Te

Sa

b

c

a

ARRA

KTAPS

1

sMue2mat12

eHbSC

v

h0

Molecular Immunology 62 (2014) 122–128

Contents lists available at ScienceDirect

Molecular Immunology

j ourna l ho me pa g e : www.elsev ier .com/ locate /mol imm

he P2/P2′ sites affect the substrate cleavage of TNF-� convertingnzyme (TACE)

en Liua,b,∗, Song Liua,b, Yanlin Wanga,b, Zhaojiang Liaoc

Institute of Molecular Biology, China Three Gorges University, Yichang 443002, PR ChinaCollege of Medical Science, China Three Gorges University, Yichang 443002, PR ChinaCollege of Biological and Pharmaceutical Science, China Three Gorges University, Yichang 443002, PR China

r t i c l e i n f o

rticle history:eceived 14 April 2014eceived in revised form 28 May 2014ccepted 29 May 2014

eywords:NF-� converting enzyme

a b s t r a c t

Tumor necrosis factor-alpha converting enzyme (TACE) is a proteinase that releases over eighty solubleproteins from their membrane-bound forms, and it has long been an intriguing therapeutic target inauto-immune diseases, and recently, in cancers. However, a haunting question is how TACE recognizesits substrates. In this work, we applied computational and experimental methods to study the role of theP2 site and the P2′ site of the substrate peptide in the substrate cleavage of TACE. In the computationalcomplex model, the sidechains of these residues do not form key interactions with TACE, but experimen-

DAM17eptide–protein dockingubstrate recognition

tally, the mutations at these two positions largely affect the peptide cleavage efficiency in the enzymaticassay. We then showed that the P2/P2′ sites could affect the efficiency of the conformation search for thecorrect peptide orientation, which in turn affects the substrate cleavage efficiency. Our result providesnew information to the better understanding of the enzymatic mechanism of TACE, and could be usefulin the design of novel TACE inhibitors.

© 2014 Elsevier Ltd. All rights reserved.

. Introduction

TACE (TNF-� converting enzyme) is an enzyme that releasesoluble TNF-� from its membrane-bound form (Black et al., 1997;oss et al., 1997), and has been found to be able to release the sol-

ble form of about another 80 proteins (Black et al., 1997; Schellert al., 2011; Moss et al., 1997; Gooz, 2010; Stephenson and Avis,012). Therefore, TACE has been a highly attractive target in inflam-atory diseases (Lichtenthaler, 2012; Scheller et al., 2011; Saftig

nd Reiss, 2011; Bahia and Silakari, 2010) and, recently, in cancerherapy (Black et al., 1997; Guinea-Viniegra et al., 2012; Moss et al.,997; Richards et al., 2012; Scheller et al., 2011; Saftig and Reiss,011; Baumgart et al., 2010).

TACE, also called ADAM17, is one of the most well-studiednzymes in the ADAM (a disintegrin and metalloproteinase) family.owever, the mechanism of substrate recognition and processing

y TACE remains elusive (Scheller et al., 2011; Gooz, 2010;tephenson and Avis, 2012). Caescu et al. (Lichtenthaler, 2012;aescu et al., 2009; Scheller et al., 2011; Saftig and Reiss, 2011;

∗ Corresponding author at: College of Medical Science, China Three Gorges Uni-ersity, Yichang 443002, PR China. Tel.: +86 717 6397179; fax: +86 717 6397179.

E-mail address: senliu.ctgu@gmail.com (S. Liu).

ttp://dx.doi.org/10.1016/j.molimm.2014.05.017161-5890/© 2014 Elsevier Ltd. All rights reserved.

Bahia and Silakari, 2010) and Lambert et al. (2005) found that theresidue identities of the substrate peptide can affect the cleavageefficiency of TACE, but Wang et al. (2002) and Hinkle et al. (2004)suggested that the position of the cleavage site relative to the trans-membrane region and the first globular part of the protein is moredeterminant than the sequence of the cleavage site. It was even sug-gested that the substrate recognition and cleavage is governed byTACE’s interactions with some unfound adaptor proteins (Mohanet al., 2002). Therefore, a structural view of the complex betweenthe substrate peptide and TACE would be very helpful for a betterunderstanding of the mechanism (Hartmann et al., 2013), whereasan experimental complex structure is still not available.

Recent advances in computational modeling has provided agood alternative to understand the molecular mechanisms ofprotein–peptide specificity (Smith and Kortemme, 2011; Baboret al., 2011; Kaufmann et al., 2011; Smith and Kortemme, 2010;Goldschmidt et al., 2010; Walshe et al., 2009; Grigoryan et al.,2009; Mandell and Kortemme, 2009; Kota et al., 2009; Humphrisand Kortemme, 2008; Fu et al., 2007) and enzyme–peptide activity(Chaudhury and Gray, 2009; London et al., 2011). Computationally,Manzetti et al. (2003) modeled substrate-enzyme complexes forADAM-9 and ADAM-10 to conclude that the S1′ pocket and the

S2/S3 region of the enzymes dominate the substrate specificity.Therefore, when an experimental complex structure is currentlyabsent, computational modeling could be a good way to provide

muno

sr

capoetfo

2

2

TT(PeI(cRcawgrFaa

2

u1tInCpcmazaswf(w

2

sSaifAB

S. Liu et al. / Molecular Im

ome useful clues to the molecular mechanism of the substrateecognition of TACE.

