You are on page 1of 18

pubs.acs.

org/biochemistry Article

Structural, Functional, and Mutational Studies of a Potent Subtilisin


Inhibitor from Budgett’s Frog, Lepidobatrachus laevis
Mihir Rami, Mohd Shafique,* and Siddhartha P. Sarma*
Cite This: https://doi.org/10.1021/acs.biochem.3c00252 Read Online

ACCESS Metrics & More Article Recommendations *


sı Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: Subtilases play a significant role in microbial


pathogen infections by degrading the host proteins. Subtilisin
Downloaded via INDIAN INST OF SCIENCE on October 11, 2023 at 13:59:32 (UTC).

inhibitors are crucial in fighting against these harmful micro-


organisms. LL-TIL, from skin secretions of Lepidobatrachus laevis,
is a cysteine-rich peptide belonging to the I8 family of inhibitors.
Protease inhibitory assays demonstrated that LL-TIL acts as a slow-
tight binding inhibitor of subtilisin Carlsberg and proteinase K with
inhibition constants of 91 pM and 2.4 nM, respectively. The
solution structures of LL-TIL and a mutant peptide reveal that they
adopt a typical TIL-type fold with a canonical conformation of a
reactive site loop (RSL). The structure of the LL-TIL-subtilisin
complex and molecular dynamics (MD) simulations provided an
in-depth view of the structural basis of inhibition. NMR relaxation
data and molecular dynamics simulations indicated a rigid conformation of RSL, which does not alter significantly upon subtilisin
binding. The energy calculation for subtilisin inhibition predicted Ile31 as the highest contributor to the binding energy, which was
confirmed experimentally by site-directed mutagenesis. A chimeric mutant of LL-TIL broadened the inhibitory profile and
attenuated subtilisin inhibition by 2 orders of magnitude. These results provide a template to engineer more specific and potent TIL-
type subtilisin inhibitors.

■ INTRODUCTION
Being the first vertebrates to conquer the land, amphibians
protease they inhibit, viz., aspartic, cysteine, metallo, serine,
and threonine protease inhibitors.6 However, there are also
have evolved diverse strategies to fight against a multitude of known examples of proteins that can inhibit two classes of
endogenous and exogenous factors. One of the key strategies is proteases using different, nonoverlapping binding sites.7 Serine
the secretion of peptides by granular glands located in the protease inhibitors are the most studied and found in almost
dermis when the amphibian is stressed or injured. These skin every living organism. Based on the mechanism of action, three
peptides have unique chemical properties and extreme distinct types of serine protease inhibitors are known:
biochemical diversity. Based on the function, amphibian skin canonical inhibitors, noncanonical inhibitors, and serpins.7
peptides can be categorized into myotropical peptides, opioid Canonical serine protease inhibitors interact with proteases
peptides, corticotropin-releasing peptides, angiotensins, pro- through an exposed loop with a conserved backbone
tease−inhibitor peptides, neuropeptides, antioxidant peptides, conformation called the “canonical” conformation.6 This
lectins, insulin-releasing peptides, wound healing promoting conformation is found in at least 19 families of proteinaceous
peptides, immunomodulatory peptides, antimicrobial peptides, serine protease inhibitors with different structural topologies
antiviral peptides, and antitumor peptides.1 described in the MEROPS database (https://www.ebi.ac.uk/
Historically, extensive studies have been carried out on merops/).8 Remarkably, the reactive site loop (RSL) for these
amphibian skin peptides due to their potential clinical inhibitors adopts a similar canonical conformation regardless of
applications and unique chemical properties.1,2 Skin secretions the different structural folds. To date, several canonical
of frogs and toads are used for prey capture as poison darts,3 in protease inhibitors have been characterized in amphibians,
traditional Chinese medicine to regulate internal bodily
functions and fertility, and as part of ancient Egyptian
pharmacopeia to treat pain and diarrhea.2 Received: May 14, 2023
Protease inhibitors play an important role in the amphibian Revised: September 21, 2023
skin. In the literature, it is speculated that protease inhibitors
either regulate endogenous proteases residing in the skin4 and/
or provide protection from exogenous proteases secreted by
pathogens.5 Protease inhibitors can be classified by the type of

© XXXX American Chemical Society https://doi.org/10.1021/acs.biochem.3c00252


A Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

including Kunitz-type,5,9 Kazal-type,10 and Bowman−Birk-type and the LL-TIL peptides. The plasmids were transformed into
inhibitors.11−13 The trypsin inhibitor-like (TIL) disulfide-rich Escherichia coli BL21 (DE3) cells and grown at 37 °C. Protein
peptides form a distinct class of protease inhibitors and are expression was induced by the addition of IPTG to a final
found to be secreted by animals across phyla.4,14−17 These concentration of 0.5 mM when the cells reached an optical
peptides are known to inhibit the S1, S8, and M4 families of density of 0.8 to 0.9 at 600 nm.
proteases. The cells were harvested by centrifugation (6000 rpm, 15
While the structures and specific protease targets of several min, 4 °C) 5 h postinduction, resuspended in lysis buffer 50
TIL inhibitors are known,18−25 information on TIL peptides mM Tris-HCl, 150 mM NaCl, 0.1% NaN3, pH 8.0 (at 25 °C),
found in amphibian secretions is limited. To the best of our and stored at −80 °C. The cOmplete protease inhibitory
knowledge, a trypsin inhibitor isolated from Bombina bombina EDTA-free tablets from Roche were added to the resuspended
is the only TIL peptide from amphibians with a known buffer before lysis. Cell lysis was carried out using French Press
structure.24 TIL inhibitors of other protease families, such as (1100 psi). Cell debris and other insoluble matter were
the S8 (subtilisin family), have not been reported for removed by centrifugation (13000 rpm, 60 min, 4 °C).
amphibian secretions. The lysate was passed through a Ni2+-NTA column pre-
Subtilisin-like proteases play a pivotal role in the equilibrated with lysis buffer. The recombinant protein bound
colonization of the host by certain pathogenic microorganisms; to the column was then eluted using 1 M Imidazole. The
this is primarily achieved by the degradation of antimicrobial elution fraction containing the recombinant protein was
peptides secreted by hosts.26 Subtilase inhibitors provide an dialyzed against thrombin cleavage buffer (50 mM Tris-HCl,
additional layer of defense for the host by neutralizing the 10 mM CaCl2, 150 mM NaCl, and 0.01% NaN3, pH 8.0)
proteolytic activity of subtilases. Commercially, subtilisin before the proteolytic cleavage reaction. Thrombin cleavage
inhibitors are used extensively in laundry detergents and reaction was performed at 22 °C with the fusion protein to
food processing to regulate the protease activity. The TIL thrombin ratio being 50:1. The purified peptide was obtained
family of inhibitors generally inhibits the S1 family by separating the LL-TIL from fusion tag cytochrome b5 using
(chymotrypsin-like) of proteases: trypsin, chymotrypsin, and size exclusion chromatography with a Superdex S-30 column.
elastase. There are a few reports of subtilisin inhibitors from The mutants LL-TILmut1 and LL-TILmut2 were purified in
the TIL family: BmSI-7 found in tick Boophilus microplus and the same manner as the native form.
multiple variants of silkworm inhibitors from Bombyx mori.27,28 The isotope precursors for protein production were
Although inhibitory properties of these TIL-type subtilisin purchased from Sigma Chemicals or Cambridge Isotope
inhibitors have been characterized, no structural information is Laboratories. Carbon-13, nitrogen-15, and nitrogen-15-en-
available for TIL-type subtilisin inhibitors. riched protein samples were prepared by culturing cells
Here we report the structure and protease inhibitory bearing the appropriate plasmids in M9 minimal medium
properties of a TIL peptide identified from cDNA in the with 15N-enriched ammonium chloride (1 g/L) and 13C6-
skin secretions of the South American frog Lepidobatrachus enriched glucose (2 g/L) or 15N-enriched ammonium chloride
laevis,29 also known as Budgett’s frog.30 This peptide, called (1 g/L) as metabolic sources of carbon and nitrogen,
LL-TIL, belongs to the I8 family of inhibitors (according to the respectively.
MEROPS database8), incorporating a Trypsin inhibitor-like Molecular Characterization. Analytical Size Exclusion
(TIL) domain. The I8 family of inhibitors generally possesses Chromatography. Size exclusion chromatography studies
10 cysteine residues that form 5 disulfide bonds. Interestingly, were performed on a GE Healthcare Superdex S75 (10/300
the TIL domain also occurs in large glycoproteins such as the GL) analytical column at 4 °C. 100 μL of a 1.0 mg/mL
von Willebrand factor (VWF),25 mucins, and SCO-spondin. solution of LL-TIL was injected into the column and eluted
The disulfide connectivity for the I8 family is C1-C7, C2-C6, using a flow rate of 0.5 mL/min. The elution volumes of LL-
C3-C5, C4-C10, and C8-C9.15 In the first report of the TIL were compared with those of the protein molecular weight
protease inhibitory activity of this peptide, it was shown that standards.
LL-TIL exhibited moderate trypsin inhibitory activity. CD Spectroscopy. Circular dichroism (CD) spectra were
However, we demonstrate unequivocally that LL-TIL strongly acquired on a JASCO-715 spectro-polarimeter over a 190 to
inhibits the microbial protease subtilisin Carlsberg in a slow- 260 nm wavelength range. The CD data were acquired with a
tight binding manner with the Ki of 91 pM. High-resolution, scan rate of 100 nm/min and a data pitch of 1/nm using a 0.1
multinuclear solution NMR methods were used to determine cm path length cuvette. The final spectra were averaged over
the solution structure of LL-TIL. Data-driven HADDOCK three scans and baseline-corrected by subtracting the spectrum
models31 of the LL-TIL-subtilisin complex were calculated. of the appropriate blank solution. The data were smoothed
The structural stability of the LL-TIL-subtilisin complex was using a Savitzky−Golay filter33 available with the JASCO
examined by using molecular dynamics simulation. Further- spectra analysis software.
more, mutations were introduced to understand the role of the Mass Spectrometry. Matrix-assisted laser desorption
RSL residues in the inhibition profile of LL-TIL. The results of ionization (MALDI) mass spectra were acquired on an Ultraex
our findings and their implications are discussed below. time of flight (TOF)/TOF or UltraeXtreme of BrukerDal-

■ MATERIALS AND METHODS


Protein Expression and Purification. The plasmid
tonics, Germany. 1 μL of (0.1 to 1 mg/mL) sample was mixed
with an equal volume of a saturated solution of either sinapinic
acid (SA) or α-cyano-4-hydroxycinnamic acid (CCA) (in 50%
vector (pET21(a)+) containing the genes of interest (LL- acetonitrile in water with 0.1% TFA) before spotting on the
TIL and mutant variants) were synthesized as cytochrome b5 MALDI plate. The data were acquired in positive ion mode
fusion proteins32 and purchased from GenScript. The plasmid with a linear or reflectron mode detector. 500 to 1000 spectra
constructs were designed to include an N-terminal hexahisti- were co-added. The data were processed and analyzed using
dine tag and a thrombin cleavage site between cytochrome b5 data analysis software Flex Analysis 3.1.
B https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

