current topics in medicinal chemistry, 33-62 33 modulation ... · current topics in medicinal...

Current Topics in Medicinal Chemistry, 2007, 7, 33-62 33

1568-0266/07 $50.00+.00 © 2007 Bentham Science Publishers Ltd.

Modulation of Protein-Protein Interactions by Stabilizing/MimickingProtein Secondary Structure Elements

Mª Jesús Pérez de Vega, Mercedes Martín-Martínez and Rosario González-Muñiz*

Instituto de Química Médica (CSIC), Juan de la Cierva, 3, 28006 Madrid, Spain

Abstract: In view of the crucial role of protein-protein intercommunication both in biological and pathological processes,the search of modulators of protein-protein interactions (PPIs) is currently a challenging issue. The development ofrational strategies to imitate key secondary structure elements of protein interfaces is complementary to other approachesbased on the screening of synthetic or virtual libraries. In this sense, the present review provides representative examplesof compounds that are able to disturb PPIs of therapeutic relevance, through the stabilization or the imitation of peptidehot-spots detected in contact areas of the interacting proteins. The review is divided into three sections, covering mimeticsof the three main secondary structural elements found in proteins, in general, and in protein-protein interfaces, inparticular (α–helices, β–sheets, and reverse turns). Once the secondary element has been identified, the first approachtypically involves the translation of the primary peptide structure into different cyclic analogues. This is normallyfollowed by gradual decrease of the peptide nature through combination of peptide and non-peptide fragments in the samemolecule. The final step usually consists in the development of pertinent organic scaffolds for appending key functionalgroups in the right spatial disposition, as a means towards totally non-peptide small molecule PPI modulators.

Keywords: Protein-protein interactions, secondary structure mimetics, α–helices, β–sheets, reverse turns, peptidomimetics,cyclic peptides, non-peptide small molecules.

1. INTRODUCTION

Protein–protein interactions, by their crucial role in mostbiological processes —from intercellular communication toprogrammed cell death—, represent an important group oftargets for therapeutic intervention [1-6]. Many molecularpathways rely on the formation of stable or dynamic proteincomplexes, such as antigen–antibody interactions, theorganization of active sites of oligomeric enzymes orreceptors, by participating in regulatory processes —fromsignal transduction, cell–cell contacts, electron transportsystems, to DNA synthesis—, and during the formation ofskeletal cellular structures. Therefore, many human diseasescan be associated to aberrant protein–protein interactions,either through the loss of an essential interaction or throughthe formation of a protein complex at an inappropriateconcentration, time or location [7-11].

A few years ago, protein–protein systems were consi-dered high-risk targets, difficult to address with conventionalpeptides, peptidomimetics and small-molecules due to therelatively large interaction surfaces involved [4,5,7].However, it might not be necessary for a protein-proteinmodulator to cover the entire protein binding surface. In fact,it is known that in many cases only small segments of theinterface contribute to high-affinity binding (“hot-spot”)[7,12,13]. Therefore, it can be thought that cavities andprotuberances defined by these hot-spots might be filled upor mimicked, respectively, by low molecular weightmolecules. In this respect, a lot of effort is being dedicated tothe dissection of protein interaction hot-spots, theidentification of motifs common to protein interaction

*Address correspondence to this author at the Instituto de Química Médica(CSIC), Juan de la Cierva, 3, 28006 Madrid, Spain; Tel: +34 91 562 20 00;Fax: +34 91 564 48 53; E-mail: [email protected]

interfaces, to mapping protein–protein contact surfaces, andhence to the search of external modulators [12-16]. Specialattention is being paid to a number of small domains presentin various proteins with different biological functions thatare recurrent in many protein-protein interactions. Amongthem, we can cite as representative examples the SH2, SH3,PDZ, WW, RGD, PTB and NPF domains [17].

Apart from antibodies, which can block protein–proteininteractions by sequestering different members of a proteincomplex [18], the prevailing approaches for the discovery ofmodulators of protein–protein interactions are the structure-based design and combinatorial methods (synthetic librariesor virtual screening), as in classical medicinal chemistry[19]. When the three-dimensional structure of proteincomplexes is unknown the search of modulators normallystarts by the selection of peptide modulators, using phage-display or synthetic peptide libraries [20]. When key aminoacids of a protein–protein interface have been identified, thesearch for optimal peptide modulators can be performed byevaluating biased peptide libraries in which all peptidesinclude these key amino acids [21]. In general, combinatorialchemistry and particularly structurally diverse small-molecule libraries are being developed as primary efforts tomodulate cellular signaling by inhibiting, promoting, ormimicking protein–protein interactions, as it is compiled in anumber of reviews [19,22-25].

Concerning design approaches, we can choose to graftessential residues involved in binding on constrainedmolecular scaffolds followed by structure-guided optimiza-tions or, alternatively, to perform computer assisted virtualscreening from existing small-molecules data bases [26,27].Efforts to mimic features of short hot-spot peptides inextended or precise folded conformations have been quitesuccessful, and some reviews have been published to this

34 Current Topics in Medicinal Chemistry, 2007, Vol. 7, No. 1 González-Muñiz et al.

respect [28-34]. In fact, all types of secondary structure(helices, β–sheets, turns and random coil) have been found incontact areas of interacting proteins, and all of them haveindividually been identified as important hot-spots inparticular complexes, as illustrated in the following sections.

In this review, we highlight recent advances in secondarystructure peptidomimetics as modulators of a range ofbiologically significant protein–protein interactions. First, wedescribe some examples where the stabilization of reverseturn conformations was a successful method to obtainprotein ligands. Next, we illustrate the relevance ofβ–strands/β–sheets within certain protein–protein interac-tions and the opportunity of imitating these structural motifsas pharmaceutical tools. In the final section, we discussabout suitable ways to fix or to mimic α–helixconformations to target some therapeutically relevantproteins. In general, starting from peptide sequencesimportant in protein comp-lex interfaces, and afterfixing/stabilizing the corresponding structural secondaryelements through cyclization or by the incorporation ofsuitable scaffolds, is a well suited approach to protein-protein modulators. It is well known that cyclic peptideanalogues and peptidomimetics are advantageous incomparison with their linear counterparts, due to reducedconformational entropy for binding, and protection of amidebonds from proteolytic degradation. In some cases, thisrational approach followed by the gradual decrease of thepeptide character has permitted the discovery of totally non-peptide derivatives of interest.

2. APPROACHES TO GENERATE STABILIZEDREVERSE TURNS

Reverse turns have long been accepted as importantstructural elements in biomolecular recognition, includingpeptide-protein and protein-protein interactions [35]. It hasrecently been described that over one hundred G proteincoupled receptors bind to their respective peptide/proteinligands through turn motifs [36]. In a similar way, turn-likehot-spots have been recognized within several protein-protein interaction surfaces [35,37]. Sites for antigen-antibody recognition, and posttranslational modificationslike phosphorylation, glycosilation, and hydroxylation arealso frequently within turns [35]. Reverse turns are the mostprevalent type of non-repetitive secondary structure element,and can be defined as short peptide fragments that makepossible the reversal of the peptide or protein chain direction[35]. Main turn types can be classified as α–, β– and γ–turns,depending on the number of residues implicated in the bend(five, four, and three, respectively), while the subtypeswithin each category are defined by the φ and ψ dihedralangles of central residues. Turns may or may not bestabilized by an intramolecular hydrogen bond between theCO group of residue i and the i+4 (α), or i+3 (β), or i+2 (γ)amide NH. Due to their structural simplicity, we and othergroups started extensive research programs aimed atdeveloping secondary structure peptidomimetics of thesenon-repetitive motifs, specially β–turns [38-46]. Severalgeneral reviews provide interesting overviews on amino acidderivatives and peptide-derived scaffolds that, whenincorporated into peptide sequences could induce or forcethe adoption of turn-like conformations [47-55]. In this

section we will just focus on key strategies followed to fixturn structures in compounds designed to interfere withintherapeutically relevant PPIs. As mentioned above, the initialapproach usually consisted in the preparation of cyclicpeptides, followed by gradual reduction of the peptidecharacter to reach forward non-peptide compounds. Investi-gations in the integrin adhesion inhibitors field, neurotrophinloop mimetics, and GRB2-SH2 domain-binding ligands wereselected to illustrate this section with representativeexamples.

2.1. β–Turn-Based Peptidomimetics as Antagonists ofIntegrins

Integrins (INs) are a large family of α/β heterodimerictransmembrane glycoproteins that mediate cell-cell and cell-extracellular matrix proteins adhesion, and participate in awide range of physiological processes, such as embryo-genesis, haemostasis, and the immune response [56]. INs arealso implicated in many pathological events, includinginflammation, tumor cell invasion, angiogenesis andmetastasis. Based on this, researchers from both academicand industrial laboratories are focusing on the developmentof antagonists of integrins [57]. The initial efforts weredirected to RGD-containing peptide derivatives, becausemost extracellular matrix proteins and desintegrins containthe Arg-Gly-Asp (RGD) sequence as the common integrinbinding motif.

