtaPT–
To evaluate these suppositions, we performed mutagenesis with
DMSPP–Gen (Fig. 5c,e,f and Supplementary Fig. 14) and thus the aim of characterizing the contribution of G326 to the donor
discovered that AtaPT has multiple acceptor binding sites. DBu, selectivity of AtaPT. The site-directed AtaPT mutant G326M
LBu and Gen were bound at different sites within the spacious was constructed (Supplementary Table 2), and prenylation assays
hydrophobic pocket. The binding site of DBu consisted of E91, H88, using 1 as the acceptor showed that G326M could not accept GPP
S177, R396 and W397; that of LBu consisted of G82, N172, V174, or FPP as donors. However, the catalytic activity for the G326M
R396 and W397; that of Gen consisted of G82, S170, R396 and Y268 mutant with DMAPP was substantially enhanced compared to
and was surrounded by hydrophobic residues, including I167, E169, that of the wild type (Fig. 6a). Similar results were observed when
L323, P324 and W397.
using 21 as the acceptor; the only detectable activity of the G326M
Notably, we observed that LBu and Gen were at peripheral sites mutant enzyme was toward GPP and FPP (Fig. 6b). The conver-
relatively far from the prenyl donor-binding site (distances of ~6 Å sion efficiency from 1 to 47a catalyzed by G326M using DMAPP as
and ~8 Å, respectively). We speculate that this spacious hydropho- the donor was 32 times higher than that of the wild type, and 47a
bic pocket is able to effectively protect carbocation intermediates became the major product (47b–47e; Figs. 2b, 6a). Notably, besides
of prenyl groups during catalysis, increasing the lifetime of these restricting the size of the donor pocket, the G326M mutation
intermediates and thereby offering a strong opportunity to attack increased the extent of the hydrophobic interface for the binding
the prenyl acceptors localized at the peripheral sites to complete the of 1, which might be another factor contributing to the increased
electrophilic alkylation. It should be noted that we did not observe prenylation efficiency for 47a that was observed for the mutant.
the same conformational changes of loop 1 that occurred in the
structure of AtaPT–GSPP–DBu when we examined the crystal Alteration of the prenylation promiscuity via mutagenesis
structures of AtaPT–DMSPP–LBu or AtaPT–DMSPP–Gen.
Using the crystal structure of AtaPT–GSPP–DBu, we performed
The above crystallographic observations and molecular simu- further mutagenesis experiments to study which residues serve as
lation studies indicate that AtaPT exhibits multiple states with the determinants of prenylation promiscuity. We selected H88, E91,
substantial fluctuations at both the loop 1 and tyrosine shield S177, R396 and W397, all of which are involved in the interaction
region and has multiple sites for acceptor binding. These properties with DBu (Fig. 6c and Supplementary Fig. 14a), as the targets for
suggest likely mechanisms that may explain the enzymatic promis- site-directed mutagenesis. We then assayed the catalytic activities
cuity of AtaPT.
of the mutants using DBu as the aromatic acceptor and GPP as the
prenyl donor.
ꢂutagenesis-based alteration of the donor selectivity
The R396A mutant resulted in an inclusion body during expres-
The aforementioned structural comparisons revealed that the sion; thus, we did not investigate this mutant further. The S177A
spacious hydrophobic substrate binding pocket of AtaPT consists mutant showed no obvious differences in enzymatic activity or
6
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.ꢀaꢁꢂrꢃ.ꢄꢅm/ꢀaꢁꢂrꢃꢄhꢃmꢆꢄaꢇꢈꢆꢅꢇꢅgy

NATURE ꢀꢁEꢂIꢀAL ꢃIOLOꢄꢅ dOI: 10.1038/nCHeMBIO.2263
article
a
DMAPP
GPP
FPP
49
c
e
Fig. 16). Markedly, the orientation of bound DBu in the mutant
structure was rotated 180 degrees, with the C4″ position (Fig. 6e)
inserted into the pocket. In the wild-type structure, the C4′ position
(Fig. 6c) was inserted into the pocket.
5
4
3
2
1
00
00
00
00
00
0
47e 160
R104
S93
4
48e
8f
4
4
7d
7c
S177
DBu
1
20
GSPP
L105
S177
GSPP
4
4
8d
8c
E91
DBu
47b
A91
8
0
0
0
R396
H88
47a
48b
8a
R396
L83
We also measured the prenylation activities of wild-type AtaPT
and the H88A, E91A, E91Q, E91D, S177A and W397A variants
using 21 and 22 as the acceptors and GPP as the prenyl donor.
Notably, we found that, similar to reactions with 1 as the acceptor,
the E91A and E91Q mutants exhibited significant differences in
product diversity when using 21 and 22 as acceptors; both mutant
enzymes primarily yielded monoprenylated products (51b and
57a) and showed increased conversion efficiency (Supplementary
Fig. 17). These seemingly spurious properties of the E91A and
E91Q mutants, as well as the molecular mechanism (or mecha-
nisms) responsible for the multiple prenylation capacity of AtaPT,
will require further investigation in the future.
4
4
W397
W397
1
1–424
AtaPT
–GSPP–DBu
E91A–GSPP–DBu
b 160 DMAPP
GPP
FPP
d 160
5
5
5
0c
0b
0a
51d
51c
51b
52d
52c
52b
52a
48f
1
20
48e
48d
48c
4
48a
1
20
5
1a
8
0
0
0
80
40
0
8b
4
With an appropriately deep understanding of the molecular
mechanisms of AtaPT–catalyzed prenylations, it should be pos-
sible to use structure-guided mutagenesis to fine tune both the
prenyl donor and the prenylation selectivity of AtaPT. This would
increase the utility of AtaPT in the synthesis of new, possibly
rationally designed prenylated derivatives for applications in drug
discovery programs.