Protease Inhibition Assays. Bovine trypsin and chymo- Scheme 3


trypsin, elastase from porcine pancreas, subtilisin Carlsberg
from Bacillus licheniformis, and proteinase K from Tritirachium
album were purchased from SIGMA. According to the
manufacturer’s instructions, the protease−inhibitor activities
of trypsin, chymotrypsin, and elastase were determined using
EnzChek Protease assay kits (E6638, Molecular probes). 50 Scheme 2 mechanism predict that Kobs increases hyperbolically
nM concentration of each protease (Trypsin, chymotrypsin, with inhibitor concentration, and the equations for the Scheme
elastase) was incubated in 100 mM Tris buffer (pH = 7.4) with 3 mechanism predict that Kobs decreases hyperbolically with
different concentrations of LL-TIL and LL-TIL mutants for 15 inhibitor concentration. For the Scheme 1 mechanism, eq 4
min. The reaction was initiated by adding 100 μL of BODIPY was used to extract kon, koff, and Kobs values.
FL casein to the protease−inhibitor mixture. Fluorescence
(vi vs)(1 ) ij 1 exp( Kobst ) yz
intensity (485 nm excitation and 510 nm emission wave- A = vst + lnjjj zz
z
lengths) was measured continuously for 10 min at 37 °C on a Kobs k 1 {
96-well plate reader (Thermo Scientific Verioscan reader). + Ao
Subtilisin and proteinase K inhibition was characterized by
2
using the synthetic substrate N-suc-AAPF-pNA (S7388, [E ] ij vs yzz
where = t jjj1 z
vi zz{
SIGMA). Subtilisin (2 nM) and proteinase K (10 nM) were
[I ] jk
incubated with LL-TIL and LL-TIL mutants in 50 mM HEPES
buffer (100 mM NaCl, 5 mM CaCl2, pH = 7.4) for 15 min, (3)
followed by the addition of N-suc-AAPF-pNA. Initial velocities k1[I ]
were measured at 405 nm continuously for 5 min in a 96-well Kobs = k 1 +
[S ]
plate reader. Apparent inhibition constant (Kapp i ) and active 1+
enzyme concentration (Eo) were determined by nonlinear Km (4)
regression curve fitting using the Morrison equation for tight where A is the absorbance, Kobs is the first-order rate
binding.34,35 constant, t is time, Ao is the initial absorbance, k−1 is the
vs E Io K iapp + [(Eo + Io + K iapp)2 4EoIo]1/2 reverse rate constant, k1 is the forward rate constant, and Km is
= o the Michaelis constant. All of the data fitting was carried out
vo 2Eo using ORIGIN software version 10.5.117.
(1) NMR Spectroscopy. Sample Preparation. Data Acquis-
The inhibition constants (Ki) were extracted using the ition. All NMR experiments for structure determination were
Cheng−Prusoff equation for competitive inhibition.36 performed on an Agilent 600 MHz spectrometer equipped
with a cryogenically cooled, triple-resonance (z-axis) pulsed
ij [S ] yzz field gradient probe. NMR titration studies were performed on
K iapp = K ijjj1 z
j
k K m zz{ (2) a Bruker 700 MHz spectrometer equipped with a triple-
resonance (z-axis) pulsed field gradient probe. NMR data were
where Eo is the active enzyme concentration, Io is the total recorded at 25 °C. Backbone and side-chain resonance
inhibitor concentration, vs is the steady-state velocity, and vo is assignments for 1H, 15N, and 13C nuclei were obtained from
the initial velocity. double- and triple-resonance NMR experiments that transfer
Slow binding kinetics was analyzed using equations magnetization via scalar couplings.39 1H−15N correlations were
described by Morrison and Walsh for subtilisin inhibition.34 determined from the 15N-HSQC experiment. Triple-resonance
Inhibition progress curves were generated by initiating the experiments, viz., CBCA(CO)NH, HNCACB, HNCO, and
reaction by adding 1 nM subtilisin to LL-TIL (1−7 nM) and HN(CA)CO experiments, were carried out to obtain
N-suc-AAPF-pNA (100 μM). The reaction was monitored at sequence-specific assignments. Side-chain assignments were
405 nm for 45 min at 37 °C. Data curves for each inhibitor obtained from HCCCONH, HCCH-TOCSY, CCCONH,
concentration were fitted by nonlinear regression analysis to HBHACONH, 13C-HSQC, and TOCSY-HSQC spectra.
the integrated rate equation for the slow-tight binding three-dimensional (3D)-HNHA experiment was carried out
inhibitors.37 The resultant Kobs values were plotted against to obtain scaler coupling constant 3JHN − Hα. HNHB40 and
inhibitor concentration to determine the inhibition mecha- HN(CO)HB41 spectra were acquired to determine the χ1
nism: Schemes 1, 2, or 3.34,35,38 angles (Figure S8a). Long-range HNCO42 experiment was
performed to detect the presence of hydrogen bonds. NOE
Scheme 1 correlations were established from simultaneous 13C and 15N
edited NOESY-HSQC spectra acquired with a mixing time
(τc) of 175 ms. 15N relaxation data was acquired to estimate R1
(logitudinal relaxation rates), R1ρ (transverse relaxation rates),
and 1H−15N heteronuclear NOE. NMR experiments and
Scheme 2 acquisition parameters are listed in Table S1. Model-free
analysis43 of the relaxation data was carried out to probe the
internal dynamics of the individual amino acid residues using
Model-free4 program44 incorporated in FAST-MF.45
Data Analysis. NMR data were processed on an Intel PC
The equations for Scheme 1 predict that Kobs increases workstation running SuSe Linux version 42.1 using NMRPipe/
linearly with inhibitor concentration, the equations for the NMRDraw processing software.46 The time domain data in
C https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

Figure 1. (a) Multiple sequence alignment of LL-TIL, LL-TIL mutants, and other protease inhibitors of the TIL family. The position of the RSL is
indicated by a red box. Disulfide-bonded cysteines are denoted by same numbers below in magenta. The sequence alignment was performed on
ESPript server.62 (b) Sequence specifically assigned 2D 1H,15NHSQC spectrum of LL-TIL acquired at 600 MHz. A total of 57 well-resolved
correlation peaks (49 from backbone, 6 from side chains of N29, N47, N48, and 2 from side chains of E56) were assigned. Side-chain correlations are
connected by horizontal dotted lines. (c) 15N backbone dynamics of LL-TIL (blue) and LL-TILmut1 (green). The RSL residues are highlighted in
gray strip.

both the direct and indirect dimensions were apodized with a in the HNHB experiment and 1Hβ-CO J coupling in the
90° phase-shifted sine bell with a power of one or two. Zero HN(CO)HB experiment. Hydrogen-bond restraints between
filling was carried out before the Fourier transformation in donor−acceptor pairs were derived from the long-range
each dimension. Data were analyzed using CcpNmr Analysis HNCO experiment. The restraints were included as the
version 2.4.2.47 distance of acceptor oxygen from the donor nitrogen (dNiC′i) d d

Structure Determination. Structural Restraints. Distance and proton (dNiC′i) with upper limits of 3.5 and 2.5 Å,
d d

restraints were generated from the integrated peak intensities respectively. Disulfide connectivity was established by
of unambiguously and ambiguously assigned interproton assigning unique NOEs between protons on disulfide-paired
correlations in 15N, and 13C edited NOESY-HSQC spectra cysteines. Two disulfide-bonded cysteines, viz., i and j, were
using the “Make Distance Restraints” routine in the CcpNmr restrained by distances between Sγi −Sγj , Cβi −Sγj , and Sγi −Cβj
suite. Upper distance limits were set to an absolute maximum atoms with both the upper and lower limits at 2.05, 3.05, and
of 6.0 Å, while the lower limits were set to 1.8 Å. Restraints for 3.05 Å respectively.
ϕ angles were generated using the 3JHN − Hα coupling constants Structure Calculation. Structures were calculated using
derived from the intensities of the cross and diagonal peaks. torsion angle dynamics protocol incorporated in CYANA
The ψ angle restraints were generated from the prediction version 3.98.13.49 NOE-based distances, hydrogen-bond
program TALOS+.48 Side-chain χ1 torsion angles were distances, disulfide bond restraints, and dihedral angles were
predicted from the qualitative analysis of 15N−1Hβ J coupling used for the structure calculations. In total, 500 structures were
D https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

Figure 2. Characterization of the protease−inhibitor activities of LL-TIL and mutants. (a) Inhibition of elastase, chymotrypsin, and trypsin by LL-
TIL. The wild-type peptide has no or minimal inhibitory activity against these proteases. (b) Inhibition of elastase, chymotrypsin, and trypsin by
LL-TILmut1. The mutant peptide has nominal inhibitory activity against these proteases and does not reach 100% inhibition at the LL-TILmut1
concentrations used in these studies. (c) Inhibition of subtilisin by LL-TIL and LL-TILmut2. The wild-type peptide inhibits subtilisin with a Kapp i of
131 pm, while the LL-TILmut2 has an inhibition constant 5-fold lower. The solid line passing through data points (blue for LL-TIL, red for LL-
TILmut2) reflects the best fit obtained from fitting the quadratic Morrison equation. (d) Inhibition of subtilisin by LL-TILmut1. The LL-TILmut1
peptide exhibits an inhibition constant that is ≈100-fold lower compared to the wild type. The solid line passing through the data points reflects the
best fit obtained from fitting the quadratic Morrison equation. (e) Progress curve analysis of LL-TIL-subtilis in inhibition, solid lines passing via
data points for each inhibitor concentration reflect the best fit obtained by fitting eq 3. (f) Kobs values obtained from progress curve analysis plotted
against inhibitor concentration of 0 to 11 nM. The solid line represents the linear fitting of the data to eq 4. All experiments were carried out in
duplicates and repeated three times (n = 3).

E https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

calculated from CYANA. The top 50 structures were selected minimized and equilibrated before production runs. The MD
for further refinement based on target function criteria. Explicit simulations were conducted for 200 ns under a constant
water refinement was conducted using XPLOR-NIH.50 Finally, temperature of 298 K and a constant pressure of 1 atm. The
20 structures were selected for further analysis based on the simulation trajectory was displayed, animated, and analyzed
least energy criteria. Restraints from CYANA to XPLOR-NIH with the visualization program PyMOL.
were converted using pdbstat.51 Structures were analyzed in The MM-PBSA method60 was used to calculate the binding
MOLMOL52 and PyMOL.53 free energy of the LL-TIL-subtilisin complex for each frame of
LL-TIL Subtilisin Binding. The lyophilized subtilisin MD simulation. In the MM-PBSA method, the binding free
powder from SIGMA was dissolved in 50 mM HEPES buffer energy is described as the sum of the potential energy in a
(100 mM NaCl, 5 mM CaCl2, pH 7.4) followed by purification vacuum, the polar and nonpolar solvation-free energies, and
using size exclusion chromatography. The purified LL-TIL and the entropy contribution. Here, as the binding interface
subtilisin were incubated in a 2:1 ratio, and the complex remains rigid throughout the simulation, the calculation does
formation was monitored by size exclusion chromatography. not include the entropic contribution. MM-PBSA calculations
The binding of subtilisin to LL-TIL and LL-TILmut1 was were carried out using g_mmpbsa tool61 implemented within
further investigated by NMR titration. This was achieved by GROMACS. Solvent and solute dielectric constants of 80 and
acquiring 15N HSQC spectra of 15N-labeled, 100 μM LL-TIL 2, respectively, and salt concentrations of 150 mM were set in
and LL-TILmut1 with increasing concentrations of UL- the PB calculations. The binding energy contribution for each
subtilisin. The combined and weighted backbone amide residue was also obtained from g_mmpbsa. Bootstrap analysis
chemical shift perturbations (Δδ) were calculated using the with 2000 steps was carried out to estimate free energy
below equation39 calculation errors.

