International Journal of Molecular Sciences

Article Structure–Activity Relationship of RGD-Containing Cyclic Octapeptide and αvβ3 Integrin Allows for Rapid Identification of a New Peptide Antagonist

Aaron Silva 1, Wenwu Xiao 2, Yan Wang 3, Wei Wang 3, Heng Wei Chang 3, James B. Ames 4, Kit S. Lam 2 and Yonghong Zhang 1,*

1 Department of Chemistry, The University of Texas Rio Grande Valley, Edinburg, TX 78539, USA; [email protected] 2 Department of Biochemistry and Molecular Medicine, University of California, Davis Cancer Center, Sacramento, CA 95616, USA; [email protected] (W.X.); [email protected] (K.S.L.) 3 CSBio Company Inc., Menlo Park, CA 94025, USA; [email protected] (Y.W.); [email protected] (W.W.); [email protected] (H.W.C.) 4 Department of Chemistry, University of California, Davis, CA 95616, USA; [email protected] * Correspondence: [email protected]; Tel.: +1-956-665-2288 (O)



Received: 22 March 2020; Accepted: 24 April 2020; Published: 27 April 2020


Abstract: The αvβ3 integrin, a receptor for many extracellular matrix proteins with RGD-sequence motif, is involved in multiple physiological processes and highly expressed in tumor cells, therefore making it a target for cancer therapy and tumor imaging. Several RGD-containing cyclic octapeptide (named LXW analogs) were screened as αvβ3 antagonists with dramatically different binding affinity, and their structure–activity relationship (SAR) remains elusive. We performed systematic SAR studies and optimized LXW analogs to improve antagonistic potency. The NMR structure of LXW64 was determined and docked to the integrin. Structural comparison and docking studies suggested that the hydrophobicity and aromaticity of the X7 amino acid are highly important for LXW analogs binding to the integrin, a potential hydrophobic pocket on the integrin surface was proposed to play a role in stabilizing the peptide binding. To develop a cost-efficient and fast screening method, computational docking was performed on LXW analogs and compared with in vitro screening. A consistency within the results of both methods was found, leading to the continuous optimization and testing of LXW mutants via in silico screening. Several new LXW analogs were predicted as the integrin antagonists, one of which—LXZ2—was validated by in vitro examination. Our study provides new insight into the RGD recognition specificity and valuable clues for rational design of novel αvβ3 antagonists.

Keywords: integrin αvβ3 antagonists; RGD peptides; structure–activity relationship; in silico screening; in vitro binding

1. Introduction Integrin αvβ3 known as the vitronectin receptor is a member of the integrin superfamily and a heterodimeric transmembrane protein formed by non-covalent association of αv and β3 subunits. Each subunit consists of a large extracellular domain, a single transmembrane domain, and a short cytoplasmic domain, through which the integrin modulates bi-directional cell signaling over the plasma membrane [1]. As a cell surface receptor of the extracellular matrix (ECM), it binds a wide variety of ECM ligands with RGD motif implicated in many normal and pathological cell functions including cell survival, angiogenesis, tumor invasion, etc. [2]. Unlike other integrins ubiquitously expressed in adult tissues, αvβ3 is most abundantly expressed on angiogenic endothelial cells in pathological tissues [3].

Int. J. Mol. Sci. 2020, 21, 3076; doi:10.3390/ijms21093076 www.mdpi.com/journal/ijms Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 2 of 17

Int.angiogenic J. Mol. Sci. 2020 therapy, 21, 3076 with growing interest in developing inhibitors specifically targeting2 ofαv 17β3 integrin in the past decade [4]. RGD peptides are well-known to bind to the integrins including αvβ3 as revealed by the crystal In fact, the inhibition of αvβ3 has been widely used in clinical trials as anti-angiogenic therapy with structure of αvβ3 ectodomain in complex with the cyclic RGD peptide—cilengitide [5]—which growing interest in developing inhibitors specifically targeting αvβ3 integrin in the past decade [4]. provided the structural basis for development of αvβ3 antagonists. Targeting tumor cells or tumor RGD peptides are well-known to bind to the integrins including αvβ3 as revealed by the crystal vasculature by RGD-based strategies is a promising approach for delivering anticancer drugs or structure of αvβ3 ectodomain in complex with the cyclic RGD peptide—cilengitide [5]—which provided contrast agents for cancer therapy and diagnosis. RGD-based strategies include RGD peptide or the structural basis for development of αvβ3 antagonists. Targeting tumor cells or tumor vasculature peptidomimetic drugs, RGD-conjugates, and the grafting of the RGD peptide or peptidomimetic, as by RGD-based strategies is a promising approach for delivering anticancer drugs or contrast agents for targeting ligand, at the surface of nanocarriers [6]. A series of RGD-containing cyclic octapeptides— cancer therapy and diagnosis. RGD-based strategies include RGD peptide or peptidomimetic drugs, LXW analogs—have been reported using the one-bead-one-compound (OBOC) combinatorial library RGD-conjugates, and the grafting of the RGD peptide or peptidomimetic, as targeting ligand, at the technology (Table 1) [7,8]. LXW7 was identified as a leading ligand that binds specifically to αvβ3 surface of nanocarriers [6]. A series of RGD-containing cyclic octapeptides—LXW analogs—have been integrin with a comparable binding affinity to those well-known RGD-cyclic pentapeptide ligands reported using the one-bead-one-compound (OBOC) combinatorial library technology (Table1)[ 7,8]. [7]. A further systematic optimization of LXW analogs was conducted and led to identification of LXW7 was identified as a leading ligand that binds specifically to αvβ3 integrin with a comparable several more potent LXW peptides [8]. One of the best ligands-LXW64 demonstrated 6-fold higher binding affinity to those well-known RGD-cyclic pentapeptide ligands [7]. A further systematic binding affinity than LXW7 and identified as the new lead [8]. However, the SAR remains elusive as optimization of LXW analogs was conducted and led to identification of several more potent LXW these LXW analogs share similar structures but exhibit substantially different integrin binding peptides [8]. One of the best ligands-LXW64 demonstrated 6-fold higher binding affinity than LXW7 affinities. Furthermore, this in vitro screening procedure requires considerable efforts such as and identified as the new lead [8]. However, the SAR remains elusive as these LXW analogs share synthesis of OBOC libraries, on-bead whole-cell screening assays, etc. which is often time-consuming similar structures but exhibit substantially different integrin binding affinities. Furthermore, this and costly. To develop and apply a rapid, low-cost in silico screening method combining with in vitro screening procedure requires considerable efforts such as synthesis of OBOC libraries, on-bead selective in vitro validation would be a better way for this purpose. As a continuation of our previous whole-cell screening assays, etc. which is often time-consuming and costly. To develop and apply a efforts in developing αvβ3 antagonists, we herein introduce a combinatorial method, report rapid, low-cost in silico screening method combining with selective in vitro validation would be a identification of a new RGD-containing cyclic octapeptide against αvβ3 integrin. Its high binding better way for this purpose. As a continuation of our previous efforts in developing αvβ3 antagonists, affinity to the integrin has been validated using the competition binding assay on αvβ3 integrin- we herein introduce a combinatorial method, report identification of a new RGD-containing cyclic transfected cells (K562/αvβ3+). octapeptide against αvβ3 integrin. Its high binding affinity to the integrin has been validated using the competition binding assay on αvβ3 integrin-transfected cells (K562/αvβ3+). Table 1. Representative LXW analogs and their IC50s*.

Table 1. Representative LXW analogs and their IC s *.IC50 Peptide Amino Acid Sequence# 50 (μmoL/L) Peptide Amino Acid Sequence # IC (µmoL/L) LXW7 cGRGDdvc-NH2 50 0.46 LXW11LXW7 cGRGDdvc-NHCGRGDdvC-NH2 2 0.46>20 LXW64LXW11 cGRGDd- CGRGDdvC-NHDNal1-c-NH2 2 >200.07 LXW64 cGRGDd-DNal1-c-NH 0.07 * IC50 of the peptide is the concentration of peptide required for2 inhibition of 0.5 µmoL/L biotinylated * ICLXW750 of the binding peptide to is theK562/ concentrationαvβ3+ cells of peptideby half. required # The lowercase for inhibition letters of 0.5 indicateµmoL/L biotinylated D-amino acids, LXW7 bindingwhereas to K562/αvβ3+ cells by half. # The lowercase letters indicate D-amino acids, whereas the uppercase letters denote L-aminothe uppercase acids. letters denote L-amino acids.

2. Results 2. Results

2.1.2.1. NMR NMR Assignments Assignments of of LXW64 LXW64 and and Verification Verification of of Disulfide Disulfide Bond Bond

LXW64LXW64 ( ) contains contains 8 amino 8 amino acids acidsincluding including non-proteinogenic non-proteinogenic amino acid—3-(1- amino 1 acid—3-(1-naphthyl)-D-alaninenaphthyl)-D-alanine (D-Nal1). (D-Nal1).The 1H-coupling The H-coupling spin system spin for systemeach residue for each type residue was unique type was and 1 1 uniqueeasily distinguishable and easily distinguishable from 1H-1H fromTOCSY,H- forH example, TOCSY, forthe example,residue, arginine, the residue, was unambiguously arginine, was 1 β γ unambiguouslyassigned based assigned on its unique based 1H on resonances its unique ofH H resonancesβ and Hγ (1–2 of Hppm).and The H other(1–2 ppm).three types The other of residues, three typestwo glycines, of residues, two two aspartates, glycines, and two two aspartates, cysteines, and we twore also cysteines, identified were and also assigned identified with and their assigned HN, Hα, N α β withand their Hβ. The H ,Hsequential, and H assignment. The sequential was then assignment completed was through then completed the connectivity through of the NOEs connectivity observed 1 ofbetween NOEs observed the amide between protons the in amide 2D NOESY protons (Figure in 2D NOESY1). Thus, (Figure all 1H1 resonances). Thus, all wereH resonances unambiguously were unambiguouslyassigned for LXW64. assigned The for proton LXW64. assignments The proton al assignmentslowed unambiguous allowed unambiguous assignments assignmentsof all proton- 13 1 13 ofattached all proton-attached 13C resonancesC using resonances 1H-13C HMQC using H-spectrum,C HMQC while spectrum, the chemical while shifts the of chemical non-pronated shifts of13C 13 1 13 non-pronatedwere assignedC from were 1H- assigned13C HMBC from spectrumH- C HMBC due to spectrum their long-r dueange to their couplings long-range with couplings other assigned with 15 otherprotons. assigned Native protons. abundance Native gradient abundance 15N HSQC gradient spectrumN HSQC was obtained spectrum for was LXW64 obtained on the for 600 LXW64 MHz onBruker the 600 spectrometer. MHz Bruker The spectrometer. high-signal-to-noise The high-signal-to-noise quality of this qualityspectrum of thisenabled spectrum unambiguous enabled 15 1 13 15 unambiguousassignments for assignments all 15N chemical for all shifts.N chemical Chemical shifts. shifts Chemical of all 1H, shifts13C and of 15 allN wereH, Cfully and assignedN were for fully assigned for LXW64 peptide and are available as Supporting Information in Table S1. The

Int. J. Mol. Sci. 2020, 21, 3076 3 of 17

Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 3 of 17 intramolecular disulfide bond of the peptide was confirmed via the NOEs between two cysteines (Cys1 LXW64 peptide and are available as Supporting Information in Table S1. The intramolecular disulfide and Cys8). Moreover, the disulfide bridge connectivity was identified by MS and 13Cβ chemical shifts bond of the peptide was confirmed via the NOEs between two cysteines (Cys1 and Cys8). Moreover, of two cysteines [9]. the disulfide bridge connectivity was identified by MS and 13Cβ chemical shifts of two cysteines [9].