In this study, we computationally modeled and optimized theomplex structure of TACEcat (the catalytic domain of TACE) and

substrate peptide, based on a preliminary docking analysis by usreviously (Liu, 2012). We noticed that the P2 and the P2′ sitesf the substrate peptide distinctly affect the docking results, andxperimentally verified that the P2/P2′ sites have big impacts onhe substrate cleavage by TACE. Finally, we showed that the con-ormation search of the substrate peptide for the correct bindingrientation could be the reason.

. Methods

.1. TACEcat-peptide docking

The substrate peptide was docked to the catalytic domain ofACE (TACEcat) in the protein design software Rosetta (version 3.4).he peptide docking process was similar to previously describedLiu, 2012). Briefly, a 9-residue peptide (corresponding to the5–P4′ sites of the recognizing peptide) was prepared as a lin-ar chain and then docked to the structure of TACEcat (from PDBD: 1BKC) using the FlexPepDocking-AbInitio protocol in RosettaRaveh et al., 2011). During the docking steps, the catalytic zinc wasoordinated by three His residues (the residues 187, 191, and 197 inosetta numbering). To keep the geometry of the catalytic center,onstraints were applied on the distances between the zinc atomnd the NE2 atoms of the three coordinating His residues (2.4 Aith a variation of 0.2 A respectively), as well as the hydroxyl oxy-

en atom of the catalytic-water-coordinating glutamate acid (theesidue 188 in Rosetta numbering; 4.6 A with a variation of 0.3 A).inally, 54,000 docking models were generated for each peptide,nd the 500 lowest-score models were clustered for model evalu-tion.

.2. Molecular dynamics optimization of the complex model

The lowest-score model in the largest cluster from Rosetta wassed for molecular dynamics simulation in NAMD (GPU version.9) (Phillips et al., 2005). Periodic water (TIP3P) box was addedo wrap the complex structure with 10 A of boundary distances.ons (Na+ and Cl−) were added to 0.15 mol/L and counteracted theet charger of the water box including the protein complex. Theharmm parameters from c35b2 c36a2 were used, and the smootharticle-mesh Ewald (PME) method was enabled. To minimize theomplex model, a 3-step procedure was applied. First, 2000 steps ofinimization were applied with constraints applied on the heavy

toms of the proteins, followed by another 2000 steps of minimi-ation with constraints only applied on the C-alpha atoms. Finally,nother 2000 steps of minimization were run without any con-traints on the system. After the minimization step, the systemas equilibrated and a 10-ns molecular dynamics run was per-

ormed, and the atom coordinates were recorded per picosecondps). The analysis of the molecular dynamics trajectory and modelsere done in VMD (version 1.9.1) (Humphrey et al., 1996).

.3. In vitro real-time binding assay

Peptides were synthesized and covalently linked to bovineerum albumin (BSA) at the C-termini (ChinaPeptides Co., Ltd,hanghai, China) with a Cystein residue. The real-time bindingssay was performed on the BLItz system (ForteBio, PALL), which

s powered by BLI (bio-layer interferometry), a powerful label-ree assay technology. The catalytic domain of TACE (residuesrg215-Asn671; R&D Systems) was biotinylated using NHS-LC-iotin (succinimidyl-6-(biotinamido)hexanoate; Sangon Biotech

logy 62 (2014) 122–128 123

(Shanghai) Co., Ltd.) at a 1:2 molar ratio for 2 h at 4 ◦C in phosphate-buffered saline (PBS: 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4,2 mM KH2PO4, pH 7.4). A streptavidin-coated (SA) biosensor (Forte-Bio, PALL) was mounted to the BLItz system and used to immobilizebiotinylated TACEcat after being hydrated for 10 min in PBS. Thebinding buffer (10 mM Hepes, 100 mM NaCl, pH 7.4, 0.05% (v/v)Tween-20) was used to dissolve the peptide samples as well as forthe real-time binding assay. The binding assay was performed at25 ◦C with 600 s of association and 420 s of disassociation. The con-centration of the peptides was 10 �M, and BSA was used as theblank control, and the response signal was subtracted in the finalanalysis.