Figure 6 | Structure-guided mutagenesis of AtaPT to alter its substrate
promiscuity. (a,b) eꢀzymaꢁꢆꢄ assays ꢅf wꢆꢇꢊ-ꢁyꢋꢃ (Wt) Aꢁapt aꢀꢊ ꢁhꢃ
mꢂꢁaꢀꢁ G326M ꢂsꢆꢀg (a) (+)-ꢈꢂꢁyrꢅꢇaꢄꢁꢅꢀꢃ ii (dbꢂ, 1) aꢀꢊ (b) gꢃꢀꢆsꢁꢃꢆꢀ
(Gꢃꢀ, 21) as ꢁhꢃ arꢅmaꢁꢆꢄ aꢄꢄꢃꢋꢁꢅrs aꢀꢊ ꢂsꢆꢀg dMApp, Gpp aꢀꢊ Fpp
as ꢁhꢃ ꢋrꢃꢀyꢇ ꢊꢅꢀꢅrs. (c) Sꢂꢈsꢁraꢁꢃ-ꢈꢆꢀꢊꢆꢀg ꢋꢅꢄkꢃꢁ ꢅf ꢁhꢃ ꢁꢃrꢀary
ꢄꢅmꢋꢇꢃx Aꢁapt–GSpp–dbꢂ shꢅwꢆꢀg ꢁhꢃ kꢃy rꢃsꢆꢊꢂꢃs ꢄꢅꢅrꢊꢆꢀaꢁꢆꢀg dbꢂ.
(d) eꢀzymaꢁꢆꢄ assays ꢅf wꢆꢇꢊ-ꢁyꢋꢃ Aꢁapt aꢀꢊ ꢁhꢃ mꢂꢁaꢀꢁs H88A, e91A,
e91Q, e91d, S177A aꢀꢊ W397A ꢂsꢆꢀg 1 as ꢁhꢃ arꢅmaꢁꢆꢄ aꢄꢄꢃꢋꢁꢅr aꢀꢊ
Gpp as ꢁhꢃ ꢋrꢃꢀyꢇ ꢊꢅꢀꢅr. (e) thꢃ sꢂꢈsꢁraꢁꢃ-ꢈꢆꢀꢊꢆꢀg ꢋꢅꢄkꢃꢁ ꢅf ꢁhꢃ ꢁꢃrꢀary
ꢄꢅmꢋꢇꢃx e91A–GSpp–dbꢂ shꢅwꢆꢀg ꢁhꢃ kꢃy rꢃsꢆꢊꢂꢃs ꢄꢅꢅrꢊꢆꢀaꢁꢆꢀg dbꢂ.
Fꢅr assays ꢂsꢆꢀg 1 as ꢁhꢃ aꢄꢄꢃꢋꢁꢅr, ꢁhꢃ ꢁꢅꢁaꢇ ꢋrꢅꢊꢂꢄꢁ yꢆꢃꢇꢊ ꢅf Wt wꢆꢁh
dMApp (4.9 ± 0.1%), Gpp (61 ± 2%) aꢀꢊ Fpp (100 ± 0%) was ꢊꢃfꢆꢀꢃꢊ
as 100% rꢃꢇaꢁꢆꢉꢃ aꢄꢁꢆꢉꢆꢁy. Fꢅr assays ꢂsꢆꢀg 21 as ꢁhꢃ aꢄꢄꢃꢋꢁꢅr, ꢁhꢃ ꢁꢅꢁaꢇ
ꢋrꢅꢊꢂꢄꢁ yꢆꢃꢇꢊ ꢅf Wt wꢆꢁh dMApp (50 ± 2%), Gpp (95 ± 3%) aꢀꢊ
Fpp (99 ± 1%) was ꢊꢃfꢆꢀꢃꢊ as 100% rꢃꢇaꢁꢆꢉꢃ aꢄꢁꢆꢉꢆꢁy. Aꢇꢇ ꢅf ꢁhꢃ ꢃꢀzymaꢁꢆꢄ
ꢃxꢋꢃrꢆmꢃꢀꢁs wꢃrꢃ ꢋꢃrfꢅrmꢃꢊ ꢆꢀ ꢁrꢆꢋꢇꢆꢄaꢁꢃ, wꢆꢁh ꢁhꢃ s.ꢊ. ꢀꢅꢁꢃꢊ.
DISꢀUSSION
The molecular mechanisms of enzyme promiscuity have been
32
discussed in different subject areas . However, structural results
are fairly scarce, especially for complex structures with a single
enzyme bound to multiple or numerous substrates. This lack of data
has, to date, limited the characterization of mechanisms relating
to enzyme promiscuity. In the present study, we characterized
multiple crystal structures of AtaPT in complex with different
acceptors and/or donors, providing a solid structural basis to explore
the mechanistic basis of the extreme promiscuity of AtaPT. Thus,
product diversity compared to the wild type. The H88A mutant our work adds important insights and empirical data that promise
showed reduced enzymatic activity but produced a similar diversity to help elucidate the molecular basis of enzyme promiscuity.