Figure 1. The observed NOEs of LXW64 in 2D 1H-1H NOESY (dashed lines for sequential NOEs; solid Figurelines for 1. long-rangeThe observed NOEs). NOEs of LXW64 in 2D 1H-1H NOESY (dashed lines for sequential NOEs; solid lines for long-range NOEs). 2.2. Structure Determination of LXW64 and Structural Comparison with Other LXW Peptides

2.2. Structure Determination of LXW64 and Structural Comparisonα with Other LXW Peptides N All sequential NOEs between α proton of residue i (H i) and amide proton of residue i+1 (H i+1) wereAll observed sequential in 2D NOEs NOESY, between as shown α proton in the of molecularresidue i (H structureαi) and amide of LXW64 proton (Figure of residue1). Interestingly, i+1 (HNi+1) α δ2 ε2 wereNOEs observed between in H2D5 NOESY,and the as aromatic shown in side-chain the molecular protons structure (H andof LXW64 H ) of(FigureD-Nal1 1). wasInterestingly, observed, NOEswhich between is similar Hα toLXW75 and the but aromatic missed side-chain in LXW7 protons isomer, (H i.e.,δ2 and LXW11 Hε2) [of8]. D-Nal1 The NMR-derived was observed, distance which isconstraints similar toLXW7 from2D but NOESY missed asin LX wellW7 as isomer, the dihedral i.e., LXW11 constraints [8]. The from NMR-derivedJ coupling constants distance constraints and carbon fromchemical 2D NOESY shifts (Tableas well2), as allow the dihedral us to perform constraints atomic from resolution J coupling structure constants calculation. and carbon Thechemical final shiftsNMR-derived (Table 2), structuresallow us to are perform illustrated atomic in Figure resolu2 tionand structure summarized calculation. in Table 2The. The final 10 lowest-energyNMR-derived structuresconformers are when illustrated superimposed in Figure have 2 and an overallsummarized main chainin Table root-mean-squared 2. The 10 lowest-energy derivation conformers (RMSD) of when0.34 superimposed0.072 Å. The energy-minimized have an overall main average chain structureroot-mean-squared of LXW64 derivation is shown in (RMSD) Figure2 ofB 0.34 (see ± for 0.072 the ± Å.structure The energy-minimized coordinate (S2 LXW64 average Coordinates)). structure of LXW64 The peptide is shown adopts in Figure a bowl-shape 2B (see Appendix and open B circularfor the structurestructure coordinate with all side (S2 chains LXW64 pointing Coordinates)). toward the The outside. peptide The adopts structure a bowl-shape (Figure2C) containsand open positively circular structure(L-Arg3) andwith negatively all side chains (L-Asp5 pointing and D-Asp6) toward charged the outside. portions The and structure hydrophobic (Figure moiety 2C) (containsD-Nal1), positivelywhich may (L-Arg3) contribute and tonegatively hydrophilic, (L-Asp5 electrostatic and D-Asp6) and hydrophobiccharged portions interactions and hydrophobic when binding moiety to (αDv-Nal1),β3 integrin. which may contribute to hydrophilic, electrostatic and hydrophobic interactions when bindingThe to structure αvβ3 integrin. of LXW64 is less compact and very similar to that of LXW7 since there is only oneThe amino structure acid di offference LXW64 at is position less compact 7 [8]. and The very backbone similar RMSD to that of of twoLXW7 peptides since there when is alignedonly one is amino1.4 Å, acid and alldifference side chains at position overlap 7 very[8]. The well backbone (Figure3 A).RMSD Both of peptides two peptides adopt when an open aligned ring-shaped is 1.4 Å, andstructure, all side in chains which overlap the side very chains well of(Figure two Asp 3A). residuesBoth peptides point adopt toward an the open outside ring-shaped ring plane, structure, while inArg3 which and theD -Val7side chains/D-Nal1 of protrude two Asp fromresidues the poin ringt structure toward the in oppositeoutside ring directions. plane, while However, Arg3 and LXW64 D- Val7/and LXW7D-Nal1 are protrude considerably from the diff ringerent structure from LXW11, in op withposite a directions. backbone RMSDHowever, of 2.5 LXW64 Å after and alignment. LXW7 areThe considerably bowl-like structures different offrom LXW64 LXW11, and with LXW7 a ba areckbone twisted RMSD in LXW11, of 2.5 resultingÅ after alignment. in drastic The changes bowl- of like structures of LXW64 and LXW7 are twisted in LXW11, resulting in drastic changes of side-chain orientations of amino acids (Figure 3B). The side chains of Asp5 and D-Val7 in LXW11 are totally buried inside the circular structure, quite close to the opposite amino acids—Gly2 and Arg3. The side

Int. J. Mol. Sci. 2020, 21, 3076 4 of 17 side-chain orientations of amino acids (Figure3B). The side chains of Asp5 and D-Val7 in LXW11 are Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 4 of 17 totally buried inside the circular structure, quite close to the opposite amino acids—Gly2 and Arg3. The side chain of D-Asp6 shows closer to the peptide backbone, and less protruding in comparison to chain of D-Asp6 shows closer to the peptide backbone, and less protruding in comparison to LXW64 LXW64and LXW7 and (Figure LXW7 3B). (Figure In addition,3B). In addition, two cysteines two cysteines (Cys1 and (Cys1 Cys8) and in Cys8) LXW11 in LXW11 are much are closer much to closer each toother each than other that than of that LXW64 of LXW64 and LXW7 and LXW7 (Figure (Figure 3C,D),3C,D), indicating indicating that that LXW11 LXW11 is is more more structurallystructurally constrained than LXW64 and LXW7. These significant significant conformational differences differences are likely to cause α β dramatic changes in their binding aaffinitiesffinities toto αvvβ3 integrin.

Table 2. Structure Statistics for the Ensembles of 10 Calculated Structures of LXW64. Table 2. Structure Statistics for the Ensembles of 10 Calculated Structures of LXW64.

NOENOE Restraints Restraints (Total) (Total) 58 dihedraldihedral angle angle restraints restraints 14 RMSDRMSD from from ideal ideal geometry geometry bondbond length length (Å) (Å) 0.0094 ± 0.00041 ± bond angles (degree) 2.27 0.04 bond angles (degree) 2.27 ± 0.04 RamachandranRamachandran plot plot allowed region (%) 100 allowed region (%) 100 disallowed region (%) 0 disallowed region (%) 0 RMSD of atom position from average structure RMSD of atom position from average structure main chain (Å) 0.34 0.072 main chain (Å) 0.34 ± 0.072 non-hydrogen (Å) 1.47 0.16 non-hydrogen (Å) 1.47 ± 0.16

Figure 2. NMR-derived structures of LXW64. (A) An ensemble of 10 superimposed minimum energy structures. (B) Sticks representation of the energy-minimized average structure, in which L-Arg3, Figure 2. NMR-derived structures of LXW64. (A) An ensemble of 10 superimposed minimum energy L-Asp5, D-Asp6 and D-Nal1 (X7) are labeled. (C) Representation of the molecular surface of LXW64, structures. (B) Sticks representation of the energy-minimized average structure, in which L-Arg3, L- colored by electrostatic potential. Blue and red color represents positive (Arg) and negative (Asp) Asp5, D-Asp6 and D-Nal1 (X7) are labeled. (C) Representation of the molecular surface of LXW64, potential, respectively. The hydrophobic portion is highlighted in yellow. colored by electrostatic potential. Blue and red color represents positive (Arg) and negative (Asp) potential, respectively. The hydrophobic portion is highlighted in yellow.

Int. J. Mol. Sci. 2020, 21, 3076 5 of 17 Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 5 of 17

FigureFigure 3.3. (A) Superimposed structuresstructures of LXW64 (grey) andand LXW7LXW7 (cyan)(cyan) withwith labeledlabeled sideside chainschains ofof L-Arg3,L-Arg3, L-Asp5,L-Asp5, D-Asp6D-Asp6 andand D-Nal1 (X7) (LXW64) or D-Val7 (LXW7);(LXW7); ((BB)) SticksSticks representationrepresentation ofof thethe energy-minimizedenergy-minimized averageaverage structurestructure ofof LXW11,LXW11, L-Arg3,L-Arg3, L-Asp5, D-Asp6, and D-Val7 are labeled; (C) α TheThe energy-minimized average structure structure of of LXW LXW6464 with with labeled labeled distance distance between between D-Cys1 D-Cys1 H Hα andand D- α D-Cys8Cys8 Hα H; (D;() DThe) The energy-minimized energy-minimized average average structure structure of LXW11 of LXW11 with with labeled labeled distances distances between between D- α α N α D-Cys1Cys1 Hα H andand D-Cys8 D-Cys8 Hα, HD-Val7, D-Val7 methyl methyl protons protons and L-Gly2 and L-Gly2 HN/L-Arg3 H /L-Arg3 Hα, L-Asp5 H , L-Asp5 carboxyl carboxyl group, N group,and L-Gly2 and L-Gly2HN. H . 2.3. Complex Structure Models of LXW Peptides and αvβ3 Integrin 2.3. Complex Structure Models of LXW Peptides and αvβ3 Integrin To further examine the SARs of LXW analogs binding to the αvβ3 integrin, the computational To further examine the SARs of LXW analogs binding to the αvβ3 integrin, the computational models of three LXW analogs (LXW64, LXW7, LXW11) bound with the integrin were generated by models of three LXW analogs (LXW64, LXW7, LXW11) bound with the integrin were generated by means of docking calculation using AutoDock Vina [10] (see Materials and Methods for a detailed means of docking calculation using AutoDock Vina [10] (see Materials and Methods for a detailed description), starting from the crystal structure of the extracellular segment of integrin αvβ3 in complex description), starting from the crystal structure of the extracellular segment of integrin αvβ3 in with cilengitide (PDB ID 1L5G). The energy-minimized average structures of LXW64, LXW7, and complex with cilengitide (PDB ID 1L5G). The energy-minimized average structures of LXW64, LXW11 were docked into the metal ion-dependent adhesion site (MIDAS), as shown in Figure4A. In LXW7, and LXW11 were docked into the metal ion-dependent adhesion site (MIDAS), as shown in contrast to the crystal structure, the Arg3 side chain of LXW64 is able to form salt bridges with Asp150 Figure 4A. In contrast to the crystal structure, the Arg3 side chain of LXW64 is able to form salt and Asp148, but not with Asp218 in the complex model, indicating that Asp218 in the integrin is not bridges with Asp150 and Asp148, but not with Asp218 in the complex model, indicating that Asp218 optimal for LXW64. The carbonyl oxygen of Arg3 also forms a hydrogen bond with Tyr178 of the in the integrin is not optimal for LXW64. The carbonyl oxygen of Arg3 also forms a hydrogen bond α-subunit. Both carboxylates of D-Asp6 side chain and D-Cys8 of LXW64 interact with Mn2+ ions with Tyr178 of the α-subunit. Both carboxylates of D-Asp6 side chain and D-Cys8 of LXW64 interact within the MIDAS domain of the β-subunit. D-Asp6 is also close to Arg214 of the β-subunit and forms with Mn2+ ions within the MIDAS domain of the β-subunit. D-Asp6 is also close to Arg214 of the β- a salt bridge. In addition, a salt bridge is formed between LXW64 Asp5 and Lys253 of the β-subunit. subunit and forms a salt bridge. In addition, a salt bridge is formed between LXW64 Asp5 and Lys253 Intriguingly, D-Nal1 (X7) side chain—naphthalene—is located in a potential hydrophobic pocket above of the β-subunit. Intriguingly, D-Nal1 (X7) side chain—naphthalene—is located in a potential Lys253 of the β-subunit, formed by nine hydrophobic amino acids from both α- (Ala215, Ile216, Phe217, hydrophobic pocket above Lys253 of the β-subunit, formed by nine hydrophobic amino acids from Ala246, Ala247) and β-subunit (Val314, Leu317, Ala218, Ala252) adjacent to the MIDAS site (Figure4B), both α- (Ala215, Ile216, Phe217, Ala246, Ala247) and β-subunit (Val314, Leu317, Ala218, Ala252) which has never been reported. The binding pattern of LXW7 with the integrin is almost identical adjacent to the MIDAS site (Figure 4B), which has never been reported. The binding pattern of LXW7 to LXW64 (Figure4C). However, instead of only interacting with one of the carboxylate oxygens with the integrin is almost identical to LXW64 (Figure 4C). However, instead of only interacting with (Asp218 in α-subunit) in the previous docking simulation studies [7], LXW7 Arg3 is able to interact one of the carboxylate oxygens (Asp218 in α-subunit) in the previous docking simulation studies [7], with both Asp148 and Asp150 in α-subunit (Figure4C). D-Val7 side chain is also in the hydrophobic LXW7 Arg3 is able to interact with both Asp148 and Asp150 in α-subunit (Figure 4C). D-Val7 side pocket, but much less extended than D-Nal1 of LXW64, indicative of a weaker binding affinity. This is chain is also in the hydrophobic pocket, but much less extended than D-Nal1 of LXW64, indicative of consistent with docking studies that LXW7 calculated free energy ( 7.7 kcal/mol) is less than LXW64 a weaker binding affinity. This is consistent with docking studies −that LXW7 calculated free energy (−7.7 kcal/mol) is less than LXW64 (−9.0 kcal/mol). Although binding to the same site, LXW11 loses

Int. J. Mol. Sci. 2020, 21, 3076 6 of 17

Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 6 of 17 ( 9.0 kcal/mol). Although binding to the same site, LXW11 loses some critical interactions with the − integrinsome critical compared interactions to LXW64 with and the LXW7 integrin (Figure comp4D),ared even to LXW64 though and the ionicLXW7 interactions (Figure 4D), with even Asp though 148, Asp150,the ionic Lys253, interactions Arg214 with still Asp remain. 148, Asp150, These missing Lys253, interactions Arg214 still include remain. the These electrostatic missing interactions interaction 2+ betweeninclude Asp5the electrostatic side-chain carboxylateinteraction andbetw Mneen Asp5ions, theside-chain hydrophobic carboxylate interaction and betweenMn2+ ions,D-Val7 the sidehydrophobic chain and interaction the hydrophobic between groove D-Val7 as bothside chain are buried and the inside hydrophobic the circular groove structure. as both Furthermore, are buried theinside configuration the circular change structure. from FurthermD-Cys (LXW64ore, the configuration and LXW7) to changeL-Cys from in LXW11 D-Cys alters (LXW64 the orientationand LXW7) 2+ ofto LL-Cys8-Cys in carboxylate LXW11 alters resulting the orientation in a much of L weaker-Cys8 carboxylate interaction resulting with Mn in a ions.much Itsweaker calculated interaction free energy ( 7.42+ kcaL/moL) is much lower than other two peptides, indicating a significantly lower affinity with Mn− ions. Its calculated free energy (−7.4 kcaL/moL) is much lower than other two peptides, bindingindicating (IC 50a significantly). These results lower imply affinity that the binding structures (IC50). of These LXW results analog imply predominate that the their structures bindings of LXW with theanalogαvβ 3predominate integrin and their a D-amino bindings acid with with the a α hydrophobicvβ3 integrin sideand chaina D-amino at position acid with 7 is preferreda hydrophobic for a higherside chain binding at position affinity. 7 is preferred for a higher binding affinity.

