6
J. Xu et al. / Phytochemistry xxx (2017) 1e8
Fig. 5. Structural basis for product selectivity. a. Polar residues close to the C7 atom of FLPP are shown and colored according to atom types (green: carbon; red: oxygen; blue:
nitrogen; yellow: sulfur). FLPP is also shown and its carbon atom is in cyan, phosphorus is in orange, oxygen is in red and fluorine is in light blue. b. The ribbon of M. spicata
limonene synthase is in orange. The Van der Waals surfaces of N345, L423 and S454 are presented and colored in the same way as the residues (green: carbon; red: oxygen; blue:
nitrogen). The FLPP is colored in the same way as in a. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
stabilizing the initially formed carbocation a polar residue may
prevent its migration towards the pyrophosphate (Wilderman and
Peters, 2007; Xu et al., 2007; Morrone et al., 2008; Zhou and Peters,
2011; Gao et al., 2012). On the other hand, if this residue is replaced
with a none-polar amino acid, this stabilization effect is lost and
carbocation migration occurs, leading to the formation of more
complicated products. For example, in the rice ent-kaurene syn-
thase, by mutating an Ile to Thr one can switch its activity to that of
ent-pimaradiene synthase (Xu et al., 2007).
active pocket than S454A. Although L423 is not a residue directly
residing in the active pocket, as a second-tier residue it contacts
S454. Therefore, its mutation to Ala may lead to an outward
movement of the loop containing resides I453, S454 and G455 and
enlarge the active pocket (Fig. 5b). Changing of product profile by
mutating second-tier residues has been well documented by
Greenhagen and colleagues (Greenhagen et al., 2006). On the other
hand, hydride shifts may be less affected by steric hindrance. This
explains why N345I produces much more phellandrene than
pinene. The behavior of M6 may also be explained this way. Its
products are also composed of 60% sabinene (Table 1). Probably, the
bulky side chain of Y321 causes steric hindrance that prevents 2,7-
ring closure but still allows 6,7-hydride shift and 2,6-ring closure.
Phylogenetic analysis on terpene synthases recognizes seven
major clades, including TPS-a, b, c, d, e/f, g and h (Chen et al., 2011).
The clade c is composed of copalyl synthase/kaurine synthase (CPS/
KS) in P. patens as well as CPS from both gymnosperms and an-
giosperms. This clade is supposed to be the base of the tree. The
clade b contains cyclic monoterpene synthases from angiosperms,
while the clade d comprises gymnospermic mono-, sesqui- and di-
terpene synthases. Therefore, it seems that monoterpene synthases
from angiosperms and gymnosperms evolved independently. After
we constructed a phylogenetic tree of limonene, pinene and phel-
landrene synthases, it is clear that the majority of limonene syn-
thases in angiosperms form a distinct branch within the clade b
(Fig. 4). The member within this branch has a more conserved
“polar pocket”. On the other hand, a few angiospermic limonene
synthases outside this branch (e.g., from L. angustifolia and
R. officinalis) as well as all gymnospermic limonene synthases
belong to many branches intermingling with branches leading to
pinene and phellandrene synthases and have more variations in
“polar pocket” residues (Fig. 4). Therefore, we hypothesize that
during the evolution of angiosperms, this “polar pocket” was “fine-
tuned”, which led to the appearance of the branch containing most
angiospermic limonene synthases from ancestral limonene/
pinene/phellandrene synthases. This scenario is consistent with the
notion that specific activities may evolve from catalytically pro-
miscuous ancestors (Bohlmann et al., 1998; O'Brien and Herschlag,
1999; Copley, 2003; O'Maille et al., 2008).
In M. spicata limonene synthase, it seems that the polar side
chain of N345 plays the same stabilizing role. Therefore, instead of
migrating towards the pyrophosphate, the terpinyl cation is
directly deprotonated to limonene. If N345 is replaced with Ser,
because of its shorter side chain, this effect is weaker. This explains
the decrease in limonene production from more than 90% to less
than 70% in N345S (Fig. 3d; Table 1). Replacing N345 with none-
polar residues, such as Gly, Ala, Leu and Ile, abolishes this effect
and allows the migration of carbocation, which leads to the for-
mation of more complicated products. As mentioned in 2.3, there
are many polar residues in the area that may be collectively
responsible for limonene formation. Here we propose that all these
polar residues may form a “polar pocket” that contributes to the
stabilization of the terpinyl cation. Therefore, mutating any of them
to non-polar residues may disrupt the “polar pocket” and allow the
migration of carbocation, which leads to the decrease of limonene
production and increase of side products. Among these residues,
W324 is also an aromatic residue. Therefore, its effect may also be
caused by the aromaticity of its side chain. Furthermore, this res-
idue might play more roles than stabilizing carbocation, as its
mutation to Ala leads to a significant increase in linalool (Srividya
et al., 2015). For example, W324 might be required for bringing
C1 and C6 in close proximity for cyclization. W324A mutation
might also cause a sufficient increase in the volume of the active
site, which allows the presence of a water molecule necessary to
produce linalool.
The final product profile, however, is influenced by the geom-
etry of the active pocket. For example, steric hindrance may place a
severe limitation on 2,7-ring closure. Therefore, enlarging the
active pocket by N345A, S454A or S454G mutations can create
more steric freedoms and significantly enhance the pinene pro-
duction. This also explains why S454G produces more pinenes than
S454A, because with a smaller side chain S454G possesses a larger
Our data support the previous hypothesis that new terpene
synthases evolved from existing enzymes by changing a very small
number of amino acids (some time a single residue) (Wilderman
Please cite this article in press as: Xu, J., et al., Converting S-limonene synthase to pinene or phellandrene synthases reveals the plasticity of the