2.4. Enzymatic cleavage of the peptides

The peptides were synthesized and assayed for cleavage by TACEwith LC–MS. Synthetic peptide stock solutions were prepared inAssay buffer (25 mM Tris–HCl, 2.5 �M ZnCl2, 0.005% Brij-35; pH9.0) at a concentration of 1 mM respectively. The catalytic domain ofTACE (residues Arg215-Asn671) was purchased from R&D Systems,Inc., Minneapolis, USA. Both protein and peptide substrates wereincubated at 500 �M final concentrations with TACE at 2 ng/�L inAssay buffer at 37 ◦C. Aliquots from each cleavage reaction weretaken at 3 h after incubation, and stopped by ice incubation. Thealiquots (20 �L/test) were analyzed by RP-HPLC on a SinochromODS-BP 4.6 mm × 150 mm column (Dalian Elite Analytical Instru-ments Co., Ltd.) at 25 ◦C, using a 15 min linear gradient from 0.1%TFA in water (Buffer A) to 0.1% TFA in acetonitrile (Buffer B) for eachpeptide.

3. Results

3.1. The peptide–protein docking analysis

As described in Section 2, a nine-residue peptide from proTNF-�(referred as TNFtide hereafter), PLAQA|VRSS, was docked to TACE-cat in Rosetta. Bertini et al. (2006) had previously done a beautifulwork in which they captured several transition states of a matrixmetalloproteinase, MMP-12, with its peptide product. Matrix met-alloproteinases (MMPs) are a family of proteins involved in cellsignaling and tissue remodeling, and the zinc-dependent activecenter of MMPs shows remarkable homology with that of TACE,including the conserved amino acid sequence of -HExGHxxGxxH-(Murphy and Lee, 2005). Therefore, the enzymatic mechanism ofTACE could be similar to that of MMPs, and indeed, some substratescan be cleaved by both TACE and MMPs, and some MMP inhibitorscan also inhibit the activity of TACE (de Meijer et al., 2010; DasGuptaet al., 2009; Pirard and Matter, 2006). So we reasoned that the pre-sented models by Bertini et al. should be a good reference for TACEtoo. In their model of the enzyme/peptide complex, the carbonicoxygen of the P1 residue of the peptide substrate was coordinatedby the zinc atom with a distance of around 2.4 A between them,which is also true in the available complex structures of TACE withother molecules in PDB and other similar proteases (Lingott et al.,2009). So, in addition to the Rosetta scores of the models (Ravehet al., 2011), the following extra two rules were applied to pick outthe reasonable complex model: first, the distance between the zincatom and the carbonic oxygen of the P1 residue of the peptide isless than 2.8 A; second, the P1′ residue of the peptide locates in theS1′ pocket of TACEcat, which is known well as the requirement of

the activity (Black et al., 2003; Caescu et al., 2009; Lambert et al.,2005; Yang et al., 2010; DasGupta et al., 2009). The docked complexmodel was further optimized with molecular dynamics modeling,which produced restrain-free complex models (Fig. 1).

124 S. Liu et al. / Molecular Immunology 62 (2014) 122–128

Fig. 1. (A) The 100 complex models from the molecular dynamics simulation are shown in gray. The starting docked model is shown in magenta. The TACEcat chains ares in sphr rom ler

csi(ome2cmaetLtt

3

stttancTarctrrcw

cleavage by TACE.

0 200 400 600 800 1000

0

0.1

0.2

0.3

0.4

0.5

Res

po

nse

Un

it (

RU

)

TNFtide (PLAQA−VRSS)

TENtide (PRYEA−YKMG)

Pep1 (PLAAA−VRSS)

Pep2 (PLAQA−VFSS)

hown in lines, and the TNFtide chains shown in sticks. The zinc atoms are shownepresentative model from the molecular dynamics simulation. The TNFtide runs feferences to color in text, the reader is referred to the web version of this article.)

Structurally, the peptide fits in the crevice running across theatalytic site of TACEcat. The peptide adopts an anti-parallel �-heet conformation against a nearby �-sheet of TACEcat, whichs quite similar to the docked models in ADAM-9 and ADAM-10Manzetti et al., 2003). As expected, the sidechain of the P1′ residuef the peptide goes into the S1′ pocket deeply, which mainly deter-ines the specificity of the substrate processing by TACE (Black

t al., 2003; Caescu et al., 2009; Lambert et al., 2005; Yang et al.,010; DasGupta et al., 2009). What interested us was the sidechainonformation of the P2 residue and the P2′ residue. In the complexodel, the sidechains of these two residues are largely exposed,

nd do not form highly favorable interactions with TACEcat. How-ver, it has been shown that distinct residue preferences exist athese two sites from peptide library screening (Caescu et al., 2009;ambert et al., 2005). Therefore, we were interested to know if thesewo sites really affect the substrate processing by TACE, and why ifhey do.