FigureFigure 4. 4.Docking Docking models models of LXWof LXW analogs analogs (LXW64, (LXW64, LXW7 LXW7 and LXW11) and LXW11) with α vwithβ3 integrin. αvβ3 integrin. (A) LXW64 (A) (green)LXW64 bound (green) to boundαvβ3 integrinto αvβ3 integrin (α-subunit: (α-subunit: blue, β-subunit: blue, β-subunit: grey, Mn grey,2+: purple);Mn2+: purple); (B) Closeup (B) Closeup view ofview LXW64 of LXW64 (green (green sticks) sticks) in the inα vtheβ3 α integrinvβ3 integrin binding binding site. site. Key Key amino amino acids acids are labeledare labeled with with blue blue in αin-subunit, α-subunit, black black in β in-subunit, β-subunit, and and red inred LXW64; in LXW64; (C) Closeup (C) Closeup view view of LXW7 of LXW7 (green (green sticks) sticks) in the αinv βthe3 integrinαvβ3 integrin binding binding site. Key site. amino Key acidsamino are acids highlighted are highlighted with the with same the labels same as labels in (B,D as) in Closeup Figure view4 (B, ofD) LXW11Closeup (green view sticks) of LXW11 in the α(greenvβ3 integrin sticks) bindingin the site.αvβ3 Key integrin amino binding acids are site. highlighted Key amino with theacids same are labelshighlighted as in Figure with 4theB. same labels as in Figure 4B.

2.4. New LXW-Analogous Peptide Screening by Autodock To further develop more potent LXW-analogous peptides with a high affinity to αvβ3 integrin, we conducted in silico screening. Our strategy is to design new LXW64 mutants by replacing the X7 amino acid with hydrophobic residues, and then perform molecular docking by AutoDock Vina to

Int. J. Mol. Sci. 2020, 21, 3076 7 of 17

2.4. New LXW-Analogous Peptide Screening by Autodock Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 7 of 17 To further develop more potent LXW-analogous peptides with a high affinity to αvβ3 integrin, we conductedscreen new in silicoLXW screening.analogs based Our on strategy the docking is to design scores. new Five LXW64 representative mutants byLXW replacing analogs the (LXW7, X7 amino acidLXW11, with LXW26, hydrophobic LXW64, residues, LXW65) andwith then known perform IC50s fr molecularom previously docking in vitro by screening AutoDock were Vina selected to screen newto test LXW theanalogs feasibility based of our on strategy. the docking Comparison scores. between Five representative in vitro screening LXW and analogs in silico (LXW7, molecular LXW11, LXW26,docking LXW64, (Table 3 LXW65) and Figure with 5) known demonstrated IC50s from a co previouslynsistency withinin vitro thescreening results of were both selected methods, to test thesuggesting feasibility that of our the strategy. in silico Comparisonmolecular docking between canin vitroproducescreening reliable and results in silico and molecular is feasible docking for (Tablescreening.3 and FigureWe designed5) demonstrated a series of a consistencyLXW64 analog withinous peptides the results in ofwhich both the methods, X7 amino suggesting acid was that thereplaced in silico by molecular D-amino dockingacids in the can SwissSidechain produce reliable database results and(https://www.swisss is feasible for screening.idechain.ch/) We designed[11]. In a seriessilico of screening LXW64 analogouswas performed peptides by docking in which new the X7des aminoigned LXW acid wasanalogs replaced to the by crystal D-amino structure acids inof the SwissSidechainαvβ3 integrin (PDB database ID: 1L5G). (https: //Totalwww.swisssidechain.ch 20 new LXW-analogous/)[11 peptides]. In silico were screening identified was with performed a high by dockingbinding new affinity designed to the α LXWvβ3 integrin analogs referred to the crystal to the new structure lead a ofccordingαvβ3 integrinto AutoDock (PDB prediction. ID: 1L5G). All Total 20the new new LXW-analogous LXW analogs contain peptides a X7 werenon-natural identified amino with acid a with high either binding a cyclic affinity (total to 13) the orα non-cyclicvβ3 integrin referred(total 7) toside the chain new as lead shown according in Table to 4. AutoDockFor these X7 prediction. amino acids, All the the side new chains LXW with analogs cyclic structures contain a X7 non-naturalwere predicted amino with acid higher with binding either aaffinity cyclic than (total that 13) of or non-cyclic non-cyclic ones. (total Seven 7) side new chain LXW as analogs shown in with DNTL, DTRP, DPHE, DQ36, D5MW, D6MW, and DQX3 as the amino acid at position 7 were Table4. For these X7 amino acids, the side chains with cyclic structures were predicted with higher predicted with a binding affinity (Kd ≤ 0.5 µM) comparable to the new lead—LXW64 (0.25 µM). binding affinity than that of non-cyclic ones. Seven new LXW analogs with DNTL, DTRP, DPHE, DQ36, Intriguingly, one of the best X7 residues, 3-(9-anthryl)-D-alanine (DNTL) is structurally similar to the D5MW, D6MW, and DQX3 as the amino acid at position 7 were predicted with a binding affinity (Kd X7 of the new lead. This DNTL-containing LXW-analogous peptide (LXZ2) was predicted with a high 0.5 µM) comparable to the new lead—LXW64 (0.25 µM). Intriguingly, one of the best X7 residues, ≤binding affinity to the αvβ3 integrin. 3-(9-anthryl)-D-alanine (DNTL) is structurally similar to the X7 of the new lead. This DNTL-containing

LXW-analogousTable 3. The peptide experimentally (LXZ2) was measured predicted half-max with aimal high bindinginhibitory a fficoncentrationnity to the α(ICvβ503) integrin.and computationally calculated binding free energy (kcal/mol) and equilibrium dissociation constant (Kd) Tableby Autodock 3. The experimentally of LXW analogs measured binding half-maximalto αvβ3 integrin. inhibitory concentration (IC50) and computationally calculated binding free energy (kcal/mol) and equilibrium dissociation constant (Kd) by Autodock of LXW IC50 Binding Free analogsRGD binding to αXv Aminoβ3 integrin. Acids in LXW Kd (μmoL/L) (μmoL/L) Energy Peptide Analogs (CGRGDdXc-NH2) by Autodock# (kcaL/moL) RGD Peptide X Amino Acids in LXW IC (µmoL/L) Binding Free Energy Kd (µmoL/L) by 50 # LXW7 Analogs (CGRGDdXc-NHD-Val 2) 0.46 (kcaL-7.7 /moL) 2.27Autodock LXW7LXW11* D-Val D-Val and L-Cys 0.46 7.7 2.27 >20 -7.4− 3.70 LXW11 * D-Val and L-Cys >20 7.4 3.70 − LXW26LXW26 D-IleD-Ile 1.841.84 -7.6 7.6 2.67 2.67 − LXW64 D-Nal1 0.07 9.0 0.24 LXW64 D-Nal1 0.07 -9.0− 0.24 LXW65 D-Nal2 0.13 9.3 0.15 LXW65 D-Nal2 − * two cysteines are L-configuration. # According0.13 to calculation-9.3 of Ki in Autodock 4. 0.15 * two cysteines are L-configuration. # According to calculation of Ki in Autodock 4.

Figure 5. Comparison between the determined IC50 and Autodock-calculated Kd of representative Figure 5. Comparison between the determined IC50 and Autodock-calculated Kd of representative LXW analogs. LXW analogs.

Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 8 of 17 Int. J. Mol. Sci.Int. 2020 J. Mol., 21 Sci., x FOR2020 PEER, 21, x REVIEWFOR PEER REVIEW 8 of 17 8 of 17