Kessler´s group has developed different series of cyclicRGD-containing pentapeptides as highly active and selectiveligands for the αVβ3 integrin receptor. The prototype,cyclo(RGDfV) (1, Figure 1), displayed strongly increasedbinding affinity and selectivity with respect to thecorresponding linear peptide, and was able to suppresstumor-induced angiogenesis. Compound 1 adopts a quitewell defined conformation in solution, characterized by aβII’/γ–turn arrangement with D-Phe in the i+1 position of theβII’–turn and Gly as the central residue of the reverse γ–turn,an spatial arrangement that was in some way associated withthe bioactive conformation [58]. However, the incorporationof local modifications, as the reduced amide bond in 2 or theN-Me group in 3, led to important changes in the peptidebackbone conformation, while maintaining or even impro-ving the antagonist activity, respectively. Thus, compound 2exhibits the typical βII’/γ–turn conformation, but with the D-Phe residue at the central position of the γ–turn [59]. Inaqueous solution, cyclopeptide 3 adopts a conformation thatis consistent with a fast equilibrium between two inverseγ–turns at Arg and Asp and a γ–turn at Gly [60]. Furtherreduction of the flexibility of 1 was achieved byincorporating different turn mimetics, as in 4-11 [61,62].From cyclic peptides 4-6, only the spiro-derivative 6 led tothe desired βII’/γ–turn conformation, while in compounds 4and 5 the Gly residue occupied the i+1 position of theβII’–turn, and the turn motifs were shifted to the i+3 and i+4position. The RGD sequence in compound 7 arranges in anensemble of different conformations, probably due to theflexibility of the alkene moiety. The structure-activityrelationships showed inhomogeneous results, with comp-ounds showing similar conformation having very differentactivity, and vice versa. These findings suggest that thestructure determined for these peptides in solution may differ

Modulation of Protein-Protein Interactions by Stabilizing/Mimicking Current Topics in Medicinal Chemistry, 2007, Vol. 7, No. 1 35

from the receptor bound conformation, which is probablyadopted upon binding by an induced fit. Another attempt tomatch the steric demands of integrin receptors involved theincorporation of sugar amino acids as β–turn mimeticreplacements of the D-Phe-Val fragment in the lead structurec(RGDfV) [63]. As hoped, the sugar-modified cyclic peptide8 exhibited high αVβ3 activity, but unexpectedly it alsodisplayed nanomolar affinity at the αIIbβ3 receptor. This wasexplained by a dynamic behavior in which peptide 8 isflexible enough to adopt the kinked and stretchedconformations characteristic of selective binding to αVβ3 andαIIbβ3 receptors.

In a similar way, Scolastico and co-workers described theincorporation of azabicycloalkane and γ–cyclopentane aminoacids as turn mimetics. A small library of RGD-containingcyclopentapeptide mimics, incorporating stereoisomeric 6,5-and 7,5-fused bicyclic lactams, was found to contain high-affinity ligands for the αVβ3 receptor (compounds 9 and 10,Fig. 1) [64,65]. The highly selective αVβ3/αVβ5 antagonist 10also revealed the highest ability to adopt the proper RGDorientation required for binding to αVβ3 integrin, as deducedfrom comparison with the crystal structure of the extra-cellular fragment of this integrin in complex with a cyclicpentapeptide ligand [66]. This conformation comprised aninverse γ–turn with Asp at position i+1 and a distortedβII’–turn with Gly and Asp at i+1 and i+2 positions. A seriesof dihydroxy-, hydroxy- and deoxy γ–aminocyclopentanecarboxylic acid derivatives, grafted onto an RGD tripeptideframework, displayed nanomolar dual binding capabilitiestoward αVβ3 and αVβ5 integrin receptors, with compound 11being the best candidate [67]. Solution NMR structuralanalysis and docking to the extracellular segment of integrinαVβ3 proved that the aminocyclopentane spacer, even if it isa non-isosteric dipeptide replacement, allowed the tetrapep-tide macrocycle to adopt appropriate conformations todevelop potent integrin blocking agents.

It has been demonstrated that a radiolabeled analogue ofcompound 1, in which the Val residue was changed by a Lysmoiety functionalized at the side-chain with a (18F)-sugar, issuitable for imaging of αVβ3 expresion and blockademonitoring using positron emission tomography [68]. Thisseems to anticipate the use of this kind of cyclic peptidederivatives as non-invasive tools to evaluate the role of αVβ3during tumor progression and angiogenesis, both in basicresearch and in clinical studies.

Scientists at SmithKline Beecham reported on the cyclicRGD-containing disulfide derivative 12, a highly activeantagonist of the αIIbβ3, with crystal and solution structurescharacterized by an extended conformation at the Glyresidue, and C7-like conformations in the regions around Argand Asp residues [69]. To test this conformational features,γ–turn mimetics were incorporated at the Asp level in non-peptide analogues. After initial attempts with 2-azepinones[70], extensive investigations on differently substituted 1,4-benzodiazepine non-peptide analogues were carried out.Several potent peptidomimetics were discovered, as exem-plified here by the orally active αIIbβ3 antagonist 13 [71,72].A similar approach led to the identification of non-peptideαVβ3 antagonists, such as 14, revealing the versatility of thebenzodiazepine nucleus as a Gly-Asp mimetic [73,74].

The doubly cyclized peptide ACDCRGDCFCG (15),with disulfide bonds between 1-4 and 2-3 Cys residues,bound the αVβ3 and αVβ5 integrin receptors but did not bindto other closely related integrins [75]. The solution structureof compound 15 displayed a modified type I β–turn at theRGD sequence, while the less potent 1-3, 2-4 cyclizedanalogue formed a type βII’–turn at the same segment,indicating that the presentation of the RGD motif is criticalfor integrin recognition. A PANAM dendrimer RGD conju-gate based on compound 15 has been used for identifyingintegrin receptor expressing cells, through flow cytometryand confocal microscopy [76]. It is expected that thisconjugate could be used to direct imaging agents orchemotherapeutics to angiogenic tumor vasculature.

In the same line of action, and based on the Leu-Asp-Thr(LDT) motif present in the N-terminal Ig-domain requiredfor binding to α4β7 integrins, Kesler’s group designed aseries of cyclic penta- and hexapeptides with definedconformations. Compound 16, as well as other analoguesresulting from systematic exchange of amino acid residueswhile maintaining the backbone structure, effectively inhibi-ted the α4β7 integrin mediated cell adhesion to MAdCAM-1[77]. A biased library of functionalized carbohydrates, asrigid scaffolds to allow the display of the essential sidechains of peptide 16, resulted in the mannose-based antago-nist 17, which mimicked the active conformation of the α4β7selective peptides [78]. Compound 17 have improvedpharmacokinetic parameters and could represent a promisingcandidate for drug development in inflammatory diseasesand autoimmune diabetes.

A bicyclic lactam, combining seven- and four-memberedrings, was envisaged as a non-peptidic scaffold mimickingthe RGD reverse turn topology, to maintain the Arg-to-Aspspatial relationships similar to that of the RGD motif inRGD-containing proteins [79]. Compound trans-18 was ableto disociate the α5β1 IN/fibronectin complex with an IC50value close to that observed for RGDs. It is claimed that, dueto its non-peptide character, derivative 18 could offer a realadvantage from a therapeutic perspective [80]. Similarly, thereduction of the peptide character of the RGD sequencethrough incorporation of the carboxylic acid and aminegroups into piperidine-containing derivatives, while retainingthe β–turn-like structure of fibrinogen, led to the discoveryof the platelet aggregation inhibitor 19 [81]. This injectableαIIbβ3 antagonist is now under clinical trials, althoughresearch is being continued to develop orally activeanalogues.

2.2. Mimetics of Neurotrophins

Neurotrophins (NTs) are highly homologous homodi-meric growth factors involved in the control of cell survival,differentiation, growth cessation, and apoptosis of sensoryneurons [82,83]. Members of this family of proteins include,nerve growth factor (NGF), brain-derived neurotrophicfactor (BDNF), neurotrofin-3 (NT-3), neurotrophin-4/5 (NT-4/5), and neurotrophin-6. Their biological actions aremediated by two classes of cell surface receptors: threetyrosine kinase (Trk) receptors that recognize in a selectivebut not specific way the different neurotrophins (NGF forTrkA, BDNF for TrkB, and NT-3 for TrkC), and the p75receptor at which all neurotrophins bind with similar


affinities [84]. It has been postulated that abnormal NTactions are relevant to neurodegeneration, pain and cancer,providing a foundation for therapeutic intervention. How-ever, clinical trials with these NTs have been disappointing,due to in vivo instability and lack of selectivity that causednon-desired effects. Following the discovery that the Trkbinding is mediated by certain discrete β–turn regions ofNTs, several approaches to develop pharmacological agentsto target neurotrophin receptors, through the search of smalland selective neurotrophin-derived peptidomimetics, havebeen described [83,84].

In the late nineties, two independent groups endeavorsimilar strategies which consisted in the preparation ofdisulfide-bridged cyclic peptides corresponding to β–loopregions of NGF. The group of Saragovi prepared and studiedthe generation of monomeric cyclic mimetics, such ascompound 20 (Fig. 2), which displaced [125I]NGF binding toTrkA receptors and specifically inhibited optimal NGF-mediated neurite outgrowth in PC12 cells [85]. NMR studiesof this peptide antagonist showed that its solution structure ischaracterized by a β–turn conformation at Asp93-Glu-Lys-Gln96 residues, strikingly similar to that of the correspondingregion of NGF [86,87]. After the first proof-of-principle,Longo and co-workers described a new series of NGF pep-tide derivatives, exemplified by the 29-35 region analogue21, that promoted neuronal survival via a p75-dependentmechanism [88].