= [( H)
2
+( 2 1/2
N /5) ] (5) ■ RESULTS AND DISCUSSION
Biophysical Characterization. The sequences of the LL-
where ΔδH and ΔδN are the chemical shift changes (in ppm) at TIL peptide and two mutant variants used in our studies are
the beginning and end of the titration, respectively. Two- given in Figure 1a. Overexpression of fusion LL-TIL protein
dimensional and one-dimensional line shape analysis of NMR was good, and the yields were high enough to prepare
titration data were performed using TITAN software package54 isotopically enriched samples for NMR. Analytical size
and NmrLineGuru55 respectively. All of the 1H−15N-HSQC exclusion chromatography shows LL-TIL elutes as a single
spectra of LL-TIL and LL-TILmut1 from the titration series peak with the elution volume corresponding to the molecular
were processed identically. For LL-TIL and LL-TILmut1, a mass of 8 kDa (Figure S1a). This is indicative of the
two-state binding model and a three-state induced fit binding monomeric state of LL-TIL in solution. The molecular mass
model were used, respectively, to track changes in chemical obtained from MALDI spectra matches the calculated LL-TIL
shift and intensities. mass (Figure S1b). CD spectrum of LL-TIL (Figure S2a)
Structure Calculation of LL-TIL-Subtilisin Complex. shows a minima at 210 nm and predicts only the antiparallel β-
Data-driven structural models of the subtilisin complex with sheets as the secondary structure in the peptide.63 The one-
LL-TIL were generated by HADDOCK 2.4. The ensemble of dimensional proton NMR spectrum of LL-TIL at 25 °C
the 20 lowest-energy structures of LL-TIL was submitted for exhibits upfield-shifted resonances and large chemical shift
docking with subtilisin. Subtilisin structure was extracted from dispersion in the spectral region between 6.5 and 10 ppm,
subtilisin-eglinC complex (1cse.pdb).56 From NMR titration of indicative of a well-folded peptide (Figure S2b). The CD and
15
N-LL-TIL with subtilisin, residues that exhibited CSP values one-dimensional NMR spectra of LL-TILmut1 and LL-
above “average CSP + Standard deviation” were selected as TILmut2 peptides showed similar spectral properties.
active site residues for LL-TIL: Cys19, Thr27, Asn29, Cys30, Ile31, Functional Characterization. Trypsin, Chymotrypsin,
and Ala32. For subtilisin, residues forming hydrogen bonds with and Elastase Inhibition. The majority of the protease
eglinC reactive site loop in 1CSE were selected as active inhibitors of the TIL family inhibit proteases from the S1
residues: Gly100, Gly102, Tyr104, Gly127, Ala129, Ser130, Asn155, family (chymotrypsin-like): trypsin, chymotrypsin, and elas-
Asn218, Ser221. The active residues of LL-TIL were introduced tase. Consequently, the protease inhibitory activity of LL-TIL
as semiflexible. The resulting clusters were manually analyzed, was tested against trypsin, chymotrypsin, and elastase using
and the top four structures of each cluster were submitted to EnzChek Protease assay kits (E6638, Molecular probes).
the PRODIGY server to predict the binding affinity. The final Surprisingly, LL-TIL did not inhibit trypsin, chymotrypsin, or
four least energy structures were analyzed in PyMOL and elastase, even at a maximum LL-TIL to protease ratio of 200
LigPlot+.57 (Figure 2a). The putative reactive site loop (RSL), which is
Molecular Dynamics Simulation. Molecular dynamics responsible for inhibiting the proteases, was identified from the
(MD) simulations58 were carried out for LL-TIL, LL- sequence alignment of LL-TIL and other TIL-type protease
TILmut1, and LL-TIL-subtilisin complex with GROMACS inhibitors (Figure 1a). The reactive site loop comprises Cys30,
version 2021 employing the AMBER99sb all-atom force Ile31, Ala32, Val33, Cys34, and Lys35, occupying positions P3, P2,
field.59 The protein molecules were solvated in a water box P1, P1′, P2′, and P3′ respectively.64 Previously identified
extending 10 Å from the outermost protein atom. The ionic trypsin inhibitors of the TIL family show a preference for
strength of the solvating solution was kept at 100 mM. The charged residues in RSL (Figure 1a). To study the implication
starting structures of LL-TIL and LL-TILmut1 were selected of polar residues in the RSL on the specificity of LL-TIL, we
based on the least energy criteria from the final ensemble of mutated three residues at position P2, P1, and P1′ to residues
NMR structures. The initial structure of the LL-TIL-subtilisin found in RSL of Ascaris trypsin inhibitor (ATI), viz., Thr31,
complex was selected based on the lowest predicted ΔG from Arg32, and Glu33. This chimeric LL-TIL mutant will be referred
cluster 1 of the HADDOCK run. The structures were energy- to as LL-TILmut1 from here on. LL-TILmut1 weakly inhibited
F https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

trypsin and chymotrypsin but did not inhibit elastase (Figure 2d). Since the fitted Kapp
i value was higher than [E]T in the case
2b). of the LL-TILmut1, the [E]T value was treated as constant as
Subtilisin Inhibition. Sequence alignment of the TIL-type suggested by Kuzmic and colleagues.66 Finally, the Cheng−
subtilisin inhibitors FPI-F and BmSPI7 from B. mori65 showed Prusoff equation yielded a Ki value of 6.93 nM. All inhibition
similar hydrophobic residues as LL-TIL in their RSL (Figure constants and kinetic parameters for subtilisin inhibition are
1a). This prompted us to investigate the inhibitory profile of tabulated in Table 2.
LL-TIL against subtilisin. Subtilisin inhibition was charac- Proteinase K Inhibition. LL-TIL also shows strong
terized using the N-suc-AAPF-pNA substrate. LL-TIL strongly inhibitory activity against fungal protease proteinase K.
inhibits subtilisin and exhibits prolonged onset of inhibition, a Proteinase K from T. album belongs to the subtilisin-like
defining feature of slow-tight binding inhibitors. The nonlinear family (S8), and it possesses keratin hydrolyzing activity.67 The
curve fitting of the Morrison equation for LL-TIL-subtilisin nonlinear curve fitting of the Morrison equation for LL-TIL-
inhibition data yielded Kappi of 131 pM and active enzyme proteinase K inhibition data yielded Kapp
i of 3.45 nM and active
concentration ([E]o) of 1.98 nM (Figure 2c). Considering enzyme concentration ([E]o) of 9.55 nM (Figure S3).
substrate concentration, the Cheng−Prusoff equation gener- Considering substrate concentration, the Cheng−Prusoff
ated a Ki of 91 pM, comparable to the well-recognized potent equation generated Ki of 2.44 nM. The inhibition of proteinase
inhibitors like Streptomyces subtilisin inhibitor (SSI), hiproly K by LL-TIL was not surprising as the active sites of proteinase
barley chymotrypsin inhibitor 2 (CI-2), turkey ovomucoid K and subtilisin are almost identical.68 Since LL-TIL inhibits
third domain (OMTKY3), and leech eglin c (Table 1). subtilisin more strongly than proteinase K, we chose to focus
on the LL-TIL-subtilisin interaction in this study.
Table 1. Potent Subtilisin Inhibitors Solution Structure of LL-TIL. Sequence-Specific Reso-
Ki nance Assignments. The sequence-specific resonance assign-
no. name specificity RSL (P3−P3′) (nM) ments of LL-TIL were obtained by using a set of double- and
1 SSI subtilisin Cys-Pro-Met-Val-Tyr-Asp 0.005 triple-resonance NMR experiments. Strip plots of HNCACB
2 CI-2 subtilisin, Val-Thr-Met-Glu-Tyr-Arg 0.012 and CBCA(CO)NH experiments show the sequential assign-
chymotrypsin ments of residues ranging from His49 to Cys58 (Figure S4).
3 OMTKY3 subtilisin, Cys-Thr-Leu-Glu-Tyr-Arg 0.023
chymotrypsin Except for Gly43, the backbone amide correlation for each
4 LL-TILa subtilisin Cys-Ile-Ala-Val-Cys-Lys 0.091 residue of LL-TIL was observed in the 15N-HSQC spectrum
5 eglin C subtilisin, Val-Thr-Leu-Asp-Leu-Arg 0.15 (Figure 1b). The single set of peaks observed for the peptide
chymotrypsin indicates an absence of chemical exchange on the NMR time
6 FPI-Fa subtilisin Cys-Ile-Thr-Val-Ile-Arg 0.148 scale. The 13Cβ chemical shifts of cysteines fall between 35 and
7 BmSI-7a subtilisin, Cys-Thr-Ala-Asp-Cys-Ile 1.4 45 ppm (Figure 3a), suggestive of their oxidized nature.69 The
elastase
a
values of Δδ 13Cβγ for all proline residues lie between 3 and 6
TIL peptides. ppm (Figure S5), a characteristic feature of “trans” X-Pro
peptide bond conformation.70 Chemical shift assignments were
The kinetics of the slow onset of subtilisin inhibition was obtained for 95% HN, 92% 15N, 96.5% 13Cα, and 94.8% CO of
analyzed to get insight into the mechanism of inhibition. the backbone nuclei and 98.1% 13Cβ of the side-chain nuclei.
Fitting of the progress curve data to eqs 2 and 3 estimated Near-complete chemical shift assignment of LL-TIL and LL-
values for vs, vi, and Kobs for each curve with different LL-TIL
TILmut1 is deposited in the BioMagResBank (http://www.
concentrations (Figure 2e). The Kobs values show a linear
bmrb.wisc.edu/) under accession numbers 36536 and 36559,
increase when plotted against LL-TIL concentration, in
respectively.
accordance with the inhibition mechanism of Scheme 1,
indicating a single-step binding. However, this linearity can Secondary Structure and Disulfide Connectivity. Secon-
also account for Scheme 2 (induced fit), when Ki is much dary structure was determined based on short- and medium-
greater than Ki*, which reduces the hyperbolic equation for range NOEs, 3JHN − Hα coupling constants, and chemical shifts
Scheme 2 to a linear relationship.35 Consequently, it has not of Cβ, Cα, and Hα atoms as shown in Figure S6. It consists of
been possible to distinguish which scheme is operational in the four β strands and extensive loop regions. The disulfide
case of LL-TIL-subtilisin inhibition. However, the slow-tight connectivity was deduced from the 15N and 13C edited NOESY
binding inhibitors almost always follow the Scheme 2 spectra. Based on the NOEs found across cysteine residues,
mechanism.35 Subsequent linear fitting with eq 4, yields kon Cys6-Cys38, Cys15-Cys34, Cys19-Cys30, and Cys40-Cys52 were
and koff of (1.08 ± 0.08) 106 M−1 s−1 and (7.53 ± 6.5) 10−5 established to be disulfide-paired (Figure 3b). Cys23-Cys58
s−1, respectively (Figure 2f). The Ki calculated (69 pM) from disulfide pairing was performed. No NOE correlations were
the ratio of kon and koff agrees well with the value obtained found across unpaired cysteine residues. Furthermore, later
from the Morrison plot (91 pM). stages of structure calculations did not detect significant
The fitting of the Morrison equation for LL-TILmut1- proximity between the nonbonded cysteine pairs, confirming
subtilisin inhibition generated a Kapp
i value of 9.75 nM (Figure the disulfide connectivity.