Int. J. Mol.Table Sci. 20204., 21, x20 FOR X7PEER REVIEWamino acids screened from the SwissSidechain database8 of 17 Int. J. TableMol. Sci. 20204.Table , 2120, x FOR4.X7 PEER 20amino REVIEWX7 acidsamino screenedacids screenedfrom thefrom SwissSidechainthe SwissSidechain database 8 databaseof 17 (https://www.swisssidechain.ch/)Int. J. Mol. Sci. 2020, 21, x FOR PEER in LXW-an REVIEWalogous cyclic octapeptides (cGRGDdXc-NH2) and the 8 of 17 Int. J. Mol.(https://www.swisssidechain.ch/) Sci. 2020(https://www.swisssidechain.ch/), 21, x FOR PEER REVIEW in LXW-an inalogous LXW-an cyclicalogous octapeptides cyclic octapeptides (cGRGDd X(cGRGDdc-NH2) andXc-NH the8 2of) and17 the Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW α β 8 of 17 Tablepeptides’Int. J.4. Mol.binding 20Sci. 2020 affiniX7, 21ties, xamino FOR(Kd, PEERµmoL/L) acids REVIEW to αscreened vβ 3 integrin α βfrom calculated the by AutoSwissSidechaindock 4.2. (note: database Residues 8 of 17 peptides’Table binding4.peptides’Table 20 affini binding4.X7 ties 20 (Kd,amino affini X7µmoL/L)ties acids(Kd,amino to µmoL/L) vscreened3acids integrin to vscreened calculated3from integrin the calculated byfrom AutoSwissSidechain dockthe by Auto4.2.SwissSidechain (note:dock Residues4.2.database (note: Residuesdatabase (https://www.swisssidechain.ch/)highlighted in bold are comparable in LXW-an to LXW64alogous listed cyclic as a referenceoctapeptides and (cGRGDdall non-naturalXc-NH amino2) and acidthe Table(https://www.swisssidechain.ch/) 4.highlighted 20 X7 in boldamino are comparableinacids LXW-an screenedalogous to LXW64 cyclicfrom list octapeptidesed asthe a referenceSwissSidechain (cGRGDd and alXlc-NH non-natural database2) and the amino acid highlightedTable 4.(https://www.swisssidechain.ch/) in bold20 areX7 comparable amino toacids LXW64 inscreened listLXW-aned as alogousa referencefrom cyclicthe and octapeptides alSwissSidechainl non-natural (cGRGDd amino databaseX acidc-NH 2 ) and the Int. peptides’J.codes Mol. Sci. are binding2020 Tablefrom, 21 the , affinix FORSwissSidechain4. ties PEER20 (Kd, REVIEWX7 µmoL/L) database).amino to α v βacids3 integrin screened calculated byfrom Auto dockthe 4.2.SwissSidechain (note: Residues 8database of 17 Int. J. Mol.(https://www.swisssidechain.ch/)codespeptides’ Sci.Int. are 2020 J. Mol. fromcodesbinding, 21 Sci., xthe areFOR 2020 affiniSwissSidechain from PEER, 21ties, xthe REVIEWFOR (Kd, SwissSidechain PEER µmoL/L) in database). LXW-an REVIEW to αalogous database). vβ3 integrin cyclicα β calculated octapeptides by Auto (cGRGDddock 4.2.Xc-NH (note:2) andResidues the8 of 17 8 of 17 (https://www.swisssidechain.ch/)peptides’ binding affinities in LXW-an(Kd, µmoL/L)alogous to cyclicv 3 integrin octapeptides calculated (cGRGDd by AutoXc-NHdock2 )4.2. and (note: the Residues highlighted(https://www.swisssidechain.ch/) in bold are comparable to LXW64α inβ LXW-anlisted asalogous a reference cyclic and octapeptides all non-natural (cGRGDd aminoX c-NHacid 2) and the peptides’highlightedpeptides’ binding bindinghighlighted in bold affini affini areties in tiescomparable (Kd,bold (Kd, µmoL/L)are µmoL/L) comparable to LXW64to tov αv3 β tolistintegrin3 integrinLXW64ed as calculateda calculatedlistreferenceedKd as bya andby referenceAuto Auto aldockl non-naturaldock and 4.2. 4.2. al(note:l (note: non-natural amino Residues Residues Kdacid amino acid codesPeptide/Table are from peptides’4. the20 SwissSidechain bindingX7 affiniaminoties database). (Kd,acids µmoL/L) Peptide/screened to αv β3 fromintegrinKd thecalculated KdSwissSidechain Peptide/ by Auto dock 4.2.database (note:Kd Residues Kd Int. J.highlightedTable Mol.codesPeptide/ Sci. are 20204. Table fromPeptide/ in, 21 20bold , the x FOR 4. SwissSidechainareX7Kd PEER comparable(20μmoL/L)amino REVIEWX7 database).acids toamino LXW64 screenedPeptide/ acids listed as Peptide/screened afrom reference(μ moL/ the fromand SwissSidechain all thenon-naturalPeptide/ SwissSidechain Peptide/ aminodatabase( μacid moL/ 8database of 17 Int. J. Mol.highlighted(https://www.swisssidechain.ch/) Sci.Int. 2020 J. Mol.,codes 21 in,Sci. x FORbold 2020are PEER from,are 21, xcomparable REVIEW theFOR SwissSidechain PEER in REVIEW toLXW-an LXW64 database).alogous listed cyclicas a reference octapeptides and al(cGRGDdl non-naturalXc-NH amino2) and acid the8 of 17 8 of 17 7 highlightedKd in (μ boldmoL/L)Kd are ( μcomparablemoL/L) X7 Residueto LXW64 list(edμmoL/ as a reference(μmoL/X7 Residueand all non-natural (μmoL/ amino(μ moL/acid codes(https://www.swisssidechain.ch/)X Residue are from(https://www.swisssidechain.ch/) the SwissSidechain in database). LXW-an in alogous LXW-an cyclicalogous octapeptides cyclic octapeptides (cGRGDd X(cGRGDdc-NH2) andXc-NH the 2) and the 7 X7 Residue X7 ResidueX 7 ResidueL) X7 ResidueX 7 Residue L) X codespeptides’Residue are bindingfrom the affini SwissSidechainties (Kd, µmoL/L) database). to α vβ3 integrin calculatedKd by Autodock 4.2. (note: ResiduesKdL) L) Int. J.peptides’ Mol.Peptide/Table Sci. 2020 bindingpeptides’codes4., 21, 20x are affiniFOR binding fromX7 PEERties the (Kd, REVIEWaminoaffini SwissSidechain µmoL/L)ties (Kd,acids to µmoL/L) αPeptide/v database).βscreened3 integrin to αv β calculated3from integrinL) Kd the calculatedby AutoL)SwissSidechainKd dockPeptide/ by Auto4.2. (note: dock Residues4.2.database (note:Kd 8 Residuesof 17 Kd DNAL1TablePeptide/ 4. Table Peptide/20 Kd4.X7 (μ20moL/L)amino X7 acidsamino screenedPeptide/acids screenedfrom(μ moL/ thefrom SwissSidechain thePeptide/ SwissSidechain database(μmoL/ database Int. J. Mol.highlighted Sci.Int. 2020 J. Mol., 21 in,Sci. x boldFOR 2020Kd PEER,are 21 (μ, comparablexmoL/L) REVIEW FOR PEER REVIEW to LXW64 listed asPeptide/ a reference(μKdmoL/ and all non-naturalPeptide/ amino(μKdmoL/ acid 8 of 17 8 of 17 XDNAL1highlightedPeptide/7(https://www.swisssidechain.ch/) Residue DNAL1highlighted in bold are in comparable bold areKd comparable (inμ tomoL/L) LXW-an LXW64DCPEX 7 Residue alogous tolist LXW64ed as cyclic a listreference edoctapeptides asKd a andreference( μal moL/(cGRGDdlX non-natural7 Residue and alXl c-NH non-natural amino2) and acid Kd theamino ( μacidmoL/ X7Peptide/ Residue(https://www.swisssidechain.ch/) inPeptide/X7 ResidueLXW-an alogous 7 L) cyclic octapeptidesKd Peptide/X7 Residue (cGRGDd 7 2 Xc-NHL) 2) and the (LXW64)(https://www.swisssidechain.ch/)codes are XPeptide/ from7 Residue theKd SwissSidechain (μmoL/L) in LXW-andatabase).DCPEalogousPeptide/ DCPE cyclic X Residue octapeptides(μmoL/L) (cGRGDdDNLEPeptide/Xc-NH X Residue) and(μmoL/ theL) Kd Int.Int. J. J. codesMol. TableMol.peptides’ Sci. Sci. are 2020 2020 from4.codes binding , ,21 21 the,20 ,xare x FOR FOR SwissSidechainaffini fromX7Kd PEER PEER ties (theμamino moL/L)REVIEW (Kd,REVIEW SwissSidechain µmoL/L) database). acids to α database).screenedvβ3 integrin Peptide/from calculated( μ moL/the by AutoSwissSidechainL) dock 4.2. (note:Peptide/ database Residues( μmoL/8 8 of of 17 17 L) X(LXW64)7 Residue (LXW64) peptides’ binding affinities (Kd, µmoL/L)Xα7 Residueβ to α vβ3 integrin calculatedX7 Residue byDNLE Auto dockDNLE 4.2. (note: Residues(μmoL/ DNAL1peptides’TableX7 ResidueInt. J. binding4. Mol.Table Sci.20 affini 20204.X7 , ties 21,20 x amino(Kd, FORKd X7µmoL/L) PEER ( μmoL/L)acids aminoREVIEW to vscreenedX 73acids Residue integrin screened calculatedfrom the byfrom Auto(μSwissSidechainmoL/ docktheX7 Residue 4.2.SwissSidechain (note: ResiduesdatabaseL) database8 of 17 DNAL1(https://www.swisssidechain.ch/) 7 in LXW-analogous Xcyclic7 Residue octapeptidesL) (cGRGDdXc-NHX7 Residue2) and theL) highlightedXDNAL1Residue in bold are comparable to LXW64DCPE listed as a referenceKdL) and all non-natural amino acidKd L) highlighted(https://www.swisssidechain.ch/)Peptide/highlighted(https://www.swisssidechain.ch/) in bold are in comparable bold0.25 are incomparable toLXW-an LXW64DCPEα inalogous Peptide/β tolistLXW-an LXW64ed ascyclic alogousa listreference octapeptidesedKd0.70 ascyclic a andreference octapeptides Kd al(cGRGDdL)l non-natural Peptide/and Xal (cGRGDdc-NHl non-natural amino2) andX c-NHKdacid1.63 the amino 2) and Kdacid the Int. J.DNAL1(LXW64) Mol.Tablepeptides’Peptide/codesTable Sci. are 2020 4.binding 4.Peptide/from , 212020 the, 3076 affini SwissSidechainX7X7 ties 0.25amino (Kd,amino µmoL/L) database).acids0.25acids to vscreened screened3 integrinDCPE calculatedfromfrom0.70 the the by Auto0.70SwissSidechainSwissSidechain dockDNLE 4.2. (note: Residuesdatabasedatabase1.63 1.63 8 of 17 codes(LXW64)peptides’DNAL1 are frombindingcodes(LXW64)peptides’ the are affini SwissSidechain frombindingKdties (theμ (Kd,moL/L) affini SwissSidechain µmoL/L)ties database). (Kd, to µmoL/L) αPeptide/ vdatabase).β3 integrin to αPeptide/ v βcalculated3 integrin(μ moL/ bycalculated AutodockPeptide/ byDNLE 4.2.Auto (note:dockPeptide/DNLE Residues 4.2. ((note:μmoL/ Residues 7 DNAL1Table Kd4. (μ20moL/L) KdX7 ( μmoL/L)aminoDCPE Xacids7 Residue screened (μmoL/ from(μmoL/ theX7 ResidueSwissSidechain (μmoL/ database(μmoL/ highlighted(https://www.swisssidechain.ch/)X Residue in bold are comparable in toLXW-an LXW64DCPEalogous listed ascyclic a reference octapeptides and al (cGRGDdl non-naturalXc-NH amino22) and acid the (LXW64)(https://www.swisssidechain.ch/)7 X7 Residue in LXW-anX7 alogousResidueDCPE Xcyclic 7 Residue octapeptides L) (cGRGDdX7DNLE ResidueXc-NHX 7 Residue) and theL) XhighlightedResidue highlighted in bold are in comparable bold are comparable to LXW64 tolist LXW64ed as a listreferenceed as a and reference all non-natural and all non-natural amino acid amino 2 acid codes(LXW64) are from(https://www.swisssidechain.ch/) the SwissSidechain0.25 database). αinα βLXW-anβ alogous0.70L) Kdcyclic octapeptidesL) DNLE (cGRGDd X1.63c-NHL)Kd ) andL) the peptides’peptides’Peptide/(LXW64) binding binding affini affinitiesties (Kd,0.25 (Kd, µmoL/L) µmoL/L) to to vPeptide/v 33 integrin integrin calculated calculatedKd0.70 by by Auto AutoKd dockdockPeptide/ 4.2. 4.2. (note: (note: DNLE Residues ResiduesKd1.63 Kd codesPeptide/DNAL1 are frompeptides’codesPeptide/ the are SwissSidechain bindingfrom the affini SwissSidechainties database). (Kd,0.25 µmoL/L)Peptide/ database). to αPeptide/v β 3 integrin calculated0.70Peptide/ by Auto dockPeptide/ 4.2. (note: Residues1.63 DNAL1 DNAL1 Kd (μmoL/L) (μmoL/ (μmoL/ (μmoL/ Tablehighlightedhighlighted7 4. 20 X7 inin boldamino boldKd areare ( acidsμ comparablemoL/L)comparable 0.25Kd screened (μmoL/L) toto LXW64 fromLXW64 DCPEX7 the Residue listlist SwissSidechaineded asas aa referencereference(μ0.70moL/ databaseandand(μmoL/ alall lX non-naturalnon-natural7 Residue (https:// www.swisssidechain.ch aminoamino(μ1.63moL/ acidacid /) X Residue 0.25 DCPE DCPE7 Kd0.70 7 1.63 X7(LXW64)Peptide/ ResidueX highlighted7 Residue in bold are comparable0.25 X7 Residue to LXW64X Residue listed L)as a reference0.70X 7 Residue andDNLE allX non-naturalResidue Kd L) amino 1.63acid (LXW64)incodescodes LXW-analogous are are (LXW64) from from the the SwissSidechain SwissSidechain cyclic octapeptides database). database).DHL1Peptide/ (cGRGDd Xc-NHKdL) ) andL)Kd the Peptide/DNLEDNVA peptides’ DNLE binding KdL) a nitiesL)Kd (Kd, DNTLPeptide/ (LXZ2) codesPeptide/ are Kdfrom (μ themoL/L) SwissSidechain DHL1Peptide/ database).DHL1 Peptide/ (μmoL/ 2 Peptide/DNVA Peptide/DNVA(μ moL/ ffi DNAL1 X7 Residue X7 Residue DNAL1DNTLX7 Residue (LXZ2)DNAL1DNTL (LXZ2) Kd ( μmoL/L) Kd (μmoL/L) (μmoL/ (μmoL/ (μmoL/ (μmoL/ µmoL/L) to αvβ3 integrin0.25 calculated DCPE7 by AutodockX7 Residue4.2.L)0.70 (note: Residues7 X7 highlightedResidue L) 1.63 in bold are X7 ResidueX 7 Residue 0.25 DCPEX ResidueDCPE KdKd 0.70X Residue KdKd 1.63 (LXW64)Peptide/Peptide/ 0.25 DHL1Peptide/Peptide/ 0.70L) KdL) Peptide/DNVAPeptide/DNLE 1.63L) KdL) (LXW64)DNTLcomparableDNAL1 (LXZ2) (LXW64)Peptide/ to LXW64 KdKd ( (μμmoL/L) moL/L)0.29 listed as a referenceDHL1 DHL1 andPeptide/ all((μμ1.38moL/ non-naturalmoL/ DNLEDNVA aminoPeptide/ acidDNLEDNVA codes((μμ 1.16moL/moL/ are from the DNTL (LXZ2) 0.29 0.29 7 7 1.38 1.38 7 7 1.16 1.16 DNAL1XX7 7Residue Residue DNAL1DNTL (LXZ2) Kd (μmoL/L)DHL1DCPE XX ResidueResidue (μmoL/XXDNVAResidueResidue (μmoL/ SwissSidechainX7 Residue database). DHL1 X7 Residue L)L) DNVAX7 Residue L) L) DNTL(LXW64) (LXZ2) 0.25 DCPE DHL1DCPE 0.70 L) DNLE DNVA 1.63 L) (LXW64)DNTL (LXZ2)(LXW64) 0.25 0.25 0.70 0.70 DNLE DNLE1.63 1.63 DNAL1DNAL1 DNTL (LXZ2) 0.290.29 1.381.38 1.161.16 Peptide/X ResidueDNAL1 Kd ( µmoL/L)0.29 PeptideDCPE DCPE/X Residue Kd (µmoL1.38/L) Peptide/X Residue Kd1.16 (µ moL/L) 7 0.25 DHL17 0.70 DNVA 7 1.63 (LXW64)(LXW64) 0.290.29 DALC DCPE 1.381.38 DNLEDNLE 1.161.16 DNTL (LXZ2)(LXW64) 0.25 0.290.25 DHL1 DHL1 0.70 1.380.70 DNVA DNVADNLE 1.63 1.161.63 DNAL1DNTLDTRP (LXZ2) (LXW64)DNTL (LXZ2) DALC DALC DQ33 DTRP DTRP DCPE DQ33 DNLEDQ33 0.25 0.70 1.63 0.25 0.70 1.63 0.250.29 0.25 DALCDHL1 1.38 0.70 0.70 DNVA 1.16 1.63 1.63 DTRPDNTL (LXZ2) 0.290.30 0.29 DHL1DALC DHL1DALC 1.380.50 1.38 DNVADQ33 DNVA 1.160.50 1.16 DNTLDTRP (LXZ2) DNTL (LXZ2) 0.30 0.30 0.50 0.50 DQ33 0.50 0.50 DTRP DALC DQ33 DHL1DALC DNVA DTRP DALC DQ33 DNTLDTRP (LXZ2) 0.300.29 DHL1 DHL1 0.501.38 DNVADQ33 DNVA 0.501.16 DTRP 0.290.30 0.29 1.380.50 1.38 DQ33 1.160.50 1.16 DNTLDNTL (LXZ2) (LXZ2)DNTL (LXZ2) 0.30 0.50 0.50 DALC DNVA 0.300.30 DHL1DLVGDHL1DHL1 0.500.50 DNVADNVA 0.500.50 DNTLDNTLDTRP (LXZ2) (LXZ2) 0.29 0.30 DALC DALCDHL1 1.38 0.50 DQ33 DNVA 1.16 0.50 DTRPDPZ4 DTRP 0.29 0.29 DLVG DLVG 1.38 1.38 DQ33DQ36 DQ33 1.16 1.16 DPZ4 DPZ4DNTL (LXZ2) 0.29 1.38 DQ36 DQ36 1.16