of product compounds. The W397A mutant showed both reduced
The molecular mechanism of the multiple-site prenylations
enzymatic activity and altered product diversity (Fig. 6d). Notably, on the aromatic rings that led to AtaPT’s product diversity has not
the conversion efficiencies from 1 to 48c catalyzed by both the been fully elucidated in the present study. Rather, we speculate
W397A mutant and the wild type were similar (Fig. 6d). It is worth that the spacious hydrophobic pocket allows AtaPT to accommo-
noting that the position of the prenylation of 1 for 48c is the C4′ date various aromatics and even some prenylated products. Each
position (Fig. 2b), which is the closest position relative to the pyro- aromatic substrate may bind AtaPT at different sites with varied
phosphate end of the prenyl donor GSPP in the complex structure orientations, although the crystallographic approach can only
of AtaPT–GSPP–DBu (Fig. 6c).
capture the most stable state of a substrate-bound complex. This
We found that the most interesting position of AtaPT in terms multiple binding scenario would allow the prenylation to occur
of its acceptor promiscuity was E91. In the complex structure at different positions. Furthermore, the prenylated products can
of AtaPT–GSPP–DBu, E91 formed a hydrogen bond with DBu, also bind the spacious hydrophobic pocket again, facilitating a
and thus seems to have a role in coordinating the binding of DBu subsequent prenylation.
(Fig. 6c and Supplementary Fig. 14a). We were therefore able
For the alteration of prenylation patterns between the E91A
to predict that a mutation from E91 to D91 would not change the and E91Q mutants, we offer the following speculations. Wild-type
catalytic activity. Results from our enzyme assays largely confirmed AtaPT could accept 1 in two orientations (one for the insertion of
this prediction, although it must be noted that the distribution C4′, the other for the insertion of C4″), and 1, in both orientations,
of the various products was slightly different between the E91D could be prenylated with different binding affinities and catalytic
mutant and the wild-type enzyme (Fig. 6d).
efficiencies. The prenylated 1 could bind AtaPT again with an alter-
However, to our surprise, the E91A and E91Q mutants did not native orientation for the second prenylation. The crystal structure
lose their prenylation catalytic activities at all but rather exhibited of AtaPT–GSPP–DBu (Fig. 4) that we solved might represent one
substantial alterations in the diversity of their products (Fig. 6d). orientation of 1 with a relatively high binding affinity. Whereas the
Only monoprenylated products (48b and 48c) could be detected for mutation of alanine or glutamine at E91 eliminates the binding of
these two mutants, and they had high conversion rates. Such nota- 1 in this orientation, in the crystal structure of E91A–GSPP–DBu,
ble changes persuaded us to solve the crystal structure of the mutant the second orientation of 1, which had a higher binding affinity,
E91A with DBu and GSPP (Fig. 6e). As we had hypothesized, could be captured (Fig. 6e). However, 1 bound in the orientation
the complex structure of E91A–GSPP–DBu exhibited substantial amenable to the insertion of C4″ could not be prenylated owing to
differences compared to that of AtaPT–GSPP–DBu. There was no steric hindrance. Thus, the products 48b and 48c with the prenyla-
conformational change for loop 1 observed in the mutant structure, tion sites at C3′ and C4′ could come from another position with low
but the binding pocket of DBu was altered to comprise residues L83, binding affinity but high catalytic efficiency. As a result, even if the
S93, K104, L105, S177, R396 and W397 (Fig. 6e and Supplementary prenylated 1 could reenter the acceptor-binding pocket, the second
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.ꢀaꢁꢂrꢃ.ꢄꢅm/ꢀaꢁꢂrꢃꢄhꢃmꢆꢄaꢇꢈꢆꢅꢇꢅgy
7

NATURE ꢀꢁEꢂIꢀAL ꢃIOLOꢄꢅ dOI: 10.1038/nCHeMBIO.2263
article
prenylation could not be catalyzed. This scenario explains how the 13. Cheng, W. & Li, W. Structural insights into ubiquinone biosynthesis in
membranes. Science 343, 878–881 (2014).
prenylation profiles of the mutants could be altered substantially.
1
4. ꢁumano, T., ꢂichard, S.B., Noel, J.P., Nishiyama, M. & ꢁuzuyama, T.
Chemoenzymatic syntheses of prenylated aromatic small molecules using
Streptomyces prenyltransferases with relaxed substrate specificities.
Bioorg. Med. Chem. 16, 8117–8126 (2008).
Traditionally, enzymes have been held to have strict specific-
32
ity for their substrates and for the reactions that they catalyze .
However, it is increasingly accepted that enzyme promiscuity is not
a rare phenomenon; rather, it seems to be widespread and may even 15. Zhou, ꢁ., Yu, X., Xie, X. & Li, S.M. Complementary flavonoid prenylations
by fungal indole prenyltransferases. J. Nat. Prod. 78, 2229–2235 (2015).
be an inherent feature in the biology of, for example, glycosyltrans-
1
6. Yin, S., Yu, X., Wang, Q., Liu, X.Q. & Li, S.M. Identification of a
brevianamide F reverse prenyltransferase BrePT from Aspergillus versicolor
with a broad substrate specificity towards tryptophan-containing cyclic
dipeptides. Appl. Microbiol. Biotechnol. 97, 1649–1660 (2013).