Table 2. Inhibition Constants and Kinetic Parameters for Subtilisin Inhibition

name Kapp
i (nM) Ki (nM) kon (M−1 s−1) koff (s−1)
LL-TIL 0.131 ± 0.03 0.091 (1.08 ± 0.08) 10 6
(7.53 ± 6.5) 10−5
LL-TILmut1 9.75 ± 0.42 6.93
LL-TILmut2 0.78 ± 0.12 0.54

G https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

Figure 3. (a) Strip plots from the 3D-HNCACB spectrum showing the chemical shifts of Cβ nuclei (in red) of the 10 cysteine residues in LL-TIL.
Dashed lines indicate the upper and lower limits of Cβ chemical shifts. The values of the chemical shifts indicate that all 10 cysteine residues are in
the oxidized state. (b) Strip plots of 13C edited NOESY spectrum depicting NOEs found across disulfide-paired cysteine residues. (c) Strip plots
from the long-range H-bond 3D-HNCO (blue) and reference 3D-HNCO (black) spectra showing the 3JNiC′j scalar coupling correlations across the
d d

hydrogen bonds. The long-range carbonyl correlations obtained from the H-bond HNCO spectrum are connected by horizontal lines to the
corresponding correlations in the reference HNCO experiment for the hydrogen-bond donor−acceptor pairs found across β strands of LL-TIL. (d)
Schematic diagram depicting the β strand topology of LL-TIL with experimentally observed interstrand hydrogen bonds.

NMR Restraints. The NOE-based distances were obtained Lys54. The schematic diagram depicting the topology of the β
from 3D 15N and 13C edited NOESY-HSQC spectra. strands, along with assigned hydrogen bonds, is presented in
Sequential NOE correlations (Hαi to HNi+1 and Hαi to Hδi+1 for Figure 3d. Strong Hiα to Hi+1δ NOE correlations observed
prolines) helped to corroborate the sequence-specific assign- across the X-Pro7, Pro18, Pro20, and Pro28 peptide bonds
ments determined from three-dimensional through bond NMR provided additional evidence for the trans peptide bond
experiments. The hydrogen bonds deduced from the long- conformation. 95 backbone dihedral angles were predicted
range HNCO experiment were crucial to establish comple- using TALOS+ and incorporated into the structure calculation.
mentarity between β strands (Figure 3c). Long-range NOEs Analysis of HNHB and HN(CO)HB experiments provided χ1
observed across the β strands confirm the β strand registry angles for 28 residues: 12 residues with t2g3 (−60°)
(Figure S7) with β-sheet 1 consisting of Lys10-Gly16 and Lys35- conformation, 9 residues with g2g3 (+60°) conformation and
Cys40, while the β-sheet 2 comprises Thr44-Asp46 and Cys52- 7 residues with g2t3 (180°) conformation. Strip plots for
H https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

HNHB and HN(CO)HB for cysteine residues are shown in antiparallel to each other. The hydrogen bonds in the β-
Figure S8b. sheet 1 are disrupted to accommodate a wide type β-bulge.71
Tertiary Structure. In total, 1399 NOE-based distances, 24 Phe14 and Cys15 from strand-1 and Lys35 from strand-3 are
hydrogen bonds (two for each H-bond), 5 disulfide bonds, and involved in the formation of this β-bulge. The relatively shorter
122 dihedral angles were used as restraints for the structure β-sheet 2 consists of two strands: β3 (residues 44−46) and β4
calculations of LL-TIL. The NMR restraints and structural (residues 52−54), which are antiparallel. Loop2, connecting
statistics are listed in Table 3. The final ensemble of the 20 the first and second β strand, is the longest loop in LL-TIL
(residue Ser17 to Cys34), harboring the reactive site loop. The
Table 3. NMR Restraints and Structural Statistics for LL- RSL acquires a characteristic canonical shape that is observed
TIL in other canonical protease inhibitors. Disulfide bonds between
Cys19-Cys30, Cys23-Cys58, and Cys15-Cys34 play a substantial
NMR restraints LL-TIL LL-TILmut1
role in stabilizing loop2. Loop3 (residues 41−44) protrudes
distance constraints out from the plane of the disk-shaped peptide, resulting in a
total 1399 1103 highly solvent-exposed region. This observation explains the
intra-residue (|i−j| = 0) 685 489 weak intensity of backbone amide peaks for residues Arg41 and
inter-residue Asp42, as well as the absence of a Gly43 resonance in the
sequential (|i−j| = 1) 318 264 1
H−15N HSQC spectrum. β3 and β4 strands are connected by
medium-range (|i−j| ≤ 4) 140 114 loop 4 (type I turn), forming a β hairpin. Finally, the structure
long-range (|i−j| ≥ 5) 256 236
of LL-TIL ends with a 310-helix formed by residues Lys55-
hydrogen-bond restraintsa 24 24
Asn57.
total dihedral angle restraints 122 125
The structure comparison server DALI72 showed that LL-
backbone dihedral angles (ϕ and ψ) 95 97
TIL is structurally similar to other protease inhibitors of this
side-chain dihedral angles (χ1)b 27 28
family, namely, Apis mellifera chymotrypsin/cathepsin G
structure statistics
inhibitor-1 (AMCI-1, 1ccv.pdb),22 ATI (Ascaris trypsin
restraints violations (mean and s.d)
inhibitor, 1ata.pdb),21 and C/E-1 (Ascaris chymotrypsin/
distance violations > 0.1 Å 0 0
elastase inhibitor, 1eai.pdb),20 with z-scores of 6.1, 5.8, and
dihedral angles > 5° 0 0
5.1, respectively. Interestingly, LL-TIL also resembles the
Ramachandran map statistics
recently determined structure of TIL′ domain (6n29.pdb) in
PROCHECK
assembly with the Human Von Willebrand factor73 (z-score:
most favored regions (%) 90.9 87.4
6.0). Pairwise structural alignment of LL-TIL with AMCI-1,
additionally allowed regions (%) 9.2 12.6
ATI, C/E-1, and BSTI (Bombina skin trypsin/thrombin
generously allowed regions (%) 0 0
inhibitor) reveals that the differences primarily lie in the
disallowed regions (%) 0 0
conformation and length of the extended loop region situated
Richardson MolProbity
between the first and second β strand, as shown in Figure S15.
favored regions (%) 99.0 97.8
The notable difference between the loop of LL-TIL and BSTI
allowed regions (%) 1.0 2.2
arises due to the two extra residues in BSTI, which gives rise to
disallowed regions (%) 0 0
an additional turn. Also, this loop in the C/E-1-elastase
close contacts and deviations from ideal
geometryc structure forms a pore of 5 Å diameter near the RSL of C/E-1,
number of close contacts 0 1 where the elastase side-chain arginine inserts itself. LL-TIL
RMS deviation for bond angles 0.6 0.6 structure lacks this pore. The residues P1 and P′1 of RSL are
RMS deviation for bond lengths (Å) 0.008 0.008 solvent-exposed across all of the serine protease inhibitors,
r.m.s deviation from average structure (Å)d making them readily available to the enzyme. In the case of LL-
backbone 0.76 ± 0.12 0.70 ± 0.14 TIL also, the relative accessible surface area is 89 and 52% for
heavy 1.18 ± 0.11 1.13 ± 0.14 P1 and P′1, making them highly accessible to the enzyme.
a
Hydrogen-bond restraints were obtained from a long-range HNCO Solution Structure of LL-TILmut1 and Comparison
experiment. bχ1 angles were obtained from analysis of HNHB and with LL-TIL. In total, 1103 NOE-based distances, 24 hydrogen
HN(CO)HB experiments. cInformation on close contacts and bonds (two for each H-bond), 5 disulfide bonds, and 125
deviations from ideal geometry was acquired from PSVS server. dihedral angle restraints were used for the structure
d
Root mean square deviation was calculated in PSVS server. calculations of LL-TILmut1. The NMR restraints and
structural statistics are tabulated in Table 3. The final ensemble
lowest-energy structures is shown in Figure 4c,d. The ensemble of the 20 lowest-energy structures is shown in Figure S10. The
of 20 structures shows a backbone root-mean-square deviation structure is similar to that of LL-TIL, with identical disulfide
(RMSD) of 0.7 Å and heavy atom RMSD of 1.1 Å for residues connectivity and secondary structure. The RMSD between the
6−58. PROCHECK-NMR analysis found 90.6 and 9.4% of the lowest-energy structure of LL-TIL and LL-TILmut1 is 1.46 Å
backbone ϕ and ψ dihedral angles lying in the most favored when superposed on Cα atoms. The overlay of the lowest-
region and additionally allowed region of the Ramachandran energy structures of LL-TIL and LL-TILmut1 is shown in
map, respectively (Figure S9). The solution structure of LL- Figure S11. Significant structural differences are observed in
TIL acquires a flat disk shape and lacks a hydrophobic core the mutated RSL region of the peptide. In the LL-TILmut1
(Figure 4b). The overall structure of LL-TIL consists of two structure, the backbone conformation of the RSL changes as
well-defined antiparallel β sheets and five loops held together Arg32 faces inward with relative accessible surface area reducing
by five disulfide bonds (Figure 4a). β1 strand is preceded by a from 0.91 (for Ala32 in LL-TIL) to 0.5. This can be attributed
type I turn formed by residues Pro7-Lys10. β-sheet 1 consists of to the hydrogen-bond formation between the side chain of
β1 (residues 10−16) and β2 (residues 35−40) strands, Arg32 to the backbone carbonyl of Glu33, also resulting in
I https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

Figure 4. Structure of LL-TIL. (a) Superposed cartoon and surface representations of the lowest energy structure of LL-TIL. Disulfide bonds are
colored pink and β strands are colored blue. The position of RSL and loop2 of LL-TIL are also indicated. (b) Side view of LL-TIL. (c, d) Ensemble
of 20 lowest energy structures of LL-TIL. The color scheme is the same as in (a).

distortion of the canonical shape of the binding loop (8a). As increasing subtilisin concentration, the free LL-TIL (F)
we will see further, this alteration in the conformation may resonances disappeared, while the peaks corresponding to
have some relevance in subtilisin inhibition. The structural subtilisin-bound LL-TIL started to appear. There was an
coordinates for LL-TIL and LL-TILmut1 have been submitted abrupt change in chemical shift between the free and bound
to the Protein Data Bank (PDB) with the accession numbers states with increasing subtilisin/LL-TIL molar ratio, character-
8HWU and 8J3Q, respectively. istic of a slow exchange. The fitting and extraction of the
15
N Backbone Dynamics of LL-TIL and LL-TILmut1. kinetic and thermodynamic parameters were performed by
The backbone amide 15N T1, T2, and {1H}-15N steady-state using TITAN and NmrLineGuru. Both programs calculated
heteronuclear NOE values of LL-TIL and LL-TILmut1 are very similar values of 2.122 s−1 (TITAN) and 1.086 s−1
shown in Figure 1c. The {1H}-15N steady-state heteronuclear (NmrLineGuru) for the koff. The Kd values were calculated
NOE values predominantly lie above 0.6, suggesting that the to be 4.56 and 0.030 μM. However, as the LL-TIL
peptides are rigid and ordered. The average order parameters concentration used in the titration study (100 μM) ≫ Kd
for LL-TIL and LL-TILmut1 were 0.69 and 0.67, respectively. (extracted from NMR titration studies) and Ki (extracted from
The extracted order parameters are plotted against the residue inhibition studies), the Kd and koff values extracted from NMR
numbers in Figure 1c. Notably, the reactive site loop residues titration data should only be taken as the upper limits and not
Cys30 and Val33 for LL-TIL and Cys30 and Glu33 for LL- as the absolute values. As seen in Figure 5a,b, the RSL region
TILmut1 yield order parameters below 0.6, suggesting a and the residues structurally proximal to it exhibit significant
relatively more dynamic nature of both residues. chemical shift perturbation, while the chemical shifts of other
LL-TIL Subtilisin Binding. LL-TIL-subtilisin binding was peptide regions are largely unperturbed.
15
studied by size exclusion chromatography and NMR titration. N-labeled LL-TILmut1 also indicated slow exchange but
The overlay of gel filtration chromatograms for LL-TIL, exhibited a different pattern of chemical shift perturbation with
subtilisin, and LL-TIL-subtilisin is shown in Figure S1a. When the addition of subtilisin (Figure S12a). The free LL-TILmut1
LL-TIL and subtilisin are injected into the column in a 2:1 peak disappeared, and two resonances appeared at different
ratio, a new peak appears at a lower retention volume, chemical shift positions, presumably corresponding to two
corresponding to the complex. subtilisin-bound states (B1 and B2). All of these states are in
NMR titration data was instrumental in identifying the slow exchange with each other on the NMR time scale. The
residues residing at the binding interface of the LL-TIL CSPs between the free state and B1 state resonances are
subtilisin complex. The titration was performed by adding negligible. Interestingly, the CSPs between the B2 state and
unlabeled subtilisin to the 15N-labeled LL-TIL sample. With free state resonances are greater and agree well with the CSPs
J https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