.2. The P2 and P2′ sites affect the substrate processing by TACE

To experimentally test the effect of the P2/P2′ sites on the sub-trate processing of TACE, we firstly tried introducing mutations athese sites and evaluating the binding affinities of different peptideso TACEcat. In one peptide, a Gln → Ala mutation was introduced athe P2 site (PLAAAVRSS; named as TNFtide4A); in another peptide,n Arg → Phe mutation was made at the P2′ position (PLAQAVFSS;amed as TNFtide7F). The TNFtide peptide was used as the positiveontrol, and a reported peptide that cannot be cleaved by TACE,ENtide (PRYEAYKMG) (Caescu et al., 2009), was used as the neg-tive control. The peptides were synthesized and subjected to theeal-time binding assay (Section 2). As shown in Fig. 2, at the sameoncentration, these peptides have different binding curves againsthe immobilized TACEcat. The negative control, TENtide, has a

ather low response, but TNFtide4A and TNFtide7F have higheresponses comparable to TNFtide. This result qualitatively indi-ates that, with one mutation at the P2 site or the P2′ site, TNFtideill have changed binding behaviors against TACEcat. Noticing that

eres. (B) The close view of the TNFtide located in the binding site of TACEcat in aft to right over the substrate-binding tunnel of TACEcat. (For interpretation of the

the dramatic difference between TENtide and the other peptides atthe turning point (600 s) of association/disassociation, we proposethat this difference is mainly caused by the cleavage of the boundpeptide by TACEcat, but this is subjected to further investigation.

We then tested the enzymatic cleavage of the peptides byTACEcat (Section 2). As shown in Fig. 3, TNFtide can be cleaved effi-ciently by TACEcat, but TENtide cannot, which is consistent with thereported result (Caescu et al., 2009). With one mutation in TNFtidesequence, both TNFtide4A and TNFtide7F show lower cleavage effi-ciency. Additionally, the other two peptides (PLAAAVFSS, named asTNFtide4A7F; PLAMAVMSS, named as TNFtide4M7M), which havemutations at both sites, show even lower activity than both single-mutation peptides. These results prove that the P2 site and the P2′

site are indeed very important for the substrate recognition and

Time (second)

Fig. 2. The binding curves of synthetic peptides to immobilized TACEcat. TNFtide isa most efficient substrate of TACE, and TENtide is a negative control peptide, whichis barely cleaved by TACE (Caescu et al., 2009).

S. Liu et al. / Molecular Immunology 62 (2014) 122–128 125

DC

BA

FE

Fig. 3. The enzymatic cleavage of synthetic peptides by TACEcat was evaluated by LC–MS. The calculated cleavage percentage: (A) TNFtide (PLAQAVRSS), 39.08%; (B) TENtide( VFSS),2

3

msptCsptwfdBsc3tiPwifcbc

4

i8

PRYEAYKMG), 2.67%; (C) TNFtide4A (PLAAAVRSS), 27.45%; (D) TNFtide7F (PLAQA.31%.

.3. Binding conformation search could play an important role

Since the cleavage position is in the middle of the P1/P1′ sites,eanwhile the P2/P2′ residues are relatively far away and the

idechains do not form clear interactions with TACEcat in the com-lex model, we supposed that the activity difference could be dueo the binding affinity difference of these peptides with TACEcat.omputationally, the affinities of similar peptide–protein complextructures could be compared by the interface score of the com-lex (Raveh et al., 2011; London et al., 2011). So we tried to see ifhe affinities of these peptide–TACEcat complexes are correlatedith the observed activities. As shown in Fig. 4, although the inter-

ace score distributions and scales of the peptide–TACEcat dockingecoys are similar, the energy funnels have distinct differences.efore docking, the extended peptide was placed around the activeite of TACEcat, and the RMSD value between the peptide in the finalomplex model and the initial peptide is calculated to be around.0–5.0 A. Compared to TNFtide, the low-score conformations ofhe mutated peptides shift to RMSD values larger than 5.0 A, thats, far away from the binding conformation. When another peptide,RAAAVKSP (named as TACEtide, which can be cleaved by TACEith higher cleavage efficiency than TNFtide (Caescu et al., 2009)),

s docked to TACEcat, this trend is more obvious. Therefore, the con-ormation search behavior of the peptide could be correlated to itsleavage efficiency. When the peptide is harder to find the correctinding conformation, the relative binding rate is slower, and theleavage efficiency is lower.

. Discussion

TACE is one of the most well studied enzymes in the ADAM fam-ly, and has been found to be responsible for the releasing of around0 proteins from their membrane-bound forms, such as TNF-�,

30.62%; (E) TNFtide4A7F (PLAAAVFSS), 26.94%; (F) TNFtide4M7M (PLAMAVMSS),

TGF-�, TNF receptor I, TNF receptor II, interleukin-6 receptor, l-selectin, and Notch (Scheller et al., 2011; Gooz, 2010; Stephensonand Avis, 2012). But how TACE recognizes the cleavage sequence ofthese proteins remains an unsolved question (Scheller et al., 2011;Hartmann et al., 2013).

Previous studies proved that the substrate recognition of TACEis not random, but with clear sequence preference. For example,Caescu et al. (2009) and Lambert et al. (2005) noticed obviousresidue preference at the P4–P4′ sites by using peptide libraries.However, no structural information was provided in their studiesto explain the sequence preference.