33
34
35
ferases , transaminases and cytochrome P450s . The promiscuous
enzymes are highly valuable in many synthetic applications, ranging
36
from laboratory to industry , especially in reactions that are diffi-
cult to access by chemical methods such as C–C bond formation and 17. ꢁremer, A., Westrich, L. & Li, S.M. A 7-dimethylallyltryptophan synthase
from Aspergillus fumigatus: overproduction, purification and biochemical
ring system construction. Therefore, there is a large drive to discover
characterization. Microbiology 153, 3409–3416 (2007).
or develop ‘universal’ enzymes that can be used for general chemi-
1
8. Schuller, J.M. et al. Structure and catalytic mechanism of a cyclic dipeptide
prenyltransferase with broad substrate promiscuity. J. Mol. Biol. 422,
87–99 (2012).
cal syntheses. In the present study, we showed that AtaPT not only
accepts various types of aromatic acceptors with complex structures
and various prenyl donors with different chain lengths but also con- 19. Pocꢀrandt, D., Sacꢀ, C., ꢁosiol, T. & Li, S.M. A promiscuous prenyltransferase
from Aspergillus oryzae catalyses C-prenylations of hydroxynaphthalenes
ducts diverse mono-, di- or triprenylations with C- or O-prenylated
products. AtaPT is therefore an aPTase with substantial versatility
in the presence of different prenyl donors. Appl. Microbiol. Biotechnol. 98,
4
987–4994 (2014).
for use in the development of applications requiring prenylation.
2
0. Mori, T. et al. Manipulation of prenylation reactions by structure-based
engineering of bacterial indolactam prenyltransferases. Nat. Commun. 7,
10849 (2016).
rꢀꢁꢀꢂvꢀd 17 Mꢃꢄꢁh 2016; ꢃꢁꢁꢀpꢅꢀd 2 Novꢀmbꢀꢄ 2016;
pubꢆꢂshꢀd onꢆꢂnꢀ 19 Dꢀꢁꢀmbꢀꢄ 2016
2
1. Shen, Y. et al. Butyrolactone and cycloheptanetrione from mangrove-associated
fungus Aspergillus terreus. Chem. Pharm. Bull. (Tokyo) 60, 1437–1441 (2012).
2. Gao, B., Chen, ꢂ., Liu, X., Dai, J. & Sun, F. Expression, purification,
crystallization and crystallographic study of the Aspergillus terreus aromatic
prenyltransferase AtaPT. Acta Crystallogr. F Struct. Biol. Commun. 71,
2
ꢂETꢁODS
Methods, including statements of data availability and any associated
accession codes and references, are available in the online version
of the paper.
8
89–894 (2015).
2
3. Metzger, U. et al. e structure of dimethylallyl tryptophan synthase reveals a
common architecture of aromatic prenyltransferases in fungi and bacteria.
Proc. Natl. Acad. Sci. USA 106, 14309–14314 (2009).
Accession codes. The sequence of AtaPT has been submitted 24. Nitta, ꢁ., Fujita, N., Yoshimura, T., Arai, ꢁ. & Yamamoto, Y. Metabolic products
to GenBank with the accession number KP893683. The atomic
coordinates and the structure factor files for the crystal structures
of AtaPT (apo), AtaPT–DMSPP, AtaPT–GSPP, AtaPT–DMSPP–
LBu, AtaPT–DMSPP–Gen, AtaPT–GSPP–DBu and E91A–GSPP–
DBu have been deposited in the Protein Data Bank with the
of Aspergillus terreus. IX. Biosynthesis of butyrolactone derivatives isolated from
strains IFO 8835 and 4100. Chem. Pharm. Bull. 31, 1528–1533 (1983).
5. Guo, C.J. et al. Spatial regulation of a common precursor from two distinct
genes generates metabolite diversity. Chem. Sci. 6, 5913–5921 (2015).
6. Marꢀovits, J. et al. Inhibitory effects of the tyrosine ꢀinase inhibitor genistein
on mammalian DNA topoisomerase II. Cancer Res. 49, 5111–5117 (1989).
2
2
accession numbers 5KCG, 5KCL, 5KCQ, 5KD6, 5KDA, 5KCY 27. Pavese, J.M., ꢁrishna, S.N. & Bergan, ꢂ.C. Genistein inhibits human prostate
cancer cell detachment, invasion, and metastasis. Am. J. Clin. Nutr. 100
and 5KD0, respectively.
(
Suppl. 1), 431S–436S (2014).
2
2
3
8. Carmichael, J., DeGraff, W.G., Gazdar, A.F., Minna, J.D. & Mitchell, J.B.
Evaluation of a tetrazolium-based semiautomated colorimetric assay:
assessment of chemosensitivity testing. Cancer Res. 47, 936–942 (1987).
9. Yu, X. et al. Catalytic mechanism of stereospecific formation of cis-configured
prenylated pyrroloindoline diꢀetopiperazines by indole prenyltransferases.
Chem. Biol. 20, 1492–1501 (2013).
rꢀfꢀꢃꢀꢄcꢀꢅ
1.
Brandt, W. et al. Molecular and structural basis of metabolic diversity
mediated by prenyldiphosphate converting enzymes. Phytochemistry 70,
1
758–1775 (2009).
2
3
.
.
Heide, L. Prenyl transfer to aromatic substrates: genetics and enzymology.
Curr. Opin. Chem. Biol. 13, 171–179 (2009).
0. Jost, M. et al. Structure-function analysis of an enzymatic prenyl transfer
reaction identifies a reaction chamber with modifiable specificity. J. Am.
Chem. Soc. 132, 17849–17858 (2010).
Li, S.M. Prenylated indole derivatives from fungi: structure diversity,
biological activities, biosynthesis and chemoenzymatic synthesis.