Figure 5. Mapping the interaction site of subtilisin on LL-TIL. (a) Overlay of 15N-HSQC spectra of free 15N-LL-TIL (blue) and 15N-LL-TIL with
UL-subtilisin (green) in a 1:1 ratio. Significantly shifted residues of LL-TIL (circled) are considered to participate in the interaction with subtilisin.
(b) Plot of cumulative CSP as a function of the primary structure of LL-TIL. Average CSP and average CSP plus one standard deviation are
indicated by dotted lines. (c) CSPs are mapped on the structure of LL-TIL with color gradient indicating no change in CSP (gray) to CSP plus one
standard deviation (green).

of LL-TIL when titrated against subtilisin, indicating a similar TIL and LL-TILmut1 in the absence and the presence of
arrangement of the binding interface (Figure S12b). subtilisin are shown in Figure S16. The MALDI spectra were
Interestingly, when the subtilisin to LL-TILmut1 ratio is acquired following incubation of the inhibitor with subtilisin
under unity during the NMR titration, the B1 state intensities for 6 h. It is clear from the spectra that the masses of LL-TIL
are higher compared to the B2 state. This begins to change at and LL-TILmut1 do not alter in the presence of subtilisin,
the subtilisin to LL-TILmut1 ratio of 1, ultimately resulting in suggesting that the cleavage does not occur under experimental
higher intensities of the B2 state at the ratio of 1.4 (Figure conditions.
S13). Implications of these results are discussed below. The crystal structures of the standard mechanism protease−
Mechanism of Inhibition. The canonical inhibitors inhibitor complexes are almost exclusively found in the
interact with the proteases according to the “Standard thermodynamically favored “C” state (tight, noncovalent
mechanism” or “Laskowski mechanism”, in which the inhibitor complex with intact inhibitor), also called a Michaelis complex.
RSL inserts itself into the protease binding site in a substrate- NMR titration of 15N LL-TIL with subtilisin resulted in the
like manner. This mechanism can be written in the following emergence of only a single-bound state. Moreover, the mass
way6 spectrometry data show that an “intact” LL-TIL forms a
complex with subtilisin (Figure S16a,b). In light of this, we can
E + I F L F C F X* F L* F E + I* safely assign this state to the Michaelis complex. The absence
where E is the protease, I and I* are intact and cleaved of “L” state from the LL-TIL-subtilisin titration spectra
inhibitors, respectively, and L and L* are loose, noncovalent suggests that either LL-TIL directly forms the Michaelis
complexes of protease and intact or modified inhibitor, complex (C) or that a loose complex is formed, which rapidly
respectively. C and X* are the stable, noncovalent protease− rearranges to the Michaelis complex. This agrees with the
inhibitor complexes with intact and cleaved inhibitors, progress curve analysis for LL-TIL-subtilisin inhibition, which
respectively. The formation of the “C” complex is usually indicated either a single-step binding (Scheme 1) or an
followed by a very slow hydrolysis of the peptide bond induced fit mechanism (Scheme 2) with Ki much greater than
between P1 and P1′ residue initiated by a nucleophilic attack that of K*i .
from Oγ atom of catalytic serine residue residing in the The NMR titration of 15N LL-TILmut1 with subtilisin leads
protease active site. to the partition of LL-TILmut1 into two bound states (B1 and
As the LL-TIL RSL is flanked by two disulfide bonds, the B2). The mass spectrometry data show that 15N LL-TILmut1
hydrolysis of the peptide bond between P1 and P1′ would stays intact during complex formation with subtilisin (Figure
result in an increase in mass by 18 Da. MALDI spectra of LL- S16c,d). Therefore, according to the standard mechanism, the
K https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

two bound states can only be the loose complex (L) and the binding interface in the LL-TIL-subtilisin complex is similar to
Michaelis complex. As the B2 state resembles the subtilisin the known subtilisin bound complexes of Greglin, CI-2, Eglin
bound form of LL-TIL with regard to chemical shifts, we C, SSI, and OMTKY3.56,74−77 LL-TIL-subtilisin interaction is
speculate that this state corresponds to a Michaelis complex. predominantly driven by residues Thr27 (P6) to Cys34 (P2′) of
On the other hand, the B1 state peaks are proximal to the free LL-TIL, inserting themselves into the subtilisin active site cleft
state, which suggests the possibility of a loose complex “L”. in a substrate-like manner (Figure 6c). The LL-TIL RSL is
The NMR titration also shows that the “L” state (B1) is held together in the active site cleft of subtilisin by hydrogen
predominant at lower subtilisin:LL-TILmut1 ratios (≤1.2), bonds formed between pairs Asn29 (P4)-Gly102, Ile31 (P2)-
while the “C” state (B2) becomes the major species at a higher Gly100 above the cleft, and Cys30 (P3)-Gly127, Ala32 (P1)-Ser125
subtilisin:LL-TILmut1 ratio (1.4) (Figure S13). In other below the cleft (subtilisin residues are indicated in “italics”
words, the equilibrium seems to shift toward the B2 state as the typeface) (Figure 6d). These interactions allow the insertion of
subtilisin/LL-TILmut1 ratio increases (Figure S14). The Asn29 (P4) and Ala32 (P1) residues into the binding pockets S4
presence of the loose complex and it is comparable populations and S1 of subtilisin, respectively78 (Figure 6e,f). LL-TIL
to the Michaelis complex observed for LL-TILmut1-subtilisin residues Asn29 (P4)-Ile31 (P2) form a β-zipper79 type of
titration are in stark contrast to what is observed for the LL- interaction with the subtilisin residues Leu126-Gly128 (Figure
TIL-subtilisin titration. It suggests that the equilibrium is not 6a), which is also observed for subtilisin complexes of CI-275
heavily biased toward the Michaelis complex as the loose and greglin.74 This type of β-sheet addition, a commonly
complex does not readily transition into the Michaelis complex observed motif in protein−protein interactions, has been
in LL-TILmut1-subtilisin titration. This may also account for pointed out by Waksman and co-workers.79 Furthermore, Ala32
the weaker inhibition of LL-TILmut1 toward subtilisin. (P1) forms two hydrogen bonds with Asn155 and catalytic
HADDOCK Model of LL-TIL-Subtilisin Complex. NMR Ser221, including the one between the Oγ atom of Ser221 and
perturbation data confirm that the reactive site loop the N atom of Ala32. On the prime side of the P1 residue, a
significantly contributes to the LL-TIL-subtilisin interaction. hydrogen bond is observed between Cys34 and Asn218. The
To gain insight into the structural basis for the LL-TIL- complex’s interface area and the buried surface area were found
subtilisin interaction, data-driven docking protocols in to be 601.5 and 1229.4 Å2, respectively.
HADDOCK were applied. Residues of LL-TIL that showed The contact map shown in Figure S17 reveals the
hydrophobic contacts formed between LL-TIL with subtilisin
chemical shift perturbation (CSP) more than average plus one
(http://mapiya.lcbio.pl/).80 In particular, the RSL residues
standard deviation were found to be present near or within the
Ile31, Ala32, and Val33 form in total 10 hydrophobic contacts of
RSL region and were considered active residues (cf. Materials
less than 5 Å with subtilisin (Table S2). Besides, the nonpolar
and Methods Section). HADDOCK clustered 158 structures
buried surface area of the LL-TIL-subtilisin complex (753.9
of the LL-TIL-subtilisin complex into 13 clusters. Analysis of
Å2) accounts for 61.3% of the total buried surface area (1229.4
clusters revealed that cluster 1, with a size of 47 structures, Å2). These lines of evidence point to the significance of
showed the lowest Z-score (−2.0) and HADDOCK score hydrophobic interactions to the binding process. The distance
(−81.7 ± 0.7). They also displayed the maximum number of between the catalytic Oγ atom of Ser221 and the Ala32 carbonyl
hydrogen bonds across the interface and highest binding carbon is 3.55 Å. This distance is ∼0.7 Å shorter for other
affinities. Structural statistics of the LL-TIL subtilisin complex subtilisin inhibitors (≈2.8 Å), essential for the nucleophilic
are tabulated in Table 4. attack to occur. Interestingly, a similar long distance of 3.1 Å is
Figure 6a,b shows the representative structure of the LL-TIL also seen in the crystal structure of TIL peptide C/E-1 bound
subtilisin complex from cluster 1. The arrangement of the to elastase (1eai.pdb).20 Perhaps, this long distance precludes
the nucleophilic attack, implying that LL-TIL does not readily
Table 4. Restraints and Structural Statistics for Data-Driven get cleaved by subtilisin.
Docking of LL-TIL and Subtilisin Molecular Dynamics Simulation. MD simulation of LL-
TIL and LL-TILmut1 illustrates that the RMSD remains
ambiguous interaction restraints HADDOCK run steady with respect to the initial structure over the time span of
LL-TIL 19,27,29,30,31,32 200 ns (Figure 7a). The root-mean-square fluctuation (RMSF)
subtilisin 100,102,104,127,129,130,155,218,221 values indicate the mobility of each amino acid residue. The
restraints for LL-TIL relatively higher RMSF values for the RSL region of both LL-
NOE 1399 TIL and LL-TILmut1 can be extrapolated to higher mobility,
H-bond 24 which agrees with the NMR relaxation data (Figure 7b).
dihedral angle 122 MD simulation analysis of the LL-TIL-subtilisin complex
structural statistics for haddock- was carried out to determine the stability of the complex and
derived structures of
LL-TIL-subtilisin complex to further refine the interface interactions unbiasedly. The low
(cluster 1) RMSD values of LL-TIL-subtilisin complex trajectories (<0.3
HADDOCK score −81.0 ± 0.7 nm) and negligible change in buried surface area (14,515 ±
cluster size 47 894 Å2) for the complex during the 200 ns simulation suggest
RMSD (Å) 0.7 ± 0.5 that the LL-TIL-subtilisin complex is stable. The hydrogen-
Evdw (kcal mol−1) −61.1 ± 1.9 bond distribution for the trajectories remains narrow with an
EEE (kcal mol−1) −90.3 ± 6.2 average of 12.4 ± 1.8 hydrogen bonds per frame (Figure 7c).
EDE (kcal mol−1) −3.8 ± 2.6 Hydrogen-bond occupancies across all of the frames in the MD
ERVE (kcal mol−1) 13.0 ± 12.4 simulation are tabulated in Table 5. Close examination of the
BSA (Å2) 1307.8 ± 79.9 interface between LL-TIL and subtilisin in MD simulation
Z-score −2.0 trajectories revealed that most of the hydrogen bonds observed
L https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