From the previous data about monocyclic monomericpeptides, and taking into account that the NTs are homo-dimeric proteins, the next advance was to design appropriatedimeric mimetics that could bring about Trk receptorshomodimerization to mimic the actions of NTs. Thus, adimeric peptidomimetic related to 20 demonstrated NGF-likeneurotrophic activity, while a bicyclic dimeric peptide,designed to mimic loop 2 of BDNF, behaved as partialagonist and was particularly potent in promoting neuronalsurvival in vitro [89,90].

In an attempt to diminish the size and peptide characterof the β–loops of NTs, the group of Burgess designed afocused library of β–turn small peptidomimetics combiningpeptide and non-peptide fragments. Compound 22 wasidentified from this library as a selective and proteolyticallystable agonist of the TrkA receptor [91], with ability torescue cholinergic neurons from the cortex and the nucleusbasalis and to improve memory/learning in impaired agedrats [92]. The next step was to produce more rigid turnanalogues by incorporation of an additional phenyl ring.Among this new series of compounds, which adoptedconformations that approximate β–turns, the NGF mimetic23 showed partial agonist activity at TrkC receptor andenhanced the trophic activity of NT-3 in cell survival assays[93]. The closely related analogue 24 selectively bound theNT-3 receptor TrkC, displayed NT-3-like neurotrophicactivity, and induced tyrosine phosphorylation of the TrkCreceptor [94]. After the initial success, they constructed aminilibrary with the i+1 and i+2 residues of the key β–loopsof NGF and NT-3 anchored into the bis-phenyl skeleton.Remarkably, a high rate of hits showing neurotrophicactivities or leading to neuronal differentiation or both wereidentified [95]. Authors claimed that this kind of focused

libraries, designed specifically to target receptor hot-spots,may be a general approach to discover functional ligands forPPIs. The assembly of this kind of monomeric peptido-mimetics into dimeric structures has once again been explo-red, leading to some homo- and heterobivalent molecules forfurther investigation [96,97]. The same group at TexasUniversity has recently published the synthesis of interestingtriazole-based β–turn mimetics, as 25, able to favor type Iand type II β–turn conformational states, but no indicationabout biological activity has yet been reported [98].

2.3. Antagonists of the SH2 Domain of Grb2

Growth factor receptor bound protein 2 (Grb2) is anubiquitously expressed adaptor protein possessing a singleSrc homology 2 (SH2) domain flanked by two SH3 domains.Grb2 plays a crucial role in signaling from the receptortyrosine kynases (RTK) to the Ras-MAPK cascade, workingas a bridging element between cell surface growth factorreceptors and the Ras protein [99,100]. An aberrant Grb2-SH2 dependent Ras activation pathway has been described tocontribute to processes important for cancer development(i.e., cell proliferation, apoptosis, and metastasis) [101,102].Therefore, the discovery of compounds that selectivelyantagonize the SH2 domain of Grb2 should be attractive inoncology.

It is known that upon RTK activation the Grb2 SH2domain links phosphotyrosyl peptides with the consensussequence pYXNX, which normally adopt folded β–turnconformations, with the Asn residue in the i+2 position[103]. The pioneering research by Novartis started byevaluating phosphotyrosyl peptide libraries, which uponoptimization led to a high-affinity and selective compound(26, Fig. 3) [104]. In this phosphopeptide the α,α–cyc-loamino acid stabilized the β–turn conformation required forefficient recognition by the Grb2-SH2 domain. Furthermodifications at different parts of the molecule resulted in aseries of inhibitors (27-30), some incorporating phosphonateand malonyl phosphotyrosyl mimetics, and others indolederivatives instead of the naphthalene moiety, which showedpotent activity in cell-based assays and improved stability tocellular phosphatases [105-108].

A step ahead in this field was achieved by Burke’s groupthrough the preparation of macrocyclic variants of the linearGrb2 SH2 domain antagonists in an attempt to tightly fix theβ–turn conformation. Using olefin metathesis reactions, theyfirst prepared the tripeptide mimetic 31 that exhibited a 100-fold enhancement in binding potency with respect to thelinear analogue, but was largely ineffective in whole-cellassays [109]. However, the incorporation of an extracarboxymethyl group at the pTyr mimetic α–position of themacrocyclic ligand, as in the linear analogue 28, affordedcompound 32 with enhanced Grb2 SH2 domain binding inextracellular assays and high efficacy in whole-cell systems.Peptidomimetic 32 displayed antiproliferative effects withsubmicromolar IC50 values in breast cancer cell cultures[110]. The related 5-methylindolyl-containing macrocycle 33displayed the highest affinity yet reported for a syntheticinhibitor against the SH2 domain [111]. This compoundblocked the association of Grb2 to erbB-2 in an effectivemanner, and showed antimitogenic effects against erbB-2-dependent breast tumors at non-cytotoxic concentrations.


Fig. (1). Turn-constrained peptidomimetics to interfere with integrin functions.

NH

N

Me

O

CO2H

R

R

HN NO

NN

NHMe

O

13 14

NH

N

O

O

HN

NR

O

X

N

O

HO2C

HN

NH2

NH

H

H

1: X = O, R = H2: X = H2 , R = H3: X = O, R = Me

S S

HNONHN

NH

Me

NH

H2N

NH

O

O O

CO2H

12 H2N

HN

NH

HN

NH

O

O

O

CO2 HO HN

NH

NH2

HN O

NH

O

NH

HN

NH

CO2 H

OS

S

O

OS

S

HN

O

HO2C

15

O

OBnBnO OBn N

( )n

O

NH

N

O

O

HNO N

O

HO2C

HN

NH2

NH

H

H

TurnMimetic

N

O

N

SH

NN

O

4 5 6 7

Turn mimetic

8 9: n = 110: n = 2

11

ON

O

CO2H

HN

NHHN

NH2

NH

HN

N

HN

HN

NH

O

O

CO2H

O

OHO

OO

O O

O

O

O

HO

CO2HO

N

HN

O

NH

O

NHAcHO2C

1816 17 19


Fig. (2). Representative examples of neurotrophin derived peptidomimetics.

Good biological results were also obtained after similarmacrocyclization of other linear derivatives containingdifferent phosphotyrosine mimetics, such as 29, but they didnot improve those encountered for 32 and 33 [112,113]. Thecrystal structure of an analogue of 32, bearing a malonylphosphomimetic, with the Grb SH2 domain allowed todetermine the binding mode and the specific interactionsbetween the protein and the inhibitor, which could serve torationalize the future design of this kind of protein-proteinmodulators [114]. To facilitate the identification of theintracellular targets of these macrocyclic inhibitors, a potentbiotinylated analogue of macrocycle 32 has very recentlybeen synthesized and demonstrated to have nanomolaraffinity for the Grb2 SH2 domain [115].

A family of non-phosphorylated ligands for the Grb2SH2 domain has been developed by Roller and co-workersfrom a phage library of 10-mer thioether linked cyclicpeptides [116]. Gradual reduction of the peptide nature andsystematic SAR studies provided compound 34 as the bestGrb2 SH2 inhibitor within this series [117,118]. This cyclicpentapeptide exhibited potent Grb2 SH2 domain binding

affinity (58 nM) and remarkable antiproliferative activityagainst erbB-2-dependent breast cancer (IC50 = 19 nM).Compounds of this family bear the YXNX sequence, whichadopts β–bend type conformations facilitated by the presenceof the β–turn inducing amino acid 1- aminocyclohexa-necarboxylic acid and an R-configured sulfoxide [118,119].Although 34 and analogues do not reach the nanomolar andeven picomolar binding affinities found in the previousseries of phosphotyrosine mimetics, they provide a noveltemplate for the development on non-pTyr-containing Grb2SH2 domain antagonists with enhance bioavailability.

2.4. Miscellaneous Examples

Related to the stabilization of turns as a way for modu-lating PPIs, one of the first examples yet described refers tothe preparation of HIV protease inhibitors [120]. Thisenzyme, a member of the aspartic protease family, processeshigh molecular weight viral polyproteins into structuralproteins and enzymes. The occurrence of a Pro residue at theP1’ position of many substrates, and the frequent appearanceof Pro in turns, drove the sequential incorporation of γ–turnmimetics into model peptides. Peptidomimetic 35 (Fig. 4)

HN

HN

NH

O

HN

O

NH

OO

OHN

NH2

NH2CO2H

HN

O

S

SO

H2N

HN O

H2NOC

OH

HNO

NH

O

O

HNNH

O

O

O2N

HO2C

NH2

CO2 H

HN

NH

O

HN

O

NH

O

OHN

CONH2

SO

NH

O

HO

OH

S

HN

O

HO2 C

HO

NH

CO2H

O

O

NH

CO2H

NH2

HNO

R1NH

O

O

HN

R2

X

R3

R4 HNO

R1NH

O

O

HN

R2

NN

N

20 21

22 23: R1 = (CH2 )2CO2H, R

2 = (CH2)4NH2,

R3 = NH2, R4 = CO2H, X = S

24: R1 = (CH2 )4NH2, R2 = CH2OH,

R3 = NO2, R4 = CONH2 , X = NH

25


discovered from this research program, showed goodinhibitory activity against the HIV protease, being 44-foldbetter than the corresponding linear analogue.