Nat. Prod. Rep. 27, 57–78 (2010).
Yazaꢀi, ꢁ., Sasaꢀi, ꢁ. & Tsurumaru, Y. Prenylation of aromatic compounds,
a ꢀey diversification of plant secondary metabolites. Phytochemistry 70,
3
3
3
1. ꢁrissinel, E. & Henricꢀ, ꢁ. Inference of macromolecular assemblies from
crystalline state. J. Mol. Biol. 372, 774–797 (2007).
4
5
6
7
.
.
.
.
2. ꢁhersonsꢀy, O. & Tawfiꢀ, D.S. Enzyme promiscuity: a mechanistic and
evolutionary perspective. Annu. Rev. Biochem. 79, 471–505 (2010).
3. Chen, D. et al. Probing the catalytic promiscuity of a regio- and stereospecific
C-glycosyltransferase from Mangifera indica. Angew. Chem. Int. Ed. Engl. 54,
1
739–1745 (2009).
Botta, B., Vitali, A., Menendez, P., Misiti, D. & Delle Monache, G.
Prenylated flavonoids: pharmacology and biotechnology. Curr. Med. Chem.
1
2, 717–739 (2005).
1
2678–12682 (2015).
Li, S.M. Applications of dimethylallyltryptophan synthases and other
indole prenyltransferases for structural modification of natural products.
Appl. Microbiol. Biotechnol. 84, 631–639 (2009).
3
4. Cuetos, A. et al. Catalytic promiscuity of transaminases: preparation of
enantioenriched β-fluoroamines by formal tandem hydrodefluorination/
deamination. Angew. Chem. Int. Ed. 55, 3144–3147 (2016).
ꢁuzuyama, T., Noel, J.P. & ꢂichard, S.B. Structural basis for the promiscuous
biosynthetic prenylation of aromatic natural products. Nature 435,
3
3
3
5. Eꢀroos, M. & Sjögren, T. Structural basis for ligand promiscuity in
cytochrome P450 3A4. Proc. Natl. Acad. Sci. USA 103, 13682–13687 (2006).
6. Schmid, A. et al. Industrial biocatalysis today and tomorrow. Nature 409,
983–987 (2005).
8.
Saleh, O., Haagen, Y., Seeger, ꢁ. & Heide, L. Prenyl transfer to aromatic substrates
in the biosynthesis of aminocoumarins, meroterpenoids and phenazines: the
ABBA prenyltransferase family. Phytochemistry 70, 1728–1738 (2009).
Chooi, Y.H. et al. Discovery and characterization of a group of fungal
polycyclic polyꢀetide prenyltransferases. J. Am. Chem. Soc. 134,
2
58–268 (2001).
7. Ho, B.ꢁ. & Gruswitz, F. HOLLOW: generating accurate representations of channel
and interior surfaces in molecular structures. BMC Struct. Biol. 8, 49 (2008).
9
.
9428–9437 (2012).
ackꢄowlꢀꢈgmꢀꢄꢆꢅ
1
1
0. Yu, X. & Li, S.M. Prenyltransferases of the dimethylallyltryptophan synthase
superfamily. Methods Enzymol. 516, 259–278 (2012).
We thank S. Huang and J. He for their on-site assistance with the crystallographic
data collection at the SSRF beamline BL17U. This work was supported by grants
from the Strategic Priority Research Program of the Chinese Academy of Sciences
(XDB08030202) to F.S., the National Basic Research Program (973 Program) of Ministry
of Science and Technology of China (2014CB910700) to F.S. and (2014CB910202)
to J.L., the National Natural Science Foundation of China (81302667) to R.C. and
the Deutsche Forschungsgemeinschaft (Li844/4-1) to S.-M.L. The computational
resources were provided by the National Supercomputing Center in Tianjin and the
1. ꢂudolf, J.D., Wang, H. & Poulter, C.D. Multisite prenylation of 4-substituted
tryptophans by dimethylallyltryptophan synthase. J. Am. Chem. Soc. 135,
1895–1902 (2013).
1
2. Ding, Y. et al. Genome-based characterization of two prenylation steps in the
assembly of the stephacidin and notoamide anticancer agents in a marine-
derived Aspergillus sp. J. Am. Chem. Soc. 132, 12733–12740 (2010).
8
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.ꢀaꢁꢂrꢃ.ꢄꢅm/ꢀaꢁꢂrꢃꢄhꢃmꢆꢄaꢇꢈꢆꢅꢇꢅgy

NATURE ꢀꢁEꢂIꢀAL ꢃIOLOꢄꢅ dOI: 10.1038/nCHeMBIO.2263
article
HPC-Service Station at the Center for Biological Imaging of the Institute of Biophysics
Comꢇꢀꢆiꢄg fiꢄꢂꢄciꢂl iꢄꢆꢀꢃꢀꢅꢆꢅ
of the Chinese Academy of Sciences.
The authors declare no competing financial interests.
aꢁꢆhoꢃ coꢄꢆꢃibꢁꢆioꢄꢅ
aꢈꢈiꢆioꢄꢂl iꢄfoꢃmꢂꢆioꢄ
J.D., S.-M.L. and F.S. supervised the project. R.C., X.L., F.R., C.W. and K.F. performed
the enzymatic and chemical experiments. B.G. performed the crystallographic and
ITC experiments. Y.Z. and J.L. performed the molecular dynamics simulations.
R.C., X.L., B.G., J.D. and F.S. analyzed the data and wrote the manuscript.
