encephalitis virus (19), and ZIKV (20) have been
determined using the same construct of NS2B-
NS3 connected with an artificial glycine-rich
linker. In the absence of an inhibitor, the C-terminal part of NS2B cofactor is flexible and often disordered (open conformation) (13, 16, 18, 19).
Inhibitor-bound structures adopt a compact complex where the NS2B fragment wraps around the
NS3 protease and makes direct contacts with the
inhibitors (closed conformation) (13–15, 17–20).
Although solution nuclear magnetic resonance
(NMR) studies of DENV and WNV proteases have
suggested that the free enzyme is able to form the
closed conformation in solution (21–23), it is still
unclear whether the closed conformation is stabilized by the binding of a ligand. For ZIKV, we
have discovered that the artificial linker introduces steric hindrance and alters the substrate
(inhibitor)–binding behavior, suggesting that the
unlinked binary ZIKV NS2B-NS3 protease (bZiPro)
is preferable for studying the enzyme behavior
and for inhibitor development (24). In this regard, it is important to characterize the dynamic
behavior of ZIKV NS2B-NS3 protease in detail.
Here we report the crystal structure of bZiPro
at a resolution of 1.58 Å (Fig. 1 and table S1). The
refined model consists of four bZiPro molecules
in one unit cell labeled according to the peptide
chain IDs AB, CD, EF, and GH (fig. S1). There
are three molecules of free enzyme and one bound
to the K14K15G16E17 tetrapeptide from the neighboring NS3 N terminus in an unusual reverse
orientation (Fig. 1 and fig. S1). NS2B cofactor is
in the closed conformation—both the N- and
C-terminal regions are folded into a b-sheet
conformation—and in a complex with the NS3
protease domain in all four bZiPro molecules.
The structures of bZiPro free enzyme are virtually identical to the peptide-bound bZiPro (AB)
with root mean square deviations (RMSDs) of
0.25 to 0.34 Å after superimposition (Fig. 1D,
fig. S1B, and table S2). In addition, the temperature factor of the individual NS2B chain is also
comparable to that of the partner NS3, indicating that the NS2B and NS3 form a stable complex in the crystal (fig. S1C). Furthermore, bZiPro
is very similar to eZiPro [ZIKV NS2B-NS3 protease after self-cleavage; Protein Data Bank (PDB)
code, 5GJ4], which has a RMSD of 0.45 Å for
162 Ca atoms, and gZiPro (the single chain ZIKV
NS2B49–96-G4SG4-NS3Pro for NS2B-NS3 protease with a glycine-rich linker; PDB code, 5LC0),
which has a RMSD of 0.52 Å for 167 Ca atoms
(Fig. 1, fig. S1A, and table S2) (20, 24). The unlinked ZIKV NS2B-NS3 protease appears to contain a preformed stable substrate-binding pocket
that does not undergo further substantial conformational changes upon substrate or inhibitor binding.
Unexpectedly, one bZiPro molecule (AB) binds
to the N-terminal K14K15G16E17 tetrapeptide
sequence extended from the neighboring NS3
protease (H′ chain) (Fig. 1A and fig. S1A). We
identified the residues E12 to T18 from the neigh-
boring NS3 N-terminal region in the electron
density map and built them into the structure
unambiguously (Fig. 1F). The same residues are
disordered in the remaining three bZiPro, eZiPro,
and gZiPro structures, suggesting that the AB-H′
protease complex might be a crystallographic
artifact (20, 24). Nonetheless, the structure does
capture the protease in complex with a reverse
peptide. The tetrapeptide K14K15G16E17 folds
into a small hairpin loop to occupy the active
site. Specifically, the K14 e-amino group occupies
the S1 pocket and forms hydrogen bonds with
D129 and Y130. Residue K15 contributes the
most to the overall binding: Its side chain e-amino
group forms hydrogen bonds with S81 and D83 in
the S2 pocket, and its main chain forms hydrogen
bonds with G151 and G153 of NS3 protease. G16
does not form any interactions with the pro-
tease and leaves the S3 pocket completely empty,
similar to NS2B G129 in the eZiPro structure.
The E17 side chain stacks with the aromatic ring
of Y161 of the protease and partially occupies the
S1 pocket with K14. Notably, the hairpin is par-
tially stabilized by the intramolecular hydrogen
bonds: one between the backbone carbonyl oxy-
gen atom of K14 and the amide nitrogen atom of
E17, and two between the side chain carboxylic
group of E17 and the e-amino group of K14 and
backbone amide of T18 (Fig. 1D). The specific
conformation of the reverse peptide does not
allow the catalytic residue S135 to establish the
hydrogen-bonding relay and form the reactive
oxyanion species. It also prevents the carbonyl
1598 23 DECEMBER 2016 • VOL 354 ISSUE 6319 sciencemag.org SCIENCE
Fig. 1. Crystal structure of bZiPro. (A) The free enzyme and peptide-bound ZIKV protease structures
determined from the bZiPro crystal. In the free enzyme form, NS2B is colored in green and NS3 in cyan.
In the peptide-bound form, NS2B is colored in magenta and NS3 in yellow. The N- and C-terminal
residues of NS2B and NS3 are labeled. (B) Electrostatic view of bZiPro as free enzyme with an empty
preformed substrate-binding pocket (pockets are labeled). (C) Electrostatic view of bZiPro in complex
with the NS3 N-terminal peptide K14K15G16E17 in an unusual reversed orientation. Electrostatic
surfaces were colored using PyMOL by electrostatic potential at neutral pH from −5 k T (red) to +5 k T
(blue). (D) Close-up views of the interactions between N-terminal residues K14 to E17 from a neighboring
NS3 (in cyan) and the residues from bZiPro. (E) Superimposition of the two structures of bZiPro, shown
as ribbons. The RMSD is 0.34 Å for 152 Ca atoms. (F) A simulated annealing omit mFo-DFc map of the
KKGE reverse peptide within the bZiPro crystal structure is contoured at 3s in green mesh. A 2mFo-DFc
electron density map is contoured at 1s in blue. Throughout, residues from NS2B are underlined. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G,
Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.