0.290.290.30 DLVGDALC 1.381.380.50 1.161.160.50 DPZ4DTRP 0.300.59 0.30 DALCDLVG DALCDLVG 0.500.83 0.50 DQ36DQ33 0.500.30 0.50 DTRPDPZ4 DTRP 0.59 0.590.29 0.83 0.831.38 DQ33DQ36 DQ33 0.30 0.301.16 DPZ4 DLVG DQ36 DLVG DQ33 DPZ4DTRP DALCDALCDLVG DQ36 DTRPDPZ4 DALC DALC DQ33DQ36 DPZ4 0.590.30 0.30 0.830.50 0.50 DQ36 0.300.50 0.50 DTRP DTRP 0.300.59 0.59 0.500.83 0.50 0.83 DQ33 DQ33 0.500.30 0.300.50 0.59 DLVG 0.83 0.30 DLEU 0.59 DALCD5MWDALC 0.83 0.30 DTRPDPZ4 0.30 0.59 DLVG DLVGDALC 0.50 0.83 DQ33DQ36 0.50 0.30 DPZ4DLEUDTRP DPZ4DLEU 0.30 0.30 D5MW D5MW 0.50 0.50 DQ36DQ33DQX3 DQ36 0.50 0.50 DTRP DQX3 DQX3DQ33 DQ36 DPZ4 DLVG DLEU 0.300.59 D5MWDLVG 0.500.83 0.500.30 DLEU 0.590.301.16 0.59 DLVGD5MW DLVG 0.830.500.30 0.83 0.300.500.38 0.30 DPZ4 DLEU 0.59 1.160.30 D5MW 0.83 0.300.50 DQX3DQ36 0.380.50 0.30 DPZ4 DPZ4 1.16 0.30 DQ36DQX3 DQ36DQX30.38 DLEUDLEU D5MWD5MW DLEU DLVG D5MW DQX3 DLVG DQX3 DPZ4 1.160.59 DLVG 0.300.83 DQ36 DQX3 0.380.30 DPZ4 DPZ4 0.591.16 0.591.16 0.830.30 0.830.30 DQ36 DQ36 0.300.38 0.300.38 D5MW DQX3 DLEUDLEU 1.16 D5MW 0.30 0.38 DLEUDPHE DLEU 0.591.16 D5MWDLVGD6MWDLVG D5MW 0.830.30 DTH9 0.300.38 DPHEDPZ4DPZ4 DPHE 1.16 D6MW D6MWDLVG 0.30 DTH9DQ36DQ36DQX3 DTH9 0.38 1.160.59 0.59 0.83 0.30 0.83 DQX3 DQX3 0.30 0.30 0.38 DPZ4 DQ36

0.36 0.42 0.59 DPHEDLEU 0.360.590.591.16 0.36 D6MWD5MW 0.420.830.830.30 0.42 DTH9 0.590.300.300.38 0.59 DLEUDPHE DLEUDPHE 1.16 1.16 D5MWD6MW D5MWD6MW 0.30 0.30 DTH9DQX3 DTH90.38 0.38 0.59 0.83 DQX3 DQX3 0.30 DPHEDPHE D6MW DTH9 DTH9 DLEUDPHE D5MWD6MWD6MW DTH9 DPHE 0.36 D6MW 0.42 DTH9 0.59 DLEU DLEU 0.360.361.16 0.36D5MW D5MW 0.420.30 0.42 0.42 DQX3 0.590.38 0.590.59 DMET 1.16 1.16 0.30 0.30 DQX3 DQX30.38 0.38 DMET DMET 0.360.36 D2TH 0.420.42 DAHP 0.590.59 DLEUDLEUDPHE 0.36 D5MWD5MWD6MW 0.42 DTH9 0.59 DPHE DPHE 1.16 D6MWD2TH D6MWD2TH 0.30 DTH9DAHP DTH9DAHP 0.38 DLEU D5MW DQX3DQX3 DMETDMET 1.16 1.16 D2TH 0.30 0.30 DAHPDQX3 0.38 0.38 DMET 0.83 0.98 1.63 DMET 0.36 0.83 D2TH 0.42 0.98 DAHP 0.59 1.63 0.830.360.83 0.36 D2TH D2TH 0.420.98 0.980.42 DAHP DAHP0.591.63 0.59 1.63 DMETDPHE DPHE 1.161.16 D6MW D6MW 0.300.30 DTH9 DTH9 0.380.38 DPHEDMET 1.16 D6MW 0.30 DTH9 0.38 DMET D2THD2TH DAHPDAHP 0.83 D2TH 0.98 DAHP 1.63 DPHE 0.830.36 0.83 D6MW 0.980.42 0.98 DTH9 1.630.59 1.63 0.36 0.36 0.42 0.42 0.59 0.59 DPHEDMET DPHE 0.83 D6MW D6MW 0.98 DTH9 DTH91.63 DMET DMET 0.83 D2TH 0.98 DAHP 1.63 2.5. In Vitro Examination of New LXW-Analogous0.83 Peptides 0.98 1.63 DPHE 0.36 D2THD6MW D2TH 0.42 DAHPDTH9 DAHP 0.59 DPHE 0.36 0.36 D6MW 0.42 0.42 DTH9 0.59 0.59 DPHE D6MW DTH9 DMET 0.83 0.98 1.63 DMETTo furtherDMET verify our 0.83 in silico0.83 screening results, we0.98 examined 0.98 the binding of1.63 new LXW-analogous1.63 0.360.36 D2TH D2TH 0.420.42 DAHP DAHP 0.590.59 peptides to αvβ3 integrin in αvβ3-K5620.36 D2TH cells. LXZ2 with DNTL0.42 asDAHP the X7 residue (Figure 0.596 left) was DMET DMET DMET 0.83 D2TH 0.98 DAHP 1.63 arbitrarily chosen and tested 0.83 through0.83 flow cytometry 0.98 for competing0.98 with 1 µM1.63 biotinylated 1.63 LXW7 D2TH D2TH DAHP DAHP α β bindingDMETDMET to v 3 integrin.0.83 As shown in Figure6, LXZ2 0.98 exhibited a stronger binding1.63 a ffinity than the DMET D2THD2TH DAHPDAHP 0.83 0.83 D2TH 0.98 0.98 DAHP 1.63 1.63 α β first lead—LXW7. The IC50 of the peptide inhibiting biotinylated LXW7 binding with K562/ v 3+ cells was determined (Figure0.830.836 right). Its converted IC500.980.98 (0.09 µmol/L referred to1.631.63 LXW7 in Table3) is 0.83 0.98 1.63 comparable to the new lead-LXW64 (0.07 µmol/L, Table3) and the well-known RGD-cyclic pentapeptide

ligand—cilengitide [5,8]. The result proved LXZ2 as a new αvβ3 integrin antagonist.

Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 9 of 17

2.5. In Vitro Examination of New LXW-Analogous Peptides To further verify our in silico screening results, we examined the binding of new LXW-analogous peptides to αvβ3 integrin in αvβ3-K562 cells. LXZ2 with DNTL as the X7 residue (Figure 6 left) was arbitrarily chosen and tested through flow cytometry for competing with 1 μM biotinylated LXW7 binding to αvβ3 integrin. As shown in Figure 6, LXZ2 exhibited a stronger binding affinity than the first lead—LXW7. The IC50 of the peptide inhibiting biotinylated LXW7 binding with K562/αvβ3+ cells was determined (Figure 6 right). Its converted IC50 (0.09 µmol/L referred to LXW7 in Table 3) is Int. J. Mol. Sci. 2020, 21, 3076 9 of 17 comparable to the new lead-LXW64 (0.07 µmol/L, Table 3) and the well-known RGD-cyclic pentapeptide ligand—cilengitide [5,8]. The result proved LXZ2 as a new αvβ3 integrin antagonist. B C A Negative LXZ2 control LXW7 Positive

control Count

FigureFigure 6. 6.( (AA)) Chemical structure structure of LXZ2; of LXZ2; (B) Competitive (B) Competitive binding bindingassay. 5 µM assay. of LXW7 5 µM (green) of LXW7 or (green) or LXZ2LXZ2 (orange) (orange) competes competes withwith 1 µ1 MµM biotinylated biotinylated LXW7 LXW7 binding binding to toα vαβv3-K562β3-K562 cells,cells,LXZ2 LXZ2 demonstrates a strongerdemonstrates binding a stronger affinity binding than a LXW7.ffinity than Positive LXW7. control Positive (cyan), controlcells (cyan), were cells treated were treated with 1withµM biotinylated 1µM biotinylated LXW7 and streptavidin-PE successively, and measured with flow cytometry. LXW7 and streptavidin-PE successively, and measured with flow cytometry. Negative control (red) Negative control (red) represents cells without treatment of biotinylated LXW7. (C) IC50 C representsmeasurement cells of without LXW7 and treatment LXZ2. The of IC50s biotinylated were determined LXW7. with ( ) IC50competitive measurement binding assay of LXW7 by and LXZ2. Theusing IC50s a series were of determined concentrations with of LXW7 competitive or LXZ2 binding competing assay with by 1 µM using biotinylated a series ofLXW7 concentrations binding of LXW7 orto LXZ2 αvβ3-K562 competing cells. with 1 µM biotinylated LXW7 binding to αvβ3-K562 cells.