Also in the HIV field, an important PPI is the interactionbetween gp120 and the CD4 receptor, which is critical forthe initial steps of HIV infection. Viral entry into the targetcell is primarily mediated by the interaction of gp120 withCD4 through highly conserved regions, followed byconformational changes of gp41 that causes the fusion of celland viral membranes. The X-Ray structure of CD4 fragmentsrevealed that the Gln40-Phe43 adopted a surface-exposedβ–turn, and likewise mutagenesis studies attributed a

significant role to Phe43 residue for binding to gp120. Guidedby these facts, stable mimetics of the complementarity-determining 2-like region of CD4, as compound 36, weredesigned and found to possess low micromolar Kd forhuman T-lymphotropic viral gp120 and reduced syncytiumformation [121,122]. A target phage-displayed library wasconstructed to identify novel peptides able to inhibit gp120-CD4 recognition process. A cyclic nonameric peptide,designed from one of the best linear peptides, exhibited goodinhibitory activity (50-fold higher than that of the linearcounterpart) [123].

Fig. (3). Stabilized turn peptidomimetics recognized by the Grb2-SH2 domain.

R =N

H3 C

29 30

NH

CONH2

O

HN O

NH

O

YX

PHO

O

OH

R NH

CONH2

O

HN O

NH

O

HNHO2C

CO2 H

O

CO2H

NH

CONH2

O

HN O

NH

O

PHO

O

OH

Y

N NH

CONH2

O

HN O

NH

O

PHO

O

OH

H3C

CO2 H

HN

HN

NH

HN

CO2H

HO2C

O

O

OH

NH2

OHN

CONH2O

NHHN

OH2NOC

O

S O

O

(R)

26: X= O; Y = NHAc

27: X= CH2; Y = NHAc

28: X= CH2; Y = CH2CO2H

31: Y = H

32: Y = CH2CO2H

33

34


The interaction between cell adhesion molecules CD2and CD58 is critical for immune response in autoimmunediseases. Therefore, modulation of this interaction could betherapeutically useful, and inhibitors may function asimmunosuppressants. Based on the structure of the CD2-CD58 complex, several cyclic peptides were designed tomimic the β–turn and β–strand hot-spot regions of the CD2protein. Some of these compounds, exemplified by 37,showed T-cell adhesion in different assays and adopted welldefined β–turn conformations in solution that closely mimicthe turn structure of the surface epitopes of CD2 protein[124,125]. Further studies will be required to elucidate thereal biological potential of this approach.

Non-peptide second mitochondria-derived activator ofcaspase (Smac) mimetics, containing well-characterizedbicyclic indolicidine and perhydropyrrolo[1,2-a]azepineβ–turn inducers, were described as modulators of the X-linked inhibitor of apoptosis protein (XIAP)/caspase-9

interaction involved in the regulation of apoptosis [126,127].Compound 38 displayed a Ki of 25 nM to XIAP BIR3domain, being 23 times more potent than the natural Smacpeptide. These peptidomimetics could have therapeuticpotential as anticancer drugs for overcoming apoptosisresistance of cancer cells with high level of IAP proteins.

A conceptually different biological task, triggered byconformationally constrained peptidomimetics, can be foundin the synthetic vaccines field. In this case, mimetics shouldfunction by stimulating the immune system to produceantibodies that recognize the intact parasite. NMR andmolecular modeling studies, as well as the recently reportedcrystal structure of the NPNA-repeated region of thecircumsporozoite (CS) protein of the malaria parasitePlasmodium falciparum revealed that this tetrapeptide motifadopted a type-I β–turn structure, stabilized by hydrogenbonding between the CO of the first Asn residue and the NHof Ala [128,129]. Using this NPNA motif as a model, a few

Fig. (4). Miscelaneous examples of reverse turn-based mimetics.

S S

H-Pen-Ile-Tyr-Asp-Thr-Lys-Gly-Lys-Asn-Val-Leu-Cys-OH

NNH

O

H2 NO NH

O

HN

CO2CH3

N

HN

HN

O

O

H

HNAla

Asn

Asn Pro

AsnAla

CO2H

O

O

Ala

AsnPro

Asn

NH

NH

N

OO

O

OH

O

NH

NH2

HO

OOH

H

Bn

H2NNH

N

O

O NH

O

NH

HN

HN

HNO

O

O

OHN

CO2H

NH2

OHO-Asn-Trp-Leu-Ser

Ac-Glu-Leu-Leu-Glu

HN

O

HN

35

37

39

36

38

40


cyclic peptidomimetics, such as 39, were mounted intodifferent template carriers and evaluated for their ability toinduce antibody responses [130,131]. Some conformation-dependent very promising results were found with thesecyclic peptidomimetics, while linear peptides containing thesame sequences failed to induce a detectable cross-reactiveimmune response.

A segment of gp41, containing the sequence Glu-Leu-Asp-Lys-Trp-Ala, has been identified as the epitope of theHIV-1 neutralizing human monoclonal antibody 2F5, and thecrystal structure of the complex revealed that this epitopemainly adopt a β–turn conformation. Cyclic peptides (i.e.,40), incorporating a side-chain to side-chain lactam bridgebetween the i and i+4 residues, induced a defined reverseβ–turn, showed good antibody binding and were highlyimmunogenic, although they were incapable of stimulating aneutralizing response [132].

3. STRATEGIES TO FIX β–SHEET EPITOPES

The second major structural element found in globularproteins is the β–sheet. This structure is built up from acombination of several regions, named β–strands, of thepolypeptide chain. The β–strands are usually 5 to 10 residueslong and are in an almost fully extended conformation, withamide bonds being almost coplanar and side chainsalternating above and below the plane of the peptidebackbone. β–Strands are aligned adjacent to each other insuch a way that hydrogen bonds can be formed between COgroups of one β–strand and NH groups of an adjacentβ–strand, and vice versa, to form a pleated sheet. The aminoacids in the aligned β–strands can all run in the samebiochemical direction, parallel β–sheet, or can havealternating directions, to form an antiparallel β–sheet.β–Strands can also combine into mixed β–sheets with someβ–strand pairs parallel and some antiparallel. Withinβ–sheets, one of the simplex supersecondary structures areβ–hairpins, which are formed between two antiparallelβ–strands connected by a β-turn.

Interactions between the hydrogen-bonding edges ofβ–strands in different protein chains constitute an importantmode of PPIs, with hydrogen bonds being formed betweenβ–strands belonging to different protein chains. Theseinteractions are frequent in protein dimmers and otherquaternary structures, both in interaction among differentproteins and in protein aggregation [127]. Although thesePPIs occur in non-pathological processes, they are alsoinvolved in a wide range of diseases, such as cancer [128],parasitic, fungal and viral infections [129-132], inflam-matory, immunological and respiratory conditions [133,134],and cardiovascular [135] and degenerative disorders,including Alzheimer’s disease [136].

Although there are some chemical models of proteinβ–sheets, based on templates that allow the transplantation ofseveral β–strands onto them, there are no binding or otherbiological studies on these engineered β–sheets, so theirdescription is out of the scope of the present review. Thus, inthis section we will focus on several β–sheets mimetics thathave shown their capacity to exert a particular biologicalactivity by interfering with different PPIs. We differentiateamong peptidomimetics designed to imitate the distinct

elements of the β–sheet, from the simplest β–strandmimetics, to compounds developed to mimic β–hairpins orto contain functional groups of more complex β–sheets.

3.1. β–Strand Mimetics as Protease Inhibitors

Although isolated β–strands are not common, theyconstitute crucial structural elements recognized by, forexample, proteolytic enzymes, major histocompatibilitycomplex (MHC) proteins, and transferases. Moreover,β–strands mimetics are of interest as inhibitors of β–sheetsaggregation, a phenomenon which is associated with anumber of neurological disorders, such as Alzheimer’s,Huntington’s and Parkinson’s diseases.

The simplest approach in the design of β–strand mimeticswould be to use short peptides corresponding to strand/sheetregions of proteins. However, it is known the disadvantageof using short peptides, due to their conformationalflexibility and poor pharmacological profiles. Methods toincrease the conformational stability involve the restrictionof the peptide freedom by incorporation of conformationalconstraints into the peptide backbone, or through differenttypes of cyclizations.

A major effort in the field of β–strand mimetics is relatedto the search of protease inhibitors. Proteolytic enzymes areclassified by the nature of their active-site catalytic residuesas metallo (34%), serine (30%), cysteine (26%), aspartic(4%), and the less characterized threonine (5%) proteases.These enzymes regulate numerous biochemical, physiolo-gical and pathological processes by controlling proteinsynthesis and degradation through hydrolysis of specificamide bonds. Consequently, control over protease expressionand function can potentially be an effective strategy fortherapeutic intervention [137]. In fact, a good number ofprotease inhibitors have already been marketed.

A recent review by Fairlie’s group shows the analysis ofover 1500 three dimensional crystal (X-ray) and solution(NMR) structures of substrates, products and inhibitorsbound in the active sites of all five protease classes. Thisstudy has shown that proteases recognize peptide and non-peptide ligands in an extended β–strand conformation, withonly a few exceptions [138]. Moreover, reviews from thesame authors [139,140] have covered β–strand mimetics upto 2004. Therefore, here we just illustrate the differentapproaches to β–strand mimetics with examples of inhibitorsof the human immunodeficiency virus (HIV) protease(HIVPR), and give some details of new inhibitors developedin the last two years.