Figure 6. Structure of the LL-TIL-subtilisin complex. (a, b) Superposed cartoon and surface representative HADDOCK structure of the LL-TIL
(orange) subtilisin (green) complex. The inset shows the β zipper formed on the interface of the LL-TIL-subtilisin complex. (c) Dimplot57 of LL-
TIL-subtilisin complex depicting interactions across the interface. The hydrogen bonds (pink dotted line) and corresponding distances between
donor and acceptor atoms are indicated. Residues or atoms involved in hydrophobic interactions are bordered by eyelashes and connected by
dotted black lines. (d) Hydrogen-bond interactions between RSL of LL-TIL and residues located above and below the active site cleft of subtilisin.
(e) Insertion of Asn29 (P4) of LL-TIL into the S4 pocket of subtilisin. (f) Insertion of Ala32 (P1) of LL-TIL to S1 pocket of subtlilsin.

in the initial structure remain intact. The hydrogen bond components are tabulated in Table 6. VdW and electrostatic
between the backbone O atom of Thr27 and the side-chain Oϵ energy are the major contributors to the binding energy. A per-
atom of Tyr104 is lost during the simulation. Interestingly, the residue decomposition was performed to examine the residue-
residue occupying the P1 site (Ala32) strengthens its wise contribution of the binding energy for the LL-TIL (Figure
interaction with subtilisin by forming another hydrogen bond 7d). As expected, the RSL residues contribute substantially to
with the backbone of Thr220 (Figure 7e). the binding energy. Ile31 contributes the most, with a binding
The free energy of binding was computed using the energy of −34.8 ± 0.3 kJ/mol. Even though the backbone of
Molecular Mechanics Poisson−Boltzmann Surface Area Ile31 forms only a single hydrogen bond with Gly100, its side
(MM-PBSA) approach as implemented in g_mmpbsa. A chain forms multiple hydrophobic interactions at the interface,
total of 400 frames extracted every 0.5 ns were used for the accounting for high binding energy (Figure 6c). We generated
calculations. The average free energy of binding and its an I31A mutant to validate this experimentally, and we indeed
M https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

Figure 7. MD simulations (a) root-mean-square deviation (RMSD) of LL-TIL (black), LL-TIL-subtilisin complex (blue), and LL-TILmut1 (red)
with respect to the initial structure. (b) Residue-wise root-mean-square fluctuation (RMSF) of LL-TIL (black), LL-TIL-subtilisin complex (blue),
and LL-TILmut1 (red). RSL residues are highlighted in light gray. (c) Distribution of hydrogen bonds across the LL-TIL-subtilisin interface during
200 ns simulation. RSL residues are highlighted in light gray. (d) Per-residue contribution of LL-TIL to the binding energy. Our calculations
indicate that residue I31 imparts the highest contribution to the binding energy. (e) Hydrogen bonds formed by Ala32 (P1) of LL-TIL with
subtilisin.

observed weaker inhibition of this mutant against subtilisin RSL is preserved across the inhibitor families. The
with a Ki of 0.54 nM (Figure 2b). conformation of RSL bound to LL-TIL is similar to the free
Conformation of RSL. Although a wide variety of scaffolds form of LL-TIL with a backbone RMSD of 0.8 Å (Figure 8b).
exist for serine protease inhibitors, the conformation of the This observation reveals that there is a minimal conformational
N https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

Table 5. Hydrogen-Bond Occupancies of LL-TIL-Subtilisin the relatively lower stability of the Michaelis complex in LL-
Complex during 200 ns MD Simulation TILmut1-subtilisin interaction.
Specificity of LL-TIL. The original study by Wang et al.
no. LL-TIL subtilisin occ. (%)
had reported that LL-TIL acts as a trypsin inhibitor with a Ki of
1 T27 (O) A129 (N) 42.26 16 μM.29 This was surprising, considering trypsin prefers a
2 N29 (O) G102 (N) 73.63 positively charged residue at their P1 position. We observed
3 N29 (ND2) G102 (O) 86.81 here that LL-TIL does not inhibit trypsin even at a high LL-
4 C30 (O) G127 (N) 99.30
TIL-to-trypsin concentration ratio of 400. Moreover, LL-TIL
5 C30 (N) G127 (O) 94.91
showed no inhibition against chymotrypsin or elastase also.
6 I31 (N) G100 (O) 99.00
LL-TIL is a specific and potent inhibitor of subtilisin. Among
7 A32 (N) S125 (O) 97.40
other potent subtilisin inhibitors (Table 1), only a few show
8 A32 (O) S221 (N) 98.80
specificity for subtilisin. The specificity of canonical serine
9 A32 (O) T220 (N) 44.26
protease inhibitors is generally attributed to the P1 residue.
10 A32 (O) N155 (ND2) 89.31
Trypsin inhibitors prefer Arg or Lys at its P1 site, while
11 A32 (N) S221 (OG) 18.68
chymotrypsin inhibitors favor Leu, Met, Phe, Tyr, Trp, or Asn
12 C34 (N) N218 (O) 99.00
at its P1 site. In contrast to these mammalian proteases,
Table 6. Free Energy of Binding and Its Components for subtilisin inhibitors have a much broader preference for P1
LL-TIL-Subtilisin Complex residue,81 which also includes Ala as found in LL-TIL.
The hydrophobic RSL residues (Ile31, Ala32, and Val33) form
no. energy parameter energy (kJ mol−1) multiple hydrophobic contacts with subtilisin at the binding
1 Van der Waals energy −360.93 ± 2.19 interface, make a large contribution to the nonpolar buried
2 electrostatic energy −321.04 ± 4.42 surface area (276.6 Å2), and also contribute significantly to the
3 polar solvation energy 409.41 ± 8.50 binding energy as revealed in the MM-PBSA analysis. All of
4 SASA energy −32.42 ± 0.28 this evidence suggests that the hydrophobic nature of these
5 binding energy −303.74 ± 8.29 amino acids plays an important role in governing the specificity
of LL-TIL. Replacing Ile31 with alanine (LL-TILmut2)
change in the RSL of LL-TIL upon binding to subtilisin. The weakens the subtilisin inhibitory activity. Moreover, mutating
rigidity of RSL is a consequence of two disulfide bonds all three hydrophobic RSL residues to hydrophilic residues
flanking the RSL and the hydrophobic interactions of Val33 (LL-TILmut1) not only weakens the subtilisin inhibition but
(P1′) to Pro18 and Ile31. The bound conformation of RSL also imparts trypsin inhibitory activity, resulting in losing its
closely mimics the RSL of other canonical subtilisin inhibitors, specificity. The specificity of canonical protease inhibitors is
as shown in Figure 8b. also ascribed to the flexibility of the RSL of the inhibitor in
The solution structure of Ascaris trypsin inhibitor (ATI)21 their free state.82,83 Here, the NMR relaxation data and
solved at two different pHs (2.4 and 4.75) shows that the molecular dynamics simulation showed that the RSL region of
canonical shape of RSL gets slightly distorted at higher pH as LL-TIL is not too flexible, which could further contribute to its
the side chain of P1 residue faces inward to the main body of high specificity.
peptide. We observed that in the structure of LL-TILmut1, Conclusions. In this study, we show that LL-TIL acts as a
RSL indeed acquires a conformation similar to ATI RSL with a potent inhibitor of subtilisin and proteinase K, with no effect
backbone RMSD of 0.802 Å and differs significantly from RSL on trypsin, chymotrypsin, or elastase. Most of the NMR
of LL-TIL with a backbone RMSD of 1.678 Å, as shown in structures of TIL peptides have been determined using 1H
Figure 8a. This structural difference could be responsible for NMR spectroscopy. Here, we have determined the structures

Figure 8. (a) Superposition of RSL of LL-TIL, LL-TILmut1, and ATI using backbone atoms N, Cα, and C. LL-TIL differs from LL-TILmut1 and
ATI in the conformation of the P1 loop, which is responsible for the binding and inhibition. (b) Superposition of RSL of subtilisin bound LL-TIL,
CI-2, OMTKY3, eglin C, and SSI and free LL-TIL using backbone atoms N, Cα, and C. The RMSD values of RSL of LL-TIL and subtilisin-bound
LL-TIL with RSL of other subtilisin inhibitors are indicated.

O https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

of LL-TIL and LL-TILmut1 using high-resolution triple- https://pubs.acs.org/10.1021/acs.biochem.3c00252


resonance NMR experiments. These experiments allowed a
deeper understanding of the structural and dynamic nature of Notes
the peptide. Mutagenesis and isotope-filtered NMR experi- The authors declare no competing financial interest.
ments have provided important mechanistic insights into the
subtilisin−inhibitor interaction. The high similarity between
subtilisin-like proteases secreted by pathogens and the ability
of LL-TIL to strongly inhibit subtilisin and proteinase K
■ ACKNOWLEDGMENTS
The authors thank the DST FIST programme for the 600 MHz
suggests that LL-TIL might be helpful in providing protection NMR and other instrumental facilities in the Molecular
against various harmful microorganisms. The involvement of Biophysics Unit, Indian Institute of Science (IISc.). The
subtilisin-like proteases as a virulence factor in the entry of authors also thank the DBT-IISc Partnership program for the
several microbial pathogens makes it an important target for Mass Spectrometric Facilities in the Division of Biological
treating several infectious diseases. Sciences (DBS), Indian Institute of Science. M.S. is thankful to
SERB, India, for NPDF grant file number PDF/2016/003833.
■ DATA DEPOSITION
Chemical shift assignments for LL-TIL and LL-TILmut1 have
The authors thank Sunita, Mass Spectrometry Facility, DBS,
IISc., for data acquisition. They also thank Sri Teja for useful
been deposited in the Biological Magnetic Resonance Bank discussions. The image of the frog used in the TOC image was
under accession numbers 36536 and 36559, respectively. 3D purchased from iStock by Getty Images.
structural data for LL-TIL and LL-TILmut1 have been
deposited in the RCSB PDB as entries 8HWU and 8J3Q,
respectively.
■ REFERENCES
(1) Demori, I.; Rashed, Z. E.; Corradino, V.; Catalano, A.; Rovegno,