Recently, computational modeling has provided an excel-lent tool to understand protein–peptide specificity (Smith andKortemme, 2010; Kaufmann et al., 2011; Walshe et al., 2009;Grigoryan et al., 2009; Goldschmidt et al., 2010; Fu et al., 2007;Mandell and Kortemme, 2009; Kota et al., 2009; Humphris andKortemme, 2008; Smith and Kortemme, 2011; Babor et al., 2011)and elucidate enzyme–peptide specificity/activity mechanisms(Chaudhury and Gray, 2009; London et al., 2011). Therefore, wewanted to see if computational modeling could help to understandthe substrate recognition mechanism of TACE.

In this study, we made a complex model of a substrate pep-tide (TNFtide) and the catalytic domain of TACE (TACEcat) (Fig. 1).Overall, the substrate peptide shows a beta-strand conformation,which nicely forms an anti-parallel beta-sheet with a neigh-bor beta-strand of TACEcat. This conformation is similar to thepeptide–enzyme complex models of TACE-homologous proteinsMMP3, ADAM9 and ADAM10 (Manzetti et al., 2003). By formingthe beta-sheet conformation with TACEcat, the substrate peptide

can therefore have high binding affinity. More importantly, in thisway, the peptide can place the sidechain of the P1′ residue deeplyin the S1′ pocket, and bring the cleavage bond to the catalytic zincatom.

126 S. Liu et al. / Molecular Immunology 62 (2014) 122–128

Fig. 4. The docking energy landscapes of different peptides with TACEcat were obtained in FlexPepDocking-AbInitio protocol in Rosetta (Raveh et al., 2011). Thepeptide–protein interface scores (I sc) are plotted against the peptide RMSD values. The RMSD value shows the conformation variation of the peptide in the complex model,c SD vaT

ttacws

at(pscmfobfet

ompared to the initial extended peptide. The correct peptide conformation has RMNFtide4A7F; (F) TNFtide4M7M; (G) TACEtide.

From the TNFtide-TACEcat complex model, we were interestedo know if the P2 residue and the P2′ residue are important forhe substrate recognition of TACEcat, since the sidechains of themre highly exposed and do not show stable interactions with TACE-at in the model. By introducing mutations into these two sites,e found that these two positions can outstandingly affect the

ubstrate cleavage efficiency (Fig. 3).To understand the possible reason that these two positions

ffect the substrate cleavage efficiency, we computationally inves-igated the binding energies of different peptides with TACEcatFig. 4). The interface scores of different peptide-TACEcat com-lex models are similar, which means that the mutations at P2/P2′

ites do not severely affect the binding energy. This is actuallyonsistent with the sidechain conformation of these two residuesentioned above. Since the P2/P2′ sites are relatively far away

rom the cleavage bond between the P1/P1′ sites, we supposed thatnce the complex structure forms, the cleavage efficiency should

e independent of the sidechains of the P2/P2′ residues. There-ore, subjected to further investigation, we reasoned that a possiblexplanation is the P2/P2′ residues affect the substrate cleavage inhe substrate binding phase. Our computational analysis supported

lues around 3.0–5.0 A. (A) TNFtide; (B) TENtide; (C) TNFtide4A; (D) TNFtide7F; (E)

this speculation, showing that the mutations at the P2/P2′ sites canshift the conformation search behavior (Fig. 4). So the rate of theconformation search of the substrate peptide to find the correctbinding orientation could play an important role in the substrateprocessing of TACE.

Structurally, the substrate-binding site of TACEcat is highlyelectro-negative as computed with APBS (Baker et al., 2001) (Fig. 5),and it has been reported that this special electrostatic potential hasa great influence on the substrate specificity of TACE (Sagi and Milla,2008; Solomon et al., 2007; Solomon, 2004; Stone et al., 1999; Milla,1999). The introduced mutations in our peptide designs eliminatedthe polar sidechains of the P2/P2′ residues, which could make a bigdifference to the electrostatic interaction between the peptide andTACEcat. Therefore, we reasoned that the activity difference of theP2/P2′ mutant peptides is largely affected by electrostatic steeredinter-molecular interactions (Wade et al., 1998), since various stepscan affect a enzyme catalytic cycle, including the formation of a

coupled interactions that bring the substrate closer, orienting itproperly, and providing a favorable electrostatic environment (Sagiand Milla, 2008; Benkovic and Hammes-Schiffer, 2003). We shouldmention that, although the P2 residue in TACEtide is Ala, it seems

S. Liu et al. / Molecular Immuno

Fig. 5. The surface electrostatic potential of TACEcat was computed with APBS(Baker et al., 2001). The surface is shown as the van der Waals surface and col-os

ta

oTlasn

A

Lcb9U

R

B

B

B

B

B

B

B

red by potential on solvent accessible surface. The catalytic zinc atom is shown inphere.

hat the polar P4 residue (Arg) compensates the electrostatic inter-ction.