A first approach to mimic extended strands is the use ofsmall carbo− and heterocyclic structures to rigidify thepeptide backbone and to arrange functional groups inapproximately the orientation they have in native β–strands[139,140]. Several examples can be found within the field ofinhibitors of HIV protease. HIV protease is an asparticprotease essential for viral replication and infectivity [141].Structurally, HIV protease functions as a C2-symmetrichomodimer with a single active site. Accordingly, the active-site symmetry has been the base for the design of potent andselective inhibitors of this enzyme. Another strategy quitefrequently applied is the replacement of the scissile amidebond by surrogates of the peptide bond. As several 3D


structures of substrates or inhibitors bound to this enzymeare known, design criteria have also taken this into accountto improve both potency and bioavailability. Thus,modification of an initial linear peptide through the use ofstructure-based design led to the potent enzyme inhibitor 41(IC50 = 6.1 nM, Fig. 5) [142]. In an alternative strategy, apyrrolinone structural motif, known to force β–strandconformations, has been introduced into the backbone of aknown peptide inhibitor providing compound 42 [143].Since the irreversibility of these inhibitors would conferlonger term efficacy in vivo, a series of inhibitors weredesigned containing a cis-epoxide instead of the scissileamide bond. The incorporation of an epoxide, as incompound 43 (IC50 = 20 nM) [144], led to inhibitors with atime-dependent irreversible pattern, as the nucleophilicoxygen of Asp side chain is added to a carbon of the epoxidering to form an ester bond with simultaneous epoxide ringopening. Finally, it is worth pointing out that the search ofβ–strand mimetics have provided a series of compoundscurrently in use for the treatment of HIV infection, asritonavir (44). This compound was developed after intensivestudies of a series of analogues whose design was based onthe active-site symmetry of HIV-1 protease. Ritonavirshowed good antiviral activity and exceptional oralpharmacokinetics [145].

Examples of inhibitors of other aspartic proteases(plasmepsin II, renin), serine proteases (trypsin, pancreaticelastase, t-plasminogen activator, coagulation factor Xa,thrombin), metalloproteases (several matrix metallo-proteinases, angitensin-converting enzyme-1, interleukin-1βconverting enzyme, adamalysin II) and cysteine proteases(cathepsin B, papain, caspase 3) have been described. Thereare also examples outside the field of proteases, as inhibitorsof farnesyl transferase (FTase), human class II MHC proteinHLA-DR1 (antigen), and protein tyrosine phosphatase(PTPase) [139,140].

In the last two years efforts have also been directed tofind other rigid scaffolds that can be used as replacements ofsegments of peptide backbones to pre-organize the moleculein its bioactive conformation, namely β–strands for proteaseligands. Mempapsin 2 (BACE), a membrane-bound asparticprotease, is a key enzyme associated with the processing ofamyloid precursor protein to generate β–amyloid peptide, thecausative agent of amyloid plaques in Alzheimer’s disease.An inspection of the X-ray structure of a heptapeptideinhibitor bound to this aspartic protease poved that the ligandadopts a β–strand conformation [138]. A cyclopentane ringwas designed as a replacement for a segment of thehydroxyethylene isostere in the heptapeptide derivative,affording compound 45 (IC50 = 39 nM, Fig. 5) [146].Previous structure-activity relationship (SAR) studies onsubsite specificity, and X-ray of the complex of the enzymewith various inhibitors, suggested the replacement of thecyclopentane for more polar analogues, and consequentlyseveral compounds incorporating a variety of heterocycleswere designed. Among them, lactam derivative 46 showedan IC50 < 10 nM, and a 7-fold increased selectivity over itshuman homologue cathepsin D (Cat D).

Another recent example of protease inhibitor is related toPlamesin (Plm) I, II and IV, the haemoglobin degrading

aspartic proteases from Plasmodium vivax and Plasmodiumfalciparum, protozoan parasites that cause malaria. Startingfrom a series of C2 symmetrical inhibitors with high affinitiesto both Plm I and II, a 1,3,4-oxadiazole heterocycle was usedas a replacement of one or two amide bonds of the parentpeptide. This lead to the symmetrical compound 47 (Fig. 5)with high affinity toward Plm IV (Ki = 35 nM) [147].

Other strategies in the search of proteases inhibitors havejust involved optimization of previously designed mimeticsthat incorporate different heterocycles. In this sense, modi-fied inhibitors have been developed for cathepsin K (Cat K)[148], a cysteine proteases crucial in bone remodelling, andplasmin [149], a serine protease that plays an important roleduring endothelial cell extracellular matrix (ECM)remodelling.

A second main approach to mimic β–strands refers to thepreparation of macrocyclic peptides able to keep the peptidebackbone or equivalent residues in an extended β–strandconformation. The design of macrocyclic compounds hasbeen based on the known 3D structure of several proteinsbound to their ligands that show close proximity ofalternating amino acids side chains from the ligand. Thestrategy followed was to condense together these side-chainsto generate several highly constrained macrocyclic mimics oftri- and tetrapeptide components of linear peptides. In thisrespect, an area of intensive research is the development ofHIV protease inhibitors. The HIV protease substrates Ac-Ser-Leu-Asn-Phe-Pro-Ile-Val and Ac-Leu-Val-Phe-Phe-Ile-Val-NH2 were easily converted to potent competitiveinhibitors by replacing the scissile amide bond (-CONH-)with non-cleavable transition state isosteres (i.e. –CHOH-CH2-) [150]. On the base of the X-ray structure of thesederivatives bound to HIVPR, Fairlie’s group designedseveral 15 and 16 member macrocycles by linking togetherthe side chains of the first and third amino acids, flankingeither the left or the right side of the scissile amide bond[150,151]. The resulting C- and N-terminal macrocyclicderivatives [48 (IC50 = 12 nM) and 49 (IC50 = 0.6 nM), Fig.6] resulted in good inhibitory potencies against HIVPR.Additionally, bis-macrocyclic inhibitors [50 (IC50 = 3 nM)][151,152], and macrocycles functionalized at both N- and C-terminus, suitable for addition of appendages at either end[51 (IC50 = 0.6 nM)] [153,154], have also been designed.Theoretically, it might be possible to independently varyeither the acyclic N- or C-terminus without affecting theinteractions between the macrocyclic C- or N-terminus,respectively, and the enzyme. In these sense, cyclictripeptide mimics have been appended to some non-peptidemoieties without loss of protease activity [155].Additionally, both N- and C-terminal macrocycles have alsobeen modified through focused combinatorial chemistry tocreate a series of potent inhibitors that bind in a predictableway within the active side of the enzyme [156]. Whenavailable, the 3D structure of the complex between theseanalogues and HIVPR showed that these inhibitors arebound in an extended conformation [138]

Following a similar approach a series of macrocyclicligands of other proteases have been designed [139,140].Among them, we can find aspartic proteases, as rennin,plasmepsin II, and penicillopepsin; metalloproteases, as


Stromelysin-1 (MMP3), angiotensin converting enzyme(ACE), aminoprotease B and tumour necrosis factor αconverting enzyme (TACE); and serine proteases as,thrombin, trypsin and streptokinase.

Other examples of macrocyclic enzyme inhibitors haveappeared in the literature during the last two years. The firstone is related to inhibitors of plasmepsin I, II and IV, themalarial aspartic proteases essential for parasite survival.Since aspartic proteases usually bind substrates/inhibitorsadopting an extended β–strand conformation, andmacrocyclization has proved to be an effective constrain,several 13 and 16-membered macrocycles were designed asinhibitors of Plm [157]. In this case a ring-closing metathesis

methodology was used to avoid incorporation of newhydrogen bond-accepting/donating amide groups into themacrocycle. Among them, the 13-membered macrocycliccompound 52 (Fig. 6) showed high affinity for Plm I, II andIV, and high selectivity over its homologous human asparticprotease Cat D. Other macrocyclic derivatives have recentlybeen designed as inhibitors of different proteases, althoughauthors do not explicitly mention if these compounds weredesigned to mimic β–strands. In general, the design metho-dology towards protease inhibitors involves the incorpora-tion of a non-hydrolizable amide bond, and/or a cyclizationbetween two close residues. This trend of thought has beenquite a general design criterion in the search of protease

Fig. (5). Structures of several β–strand mimetics as protease inhibitors.

H2N

O

NH2O

NH

CO2H

OH

HN

ONH

O

O

HN

O

OH

N

O

HN

O

OH

O

HN

HN

O

O

NH2

O

OHHNO

O

N

HN

ONH

O

O

OH2N

O

HN

N

Br

N

OO

N

O

Br

OH

OH

O

N

NH

HN

O N

O

NH

O

O

HN

O

S

OH

NH

O

N NH

OHN

O OH

NH

O

O

N

S

N

S

41 42

43 44

45 46

47


inhibitors, as mentioned above, for which the adoption ofβ–strand conformations was recognized after the crystalcomplex structure has been solved [138]. Thus, it is likelythat these new macrocyclic inhibitors also adopt this type ofsecondary structure. Examples are related to HIV-1 protease[158], memapsin 2 (BACE, β-secretase) [159], and chronichepatitis C virus (HCV) non structural 3 (NS3) protease[160-163].