Any supplementary information, chemical compound information and source data are
available in the online version of the paper. Reprints and permissions information is
available online at http://www.nature.com/reprints/index.html. Correspondence and
requests for materials should be addressed to J.D. or F.S.
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.ꢀaꢁꢂrꢃ.ꢄꢅm/ꢀaꢁꢂrꢃꢄhꢃmꢆꢄaꢇꢈꢆꢅꢇꢅgy
9

ONLINE ꢂETꢁODS
the cells were ruptured with a VCX-130 sonifier (Sonics). Debris and mem-
Chemicals. The prenyl donors dimethylallyl diphosphate (DMAPP), geranyl branes were removed by centrifugation at 12,000g for 45 min. The supernatant
diphosphate (GPP), farnesyl diphosphate (FPP), geranylgeranyl diphosphate
was applied to a Ni Sepharose 6 fast flow resin column (GE Healthcare) and
ent of 20–500 mM imidazole (in 20 mM NaH PO -Na HPO , 500 mM NaCl,
(
GGPP) and phytyl diphosphate (PPP) were chemically synthesized accord- separated according to the manufacturer’s instructions, using a linear gradi-
38
ing to the literature . DMSPP and GSPP were purchased from Echelon
2
4
2
4
Biosciences. (±)-Butyrolactone II was prepared as previously reported2
4,39
.
pH 7.4). After SDS–PAGE analysis, fractions containing pure AtaPT were
Optically active (+)-butyrolactone II (1) and (−)-butyrolactone II (2) were collected and concentrated by ultrafiltration with 30-kDa ultrafiltration
isolated from (±)-butyrolactone II via chiral HPLC using a Chiralpak AD-H tubes (Millipore). To facilitate changing of the buffer, the protein fraction
column (250 mm × 4.6 mm i.d., 5 μm, Daicel); their enantiomeric excess val- was subjected to a PD-10 column (GE Healthcare) and eluted with 50 mM
ues were over 99%. All other tested aromatic compounds were purchased from Tris–HCl, 1 mM DTT and 1.5% (v/v) glycerol, pH 7.4. The purified AtaPT
Sigma-Aldrich, GL Biochem or BioBioPha. All of the tested aromatic substrates was stored at −80 °C before being used in enzymatic assays.
displayed NMR and MS spectroscopic data consistent with those reported
in the literature.
The detailed procedures for protein expression and purification used here
for the crystallization study have been described previously by the authors .
2
2
HPLC-grade methanol and acetonitrile were purchased from Merck; water
was purified and deionized by a water purification system from Tauto Biotech. Assays for prenyltransferase activity and substrate specificity. For the enzyme
The HPLC–MS grade formic acid used in this study was purchased from assays, the reaction mixture routinely contained 0.4 mM aromatic substrate,
Sigma-Aldrich. All of the other chemicals used in this study were obtained 0.8 mM isoprenoid substrate and 45 μM AtaPT in 200 μl of reaction buffer
from standard sources for laboratory use.
(50 mM Tris–HCl, 1 mM DTT, 1.5% (v/v) glycerol, pH 7.4). Assays with heat-
inactivated enzyme that had been boiled at 100 °C for 15 min were used as
negative control reactions. After incubation for 16 h at 37 °C, the reaction was
Chemical analyses. The analyses of substrate specificity and the determination
of conversion rates were performed on an Agilent 1200 series HPLC system terminated by the addition of 400 μl of MeOH and vortexing for 2 min. The
coupled with an LCQ Fleet ion trap mass spectrometer (Thermo Electron) protein was removed by centrifugation at 12,000g for 30 min. The superna-
equipped with an electrospray ionization (ESI) source. The LC chromato- tant was directly analyzed by HPLC-DAD-MS. The conversion rates of the
graph was equipped with a quaternary pump, a diode-array detector (DAD), enzyme reactions were calculated from peak areas of prenylated products and
an auto-sampler, a column compartment and a Shiseido capcell pak C18 MG II substrates as analyzed by HPLC. To facilitate inferential statistical analysis,
column (250 mm × 4.6 mm i.d., 5 μm, Shiseido). The mobile phase consisted three parallel assays were routinely performed; mean ± s.d. from triplicate
of a gradient elution of solvent A (i.e., 0.1% (v/v) formic acid aqueous solution) analyses are reported here.
and solvent B (i.e., methanol or acetonitrile) at a flow rate of 1.0 ml/min at 30 °C.
For HPLC–MS analyses, ultra-high purity helium (He) was used as the
To investigate the substrate specificity of AtaPT, a total of 106 aromatics
were tested as potential prenyl acceptors in assays using prenyl diphosphate
collision gas, and high-purity nitrogen (N ) was used as the nebulizing gas. The compounds of differing chain-lengths, including DMAPP, GPP, FPP, GGPP
2
optimized ESI source parameters were as follows: sheath gas flow rate, 20 (arbi- and PPP. The structures of the 106 aromatics are detailed in Figure 1 and
trary units); auxiliary gas flow rate, 5 (arbitrary units); spray voltage, 5.0 kV; Supplementary Figure 2.
capillary temperature, 350 °C; source collision-induced decomposition (CID),
3
5 V; tube lens offset voltage, −75 V. For full scan MS analysis, mass spectra
Determination of kinetic parameters. For the determination of kinetic param-
were recorded in the 100–1,000 m/z range. The split ratio of effluent from the eters, 35 μM of recombinant AtaPT was incubated with either various concen-
LC to the ion source was 2:1. The data were analyzed using XCalibur software.