Taken together, our result provided a reasonable structural viewf the complex of the substrate peptide and the catalytic domain ofACE, and showed that the P2/P2′ sites affect the substrate cleavageikely through an electrostatic steered conformation search mech-nism. This study could be useful for better understanding of theubstrate recognition and specificity of TACE, and the design ofovel TACE inhibitors.

cknowledgements

The authors thank the Rosetta Community, as well as Dr.uhua Lai and Daqi Yu of Peking University for stimulating dis-ussions and general help. This work was financially supportedy the National Natural Science Foundation of China (21103098,1330113), Hubei Province (T201203), and China Three Gorgesniversity (2011071001, KJ2012B004, 2013PY054).

eferences

abor, M., Mandell, D.J., Kortemme, T., 2011. Assessment of flexible backbone pro-tein design methods for sequence library prediction in the therapeutic antibodyHerceptin-HER2 interface. Protein Sci. 20, 1082–1089.

ahia, M.S., Silakari, O., 2010. Tumor necrosis factor alpha converting enzyme: anencouraging target for various inflammatory disorders. Chem. Biol. Drug Des.75, 415–443.

aker, N.A., Sept, D., Joseph, S., Holst, M.J., McCammon, J.A., 2001. Electrostatics ofnanosystems: application to microtubules and the ribosome. Proc. Natl. Acad.Sci. U.S.A. 98, 10037–10041.

aumgart, A., Seidl, S., Vlachou, P., Michel, L., Mitova, N., Schatz, N., Specht, K., Koch, I.,Schuster, T., Grundler, R., Kremer, M., Fend, F., Siveke, J.T., Peschel, C., Duyster, J.,Dechow, T., 2010. ADAM17 regulates epidermal growth factor receptor expres-sion through the activation of notch1 in non-small cell lung cancer. Cancer Res.70, 5368–5378.

enkovic, S.J., Hammes-Schiffer, S., 2003. A perspective on enzyme catalysis. Science301, 1196–1202.

ertini, I., Calderone, V., Fragai, M., Luchinat, C., Maletta, M., Yeo, K.J., 2006. Snapshotsof the reaction mechanism of matrix metalloproteinases. Angew. Chem. Int. Ed.45, 7952–7955.

lack, R.A., Doedens, J.R., Mahimkar, R., Johnson, R., Guo, L., Wallace, A., Virca, D.,Eisenman, J., Slack, J., Castner, B., Sunnarborg, S.W., Lee, D.C., Cowling, R., Jin, G.,

logy 62 (2014) 122–128 127

Charrier, K., Peschon, J.J., Paxton, R., 2003. Substrate specificity and inducibilityof TACE (tumour necrosis factor alpha-converting enzyme) revisited: the Ala-Valpreference, and induced intrinsic activity. Biochem. Soc. Symp., 39–52.

Black, R.A., Rauch, C.T., Kozlosky, C.J., Peschon, J.J., Slack, J.L., Wolfson, M.F., Castner,B.J., Stocking, K.L., Reddy, P., Srinivasan, S., Nelson, N., Boiani, N., Schooley, K.A.,Gerhart, M., Davis, R., Fitzner, J.N., Johnson, R.S., Paxton, R.J., March, C.J., Cerretti,D.P., 1997. A metalloproteinase disintegrin that releases tumour-necrosis factor-alpha from cells. Nature 385, 729–733.

Caescu, C.I., Jeschke, G.R., Turk, B.E., 2009. Active-site determinants of substraterecognition by the metalloproteinases TACE and ADAM10. Biochem. J. 424,79–88.

Chaudhury, S., Gray, J.J., 2009. Identification of structural mechanisms of HIV-1 pro-tease specificity using computational peptide docking: implications for drugresistance. Structure 17, 1636–1648.

DasGupta, S., Murumkar, P.R., Giridhar, R., Yadav, M.R., 2009. Current perspective ofTACE inhibitors: a review. Bioorg. Med. Chem. 17, 444–459.

de Meijer, V.E., Sverdlov, D.Y., Popov, Y., Le, H.D., Meisel, J.A., Nosé, V., Schuppan,D., Puder, M., 2010. In: Rodrigues-Lima, F. (Ed.), Broad-spectrum matrix met-alloproteinase inhibition curbs inflammation and liver injury but aggravatesexperimental liver fibrosis in mice. PLoS ONE 5, e11256.

Fu, X., Apgar, J.R., Keating, A.E., 2007. Modeling backbone flexibility to achievesequence diversity: the design of novel alpha-helical ligands for Bcl-xL. J. Mol.Biol. 371, 1099–1117.

Goldschmidt, L., Teng, P.K., Riek, R., Eisenberg, D., 2010. Identifying the amylome,proteins capable of forming amyloid-like fibrils. Proc. Natl. Acad. Sci. 107,3487–3492.

Gooz, M., 2010. ADAM-17: the enzyme that does it all. Crit. Rev. Biochem. Mol. Biol.45, 146–169.