3.2. CXCR4 Antagonists by Stabilizing β–Hairpins

Chemokine receptor, CXCR4, is a seven transmembraneG-Protein Coupled Receptor (GPCR) that transduces signalsof its endogenous ligand, the stromal cell-derived factor-1(SDF-1). CXCR4 has been identified as a coreceptor that isutilized in T cell line-tropic (X4-)HIV-1 entry. Besides, it isexpressed in malignant cells from different types of cancers.Thus, this system has been involved in several diseases,including HIV infection, cancer metastasis/progression, andrheumatoid arthritis, making it an attractive therapeutictarget. Fujii and co-workers have found a CXCR4 antago-nist, compound 53 (Fig. 7), which is an 18-residue peptidethat inhibits T-cell line-trophic HIV-infection. Peptide 53 isalso able to bind specifically to both gp120 (an envelopeprotein of HIV) and CD4 (a T-cell surface protein) [164].This compound derived from chemical modifications of self-defense peptides of horseshoe crabs [165], and posses anti-HIV activity comparable to that of AZT, currently in use for

the therapy of AIDS patients. Derivative 53 has anantiparallel β–sheet structure with a type II β–turn, which ismaintained by two intramolecular disulfide bridges [166].Several studies recognize the importance of the disulfidebridges, especially the major disulfide loop, and the β–sheetstructure for the expression of high anti-HIV activity, alongwith a positive charge in the side chain at the (i+1) positionof the β–turn region. In addition, Trp3 can be replaced byother aromatic residues [Tyr, Phe and L-2-naphthylalanine(Nal)] [164]. It has been shown that a 14 residue analogue(54, Fig. 7), having only one disulfide bridge maintains boththe anti-HIV activity and the conformational structure of 53[165,167,168]. SAR studies on 54 allowed the discovery ofthe citruline (Cit) derivative 55 and its analogue 56 (Fig. 7)with increased anti-HIV-1 activity and diminished cytoto-xicity [167,168]. Conformational studies on 56 showed thatit adopts an antiparallel β–sheet conformation, with fourpharmacologically significant residues, Arg2, Nal3, Tyr 5 andArg14. Based on this, cyclic pentapeptides were used asmolecular templates to dispose these four residues intoproximity. Among this series of cyclic pentapeptides,compound 57 exhibited strong CXCR4 antagonist activity,comparable to that of 56 [170]. However, solution confor-mational studies did not show an exact correlation of thespatial disposition of the pharmacophore groups of 57 withthose of 56. Subsequent SAR studies on this series give anidea about the importance of the Arg, Nal and D-Tyr

Fig. (6). Structures of representative macrocyclic HIV protease inhibitors.

OH

O

O

HN

OH

NH

O

O

OH

NH

OHN

O

HN

O

OHHN

Ac-Leu-Val

HN

O

OH

Pro-Ile-Val-NH2

CONH2

NH

O

O

NH

OHN

O

HN

O

HN

O

NH

O

O

OH

BocHN NH

OH

HN

O

NH

O

CO2 Me

O

48

52

49

50 51


residues at position 2, 3 and 5, together with several amidebonds [171,172]. Following the discovery of peptide 56,other cyclic tetrapeptides, including a γ-amino acid, andpseudopeptides cyclized through disulfide and olefin bridges,and therefore having a smaller number of peptide bondscompared to the parent compound, have also been prepared[173]. Very recently, a diketopiperazine (DKP) mimetic(compound 58, Fig. 7), having a (Z)-alkene unit instead ofthe cis-amide bond of DKP, has been used by Fujii and co-workers as an scaffold able to support the important guanidyland naphthyl side chains of 57. Although derivative 58showed significant CXCR4 antagonist activity (IC50 = 15.1µM), it did not reach the potency of the parent cyclicpentapeptide [174].

3.3. Non-Peptide β–Sheet Mimetics as ICAM-1/LFA-1Modulators

The interaction of leukocyte function-associated antigen-1 (LFA-1) with the intracellular adhesion molecule-1(ICAM-1) is critical to lymphocyte and immune systemfunction, and has been implicated in numerous autoimmunediseases such as psoriasis, asthma and rheumatoid arthritis.Therefore, antagonists of this interaction could be of mayortherapeutic interest. The binding region of both moleculeshas been characterized by antibody binding, mutagenesis andcrystallographic studies. In particular, the binding epitote ofICAM-1 that interact with the I domain of the αL subunit ofLFA-1 is a discontinue region, encompassing residuesspanning four β–strands comprising one β–sheet [175,176].ICAM-1 mutagenesis studies identified six residues

important for the interaction, with Glu34 and Lys39 beingcritical for LFA-1 binding [175,177]. A homology modelindicated that these residues opposed each other on twostrands of an antiparallel β–sheet [175]. Thus, the simplerapproach to mimic this situation would be to traverse fromα-carbon to α-carbon crossing the H-bond network betweenthe strands. This data, together with extensive SAR studieson different inhibitors of LFA-1/ICAM binding, guidedGadek and co-workers to design a class of nanomolar small-molecule LFA-1 antagonists, illustrated by 59 and 60 (Fig. 8)[178]. A comparison of the structure and molecularfunctionality of these compounds with ICAM-1 led topropose that these derivatives are mimics of ICAM-1, andbound in a similar manner to the LFA-1 inserted domain (Idomain). In contrast, several authors, on the base of indirectbinding studies of these small molecules to LFA-1,concluded that they bind to a different epitote on LFA-1, inparticular the β2 subunit I-like domain, triggering epitoteschanges within and across LFA-1 domains [179-181]. Morerecently, a direct comparison of the binding of ICAM-1 andthat of these small molecule antagonists [182] concluded thatthere are two distinct binding sites, one of high-affinity in theαL subunit of LFA-1, overlapping the ICAM-1 binding site,and a second of lower-affinity in the β subunit. The bindingof the ICAM-1 mimetics to this low affinity site might bedue to the sequence homology between I and I-like domains.

3.4. Miscelaneous Examples

Several approximations to mimic β–hairpins involvetransplantation of the β–hairpin loop sequences from thenative protein onto a template that fixes the N- and C-termini

53 R R W C Y R K C Y K G Y C Y R K C R-NH2

54 R R W C Y R K dK P Y R K C R-NH2

55 R R W C Y R K dK P Y R Cit C R-OH

56 R R Nal C Y R K dK P Y R Cit C R-OH

Fig. (7) . Amino acid sequences of 53 and several of its analogues, aligned based on their homology. The disulfide linkages between Cysresidues are shown by solid lines. Structure of the cyclic peptide 57, and its DKP analogue 58.

N

HN

NHTFAH2.N

O

NH

NH2.TFA

NH

58

NH

HN

O

NH

NH

O

HNO

O

O

NH

NH2

HN

HN

H2 N

NH

HO

57


of the loop into a β–hairpin geometry. One of these hairpin-inducing templates is the dipeptide D-Pro-L-Pro, extensivelyused by Robinson´s group [183-188], which is able tostabilize several attached loops in β–hairpin conformation.This template has been applyed to retain the β–hairpingeometry of cyclic peptide trypsin inhibitors, and ofinhibitors of the binding of Protein A to a human antibodyFc fragment.

The premature activation of trypsin, one of the threeprincipal digestive proteinases, has been related to eventsthat lead to pancreatitis, a disease with substantial morbidityand mortality [189]. Thus, inhibitors of this protease mightbe of use in the prevention and treatment of this life-threatening illness. A 14 amino acid cyclic peptide fromsunflower seeds has proved to be a trypsin inhibitor. From itscrystal structure in a complex with trypsin it was shown thatthe inhibitor has a well-defined β–hairpin loop. To produceconformational mimetics of this natural product with correctgeometry, either 11 or 7 residues of its hairpin weretransplanted onto a D-Pro-L-Pro template [184], leading tomimetics 61 and 62 (Fig. 9). NMR studies of thesecompounds revealed clear β–hairpin conformations in bothcases, besides the longer mimetic displayed similar potencyto the natural product, whereas the shorter analogue was ninefold less active.

Another example of the utility of the D-Pro-L-Proscaffold is related to the disulfide bridged peptide FcIII(Asp-Cys-Ala-Trp-His-Leu-Gly-Glu-Leu-Val-Trp-Cys-Thr),which interacts with the Fc domain inhibiting its binding toprotein A. The hinge region on the Fc fragment of humanimmunoglobulin G interacts with at least four differentnatural protein scaffolds, including protein A [190]. Fcbinding ligands may be of value for biotechnologicalapplications, as replacement for recombinant proteins, suchas Protein-A, in the affinity chromatography of therapeuticantibodies. A combination of phage display and peptido-mimetic chemistry, directed to find peptides able to bind tothis fragment, resulted in the disulfide bridged peptide FcIII[188]. The crystal structure of this peptide in complex withthe Fc domain confirmed that FcIII is constrained into aβ–hairpin conformation by the disulfide bridge. Then,computer models, generated by transposing either nine orthirteen residues from this peptide onto the D-Pro-L-Protemplate, afforded compounds 63 and 64, respectively (Fig.

9). This design was partially successful, as only the longermimetic 64 was able to adopt the desired conformation, andit has approximately 80-fold lower affinity compared to thestarting disulfide bridged peptide.