■ ASSOCIATED CONTENT
* Supporting Information

L.; Queirolo, L.; Salvidio, S.; Biggi, E.; Zanotti-Russo, M.; Canesi, L.;
Catenazzi, A.; Grasselli, E. Peptides for Skin Protection and Healing
in Amphibians. Molecules 2019, 24, No. 347.
The Supporting Information is available free of charge at (2) Xu, X.; Lai, R. The chemistry and biological activities of peptides
https://pubs.acs.org/doi/10.1021/acs.biochem.3c00252. from amphibian skin secretions. Chem. Rev. 2015, 115, 1760−1846.
(3) Mayor, A. Greek Fire, Poison Arrows, and Scorpion Bombs;
Gel filtration spectra of LL-TIL, subtilisin, and LL-TIL- Princeton University Press, 2022.
subtilisin complex; CD and one-dimensional (1D) NMR (4) Mignogna, G.; Pascarella, S.; Amiconi, G.; Barra, D.;
spectra of LL-TIL; determination of Kapp i for LL-TIL- Wechselberger, C.; Hinterleitner, C.; Mollay, C.; Kreil, G. BSTI, a
proteinase K inhibition; sequence-specific assignments trypsin inhibitor from skin secretions of Bombina bombina related to
of LL-TIL; difference in Cβ and Cγ chemical shifts for protease inhibitors of nematodes. Protein Sci. 1996, 5, 357−362.
each proline residue of LL-TIL; secondary structure (5) Conlon, J. M.; Kim, J. B. A protease inhibitor of the Kunitz
family from skin secretions of the tomato frog, Dyscophus guineti
assignment for LL-TIL; long-distance NOE correlations
(Microhylidae). Biochem. Biophys. Res. Commun. 2000, 279, 961−964.
found across β strands of LL-TIL; determination of the (6) Laskowski, M.; Kato, I. Protein inhibitors of proteinases. Annu.
χ1 angles for LL-TIL; Ramachandran plot analysis of Rev. Biochem. 1980, 49, 593−626.
LL-TIL using PROCHECK; structure of LL-TILmut1; (7) Krowarsch, D.; Cierpicki, T.; Jelen, F.; Otlewski, J. Canonical
structural superposition of LL-TIL and LL-TILmut1; protein inhibitors of serine proteases. Cell. Mol. Life Sci. 2003, 60,
NMR titration of LL-TILmut1 with subtilisin; modu- 2427−2444.
lation in peak volumes during NMR titration of (8) Rawlings, N. D.; Barrett, A. J.; Thomas, P. D.; Huang, X.;
subtilisin:LL-TILmut1; B2:B1 peak volume ratios during Bateman, A.; Finn, R. D. The MEROPS database of proteolytic
NMR titration of subtilisin/LL-TILmut1; structural enzymes, their substrates and inhibitors in 2017 and a comparison
superposition of LL-TIL with TIL inhibitors; MALDI with peptidases in the PANTHER database. Nucleic Acids Res. 2018,
46, D624−D632.
spectra of 15N LL-TIL and LL-TILmut1 in the presence
(9) Dong, Y.; Shi, D.; Ying, Y.; Xi, X.; Chen, X.; Wang, L.; Zhou, M.;
and absence of subtilisin; contact map of LL-TIL- Wu, Q.; Ma, C.; Chen, T. A Novel Kunitzin-Like Trypsin Inhibitor
subtilisin interaction; NMR data acquisition parameters Isolated from Defensive Skin Secretion of Odorrana versabilis.
for LL-TIL, and hydrophobic contacts between RSL Biomolecules 2019, 9, No. 254.
residues of LL-TIL and subtilisin (PDF) (10) Li, R.; Wang, H.; Jiang, Y.; Yu, Y.; Wang, L.; Zhou, M.; Zhang,
Y.; Chen, T.; Shaw, C. A novel Kazal-type trypsin inhibitor from the
■ AUTHOR INFORMATION
Corresponding Authors
skin secretion of the Central American red-eyed leaf frog, Agalychnis
callidryas. Biochimie 2012, 94, 1376−1381.
(11) Song, G.; Zhou, M.; Chen, W.; Chen, T.; Walker, B.; Shaw, C.
Mohd Shafique − Molecular Biophysics Unit, Indian Institute HV-BBI−A novel amphibian skin Bowman−Birk-like trypsin
of Science, Bangalore, Karnataka 560012, India; inhibitor. Biochem. Biophys. Res. Commun. 2008, 372, 191−196.
Email: msnmbb@gmail.com (12) Yang, J.; Tong, C.; Qi, J.; Liao, X.; Li, X.; Zhang, X.; Zhou, M.;
Siddhartha P. Sarma − Molecular Biophysics Unit, Indian Wang, L.; Ma, C.; Xi, X.; Chen, T.; Gao, Y.; Wu, D. Engineering and
Institute of Science, Bangalore, Karnataka 560012, India; Structural Insights of a Novel BBI-like Protease Inhibitor Livisin from
orcid.org/0000-0001-7619-8904; Phone: +91 80 2293 the Frog Skin Secretion. Toxins 2022, 14, No. 273.
3454; Email: sidd@iisc.ac.in (13) Wang, T.; Jiang, Y.; Chen, X.; Wang, L.; Ma, C.; Xi, X.; Zhang,
Y.; Chen, T.; Shaw, C.; Zhou, M. Ranacyclin-NF, a Novel Bowman−
Author Birk Type Protease Inhibitor from the Skin Secretion of the East
Mihir Rami − Molecular Biophysics Unit, Indian Institute of Asian Frog, Pelophylax nigromaculatus. Biology 2020, 9, No. 149.
(14) Bernard, V. D.; Peanasky, R. J. The serine protease inhibitor
Science, Bangalore, Karnataka 560012, India family from Ascaris suum: chemical determination of the five disulfide
Complete contact information is available at: bridges. Arch. Biochem. Biophys. 1993, 303, 367−376.

P https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

(15) Bania, J.; Stachowiak, D.; Polanowski, A. Primary structure and (32) Mitra, A.; Chakrabarti, K. S.; Hameed, M. S.; Srinivas, K. V.;
properties of the cathepsin G/chymotrypsin inhibitor from the larval Kumar, G. S.; Sarma, S. P. High level expression of peptides and
hemolymph of Apis mellifera. Eur. J. Biochem. 1999, 262, 680−687. proteins using cytochrome b5 as a fusion host. Protein Expression
(16) Fogaça, A. C.; Almeida, I. C.; Eberlin, M. N.; Tanaka, A. S.; Purif. 2005, 41, 84−97.
Bulet, P.; Daffre, S. Ixodidin, a novel antimicrobial peptide from the (33) Savitzky, A.; Golay, M. J. Smoothing and differentiation of data
hemocytes of the cattle tick Boophilus microplus with inhibitory by simplified least squares procedures. Anal. Chem. 1964, 36, 1627−
activity against serine proteinases. Peptides 2006, 27, 667−674. 1639.
(17) Rios-Steiner, J. L.; Murakami, M. T.; Tulinsky, A.; Arni, R. K. (34) Morrison, J. F.; Walsh, C. T. The behavior and significance of
Active and exo-site inhibition of human factor Xa: structure of des-Gla slow-binding enzyme inhibitors. Adv. Enzymol. 1988, 61, 201−301.
factor Xa inhibited by NAP5, a potent nematode anticoagulant (35) Copeland, R. A. Enzymes: A Practical Introduction to Structure,
protein from Ancylostoma caninum. J. Mol. Biol. 2007, 371, 774−786. Mechanism, and Data Analysis; John Wiley & Sons, 2000.
(18) Murakami, M.; Rios-Steiner, J.; Weaver, S.; Tulinsky, A.; (36) Yung-Chi, C.; Prusoff, W. H. Relationship between the
Geiger, J.; Arni, R. Intermolecular interactions and characterization of inhibition constant (KI) and the concentration of inhibitor which
the novel factor Xa exosite involved in macromolecular recognition causes 50% inhibition (I50) of an enzymatic reaction. Biochem.
and inhibition: crystal structure of human Gla-domainless factor Xa Pharmacol. 1973, 22, 3099−3108.
complexed with the anticoagulant protein NAPc2 from the (37) Szedlacsek, S. E.; Duggleby, R. G. Methods in Enzymology;
hematophagous nematode Ancylostoma caninum. J. Mol. Biol. 2007, Elsevier, 1995; Vol. 249, pp 144−180.
366, 602−610. (38) Yang, H.; Li, X.; Li, G.; Huang, H.; Yang, W.; Jiang, X.; Sen, M.;
(19) Arolas, J. L.; Botelho, T. O.; Vilcinskas, A.; Gomis-Rüth, F. X. Liu, J.; Liu, Y.; Pan, Y.; Wang, G. Accurate quantitative determination
Structural evidence for standard-mechanism inhibition in metal- of affinity and binding kinetics for tight binding inhibition of xanthine
lopeptidases from a complex poised to resynthesize a peptide bond. oxidase. Biomed. Pharmacother. 2021, 139, No. 111664.
Angew. Chem., Int. Ed. 2011, 50, 10357−10360. (39) Cavanagh, J.; Fairbrother, W. J.; Palmer, A. G., III; Skelton, N.
(20) Huang, K.; Strynadka, N. C.; Bernard, V. D.; Peanasky, R. J.; J. Protein NMR Spectroscopy: Principles and Practice; Academic Press,
James, M. N. The molecular structure of the complex of Ascaris 1996.
chymotrypsin/elastase inhibitor with porcine elastase. Structure 1994, (40) Archer, S. J.; Ikura, M.; Torchia, D. A.; Bax, A. An alternative
2, 679−689. 3D NMR technique for correlating backbone 15N with side chain H
(21) Grasberger, B. L.; Clore, G. M.; Gronenborn, A. M. High- resonances in larger proteins. J. Magn. Reson. 1991, 95, 636−641.
resolution structure of Ascaris trypsin inhibitor in solution: direct (41) Grzesiek, S.; Ikura, M.; Clore, G.; Gronenborn, A. M.; Bax, A. A
evidence for a pH-induced conformational transition in the reactive 3D triple-resonance NMR technique for qualitative measurement of
carbonyl-Hb J couplings in isotopically enriched proteins. J. Magn.
site. Structure 1994, 2, 669−678.
Reson. (1969) 1992, 96, 215−221.
(22) Cierpicki, T.; Bania, J.; Otlewski, J. NMR solution structure of
(42) Dingley, A. J.; Nisius, L.; Cordier, F.; Grzesiek, S. Direct
Apis mellifera chymotrypsin/cathepsin G inhibitor-1 (AMCI-1):
detection of NHN hydrogen bonds in biomolecules by NMR
Structural similarity with Ascaris protease inhibitors. Protein Sci.
spectroscopy. Nat. Protoc. 2008, 3, 242−248.
2000, 9, 976−984.
(43) Lipari, G.; Szabo, A. Model-free approach to the interpretation
(23) Duggan, B. M.; Dyson, H. J.; Wright, P. E. Inherent flexibility in
of nuclear magnetic resonance relaxation in macromolecules. 1.
a potent inhibitor of blood coagulation, recombinant nematode
Theory and range of validity. J. Am. Chem. Soc. 1982, 104, 4546−
anticoagulant protein c2. Eur. J. Biochem. 1999, 265, 539−548.
4559.
(24) Rosengren, K. J.; Daly, N. L.; Scanlon, M. J.; Craik, D. J.
(44) Mandel, A. M.; Akke, M.; Palmer, A. G., III Backbone
Solution structure of BSTI: a new trypsin inhibitor from skin Dynamics of Escherichia coli Ribonuclease HI: Correlations with
secretions of Bombina bombina. Biochemistry 2001, 40, 4601−4609. Structure and Function in an Active Enzyme. J. Mol. Biol. 1995, 246,
(25) Shiltagh, N.; Kirkpatrick, J.; Cabrita, L. D.; McKinnon, T. A.; 144−163.
Thalassinos, K.; Tuddenham, E. G.; Hansen, D. F. Solution structure (45) Cole, R.; Loria, J. P. FAST-Modelfree: a program for rapid
of the major factor VIII binding region on von Willebrand factor. automated analysis of solution NMR spin-relaxation data. J. Biomol.
Blood 2014, 123, 4143−4151. NMR 2003, 26, 203−213.
(26) Thekkiniath, J. C.; Zabet-Moghaddam, M.; Francisco, S. K. S.; (46) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.;
Francisco, M. J. S. A novel subtilisin-like serine protease of Bax, A. NMRPipe: a multidimensional spectral processing system
Batrachochytrium dendrobatidis is induced by thyroid hormone and based on UNIX pipes. J. Biomol. NMR 1995, 6, 277−293.
degrades antimicrobial peptides. Fungal Biol. 2013, 117, 451−461. (47) Vranken, W. F.; Boucher, W.; Stevens, T. J.; Fogh, R. H.; Pajon,
(27) Eguchi, M.; Itoh, M.; Chou, L.-Y.; Nishino, K. Purification and A.; Llinas, M.; Ulrich, E. L.; Markley, J. L.; Ionides, J.; Laue, E. D. The
characterization of a fungal protease specific protein inhibitor (FPI-F) CCPN data model for NMR spectroscopy: development of a software
in the silkworm haemolymph. Comp. Biochem. Physiol., Part B: Comp. pipeline. Proteins: Struct., Funct., Bioinf. 2005, 59, 687−696.
Biochem. 1993, 104, 537−543. (48) Shen, Y.; Delaglio, F.; Cornilescu, G.; Bax, A. TALOS+: a
(28) Sasaki, S. D.; de Lima, C. A.; Lovato, D. V.; Juliano, M. A.; hybrid method for predicting protein backbone torsion angles from
Torquato, R. J.; Tanaka, A. S. BmSI-7, a novel subtilisin inhibitor from NMR chemical shifts. J. Biomol. NMR 2009, 44, 213−223.
Boophilus microplus, with activity toward Pr1 proteases from the (49) Güntert, P. Structure calculation of biological macromolecules
fungus Metarhizium anisopliae. Exp. Parasitol. 2008, 118, 214−220. from NMR data. Q. Rev. Biophys. 1998, 31, 145−237.
(29) Wang, Y.-W.; Tan, J.-M.; Du, C.-W.; Luan, N.; Yan, X.-W.; Lai, (50) Schwieters, C. D.; Kuszewski, J. J.; Tjandra, N.; Clore, G. M.
R.; Lu, Q.-M. A novel trypsin inhibitor-like cysteine-rich peptide from The Xplor-NIH NMR molecular structure determination package. J.
the frog Lepidobatrachus laevis containing proteinase-inhibiting Magn. Reson. 2003, 160, 65−73.
activity. Nat. Prod. Bioprospect. 2015, 5, 209−214. (51) Tejero, R.; Snyder, D.; Mao, B.; Aramini, J. M.; Montelione, G.
(30) Budgett, J. S. Notes on the Batrachians of the Paraguayan T. PDBStat: a universal restraint converter and restraint analysis
Chaco, with observations upon their breeding habits and develop- software package for protein NMR. J. Biomol. NMR 2013, 56, 337−
ment, especially with regard to Phyllomedusa hypochondrialis, Cope. 351.
Also a Description of a new Genus. Q. J. Microsc. Sci. 1899, 42, 305− (52) Koradi, R.; Billeter, M.; Wüthrich, K. MOLMOL: a program for
333. display and analysis of macromolecular structures. J. Mol. Graphics
(31) Dominguez, C.; Boelens, R.; Bonvin, A. M. HADDOCK: a 1996, 14, 51−55.
protein- protein docking approach based on biochemical or (53) Schrödinger, L.; DeLano, W. PyMOL Molecular Graphics
biophysical information. J. Am. Chem. Soc. 2003, 125, 1731−1737. System, version 2; PyMOL, 2020.