Grigoryan, G., Reinke, A.W., Keating, A.E., 2009. Design of protein-interaction speci-ficity gives selective bZIP-binding peptides. Nature 458, 859–864.

Guinea-Viniegra, J., Zenz, R., Scheuch, H., Jiménez, M., Bakiri, L., Petzelbauer, P., Wag-ner, E.F., 2012. Differentiation-induced skin cancer suppression by FOS, p53, andTACE/ADAM17. J. Clin. Invest. 122, 2898–2910.

Hartmann, M., Herrlich, A., Herrlich, P., 2013. Who decides when to cleave anectodomain? Trends Biochem. Sci. 38, 111–120.

Hinkle, C.L., Sunnarborg, S.W., Loiselle, D., Parker, C.E., Stevenson, M., Russell,W.E., Lee, D.C., 2004. Selective roles for tumor necrosis factor alpha-convertingenzyme/ADAM17 in the shedding of the epidermal growth factor receptor ligandfamily: the juxtamembrane stalk determines cleavage efficiency. J. Biol. Chem.279, 24179–24188.

Humphrey, W., Dalke, A., Schulten, K., 1996. VMD: visual molecular dynamics. J. Mol.Graph. 14, 33–8–27–8.

Humphris, E.L., Kortemme, T., 2008. Prediction of protein–protein interfacesequence diversity using flexible backbone computational protein design. Struc-ture 16, 1777–1788.

Kaufmann, K., Shen, N., Mizoue, L., Meiler, J., 2011. A physical model for PDZ-domain/peptide interactions. J. Mol. Model. 17, 315–324.

Kota, P., Summers, D.W., Ren, H.-Y., Cyr, D.M., Dokholyan, N.V., 2009. Identificationof a consensus motif in substrates bound by a Type I Hsp40. Proc. Natl. Acad. Sci.106, 11073–11078.

Lambert, M.H., Blackburn, R.K., Seaton, T.D., Kassel, D.B., Kinder, D.S., Leesnitzer,M.A., Bickett, D.M., Warner, J.R., Andersen, M.W., Badiang, J.G., Cowan, D.J., Gaul,M.D., Petrov, K.G., Rabinowitz, M.H., Wiethe, R.W., Becherer, J.D., McDougald,D.L., Musso, D.L., Andrews, R.C., Moss, M.L., 2005. Substrate specificity andnovel selective inhibitors of TNF-alpha converting enzyme (TACE) from two-dimensional substrate mapping. Comb. Chem. High Throughput Screen. 8,327–339.

Lichtenthaler, S.F., 2012. Sheddase gets guidance. Science 335, 179–180.Lingott, T., Schleberger, C., Gutiérrez, J.M., Merfort, I., 2009. High-resolution crys-

tal structure of the snake venom metalloproteinase BaP1 complexed with apeptidomimetic: insight into inhibitor binding. Biochemistry 48, 6166–6174.

Liu, S., 2012. A physical model for TACE/peptide interactions. Int. Proc. Chem. Biol.Environ. Eng. 29, 1–5.

London, N., Lamphear, C.L., Hougland, J.L., Fierke, C.A., Schueler-Furman, O., 2011.In: Keating, A.E. (Ed.), Identification of a novel class of farnesylation targets bystructure-based modeling of binding specificity. PLoS Comput. Biol. 7, e1002170.

Mandell, D.J., Kortemme, T., 2009. Backbone flexibility in computational proteindesign. Curr. Opin. Biotechnol. 20, 420–428.

Manzetti, S., McCulloch, D.R., Herington, A.C., van der Spoel, D., 2003. Modeling ofenzyme–substrate complexes for the metalloproteases MMP-3, ADAM-9 andADAM-10. J. Comput. Aided Mol. Des. 17, 551–565.

Milla, M.E., 1999. Specific sequence elements are required for the expression offunctional tumor necrosis factor-alpha-converting enzyme (TACE). J. Biol. Chem.274, 30563–30570.

Mohan, M.J., Seaton, T., Mitchell, J., Howe, A., Blackburn, K., Burkhart, W., Moyer,M., Patel, I., Waitt, G.M., Becherer, J.D., Moss, M.L., Milla, M.E., 2002. The tumornecrosis factor-� converting enzyme (TACE): a unique metalloproteinase withhighly defined substrate selectivity. Biochemistry 41, 9462–9469.

Moss, M.L., Jin, S.L., Milla, M.E., Bickett, D.M., Burkhart, W., Carter, H.L., Chen, W.J.,Clay, W.C., Didsbury, J.R., Hassler, D., Hoffman, C.R., Kost, T.A., Lambert, M.H.,Leesnitzer, M.A., McCauley, P., McGeehan, G., Mitchell, J., Moyer, M., Pahel, G.,

Rocque, W., Overton, L.K., Schoenen, F., Seaton, T., Su, J.L., Becherer, J.D., 1997.Cloning of a disintegrin metalloproteinase that processes precursor tumour-necrosis factor-alpha. Nature 385, 733–736.