In another example, Mayo and co-workers have usedβ–hairpin mimetics to replace a β–sheet from anginex, apeptide comprising a β–sheet formed by three β–strand.Anginex, compound 65 in Fig. 10, is a potent anti-angiogenic, which specifically inhibits vascular endothelialcell (EC) growth [191]. The mechanism of action of anginexis not well understood, but it has been suggested that inducesdown-regulation of adhesion receptors [191]. Anginex wasidentified from the study of several βpep peptides,amphipathic β–sheet-forming peptides, which were designedemploying basic folding principles from short β–sheetsequences of three antiangiogenic proteins, namely plateletfactor-4 (PF4), interleukin (IL)-8 and bactericidal-permea-bility increasing (BPI) protein [192,193]. To demonstrate thehypothesis that the bioactive conformation of anginex was aβ–sheet, several disulphide-linked peptide analogues wereprepared. Among them, compound 66 (Fig. 10) was able toinhibit the EC proliferation and to promote apoptosis, evenwith slightly higher efficacy than anginex [194].Peptidomimetic 67 that incorporate the β–turn mimeticdibenzofuran (DBF), has also been prepared and shown to beeffective at inhibiting the EC growth [195]. Subsequently,shorter analogues were made to identify the minimumsequence required for activity. It has been shown that neitherthe N-terminal nor the C-terminal hexapeptide wereessential, and thus the shorter derivative, Ser-Val-Gln-Met-Lys-Leu-[DBF]-Ile-Ile-Val-Lys-Leu-Asn-Asp showedcomparable affinity with that of 67. NMR studies indicatedthat the β–sheet structure is preserved in DBF analogues and,interestingly, several of these peptidomimetics retained theantiangiogenic activity and were able to inhibit tumorgrowth.

4. ATTEMPTS TO STABILIZE/MIMIC α–HELIXELEMENTS

α–Helices are the most common protein secondarystructure accounting for over 40% of polypeptide aminoacids in natural proteins. α–Helices in proteins are foundwhen a stretch of consecutive residues all have the φ, ψangle pair approximately −60º and −50º. The α–helix has 3.6residues per turn with hydrogen bonds between C=O of

Fig. (8). Structure of β–sheet mimetics of ICAM-1.

HN

O

Cl

HN

OO

OH

H2N

O

OHOH

OH

Cl

Cl

HN

OO

OH

HN

O

S

59 60


residue i and NH of residue i+4, thus all NH and C=Ogroups are connected with hydrogen bonds except for thefirst NH and the last CO groups at the ends of the α–helix.The average length of an α–helix is around 10 residues,corresponding to three turns, and in proteins is almost alwaysright-handed, according to the screw direction of the chain.All the hydrogen bonds in an α–helix point in the samedirection, so the peptide units are aligned in the sameorientation along the helical axis. Since a peptide unit has adipole moment arising from the different polarity of NH and

CO groups, these dipole moments are also aligned along thehelical axis. The overall effect is a significant net dipole forthe α–helix that gives a partial positive charge at the aminoend and a partial negative charge at the carboxy end. Theside-chains of amino acids, except for Pro, projects out fromthe α–helix, and do not interfere with it. In general, there areamino acids that are good α–helices formers, such as Ala,Glu, Leu, and Met, while Pro, Gly, Tyr, and Ser are verypoor [196].

Fig. (9). Structure of β–hairpin mimetics based on the D-Pro-L-Pro template.

HN

NH

HN

O

O

O

N

N

N

HN

NH

OO

O

O

H

OHN

O

SS

NH

HN

H2N

OH

NH

O

H2N

NH

O N

O

NOHO

L-Pro

HN

O

NN

NH

OO

OHN

O

OH

NHO

H2 N

NH

ON

O

NOHO

L-Pro

HN

O

N

N

NH

OO

O

NH

HN

NHO

NH

N

NH

O

NH

O

HN

O HN

O

NH

O HN

O

OH

O

L-Pro

HN

NH

HN

O

O

O

N

N

N

HN

NH

OO

O

O

H

O

SS

O

HO

NH

HN

NHO

NH

N

NH

O

NH

O

HN

O HN

O

NH

O HN

O

OH

O

L-Pro

61

62

6463

D-Pro

D-Pro

D-Pro D-Pro


α–Helices play pivotal roles in many PPIs, therefore thetherapeutic potential of mimetics of this protein secondarystructure element is really enormous. However, in contrast toother elements, such as β–turns, small molecule α–helixmimetics have hitherto rarely been described in the literature.

There are several therapeutically relevant biologicalpathways that involve PPIs in which the hot-spot is locatedat an α–helix motif. Especially important is the fact thatα–helices constitute the mayor secondary structure elementsin key apoptosis regulating PPIs, (i.e., Bcl-2 family proteins,p53/HDM2). Therefore, the disruption of these PPIs couldconstitute an effective and selective way for the treatment ofcancer. This prompted several groups to focus their researchwork in that direction.

4.1. Mimetics of the Amphipatic BH3 Segment of Pro-Apoptotic Members of the Bcl-2 Family

The apoptosis process requires the cooperation of a seriesof molecules, such as the caspase-cascade signaling system,which in turn is regulated by various molecules, such as theBcl-2 (B-cell lymphoma-2) family proteins. Members of thisfamily of proteins can be divided into two groups: anti-apoptotic proteins (for example Bcl-2, Bcl-xL), whichdisplay sequence conservation in all BH domains, and pro-apoptotic proteins that are divided into multidomainmembers (such as Bax and Bak), and BH3 only members(such as Bid and Bad) that display sequence similarity onlyto the BH3 α-helical domain. The BH3-only proteinsmonitor cellular wellbeing and, when activated by cytotoxicsignals, engage pro-survival relatives by inserting theamphipathic α–helix BH3 domain into a hydrophobic grooveon the surface of the anti-apoptotic members (Bcl-2, Bcl-xL),priming the cell for apoptosis [197,198]. Since the BH3domain has an α-helical conformation, preparation ofmimetics of this type of secondary structure is excitingenormous interest.

One of the strategies used to restrict the conformationalfreedom and constrain the structure of synthetic α-helicalpeptides is through the incorporation of either covalent ornon-covalent linkages between amino acid side-chains. Avery new approach, in this context of fixing helicity bycyclization, is the strategy developed by Grubbs and co-workers making use of the olefin cross-linking. They appliedthe metathesis reaction to the cyclization of helices throughO-allyl serine residues located on adjacent helical turns(Compound 68, Fig. 11), by using the methodology ofruthenium-catalized ring-closing metathesis [199]. Theprincipal advantage of this chemistry is the high functionalgroup tolerance of the catalysts.

Verdine and co-workers developed this concept further inorder to confer stability to the α–helix, connecting residues iand i+4, (or i+7) by introducing in that positions α,α-disubstituted amino acids containing olefin chains [200]. Thestudy comprised variations in the position of attachment,stereochemistry, and cross-linker length, and concludes thathydrocarbon cross-linking can greatly increase metabolicstability as well as helical propensity of peptides, as shownby circular dichroism (CD) experiments. Additional work inthis area led to the search of apoptosis promoting derivatives.With this in mind, the group of Verdine prepared peptidesthat mimicked the amphipatic BH3 segment of Bid, a pro-apoptotic member of the Bcl-2 family proteins. Derivativesof α,α–disubstituted non-natural amino acids containingolefin-bearing tethers, to generate what they named an all-hydrocarbon “staple”, showed in vitro/in vivo biologicalactivity when tested for their ability to activate apoptosis.This was assessed in a wide panel of leukemia cells, as wellas in mice bearing established human leukemia xenografts.Compound 69 (SAHBA, Fig. 11) inhibited the proliferationof leukemia cells at moderate inhibitory concentrations andduplicated survival of the treated mice [201].

Following a very similar approach of stapling residues iand i+4 of the α–helix, but mimicking the C=O…H−N

Fig. (10). Backbone conformation of anginex and two β–hairpin mimetics.

O

HN

Leu

Lys

Met

Gln

Val

Ser

Leu

Lys

Ile

Asn

Ala-NH3

Ile

Ile

Val

Lys

Leu

Asn

Asp-CONH2

O

His

Asp

His

65 (Anginex) 66 67

Lys

Trp

Lys

Ile

Ile

Val

Lys

Leu

Asn

Asp-CONH2

Leu

Ser

Leu

Glu

Arg

Gly

Arg

Lys

Phe

Leu

Lys

Met

Gln

Val

Ser

Leu

Lys

Ile

Asn

Ala-NH3

Leu

Lys

Trp

Lys

Ile

Ile

Val

Lys

Leu

Cys

Asp

Gly-CONH2

Arg

Lys

Phe

Leu

Lys

Met

Gln

Val

Cys

Leu

Lys

Ile

Asn

Ala-NH3

Leu


hydrogen bond as closely as possible, Arora and co-workersdeveloped a strategy that permits at the same time thepreservation of helix surfaces. It consists in the substitutionof the hydrogen bond, by a covalent bond of the typeC=X−Y−N, were X and Y would be respectively part of the iand i+4 residues. In general, helix stabilization methods haverelied predominantly on side-chains constraints, in methodsthat either blocked solvent-exposed surfaces of the targetα–helices or removed important side-chain functionalities.The attractiveness of this strategy of hydrogen-bondsurrogates (HBS) relies on the fact that the cross-link isplaced inside the helix, so the solvent-exposed molecularrecognition surfaces of the helix are not blocked, whileshowing a high stability [202]. Derivatives such ascompound 70 (Fig. 11) were prepared as mimetics of the

BH3 domain of Bak, in order to check this hypothesis, and tocompare the ability of HBS derivatives with lactam-basedartificial helices previously reported, that failed to bind Bcl-xL. Results of fluorescent polarization assays showed thatcompound 70 was a high-affinity binder for Bcl-xL,accessing the deep hydrophobic cleft, which validate theinitial hypothesis [203]. Further studies in order to check thereal pharmacologic potential of these derivatives remains tobe performed.