trations of prenyl acceptors (20–400 μM) or a fixed concentration of prenyl
donors (DMAPP, GPP or FPP, 800 μM) in a total volume of 200 μl at 37 °C for
30 min. All of these reactions were performed in triplicate, and enzyme activity
was evaluated via quantification of the corresponding prenylated derivatives as
Cloning of AtaPT. Total RNA was isolated from the 3-day-old mycelia of
A. terreus with Trizol reagent (Invitrogen) before being reverse transcribed with
SMARTScribe reverse transcriptase (Clontech). PCR amplification was carried nmol per mg of recombinant protein per minute. Kinetic parameters, includ-
out on an iCycler from Bio-Rad. Using TransStart FastPfu DNA Polymerase ing the Michaelis–Menten constant (K ) and the turnover number (ꢀ ), were
M
cat
(
TransGen Biotech), a PCR fragment of 1,275 bp, containing the entire coding
calculated by nonlinear regression analysis using GraphPad Prism 5 software.
sequence of AtaPT, was amplified from the cDNA template using the forward
primer CDS-F and the reverse primer CDS-R (Supplementary Table 2). The
Preparation and identification of the reaction products. To prepare the
amplification product was digested with NdeI and XhoI, ligated into pET-28a enzymatic products for structural elucidation, each reaction mixture, all of
plasmid and transformed into DH5α E. coli cells. The identity of the resulting which contained 2 mM prenyl acceptor, 4 mM prenyl donor and 45 μM of
pET-28a–AtaPT construct was confirmed by DNA sequencing. The expres- recombinant AtaPT in a total volume of 50 ml reaction buffer, was incubated
sion construct was transformed into E. coli Transetta (DE3) cells (TransGen for 16 h at 37 °C. The enzymatic product was extracted with EtOAc (10 ml × 3).
Biotech). Mutations were introduced into AtaPT following the protocols for
The organic layers of these extracts were combined and concentrated under
the Fast Mutagenesis System (TransGen Biotech). All primers used in this vacuum. The resulting residue was further dissolved in MeOH, and the enzy-
study (gene deletion, truncation and mutation) were synthesized by Generay
Biotech and are detailed in Supplementary Table 2.
matic products were then purified by semi-preparative HPLC. The structures
of the enzymatic products were characterized by MS and NMR spectral and
polarimetric analyses (Supplementary Note). Optical rotations in MeOH
were measured on a PerkinElmer Model-343 digital polarimeter (PerkinElmer).
Protein expression and purification. E. coli Transetta (DE3) cells harboring
the recombinant plasmid were cultured (10 ml) overnight in LB medium con- NMR spectra were recorded on Varian NMR System 600 and 500 spectrom-
taining 50 μg/ml kanamycin and 40 μg/ml chloramphenicol. These cultures eters (Varian). All spectra were processed with MestReNova.6.0.2 (Metrelab),
were then added into 1 L of LB medium (containing 50 μg/ml kanamycin and and chemical shifts were referenced to those of the solvent signals. Chemical
4
2
0 μg/ml chloramphenicol) and grown at 30 °C with the incubator shaking at shifts are given in p.p.m. and coupling constants (J) are given in hertz (Hz). To
00 r.p.m. When the optical density (OD600) of the culture reached 0.6, protein confirm the elemental composition of molecular ions with high mass accuracy,
overexpression was induced by the addition of 0.15 mM IPTG. The cells were
further cultured at 16 °C for 20 h for expression of His-tagged AtaPT.
The cells were pelleted by centrifugation (15 min at 4,000g) and resuspended
in a lysis buffer (20 mM imidazole, 20 mM NaH PO -Na HPO , 0.5 mM
an LC–ESI–IT–TOF–MS analysis was performed on a Shimadzu LC–MS–IT–
TOF instrument equipped with a Shimadzu Prominence HPLC system.
Crystallization, data collection and structure determination. The detailed
2
4
2
4
phenylmethylsulfonyl fluoride, 5 μg/ml DNase and 500 mM NaCl, pH 7.4) procedures for protein crystallization and diffraction data collection have
2
2
11-424
with the ratio of 3 ml lysis buffer per gram of cells (wet weight). Lysozyme was
added to a final concentration of 1 mg/ml. After incubation on ice for 30 min,
been described previously by the authors . In brief, crystals of AtaPT
were obtained using the hanging-drop vapor-diffusion method at 16 °C. The
nature CHeMICaL BIOLOGY
doi:10.1038/nchembio.2263

final conditions for AtaPT11-424 crystal growth were 100 mM Bis–Tris, pH 6.0, conditions, a 12-Å cutoff was used for van der Waals interactions, and Particle
2
3
00 mM ammonium sulfate, 17% (w/v) polyethylene glycol 3350 and
.3% (w/v) DDM. Substrates were soaked into crystals to yield complex struc- After a multistep energy minimization process undertaken to avoid possible
Mesh Ewald summation was used to calculate the electrostatic interactions.
tures. To get crystals of the ternary complex, the crystals were first soaked with clashes, the system was equilibrated in two steps: a 2-ns simulation with the
prenyl donor compounds (DMSPP or GSPP) at a final concentration of 10 mM heavy atoms of the protein constrained and a 5-ns simulation with the pro-
and then soaked with various acceptor compounds at a final concentration
tein backbone constrained. Subsequently, free molecular dynamics simula-
of 16 mM for DBu (1), 8 mM for LBu (2) and 8 mM for Gen (21). Prior to tions were performed without any constraints. Two individual simulations
diffraction experiments, all of the crystals were soaked in a cryoprotectant (named MD1 and MD2) were performed to verify repeatability; both MD1 and
solution (0.1 M Bis–Tris, pH 6.0, 0.2 M ammonium sulfate and 30% (m/v)
PGE3350) and flash frozen in a liquid nitrogen stream.