Murphy, G., Lee, M.H., 2005. What are the roles of metalloproteinases in cartilageand bone damage? Ann. Rheum. Dis. 64, iv44–iv47.

1 muno

P

P

R

R

S

S

S

S

S

S

Yang, J.S., Chun, K., Park, J.E., Cho, M., Seo, J., Song, D., Yoon, H., Park, C.-H., Joe,

28 S. Liu et al. / Molecular Im

hillips, J.C., Braun, R., Wang, W., Gumbart, J., Tajkhorshid, E., Villa, E., Chipot, C.,Skeel, R.D., Kalé, L., Schulten, K., 2005. Scalable molecular dynamics with NAMD.J. Comput. Chem. 26, 1781–1802.

irard, B., Matter, H., 2006. Matrix metalloproteinase target family landscape: achemometrical approach to ligand selectivity based on protein binding siteanalysis. J. Med. Chem. 49, 51–69.

aveh, B., London, N., Zimmerman, L., Schueler-Furman, O., 2011. Rosetta FlexPep-Dock ab-initio: simultaneous folding, docking and refinement of peptides ontotheir receptors. PLoS ONE 6, e18934.

ichards, F.M., Tape, C.J., Jodrell, D.I., Murphy, G., 2012. Anti-tumour effects of aspecific anti-ADAM17 antibody in an ovarian cancer model in vivo. PLoS ONE 7,e40597.

aftig, P., Reiss, K., 2011. The “A Disintegrin And Metalloproteases” ADAM10 andADAM17: novel drug targets with therapeutic potential? Eur. J. Cell Biol. 90,527–535.

agi, I., Milla, M.E., 2008. Application of structural dynamic approaches providenovel insights into the enzymatic mechanism of the tumor necrosis factor-�-converting enzyme. Anal. Biochem. 372, 1–10.

cheller, J., Chalaris, A., Garbers, C., Rose-John, S., 2011. ADAM17: a molecular switchto control inflammation and tissue regeneration. Trends Immunol. 32, 380–387.

mith, C.A., Kortemme, T., 2011. Predicting the tolerated sequences for proteins andprotein interfaces using RosettaBackrub flexible backbone design. PLoS ONE 6,e20451.

mith, C.A., Kortemme, T., 2010. Structure-based prediction of the peptide sequencespace recognized by natural and synthetic PDZ domains. J. Mol. Biol. 402,

460–474.

olomon, A., 2004. Pronounced diversity in electronic and chemical propertiesbetween the catalytic zinc sites of tumor necrosis factor – converting enzymeand matrix metalloproteinases despite their high structural similarity. J. Biol.Chem. 279, 31646–31654.

logy 62 (2014) 122–128

Solomon, A., Akabayov, B., Frenkel, A., Milla, M.E., Sagi, I., 2007. Key feature of thecatalytic cycle of TNF-� converting enzyme involves communication betweendistal protein sites and the enzyme catalytic core. Proc. Natl. Acad. Sci. U.S.A.104, 4931.

Stephenson, N.L., Avis, J.M., 2012. Direct observation of proteolytic cleavage at theS2 site upon forced unfolding of the Notch negative regulatory region. Proc. Natl.Acad. Sci. 109, E2757–E2765.

Stone, A.L., Kroeger, M., Sang, Q.X., 1999. Structure-function analysis of the ADAMfamily of disintegrin-like and metalloproteinase-containing proteins (review).J. Protein Chem. 18, 447–465.

Wade, R.C., Gabdoulline, R.R., Lüdemann, S.K., Lounnas, V., 1998. Electrostatic steer-ing and ionic tethering in enzyme-ligand binding: insights from simulations.Proc. Natl. Acad. Sci. U.S.A. 95, 5942–5949.

Walshe, V.A., Hattotuwagama, C.K., Doytchinova, I.A., Wong, M., Macdonald,I.K., Mulder, A., Claas, F.H.J., Pellegrino, P., Turner, J., Williams, I., Turnbull,E.L., Borrow, P., Flower, D.R., 2009. In: Ostrowski, M.A. (Ed.), Integratingin silico and in vitro analysis of peptide binding affinity to HLA-Cw*0102:a bioinformatic approach to the prediction of new epitopes. PLoS ONE 4,e8095.

Wang, X., He, K., Gerhart, M., Huang, Y., Jiang, J., Paxton, R.J., Yang, S., Lu, C.,Menon, R.K., Black, R.A., Baumann, G., Frank, S.J., 2002. Metalloprotease-mediated GH receptor proteolysis and GHBP shedding. Determinationof extracellular domain stem region cleavage site. J. Biol. Chem. 277,50510–50519.

B.-Y., Choi, J.-H., Kim, M.-H., Han, G., 2010. Structure based optimization ofchromen-based TNF-� converting enzyme (TACE) inhibitors on S1′ pocket andtheir quantitative structure-activity relationship (QSAR) study. Bioorg. Med.Chem. 18, 8618–8629.