H.J. Rozeboom et al. / Biochimica et Biophysica Acta 1844 (2014) 567–575
573
by replacing them with the equivalent CesB residues Val166 and Tyr182
[6]. In contrast, CesB hydrolyzes 1,2-O-isopropylideneglycerol esters to
(S)-IPG with 99.9% enantioselectivity. Thus both esterases are highly se-
lective, but NP is enantioselective towards the acyl/aryl part of sub-
strates, while CesB specifically recognizes the chirality of the alcohol
part, with some additional enantioselectivity towards the acyl/aryl part.
Since NP (and its homolog CesA) have a high enantiospecificity
for S-naproxenethylester (99%) [7], we tried to model the S- and R-
naproxenester substrates in the active site of NP (Fig. 6). For the model-
ing, several interactions were taken into account. For productive
binding, the substrate's ester carbonyl oxygen atom needs to make hy-
drogen bonds to the two NH-groups of the oxyanion hole (Ala64 in NP,
Gly64 in CesA, and Leu131 in both). Furthermore, the Nε2 atom of the
catalytic His274 should be at hydrogen bonding distance to the oxygen
of the substrate alcohol and to the catalytic Ser130 Oγ. In this configu-
ration the substrate S-naproxenethylester binds with the alcohol moie-
ty deep in the active site, and the naphtyl group at the hydrophobic
entrance to the active site. Note, that the entrance to the active site is
only 5.7 Å wide, measured between Cβ of Ala64 and Cζ of Phe166,
and that it functions as a slit pocket. Using the same requirements, the
model did not yield a productively bound R-naproxen ester substrate
because of steric hindrance of the R-naproxen methyl group by
Ala165, in agreement with the lack of activity of the enzyme towards
this compound [7].
Applying the same criteria for productive binding, as used to model
the naproxen ester in the active site of NP, the IPG moiety of the IPG es-
ters most likely binds in the alcohol-binding pocket. The size of this
pocket is mainly limited by Phe66 (Fig. 6), and could fit only methyl
or ethyl substrates, in agreement with the enzyme's activity on
naproxen methyl and ethyl esters [7]. However, the ability of the
carboxylesterases to also hydrolyze the larger IPG esters [5,7] suggests
that the active site alcohol-binding pocket is sufficiently flexible to ac-
commodate the IPG-moiety. Only if Phe66, located at the interface be-
tween the two domains (Fig. 6A), adopts a different conformation, the
alcohol-binding pocket would be able to accommodate the larger IPG al-
cohol moiety. Similar conformational changes have been observed in
B. subtilis lipase [45,46] where, upon binding of an IPG-like inhibitor,
the alcohol-binding pocket expanded to accommodate the IPG moiety
of the inhibitor.
substrate and product are observed. The exit route found in other ester-
ases [47] is blocked by Phe162, His163, Pro164, Asp165 and Val166, all
from the cap domain. Likewise, a possible water tunnel at the end of
the alcohol-binding pocket is blocked by Phe66 (Fig. 3c,d). As a conse-
quence, only one possible route to and from the active site is present
in NP and CesB. However, in the reaction pathway the alcohol product
is supposed to be the first to leave the active site (Scheme 1), but the
exit route is blocked by the covalently bound acid part of the substrate.
Similarly, the entrance route for the hydrolytic water molecule is
obstructed. Yet, a lid-like movement of the cap domain (in particular
of the flap made by helices α5 and α6) may be sufficient to allow the
water molecule to enter the active site, and the products to be released.
Indeed, the mainly hydrophobic interactions of the residues of helices
α5 and α6 with the core domain could facilitate such a lid-like move-
ment. Similar mechanisms of active site opening through cap domain
helix relocation occur in family I.1 lipases [48,49] as well as in other
α/β hydrolases [50,51].
3.6. Comparison to other α/β hydrolase fold esterases; similarity with
meta-cleavage product hydrolases
A search of the Dali database [40] revealed that the NP and CesB
structures are not very similar to those of other esterases, with the ex-
ception of the meta-cleavage product (MCP) hydrolases (Table S1). In
general, the active site residues and oxyanion binding site amides are
structurally conserved, but the alcohol-binding pockets of CesB and
NP are smaller than those of other esterases with known structure.
In NP the acyl part of the substrate likely binds in the entrance to the
active site as shown in Fig. 6. In contrast, in other α/β hydrolase fold es-
terases the acyl part is usually bound in a deep hydrophobic pocket at
the interface between the two domains [31,51,52]. However, at that po-
sition in NP, Ala156 and Glu157, two residues located at the domain in-
terface, prevent binding of substrates containing a large acyl moiety.
Instead, the acyl moiety binds in the entrance to the active site. This is
possible, because the helices in the cap domain have a different position,
causing the active site entrance of NP and CesB to have a different loca-
tion compared to other esterases.
A secondary structure motif (SSM) search [53] of the PDB indicated
significant structural similarity of the carboxylesterases NP and CesB
with C\C bond breaking enzymes that belong to the MCP hydrolase
family [54]. These latter enzymes have 19–20% sequence identity to NP
(Table S1) and overall root mean square (r.m.s.) deviations of less than
2.3 Å. This superfamily of cytosolic hydrolases also includes soluble
non-heme peroxidases, epoxide hydrolases, fluoroacetate dehalogenases
and haloalkane dehalogenases [55]. Several MCP hydrolase family
members also show esterase activity [56]. Intriguingly, compared to
the dual specificity ester/MCP hydrolases, CesB and NP have an Ala156
CesB is enantioselective towards IPG caprylate ester [6], but NP is
not. Indeed, our model shows that in CesB Tyr182 interacts with one
of the IPG ring oxygen atoms, thus favoring the binding of the (S) enan-
tiomer over the (R) enantiomer of IPG, where this hydrogen bond is not
possible. In contrast, in NP the residue equivalent to Tyr182 is a Phe,
which cannot make this hydrogen bonding interaction, resulting in a
less selective binding.
It is noteworthy that in the carboxylesterases NP, CesB and CesA no
separate entry and exit routes (backdoor, side-door, or tunnel) for
Fig. 6. Stereo image of the modeling of the tetrahedral intermediate of (S)- (in cyan) and (R)-naproxen ester (in magenta) in the active site of NP. Amino acids Phe66 and A156, restricting
access to the active site, are colored in yellow.