Diffraction data were collected at 100 K at a wavelength of 0.97907 Å at the
MD2 lasted more than 140 ns. The DMSPP and GSPP bound AtaPT systems
were simulated with a similar strategy; the force field parameters for ligands
were from the CHARMM General Force Field (CGenFF) . During the sim-
47
BL17U beamline of the Shanghai Synchrotron Facility (SSRF) with the ADSC ulations, the temperature was controlled at 300 K by the Langevin method,
Q315 CCD detector. The apo structure of AtaPT was solved by molecular and the pressure was controlled at 1 atm by the Langevin piston method.
replacement using PHASER with the structure of FgaPT2 (PDB code 3I4Z) The SHAKE method was used on all hydrogen-containing bonds to allow a
as the initial model. Binary complex and ternary complex structures were 2-fs time step in the simulations. The simulation trajectories were analyzed
4
0
23
48
solved by molecular replacement using PHASER with the apo structure as a with the VMD program .
starting model; the ligands were built on the basis of difference maps. All of
the structures were refined using Refmac5 (ref. 41). The statistics for the data
collection and refinement and the Ramachandran statistics are summarized
3
8. Woodside, A.B., Huang, Z. & Poulter, C.D. Trisammonium geranyl
diphosphate. [diphosphoric acid, mono(3,7-dimethyl-2,6-octadienyl) ester
in Supplementary Table 5. Structure figures were illustrated using PyMOL
(E)-, trisammonium salt]. Org. Synth. 66, 211 (1988).
4
2
(
http://www.pymol.org) and Chimera . The substrate-binding pockets were
39. Braña, M.F. et al. Synthesis and biological evaluation of analogues of
butyrolactone I and molecular model of its interaction with CDꢁ2. Org.
Biomol. Chem. 2, 1864–1871 (2004).
3
7
computed using HOLLOW and generated in PyMOL.
4
0. McCoy, A.J. et al. Phaser crystallographic sof ware. J. Appl. Crystallogr. 40,
Isothermal titration calorimetry experiment assay. Isothermal titration
calorimetry (ITC) measurements were performed with a MicroCal iTC-200
titration micro-calorimeter (GE Healthcare) at 25 °C. As a control, the sample
cell was filled with AtaPT (30 μM buffered in 20 mM Tris–HCl, pH 8.0, 100 mM
NaCl). The AtaPT concentration was determined via a BCA (bicinchoninic
acid) method. DBu (1, 300 μM) prepared in the same buffer was injected into
the sample cell in 2-min time intervals. In total, 20 injections were performed
within 40 min. DBu (1, 300 μM) was titrated into the AtaPT protein solution
containing 300 μM GSPP. All experiments were repeated three times. The data
were processed using Origin software (Version 7.0).
6
58–674 (2007).
41. Murshudov, G.N. et al. ꢂEFMAC5 for the refinement of macromolecular
crystal structures. Acta Crystallogr. D Biol. Crystallogr. 67, 355–367 (2011).
4
2. Pettersen, E.F. et al. UCSF Chimera--a visualization system for exploratory
research and analysis. J. Comput. Chem. 25, 1605–1612 (2004).
3. Phillips, J.C. et al. Scalable molecular dynamics with NAMD. J. Comput.
Chem. 26, 1781–1802 (2005).
4
44. Macꢁerell, A.D. et al. All-atom empirical potential for molecular modeling
and dynamics studies of proteins. J. Phys. Chem. B 102, 3586–3616 (1998).
4
5. Macꢁerell, A.D. Jr., Feig, M. & Brooꢀs, C.L., III. Improved treatment
of the protein bacꢀbone in empirical force fields. J. Am. Chem. Soc. 126,
6
98–699 (2004).
4
6. Best, ꢂ.B. et al. Optimization of the additive CHAꢂMM all-atom protein
force field targeting improved sampling of the bacꢀbone ϕ, ψ and side-chain
χ(1) and χ(2) dihedral angles. J. Chem. eory Comput. 8, 3257–3273 (2012).
Molecular dynamics simulations. The initial model of AtaPT (R11–V421) was
built on the basis of our crystal structures. The modeled structure was solvated
3
+
−
47. Yu, W., He, X., Vanommeslaeghe, ꢁ. & Macꢁerell, A.D. Jr. Extension of the
CHAꢂMM General Force Field to sulfonyl-containing compounds and its
utility in biomolecular simulations. J. Comput. Chem. 33, 2451–2468 (2012).
in a ~77 × 87 × 84 Å water box, which included 44 Na ions and 36 Cl ions
43
to neutralize the system. The NAMD package and the CHARMM36 all-atom
force field with the CMAP correction
4
4–46
were used for energy minimizations
4
8. Humphrey, W., Dalꢀe, A. & Schulten, ꢁ. VMD: visual molecular dynamics.
and molecular dynamics simulations, respectively. Under periodic boundary
J. Mol. Graph. 14, 33–38, 27–28 (1996).
doi:10.1038/nchembio.2263
nature CHeMICaL BIOLOGY