Q https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX
Biochemistry pubs.acs.org/biochemistry Article

(54) Waudby, C. A.; Ramos, A.; Cabrita, L. D.; Christodoulou, J. (74) Derache, C.; Epinette, C.; Roussel, A.; Gabant, G.; Cadene, M.;
Two-dimensional NMR lineshape analysis. Sci. Rep. 2016, 6, Korkmaz, B.; Gauthier, F.; Kellenberger, C. Crystal structure of
No. 24826. greglin, a novel non-classical Kazal inhibitor, in complex with
(55) Feng, C.; Kovrigin, E. L.; Post, C. B. NmrLineGuru: Standalone subtilisin. FEBS J. 2012, 279, 4466−4478.
and User-Friendly GUIs for Fast 1D NMR Lineshape Simulation and (75) McPhalen, C. A.; James, M. N. Structural comparison of two
Analysis of Multi-State Equilibrium Binding Models. Sci. Rep. 2019, 9, serine proteinase-protein inhibitor complexes: eglin-c-subtilisin
No. 16023, DOI: 10.1038/s41598-019-52451-8. Carlsberg and CI-2-subtilisin Novo. Biochemistry 1988, 27, 6582−
(56) Bode, W.; Papamokos, E.; Musil, D. The high-resolution X-ray 6598.
crystal structure of the complex formed between subtilisin Carlsberg (76) Takeuchi, Y.; Satow, Y.; Nakamura, K. T.; Mitsui, Y. Refined
and eglin c, an elastase inhibitor from the leech Hirudo medicinalis: crystal structure of the complex of subtilisin BPN and Streptomyces
Structural analysis, subtilisin structure and interface geometry. Eur. J. subtilisin inhibitor at 1 · 8Åresolution. J. Mol. Biol. 1991, 221, 309−
Biochem. 1987, 166, 673−692. 325.
(57) Laskowski, R. A.; Swindells, M. B. LigPlot+: Multiple Ligand− (77) Maynes, J. T.; Cherney, M. M.; Qasim, M.; Laskowski, M.;
Protein Interaction Diagrams for Drug Discovery. J. Chem. Inf. Model. James, M. N. Structure of the subtilisin Carlsberg−OMTKY3 complex
2011, 51, 2778−2786. reveals two different ovomucoid conformations. Acta Crystallogr., Sect.
(58) Adcock, S. A.; McCammon, J. A. Molecular dynamics: survey of D: Biol. Crystallogr. 2005, 61, 580−588.
methods for simulating the activity of proteins. Chem. Rev. 2006, 106, (78) Takeuchi, Y.; Noguchi, S.; Satow, Y.; Kojima, S.; Kumagai, I.;
1589−1615. Miura, K.-i.; Nakamura, K. T.; Mitsui, Y. Molecular recognition at the
(59) Hornak, V.; Abel, R.; Okur, A.; Strockbine, B.; Roitberg, A.; active site of subtilisin BPN’: crystallographic studies using genetically
engineered proteinaceous inhibitor SSI (Streptomyces subtilisin
Simmerling, C. Comparison of multiple Amber force fields and
inhibitor). Protein Eng., Des. Sel. 1991, 4, 501−508.
development of improved protein backbone parameters. Proteins:
(79) Remaut, H.; Waksman, G. Protein−protein interaction through
Struct., Funct., Bioinf. 2006, 65, 712−725.
β-strand addition. Trends Biochem. Sci. 2006, 31, 436−444.
(60) Homeyer, N.; Gohlke, H. Free energy calculations by the
(80) Badaczewska-Dawid, A. E.; Nithin, C.; Wroblewski, K.;
molecular mechanics Poisson- Boltzmann surface area method. Mol. Kurcinski, M.; Kmiecik, S. MAPIYA contact map server for
Inf. 2012, 31, 114−122. identification and visualization of molecular interactions in proteins
(61) Kumari, R.; Kumar, R.; Consortium, O. S. D. D.; Lynn, A. and biological complexes. Nucleic Acids Res. 2022, 50, W474−W482.
g_mmpbsa-A GROMACS tool for high-throughput MM-PBSA (81) Kojima, S.; Nishiyama, Y.; Kumagai, I.; Miura, K.-i. Inhibition
calculations. J. Chem. Inf. Model. 2014, 54, 1951−1962. of subtilisin BPN’ by reaction site P1 mutants of Streptomyces
(62) Robert, X.; Gouet, P. Deciphering key features in protein subtilisin inhibitor. J. Biochem. 1991, 109, 377−382.
structures with the new ENDscript server. Nucleic Acids Res. 2014, 42, (82) Cabrera-Muñoz, A.; Valiente, P. A.; Rojas, L.; Antigua, M. A.-
W320−W324. d.-R.; Pires, J. R. NMR structure of CmPI-II, a non-classical Kazal
(63) Micsonai, A.; Moussong, É .; Wien, F.; Boros, E.; Vadászi, H.; protease inhibitor: understanding its conformational dynamics and
Murvai, N.; Lee, Y.-H.; Molnár, T.; Réfrégiers, M.; Goto, Y.; Tantos, subtilisin A inhibition. J. Struct. Biol. 2019, 206, 280−294.
Á .; Kardos, J. BeStSel: webserver for secondary structure and fold (83) Barrette-Ng, I. H.; Ng, K. K.-S.; Cherney, M. M.; Pearce, G.;
prediction for protein CD spectroscopy. Nucleic Acids Res. 2022, 50, Ryan, C. A.; James, M. N. Structural basis of inhibition revealed by a
W90−W98. 1:2 complex of the two-headed tomato inhibitor-II and subtilisin
(64) Schechter, I.; Berger, A. On the size of the active site in Carlsberg. J. Biol. Chem. 2003, 278, 24062−24071.
proteases. I. Papain. Biochem. Biophys. Res. Commun. 1967, 27, 157−
162.
(65) Li, Y.; Zhao, P.; Liu, S.; Dong, Z.; Chen, J.; Xiang, Z.; Xia, Q. A
novel protease inhibitor in Bombyx mori is involved in defense against
Beauveria bassiana. Insect Biochem. Mol. Biol. 2012, 42, 766−775.
(66) Kuzmič, P.; Elrod, K. C.; Cregar, L. M.; Sideris, S.; Rai, R.; Janc,
J. W. High-throughput screening of enzyme inhibitors: simultaneous
determination of tight-binding inhibition constants and enzyme
concentration. Anal. Biochem. 2000, 286, 45−50.
(67) Ebeling, W.; Hennrich, N.; Klockow, M.; Metz, H.; Orth, H.
D.; Lang, H. Proteinase K from Tritirachium album limber. Eur. J.
Biochem. 1974, 47, 91−97.
(68) Betzel, C.; Bellemann, M.; Pal, G.; Bajorath, J.; Saenger, W.;
Wilson, K. X-ray and model-building studies on the specificity of the
active site of proteinase K. Proteins: Struct., Funct., Genet. 1988, 4,
157−164.
(69) Sharma, D.; Rajarathnam, K. 13C NMR chemical shifts can
predict disulfide bond formation. J. Biomol. NMR 2000, 18, 165−171.
(70) Dorman, D. E.; Bovey, F. A. Carbon-13 magnetic resonance
spectroscopy. Spectrum of proline in oligopeptides. J. Org. Chem.
1973, 38, 2379−2383.
(71) Richardson, J. S.; Getzoff, E. D.; Richardson, D. C. The beta
bulge: a common small unit of nonrepetitive protein structure. Proc.
Natl. Acad. Sci. U.S.A. 1978, 75, 2574−2578.
(72) Holm, L. Dali server: structural unification of protein families.
Nucleic Acids Res. 2022, 50, W210−W215.
(73) Dong, X.; Leksa, N. C.; Chhabra, E. S.; Arndt, J. W.; Lu, Q.;
Knockenhauer, K. E.; Peters, R. T.; Springer, T. A. The von
Willebrand factor DD3 assembly and structural principles for factor
VIII binding and concatemer biogenesis. Blood 2019, 133, 1523−
1533.

R https://doi.org/10.1021/acs.biochem.3c00252
Biochemistry XXXX, XXX, XXX−XXX

You might also like