3. Discussion3. Discussion RelativelyRelatively rapidrapid and and low-cost low-cost identification identification of RGD-containing of RGD-containing αvβ3 integrinαvβ 3antagonists integrin antagonistsThe The specificityspecificity of of the the integrins integrins (e.g., (e.g.,α αvvββ3)3) recognizingrecognizing and binding RGD RGD motif motif has has provided provided the the molecular molecular basis of integrin-targeted cancer therapy and enabled the development of several RGD- basis of integrin-targeted cancer therapy and enabled the development of several RGD-containing containing drugs for cardiovascular disease and cancer [2]. The first integrin antagonist— drugscilengitide— for cardiovascular was discovered disease on the basis and of cancer “ligand-oriented [2]. The firstdesign” integrin via the antagonist—cilengitide—optimization of RGD was discoveredpeptides by on means the basisof different of “ligand-oriented chemical approaches design” including via reduction the optimization of the conformational of RGD peptides space by means of dibyff cyclizationerent chemical and spatial approaches screening includingof cyclic pept reductionides [12]. A of similar the conformational RGD-based strategy space combining by cyclization and spatialwith different screening techniques of cyclic has peptides led to identification [12]. A similar of several RGD-based RGD peptides, strategy such as combining 1a-RGD [13], with different techniquescyclopeptide has c-Lys led [14], to identification LXW-analogous of peptides several [7,8]. RGD These peptides, screening such methods as 1a-RGD often require [13], cyclopeptide relatively expensive and time-consuming synthesis and a large amount of in vitro/in vivo cellular c-Lysassays. [14 ],Interestingly, LXW-analogous two linear peptides antagonists— [7,8]. TheseRWr and screening RWrNM methodspeptides—were often identified require relativelyrecently expensive andusing time-consuming pharmacophore-based synthesis virtual and screening a large amount[15], although of in cyclic vitro /peptidesin vivo arecellular likely assays.preferred Interestingly, due two linearto the antagonists—RWr fact that they usually and show RWrNM great peptides—werebiological activities identified compared recentlyto their linear using counterparts pharmacophore-based virtualbecause screening of their advantages [15], although including cyclic the peptides conformati areonal likely rigidity, preferred the resistance due to to thehydrolysis fact that by they usually showexopeptidases, great biological the receptor activities selectivity, compared and the ef toficient their membrane-crossing linear counterparts property because [16,17]. of In their our advantages current studies, NMR structure determination of LXW64 and structural comparison with other LXW including the conformational rigidity, the resistance to hydrolysis by exopeptidases, the receptor analogs, e.g., LXW7 and LXW11, enabled us to design new LXW-analogous peptides. Several new selectivity,LXW-analogous and the antagonists efficient membrane-crossingwere identified via in silico property screening [16 and,17]. one In of our the current best LXW studies, analogs— NMR structure determination of LXW64 and structural comparison with other LXW analogs, e.g., LXW7 and LXW11, enabled us to design new LXW-analogous peptides. Several new LXW-analogous antagonists were identified via in silico screening and one of the best LXW analogs—LXZ2—was arbitrarily chosen and verified by in vitro examination. Our results demonstrate that SAR studies provide clues for new RGD-containing αvβ3 integrin inhibitor design. Computational docking can predict the binding affinity quickly and narrow down the candidates; finally, the selected candidates can be tested by in vitro/in vivo examination. This provides a brief, rapid and relatively low-cost screening procedure for identification of αvβ3 integrin antagonists. Structure–activity relationship between LXW peptides and αvβ3 LXW analogs are a new category of RGD-containing cyclopeptides that binds specifically to αvβ3 integrin [7,8]. The LXW analogs were designed and screened using OBOC combinatorial technology. Unlike other RGD cyclopeptide antagonists (i.e., cilengitide) [12,14], LXW analogs are extended circular structures with a disulfide bond formed between two cysteines in the peptide sequence (cGRGDdXc). D-configuration of two cysteines in these analogs is essential for the antagonistic activity, which can force the cyclopeptide to adopt an open bowl-shape conformation with the side chains of Arg, Asp, D-Asp and X7 residue pointing Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 11 of 17 aliphatic residues (Kd ≥ 1.0 µM). Among the cyclic amino acids, the aromatic (mainly polycyclic excepted DPHE) structures (e.g., DNTL, DTRP, DQ36, D5MW, DQX3, D6MW) showed a Kd ≤ 0.5 µM and were identified as the most favorable X7 residues, suggesting the importance of both hydrophobicity and aromaticity for the binding. However, further cellular binding assays are still Int. J. Mol. Sci. 2020, 21, 3076 10 of 17 needed to verify their bioactivities as potent integrin antagonists. It would not be surprising to discover that some peptides might not be functional as good as the docking prediction, likely outwardsresultingfrom from the the peptide induced-fit ring [8effects]. These [19]. protruding We arbitrarily side chains chose are LXZ2 critical among to interact all possible with the candidates integrin. LXW7(i.e., X7 with = DPHE,D-Val7 wasDTRP, first D5MW, identified D6MW, as a lead DQ36, ligand DQX3) and furtherfor further optimized in vitro by OBOCstudies. technology LXZ2 was μ thatidentified led to the as a discovery new αvβ of3 antagonist. a 6-fold αvβ It3-binding has IC50 aofffi nity0.09 increasedM (this value antagonist—LXW64—with was converted to matchD-Nal1 with atthe position IC50s X7in [7Table,8]. As 3), shown which in is Figure comparable7, LXW7 to and LXW64 LXW64 (0.07 share µM) similar [8] structuraland the first patterns antagonist— with an enlargedcilengitide—with circular backbone IC50 of structure,0.25 µM [20]. an extended In comparison hydrophobic with other moiety RGD (disulfide-bonded antagonists, LXZ2 cysteines as a LXW and X7analog, residue) not as only well shows as an a extra high polar binding group affinity, (carboxyl but also group likely of D- exhibitsAsp) in the comparison binding specificity with cilengitide. against Theseαvβ3 uniqueas previous structural studies properties showed that suggest LXW that peptides LXW-analogous contain the antagonistsauxiliary binding are more motifs flexible including and mayD-Asp adopt at position better fit-in 6 and conformations the hydrophobicity and exhibit of amino better acid recognition at X7 position specificity [7,8]. and LXZ2 selectivity contains than non- otherproteinogenic RGD peptidomimetics amino acid—3-(9-Anthryl)- (i.e., cyclic pentapeptide—cilengitide)D-alanine—and its when anthracene—tricyclic binding to αvβ3 integrin. aromatic In fact,hydrocarbon—is LXW64 shows noncarcinogenic, significant positive and binding readily with biodαvegradedβ3, weak in or soil no binding and especially to αvβ5, susceptibleαIIbβ3, α5β 1to expresseddegradation on K562in the cells presence [8], while of light cilengitide [21]. Consi candering inhibit the both toxicityαvβ3 andand αenvironmentalvβ5 [18]. In addition, impact [22], the biotinylatedLXZ2 can be forms used ofas LXW a good ligands substitu showte of the LXW64 similar as binding both have strengths a similar as LXW affinity peptides to the againstintegrin.α vLikeβ3 integrinLXW64, [ 7],it whereasmay also biotinylated be used otheras an RGD excellent cyclopentapeptide candidate ligandsvehicle exhibitfor delivering much weaker drug-loaded binding ananoparticlesffinities than their for cancer free forms imaging [7]. Thus,and therapy. LWX analogous peptides are preferable as αvβ3 antagonists.

Figure 7. Structure superposition of LXW7 (light blue), LXW64 (cyan), and cilengitide from the crystal structureFigure 7. in Structure complex superposition with αvβ3 (PDB of LXW7 ID 1L5G, (light grey). blue), LXW64 (cyan), and cilengitide from the crystal structure in complex with αvβ3 (PDB ID 1L5G, grey). Despite the structural similarity, LXW64 demonstrated 6-fold higher binding affinity than that of LXW7. This suggests that the hydrophobicity and aromaticity of X7 amino acid plays a critically important role in improving the binding affinity of LXW analogs. A careful inspection of crystal structure of αvβ3 in complex with RGD ligand (PDB ID 1L5G) shows that there is a potential hydrophobic pocket next to the RGD ligand-binding area; however, two hydrophobic amino acids of the RGD ligand—D-Phenylalanine and N-methyl-Valine—make no contact with this pocket, likely due to the compact and rigid structure of the small ring-shape ligand. The hydrophobic pocket is formed by hydrophobic amino acids from both α- and β-subunits as shown in Figure8, which are located on the flexible surface area (e.g., flexible loops, short helical or β turn structures). As mentioned above, LXW-analogous peptides, especially LXW7 and LXW64 contain an extended hydrophobic moiety (i.e., disulfide-bonded D-cysteines and D-valine/3-(1-naphthyl)-D-alanine (D-Nal1)). It is highly possible that this hydrophobic moiety may undergo the hydrophobic interaction with this pocket and induce tertiary and quaternary structural changes of αvβ3 (Figure9). In fact, LXW64 with a large polycyclic aromatic and hydrophobic X7 residue exhibited much higher binding affinity for the integrin than LXW7 (Figure3A), suggesting the importance of hydrophobicity and aromaticity for the binding. It is surprising that this new hydrophobic pocket has not been identified previously [5], but likely plays a critical role in stabilizing the binding of LXW analogs to αvβ3 integrin and increase the binding affinity. Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 12 of 17 Int. J. Mol. Sci. 2020, 21, 3076 11 of 17 Int. J. Mol. Sci. 2020, 21, x FOR PEER REVIEW 12 of 17

FigureFigure 8.8. 8. (LeftLeft Left:): :The The The RGD RGD RGD ligand-bindingligand-binding ligand-binding areaarea area onon on the the the major major major surface surface surface of of ofα vααβvv3ββ integrin33 integrin integrin (PDB (PDB (PDB ID ID 1L5G) 1L5G) 2+ withwith αα -subunitα-subunit-subunit inin in grey,grey, grey, β ββ-subunit-subunit-subunit inin lightlightlight grey,grey,grey, MnMn Mn22++ in inin purple. purple.purple. Asp150/Asp218 Asp150Asp150/Asp218/Asp218 (red) (red) in αinv α subunitv subunitsubunit formform saltsalt salt bridgesbridges bridges withwith with thethethe arginineargininearginine (R)(R) guanidiniumguanidinguanidiniumium group groupgroup of ofof RGD RGDRGD ligand ligandligand (i.e., (i.e.,(i.e., cilengitide), cilengitide),cilengitide), whose whosewhose asparticaspartic acid acid (D) (D) carboxy carboxy groupgroup contactscontacts with with Arg214 Arg214 (blue), (blue),(blue), Asn215/Tyr122 Asn215Asn215/Tyr122/Tyr122 (pink) (pink)(pink) in inβin3β βsubunit.33 subunit.subunit. The The Arg216Arg216 isis alsois also also involved involvedinvolved in the inin contactthe contact with thewithwith ligand. ththee ligand.ligand. A cluster A A ofcluster cluster hydrophobic of of hydrophobic hydrophobic amino acids amino amino highlighted acids acids inhighlightedhighlighted yellow on in thein yellow yellow surface onon of thetheα- surface and β-subunit of αα-- andand form ββ-subunit-subunit a hydrophobic form form a ahydrophobic hydrophobic circle near thecircle circle RGD near near binding the the RGD RGD area, includingbindingbinding area, area, Ala215, including including Ile216, Ala215,Ala215, Phe217, Ile216, Ala246, PhPh Ala247,e217,e217, Ala246, Ala246, Met272 Ala247, Ala247, and Ala273 Met272 Met272 from and and αAla273-subunit; Ala273 from from Ala218, α-subunit; α-subunit; Ala252, Val314Ala218,Ala218, and Ala252, Ala252, Leu317 Val314Val314 from β -subunit;andand Leu317 (Right fromfrom): Closeup ββ-subunit;-subunit; ribbon RightRight view: :Closeup ofCloseup potential ribbon ribbon hydrophobic view view of ofpotential pocket potential (red dashedhydrophobichydrophobic circle) pocket pocket at the (red major(red dashed dashed interface circle) of atαv thethe (grey) major major and interface interfaceβ3 (light of of α grey)vα v(grey) (grey) subunits, and and β3 β (light3 which (light grey) isgrey) nearby subunits, subunits, RGD ligand-bindingwhichwhich is is nearby nearby areaRGD RGD (light ligand-binding ligand-binding blue). Hydrophobic area (light(light bl aminoblue).ue). Hydrophobic Hydrophobic acids (names amino labeledamino ac acids sameids (names (names as above) labeled labeled with same same side chainsasas above) above) shown with with are side side highlighted chains chains shownshown in yellow. are highlighted in in yellow. yellow.

FigureFigure 9. 9.The The hydrophobic hydrophobic aminoamino acids ( (DD-Val-Val (purple), (purple), DD-Nal1-Nal1 (3-(1-naphthyl)- (3-(1-naphthyl)-D-alanine,D-alanine, green)) green)) at at FigureX7 position 9. The of hydrophobic LXW-analogous amino cyclic acids octapeptides (D-Val (purple), (cGRGDd D-Nal1Xc-NH (3-(1-naphthyl)-2) binds at the majorD-alanine, hydrophobic green)) at X7 position of LXW-analogous cyclic octapeptides (cGRGDdXc-NH2) binds at the major hydrophobic X7 position of LXW-analogous cyclic octapeptides (cGRGDdXc-NH2) binds at the major hydrophobic interfaceinterface between between the theα α(blue) (blue) andand β (grey)(grey) subunits subunits formed formed by by Al Ala215,a215, Ile216, Ile216, Phe217, Phe217, Ala246, Ala246, Ala247 Ala247 interfacein the α betweensubunit and the Ala218,α (blue) Ala252, and β (grey) Val314, subunits Leu317 formedin the β bysubunit Ala215, of the Ile216, αvβ 3Phe217, integrin. Ala246, Ala247 in the α subunit and Ala218, Ala252, Val314, Leu317 in the β subunit of the αvβ3 integrin. in the α subunit and Ala218, Ala252, Val314, Leu317 in the β subunit of the αvβ3 integrin. Identification4. Materials of and a New Methods LXW Analog—LXZ2 4. Materials and Methods 4.1.In Synthesis our current of Peptides study, 20 new LXW analogs were predicted via in silico screening with a high binding affinity to αvβ3 integrin. As shown in Table4, all these peptides contain a hydrophobic non-natural 4.1. Synthesis of Peptides amino acid at X7 position, which is consistent with our SAR studies. These X7 amino acids can be

Int. J. Mol. Sci. 2020, 21, 3076 12 of 17

classified into two groups according to their side-chain structures—cyclic (total 13) and non-cyclic (aliphatic, total 7). All cyclic amino acids showed a lower but favorable Kd ( 1.0 µM) than aliphatic ≤ residues (Kd 1.0 µM). Among the cyclic amino acids, the aromatic (mainly polycyclic excepted ≥ DPHE) structures (e.g., DNTL, DTRP, DQ36, D5MW, DQX3, D6MW) showed a Kd 0.5 µM and ≤ were identified as the most favorable X7 residues, suggesting the importance of both hydrophobicity and aromaticity for the binding. However, further cellular binding assays are still needed to verify their bioactivities as potent integrin antagonists. It would not be surprising to discover that some peptides might not be functional as good as the docking prediction, likely resulting from the induced-fit effects [19]. We arbitrarily chose LXZ2 among all possible candidates (i.e., X7 = DPHE, DTRP, D5MW, D6MW, DQ36, DQX3) for further in vitro studies. LXZ2 was identified as a new αvβ3 antagonist. It has IC50 of 0.09 µM (this value was converted to match with the IC50s in Table3), which is comparable to LXW64 (0.07 µM) [8] and the first antagonist—cilengitide—with IC50 of 0.25 µM[20]. In comparison with other RGD antagonists, LXZ2 as a LXW analog, not only shows a high binding affinity, but also likely exhibits the binding specificity against αvβ3 as previous studies showed that LXW peptides contain the auxiliary binding motifs including D-Asp at position 6 and the hydrophobicity of amino acid at X7 position [7,8]. LXZ2 contains non-proteinogenic amino acid—3-(9-Anthryl)-D-alanine—and its anthracene—tricyclic aromatic hydrocarbon—is noncarcinogenic, and readily biodegraded in soil and especially susceptible to degradation in the presence of light [21]. Considering the toxicity and environmental impact [22], LXZ2 can be used as a good substitute of LXW64 as both have a similar affinity to the integrin. Like LXW64, it may also be used as an excellent candidate vehicle for delivering drug-loaded nanoparticles for cancer imaging and therapy.

4. Materials and Methods

Int. J. Mol.4.1. Sci. Synthesis 2020, 21, ofx FOR Peptides PEER REVIEW 13 of 17

All All RGD-cyclic RGD-cyclic peptides, peptides, where where a a lower lower case case letter letter represents representsD D-amino-amino acid acid (x is a (x is avariable variable amino amino acid) acid) and and a disulfidea disulfide bridge bridge is formedis formed between between two twoD-cysteines D-cysteines (D-Cys1 (D-Cys1 and andD-Cys8), D-Cys8),were were chemically chemically synthesized synthesized and purifiedand purified by preparative by preparative RP-HPLC) RP-HPLC) from from C S Bio C S Co Bio (Menlo Co (Menlo Park, CA, Park,USA). CA, USA). For synthesis For synthesis of LXW64 of and LXW64 LXZ2, and the non-naturalLXZ2, the aminonon-natural acids—Fmoc-3-(1-naphthyl)- amino acids—Fmoc-3-(1-D-alanine naphthyl)-and Fmoc-3-(9-anthryl)-D-alanine and Fmoc-3-(9-anthryl)-D-alanine—wereD purchased-alanine—were from purchased Chem Impex from (Wood Chem Dale, Impex IL, USA).(Wood HPLC Dale,purification IL, USA). wasHPLC performed purification using was an Agilentperformed 1200 instrumentusing an Agilent and a Phenomenex 1200 instrument Luna 5andµm C18(2)a Phenomenex100A 250 Luna4.6 5 mm μm C18(2) column. 100A The 250 peptides × 4.6 mm were column. eluted usingThe peptides a gradient were of eluted buffer Ausing (0.1% a gradient TFA in water) × of bufferand A B (0.1%(0.1% TFATFA inin acetonitrile)water) and withB (0.1% a flow TFA rate in acetonitrile) of 1 mL/min. with Each a peptideflow rate eluted of 1 mL/min. as a single Each peak via peptideHPLC eluted with as >a single97% purity, peak via was HPLC verified with by > MS. 97% The purity, theoretical was verified mass ofby LXW64 MS. The (918.98) theoretical was mass very close of LXW64to the (918.98) experimental was very value close (918.73). to the Forexperimental LXZ2, the value experimentally (918.73). For measured LXZ2, the mass experimentally of 968.72 was also measuredsimilar mass to the of expected968.72 was 969.09. also similar For NMR to samplethe expected preparation, 969.09. 20 For mg NMR peptides sample were preparation, dissolved in 20 0.6 mL mg peptidesDMSO-d 6weresolvent dissolved (Cambridge in 0.6 Isotope mL DMSO- Laboratories,d6 solvent Inc., (Cambridge Tewksbury, Isotope MA, USA). Laboratories, The peptide Inc., solution Tewksbury,was carefully MA, USA). transferred The peptide into an NMRsolution tube was after carefully centrifugation. transferred The repeated into an NMRNMR analysistube after showed centrifugation.that the peptide The repeated samples NMR were analysis stable over showed several that months the peptide in the samples chosen were DMSO stable solvent. over several months in the chosen DMSO solvent. 4.2. NMR Spectroscopy 4.2. NMR SpectroscopyAll NMR data were recorded at 295 K on a Bruker Ultrashield Plus 600 MHz spectrometer Allequipped NMR withdata awere 5 mm recorded double resonanceat 295 K broadon a Bruker band room Ultrashield temperature Plus probe600 MHz (BBO) spectrometer and a single-axis equippedpulse-field-gradient with a 5 mm double accessory resonance along the broad z-axis. band All experimentsroom temperature were performed probe (BBO) in a dimethyland a single- sulfoxide 1 1 axis pulse-field-gradientsolution to observe theaccessory amide protons.along theH- z-axisH homonuclear. All experiments two-dimensional were performed (2D) NMRin a dimethyl NOESY (τmix sulfoxide= 125 solution and 250 to ms), observe TOCSY the ((amideτmix = protons.70 ms), and1H-1H DQF-COSY homonuclear spectra two-dimensional were acquired (2D) using NMR a sweep NOESYwidth (τmix of = 14423125 and Hz 250 and ms), 1024 TOCSY complex ((τmix points = 70 ms), in F1. and The DQF-COSY transmitter spectra carrier were was acquired placed on using the water a sweepresonance. width of Gradient 14423 Hz heteronuclear and 1024 complex correlation points experiments,in F1. The transmitter (1H-13C)-HMQC, carrier was (1H- placed13C)-HMBC, on the and water( 1resonance.H-15N)-HSQC Gradient were heteronuclear carried out to correlation assign all carbon experiments, (13C) and (1H- nitrogen13C)-HMQC, (15N) (1 chemicalH-13C)-HMBC, shifts. The and (1carbonH-15N)-HSQC carrier frequency were carried was out kept to atassign 82 ppm all forcarbon HMQC (13C) and and 112 nitrogen ppm for (15 HMBC,N) chemical the spectral shifts. The width in carbon carrier frequency was kept at 82 ppm for HMQC and 112 ppm for HMBC, the spectral width in the indirect carbon dimension was set to 150 ppm and 200 ppm for HMQC and HMBC, respectively. For 1H-15N HSQC, the nitrogen channel was centered at 116 ppm with a sweep width of 32 ppm. The States-TPPI method was used for quadrature detection in all indirectly detected dimensions. 1H, 13C, and 15N NMR chemical shifts were reported using DSS and DMSO-d6 as references. NMR data were processed by NMRPipe [23] and analyzed with SPARKY (https://www.cgl.ucsf.edu/home/sparky/) [24].

4.3. Experimental Constraints and Structure Calculation The NMR structures of LXW64 were calculated based on NOE distances and dihedral angle restraints. Distance constraints were extracted from 2D NOESY recorded using two mixing times (125 ms and 250 ms). NOE cross peaks (NOEs) were classified into strong, medium and weak according to the intensities, and assigned to the interproton distances of 2.9, 3.5 and 5 Å. The upper bound distance constraints of the NOEs involving methyl and methylene groups were modified using pseudoatom correction [25]. The backbone dihedral angles (φ) were calculated from the Karplus Equation using 3JHNα coupling constants measured from DQF-COSY spectrum [26]. For side-chain dihedral angles, the χ1 was defined according to the NOE intensities between the amino proton (HN) and two β protons (Hβ) in comparison with the NOE intenwrsities between α proton (Hα) and two β protons in the same residue [27,28]. The additional constraint data from the chemical shifts of Cα, Cβ and Hα were used in the final structure refinement. The NMR-derived distances and dihedral angles then served as constraints for calculating the three-dimensional structures using distance geometry and restrained molecular dynamics. Structure calculations were performed using the YASAP protocol within X-PLOR version 2.36 [29,30] installed on a regular Linux workstation (CentOS 6), as described previously [31]. Fifty independent structures were calculated, and the 10 lowest-energy structures were selected. The average total and experimental distance energy were 185 ± 7 and 12 kcal/mol. Ramachandran analysis of the determined structures was performed through MolProbity [32] with default settings to measure the structure reliability. The average root-mean-square (rms)

Int. J. Mol. Sci. 2020, 21, 3076 13 of 17 the indirect carbon dimension was set to 150 ppm and 200 ppm for HMQC and HMBC, respectively. For 1H-15N HSQC, the nitrogen channel was centered at 116 ppm with a sweep width of 32 ppm. The States-TPPI method was used for quadrature detection in all indirectly detected dimensions. 1H, 13C, 15 and N NMR chemical shifts were reported using DSS and DMSO-d6 as references. NMR data were processed by NMRPipe [23] and analyzed with SPARKY (https://www.cgl.ucsf.edu/home/sparky/)[24].

4.3. Experimental Constraints and Structure Calculation The NMR structures of LXW64 were calculated based on NOE distances and dihedral angle restraints. Distance constraints were extracted from 2D NOESY recorded using two mixing times (125 ms and 250 ms). NOE cross peaks (NOEs) were classified into strong, medium and weak according to the intensities, and assigned to the interproton distances of 2.9, 3.5 and 5 Å. The upper bound distance constraints of the NOEs involving methyl and methylene groups were modified using pseudoatom correction [25]. The backbone dihedral angles (ϕ) were calculated from the Karplus Equation using 3 JHNα coupling constants measured from DQF-COSY spectrum [26]. For side-chain dihedral angles, the χ1 was defined according to the NOE intensities between the amino proton (HN) and two β protons (Hβ) in comparison with the NOE intenwrsities between α proton (Hα) and two β protons in the same residue [27,28]. The additional constraint data from the chemical shifts of Cα,Cβ and Hα were used in the final structure refinement. The NMR-derived distances and dihedral angles then served as constraints for calculating the three-dimensional structures using distance geometry and restrained molecular dynamics. Structure calculations were performed using the YASAP protocol within X-PLOR version 2.36 [29,30] installed on a regular Linux workstation (CentOS 6), as described previously [31]. Fifty independent structures were calculated, and the 10 lowest-energy structures were selected. The average total and experimental distance energy were 185 7 and 12 kcal/mol. ± Ramachandran analysis of the determined structures was performed through MolProbity [32] with default settings to measure the structure reliability. The average root-mean-square (rms) deviation from an idealized geometry for bonds and angles were 0.0094 Å and 2.27◦. None of the distance and angle constraints were violated by more than 0.4 Å and 4◦, respectively.

4.4. Complex Modeling and Autodock Screening Complex modeling of three representative LXW analogs (LXW7, LXW11, LXW64) and in silico screening of new LXW analogs were performed using the program AutoDock 4.20 [10] by following stepwise guidelines [33]. The NMR structures of LXW7 and LXW11 determined previously [8], and of LXW64 were used for docking. For screening of new RGD-containing peptides, all peptide mutants were generated using PyMol Mutagenesis Wizard by substituting D-Val7 of LXW7 with non-native amino acids in the SwissSidechain database (https://www.swisssidechain.ch/)[11]. Each peptide was prepared by the AutoDock Tools GUI (graphical user interface) and with the rotatable side-chain bonds set to allow rotation and docked into the ligand-binding pocket of integrin αvβ3 (PDB ID 1L5G) [5], in which the default ligand was removed using a text editing software. The modified αvβ3 structure was prepared for AutoDock Vina [34] compatibility with AutoDock Tools GUI [10]. The grid box with spacing size of 0.375 Å was placed in the center of the binding pocket in accordance with the ligand (cyclo(RGDf-N(Me))V-) on the crystal structure (PDB ID 1L5G). The grid box center was selected to include all the residues in the binding pocket and set as follows: X-center 18.792, Y-center 42.057, and Z-center 43.65. The configuration settings for AutoDock Vina were set to default except for the number of binding modes, which was set to 8. The appropriate search space parameters were determined through eBoxSize for each ligand [35]. A stochastic Lamarckian genetic algorithm was employed for computing peptide conformations within the active pocket by using the default parameters. A total of 1000 conformers were generated for the ligand in the binding pocket, which were clustered using a 2.0 Å root-mean-square deviation. The output ligand files were modified to contain only the optimal configuration through UCSF Chimera version 1.9 [36]. The receptor-ligand complexes were developed through the UCSF Chimera ViewDock extension and prepared using the Int. J. Mol. Sci. 2020, 21, 3076 14 of 17

FindHBond tool. The receptor and ligand-binding modes were joined using a text editing software and analyzed by PyMol 1.7 (open-source) [37]. The AutoDock screening generates an energy score (kcal/moL), which represents the free energy (∆G) of peptide binding to integrin αvβ3. The peptide binding affinity (Kd) is obtained according to calculation of Ki in AutoDock 4 at 298K.

4.5. Flow Cytometry The human leukemia cancer cells (K562, ATCC, Manassas, VA, USA) transfected with αvβ3-integrin were grown in the RPMI1640 medium and used for the assay as previously described [7,8]. To demonstrate the peptides binding to the integrin and compare with the binding affinity of LXW7, the cells in each sample were incubated with 1.0 µM biotinylated LXW7 in 50 µL of phosphate-buffered saline (PBS) containing 10% fetal calf serum (FBS) and 1 mM MnCl2 for 30 min on ice. The samples were washed with 1 mL PBS containing 1% FBS for three times, then incubated with a 1:500 dilution of streptavidin-PE (1 mg/mL) for 30 min on ice followed by a single wash with 1 mL of PBS containing 1% FBS. Finally, the samples were analyzed via flow cytometry (Coulter XL-MCL). To measure the half-maximal inhibitory concentration (IC50) of the peptides, various diluted peptide solutions were mixed with 1.0 µM biotinylated LXW7, then incubated with cells, and followed by streptavidin-PE incubation. The samples were run through flow cytometer and Mean Fluorescence Intensity was decided for each individual sample. The IC50s were calculated from each sample based on the readings.

5. Conclusions In this study, SARs of LXW-analogous cyclic octapeptides and αvβ3 integrin had been investigated through NMR structure determination and complex modeling. The hydrophobicity and aromaticity of the X7 amino acid in LXW-analogous sequence was found to be important for enhancing LXW analogs binding to the integrin, likely through the interaction with a potential hydrophobic pocket on the integrin surface. The SAR studies led to the identification of several new LXW-analogous peptides by in silico screening, which were predicted with high binding affinity. One of the best peptides—LXZ2—was arbitrarily chosen and verified by cell-based competitive binding assays and found to be comparable to other well-known “head-to-tail” RGD cyclopeptide—LXW64—and cilengitide. Most importantly, this new αvβ3 antagonist was identified through a brief and comparatively inexpensive screening procedure, benefited from the SAR studies. LXZ2 as an analogous peptide of LXW64 demonstrated a high binding affinity to αvβ3 integrin transfected in K562 cells, can be used a vehicle for delivery of cytotoxic payload to tumors and tumor blood vessels with overexpressing αvβ3 integrin.

Supplementary Materials: Supplementary materials can be found at http://www.mdpi.com/1422-0067/21/9/3076/ s1. Author Contributions: Conceptualization, W.X., K.S.L. and Y.Z.; methodology, K.S.L., J.B.A. and Y.Z.; resources, H.W.C., K.S.L. and Y.Z.; experiment and data analysis, A.S., W.X., Y.W., W.W. and Y.Z.; manuscript writing, A.S., W.X., J.B.A., K.S.L. and Y.Z. All authors have read and agree to the published version of the manuscript. Funding: This research received no external funding. Acknowledgments: We would like to thank Thomas Eubanks for NMR technical support at UTRGV. This work was supported by UTRGV ISRP award (2017–2018) to Y.Z, UTRGV Engaged Scholar Award to A.S., and NIH grant (EY012347) to J.B.A. The Department of Chemistry at UTRGV is grateful for the generous support provided by a Departmental Grant from the Robert A. Welch Foundation (Grant No. BX-0048). Conflicts of Interest: The authors declare no conflict of interest. Int. J. Mol. Sci. 2020, 21, 3076 15 of 17

Abbreviations

SAR Structure–Activity Relationship RGD Arginylglycylaspartic acid NMR Nuclear Magnetic Resonance NOE Nuclear Overhauser Effect HSQC Heteronuclear Single Quantum Correlation TOCSY TOtal Correlated SpectroscopY NOESY Nuclear Overhauser Effect Spectroscopy HMQC Heteronuclear Multiple-Quantum Correlation HMBC Heteronuclear Multiple Bond Correlation RMSD Root-Mean-Squared Derivation RP-HPLC Reversed Phase-High Performance Liquid Chromatography MS Mass spectrometry DMSO Dimethyl sulfoxide DSS 2,2-Dimethyl-2-silapentane-5-sulfonate PBS Phosphate-Buffered Saline FBS Fetal Calf Serum IC50 Half-Maximal Inhibitory Concentration

References

1. Horton, M.A. The αvβ3 integrin “vitronectin receptor”. Int. J. Biochem. Cell Biol. 1997, 29, 721–725. [CrossRef] 2. Liu, Z.; Wang, F.; Chen, X. Integrin αvβ3-target cancer therapy. Drug Dev. Res. 2008, 69, 329–339. [CrossRef] [PubMed] 3. Brooks, P.C.; Strömblad, S.; Klemke, R.; Visscher, D.; Sarkar, F.H.; Cheresh, D.A. Antiintegrin alpha v beta 3 blocks human breast cancer growth and angiogenesis in human skin. J. Clin. Invest. 1995, 96, 1815–1822. [CrossRef][PubMed] 4. Khan, K.A.; Bicknell, R. Anti-angiogenic alternatives to VEGF blockade. Clin. Exp. Metastasis 2016, 33, 197–210. [CrossRef][PubMed] 5. Xiong, J.P.; Stehle, T.; Zhang, R.; Joachimiak, A.; Frech, M.; Goodman, S.L.; Arnaout, M.A. Crystal structure of the extracellular segment of integrin αvβ3 in complex with an arg-gly-asp ligand. Science 2002, 296, 151–155. [CrossRef][PubMed] 6. Danhier, F.; Le Breton, A.; Préat, V. RGD-based strategies to target alpha(v) beta(3) integrin in cancer therapy and diagnosis. Mol. Pharm. 2012, 9, 2961–2973. [CrossRef][PubMed] 7. Xiao, W.; Wang, Y.; Lau, E.Y.; Luo, J.; Yao, N.; Shi, C.; Meza, L.; Tseng, H.; Maeda, Y.; Kumaresan, P.; et al. The use of one-bead one-compound combinatorial library technology to discover high-affinity αvβ3 integrin and cancer targeting arginine-glycine-aspartic acid ligands with a built-in handle. Mol. Cancer Ther. 2010, 9, 2714–2723. [CrossRef][PubMed] 8. Wang, Y.; Xiao, W.; Zhang, Y.; Meza, L.; Tseng, H.; Takada, Y.; Ames, J.B.; Lam, K.S. Optimization of RGD-Containing Cyclic Peptides against αvβ3 Integrin. Mol. Cancer Ther. 2016, 15, 232–240. [CrossRef] 9. Sharma, D.; Rajarathnam, K. 13C NMR chemical shifts can predict disulfide bond formation. J. Biomol. NMR 2000, 18, 165–171. [CrossRef] 10. Morris, G.M.; Huey, R.; Lindstrom, W.; Sanner, M.F.; Belew, R.K.; Goodsell, D.S.; Olson, A.J. Autodock4 and AutoDockTools4: Automated docking with selective receptor flexiblity. J. Comput. Chem. 2009, 16, 2785–2791. [CrossRef] 11. Gfeller, D.; Michielin, O.; Zoete, V. SwissSidechain: A molecular and structural database of non-natural sidechains. Nucleic Acids Res. 2013, 41, D327–D332. [CrossRef][PubMed] 12. Mas-Moruno, C.; Rechenmacher, F.; Kessler, H. Cilengitide: The first anti-angiogenic small molecule drug candidate. Design, synthesis and clinical evaluation. Anticancer Agents Med. Chem. 2010, 10, 753–768. [CrossRef][PubMed] 13. Russo, M.A.; Paolillo, M.; Sanchez-Hernandez, Y.; Curti, D.; Ciusani, E.; Serra, M.; Colombo, L.; Schinelli, S. A small-molecule RGD-integrin antagonist inhibits cell adhesion, cell migration and induces anoikis in glioblastoma cells. Int. J. Oncol. 2013, 42, 83–92. [CrossRef] Int. J. Mol. Sci. 2020, 21, 3076 16 of 17

14. Liu, Y.; Pan, Y.; Xu, Y. Binding investigation of integrin alphavbeta3 with its inhibitors by SPR technology and molecular docking simulation. J. Biomol. Screen. 2010, 15, 131–137. [CrossRef] 15. Ma, Y.; Ai, G.; Zhang, C.; Zhao, M.; Dong, X.; Han, Z.; Wang, Z.; Zhang, M.; Liu, Y.; Gao, W.; et al. Novel Linear Peptides with High Affinity to αvβ3 Integrin for Precise Tumor Identification. Theranostics 2017, 7, 1511–1523. [CrossRef] 16. Joo, S.H. Cyclic Peptides as Therapeutic Agents and Biochemical Tools. Biomol. Ther. (Seoul) 2012, 20, 19–26. [CrossRef] 17. Rezai, T.; Yu, B.; Millhauser, G.L.; Jacobson, M.P.; Lokey, R.S. Testing the conformational hypothesis of passive membrane permeability using synthetic cyclic peptide diastereomers. J. Am. Chem. Soc. 2006, 128, 2510–2511. [CrossRef] 18. Reardon, D.A.; Neyns, B.; Weller, M.; Tonn, J.C.; Nabors, L.B.; Stupp, R. Cilengitide: An RGD pentapeptide ανβ3 and ανβ5 integrin inhibitor in development for glioblastoma and other malignancies. Future Oncol. 2011, 7, 339–354. [CrossRef] 19. Koshland, D.E. Application of a Theory of Enzyme Specificity to Protein Synthesis. Proc. Natl. Acad. Sci. USA 1958, 44, 98–104. [CrossRef] 20. Sykoutri, D.; Geetha, N.; Hayer, S.; Mandl, P.; Smolen, S.S.; Prager, G.; Redlich, K. αvβ3 Integrin Inhibition with Cilengitide both Prevents and Treats Collagen Induced Arthritis. Ann. Rheum Dis. 2013, 72 (Suppl. 1), A1–A88. [CrossRef] 21. Collin, G.; Höke, H.; Talbiersky, J. Anthracene. Ullmann’s Encyclopedia of Industrial Chemistry, 7th ed.; John Wiley & Sons: New York, NY, USA, 2006. 22. Toxicology and carcinogenesis studies of naphthalene (cas no. 91-20-3) in F344/N rats (inhalation studies). Natl. Toxicol. Program Tech. Rep. Ser. 2000, 500, 1–173. 23. Delaglio, F.; Grzesiek, S.; Vuister, G.W.; Zhu, G.; Pfeifer, J.; Bax, A. NMRPipe: A multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 1995, 6, 277–293. [CrossRef][PubMed] 24. Goddard, T.D.; Kneller, D.G. SPARKY 3; University of California: San Francisco, CA, USA, 2008. 25. Wuthrich, K.; Billeter, M.; Braun, W. Pseudo-structures for the 20 common amino acids for use in studies of protein conformations by measurements of intramolecular proton-proton distance constraints with nuclear magnetic resonance. J. Mol. Biol. 1983, 169, 949–961. [CrossRef] 26. Wang, A.C.; Bax, A. Determination of the Backbone Dihedral Angles f in Human Ubiquitin from Reparametrized Empirical Karlpus Equations. J. Am. Chem. Soc. 1996, 118, 2483–2494. [CrossRef] 27. Assa-Munt, N.; Jia, X.; Laakkonen, P.; Ruoslahti, E. Solution structures and integrin binding activities of an RGD peptide with two isomers. Biochemistry 2001, 40, 2373–2378. [CrossRef] 28. Kessler, H.; Seip, S. NMR of Peptides. In Two-Dimensional NMR Spectroscopy: Applications for Chemists and Biochemists, 2nd ed.; Croasmun, W.R., Carlson, R.M.K., Eds.; VCH Publishers: New York, NY, USA, 1994; pp. 642–643. 29. Brunger, A.T. X-PLOR, Version 3.1: A System for X-Ray Crystallography and NMR; Yale University Press: New Haven, CT, USA, 1992. 30. Badger, J.; Kumar, R.A.; Yip, P.; Szalma, S. New features and enhancements in the X-PLOR computer program. Proteins 1999, 35, 25–33. [CrossRef] 31. Bagby, S.; Harvey, T.S.; Eagle, S.G.; Inouye, S.; Ikura, M. NMR-derived three-dimensional solution structure of protein S complexed with calcium. Structure 1994, 2, 107–122. [CrossRef] 32. Chen, V.B.; Arendall, W.B., III; Headd, J.J.; Keedy, D.A.; Immormino, R.M.; Kapral, G.J.; Murray, L.W.; Richardson, J.S.; Richardson, D.C. MolProbity: All-atom structure validation for macromolecular crystallography. Acta Crystallogr. D Biol. Crystallogr. 2010, 66, 12–21. [CrossRef] 33. Mohd, S.; Rizvi, D.; Shakil, S.; Mohd, H. A simple click by click protocol to perform docking: Autodock 4.2 made easy for non-bioinformaticians. EXCLI J. 2013, 12, 831–857. 34. Trott, O.; Olson, A.J. AutoDock Vina: Improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J. Comput. Chem. 2010, 31, 455–461. [CrossRef] 35. Feinstein, W.P.; Brylinski, M. Calculating an optimal box size for ligand docking and virtual screening against experimental and predicted binding pockets. J. Cheminform. 2015, 7, 18. [CrossRef][PubMed] Int. J. Mol. Sci. 2020, 21, 3076 17 of 17

36. Pettersen, E.F.; Goddard, T.D.; Huang, C.C.; Couch, G.S.; Greenblatt, D.M.; Meng, E.C.; Ferrin, T.E. UCSF Chimera–a visualization system for exploratory research and analysis. J. Comput. Chem. 2004, 25, 1605–1612. [CrossRef][PubMed] 37. DeLano, W.L. Pymol: An open-source molecular graphics tool. CCP4 Newsl. Protein Crystallogr. 2002, 40, 82–92.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).