Using a different approach, the group of Gellman havedescribed a series of mixed α,β-peptide derivatives showingantimicrobial activity that were design to be amphiphillic indifferent helical conformations. Extended work in this arealed to the design of Bcl-xL ligands, based on structures that

Fig. (11). Cyclic mimetics of the amphipatic BH3 segment of Bcl-2 family proteins.

NH

HN

NH

O

O

NH

O

NH

O

O

OH

O

HN

NH

O

HN

O

Glu-Asp-Ile-Ile-Arg-Asn-Ile-Ala-Arg-His-Leu

Trp-Ile-Ser-Arg-Asp

NH

O

N

NH

O

NH

O

O

NH2

O

Trp-Ile-Ser-Arg-Asp

HN

HN

BocNH

O

O

NH

O

NH

O

O

HN

MeO

O

HN

O

O

O

n

n

69: n=2 (SAHBA)

n

68: n =1 or 2

n

70 (HBS)


combine α/β and α-peptides in different ways. Bindingassays demonstrated that some of the prepared chimeric(α/β+α)-peptides displayed significant affinity for Bcl-xL(compound 71, Fig. 12) [204].

Other type of molecules that display pharmacologicalactivity towards this target corresponds to non-peptidestructures. One of the first reported non-peptide α–helixmimetic had the structure of a 1,6-disubstituted indane. Thisscaffold was designed to mimic the orientation of twoadjacent amino acid side-chains i and i+1 (compound 72,Fig. 12) [205]. Soon after, the corresponding 1,1,6-trisubs-tituted derivatives were suggested to operate as mimetics ofde i–1, i and i+1 residues, since the second substituent at the1-position of the indane overlays with the i–1 residue [206].This kind of template was used for the preparation ofmimetics of the endogenous neuropeptides of tachykininreceptors (peptide-protein interaction), in order to prove their

binding conformation. Thus, Horwell and co-workers haveshown that compound 73 (Fig. 12) bind with micromolaraffinity to both NK1 and NK3 tachykinin receptors [206].

On the other hand, and on the basis of molecularmodeling studies Jacoby proposed 2,6,3’,5’-tetrasubstitutedbiphenyls to mimic the side-chains of i, i+1, i+3, i+4residues of an α–helix [207], although the real behavior ofsuch structures as α–helix mimetics has not been determined(compound 74, Fig. 12).

In clear connexion with this, Hamilton and co-workershave reported on trisubstituted terphenyls as α-helicalproteomimetics. They are able to mimic one face of theα–helix by the spatial projection of its functionality in asimilar way to that corresponding to two turns of theα–helix. The terphenyl is expected to adopt a staggeredconformation and closely reproduce the position and angularorientation of functionality on the surface of an α–helix. In

Fig. (12). Quimeric and non-peptide mimetics of the amphipatic BH3 segment of Bcl-2 family proteins.

CO2H

O

CO2 H

N

N

O

O

R i

R i+1

Ri

R i+1

R i-1

Ri+3

RiR i+1

R i+4

O

HN O

ON

CO2Me

NH

NH

HN

HN

O

O

NH

O

NH

OHN

OO

NH2

O

COOH

HN

N

N

N

NH2

O O

O

O

O

NOH

NOH

78

72

75

Ri-1= CH2Ph

Ri = CONH(CH2)2Ph

Ri+1= CH2 Ph

73 74 Biphenyls

76

71

H-Gly-Asn-Leu-Gly-Arg-Asn-Leu-Ala-Ile

77


fact, the 3,2’,2”-trisubstituted terphenyls (coupled in para)mimic the side-chains at the i, i+3 or i+4, and i+7 positionsof an α–helix, which are the positions that frequently playkey roles in mediating protein contacts, and result inmolecules that would be able to disrupt PPIs [208]. In fact,this strategy of helix mimicry based on a terphenyl scaffoldhas been applied to the design of several modulators of PPIsin which the hot-spot is located at an α–helix. Thus, theauthors have described the capacity of the terphenyl basedα–helix derivatives (analogues of compound 75, Fig. 12) asmimetics of the Bad/Bak BH3 peptide epitope. Thesecompounds were able to inhibit the interaction between Bcl-xL (antiapoptotic protein) with the BH3 domains of thementioned pro-apoptotic proteins (Bad/Bak) [208,209].However, the challenging synthesis and physical propertiesof terphenyls prompted the search for simpler scaffolds thatcould similarly mimic the side-chain presentation on anα–helix. With this aim, a new set of inhibitors based on theterephthalamide scaffold was designed. These new deriva-tives, exemplified by compound 76 (Fig. 12), have shownhigh in vitro inhibition potencies in disrupting the Bcl-xL/Bak BH3 domain complex, as well as a significantimprovement in water solubility relative to the terphenylderivatives [210-211]. In the same context, Hamilton´s grouphave also designed a novel oligoamide foldamer, thetrispiridylamide scaffold 77 (Fig. 12), as an α–helix mimetic.The new derivatives showed affinity for Bcl-xL and inhibitedits interaction with Bak, even if only in a low micromolarrange [212].

Additionally, Dömling and co-workers, prompted byHamilton’s terphenyl α–helix mimetic concept, and in orderto obviate the disadvantage inherent to the multi-stepsynthetic approach needed for the assembly of the terphenylscaffold, investigated alternate scaffolds that would beaccessible by more efficient and faster synthetic methods,such as the multicomponent reactions (MCR) [213].Molecular modeling considerations led them to the selectionof a trisubstituted imidazole backbone (78, Fig. 12). Furtherstudies illustrated the ability of these compounds to mimicthe critical i, i+3 and i+7 residues of an α-helical Badpeptide epitope, and some of them, as compound 78, werefound to disrupt the interaction of Bcl-w with Bak-BH3peptide.

4.2. Disruption of the P53/HDM2 Protein-ProteinInteraction by α–Helix Mimetics

Protein p53 is a transcriptional activator critical forstress-induced cell cycle arrest and apoptosis. Under stressconditions, such as hypoxia or DNA damage, the p53 proteinaccumulates in the cell nucleus, and induces the transcriptionof various genes involved in cell-cycle control, apoptosis,differentiation, etc. In the absence of stress, HDM2 (humandouble minute 2) down-regulates p53 activity, interactingwith the p53 activation domain (p53AD) and exporting p53from the nucleus. Tumor cells often overspress HDM2,resulting in a loss of the cell’s primary response to stress,and dealing to unchecked cell growth.

HDM2-p53 interaction is an attractive therapeutic targetin oncology, since its inhibition with synthetic molecules is apromising approach for activating p53, leading to cell-cyclearrest or apoptosis. Additionally, p53-HDM2 association is

well characterized at both the structural and biological level[214,215].

One of the first attempts to reproduce the structural andconformational features of a naturally occurring α-helicalprotein epitope, the HDM2-binding domain on p53, is theapproach described by García-Echeverría and co-workers[216]. Short peptides (linear peptides, from 8- to 12-mer)were designed on the basis of the X-ray crystal structure ofthe N-terminal domain of HDM2 bound to a 15-mer wild-type p53-derived peptide (Ac-Gln16-Glu-Thr-Phe-Ser-Asp21-Leu-Trp22-Lys-Leu-Leu-Pro27-NH2). Starting from a 12-merderivative identified from phage display peptide libraries andconsidering molecular modeling studies, they obtained, afterseveral modifications, a truncate derivative, 79 (Fig. 13), thatexhibited interesting values of binding affinity. Theincorporation of some conformational constraints by theintroduction of α,α-disubstituted amino acid residues wasthen undertaken, as it is known that they contribute tostabilize the helix conformation in short peptide motifs.Additionally, modifications such as the introduction of aphosphonate group (Pmp residue) in order to form astabilizing salt bridge with the ε–amino group of Lys94 ofHDM2, and above all the incorporation of a Cl atom inposition 6 of Trp, led to an important increase of the bindingaffinity [216]. This was one of the first experimentalevidences about the capacity of peptide derivatives to inhibitPPIs, and shows the possible utility of peptides as a startingpoint for the development of non-peptide leads, or for the denovo design of PPIs antagonists.

Besides, Robinson and co-workers have described a newstrategy for the preparation of α–helix mimetics based on theuse of a β–hairpin scaffold [185], also inspired by theα–helix epitope of the HDM2 identified as the hot-spot forits interaction with p53. Taking as reference the structure ofthe complex described by Kussie [215], it can be realizedthat the distance between the Cα atoms of Phe

19 and Trp23 onone face of the HDM2-bound p53 α–helix is close to thedistance between the Cα atoms of residues i and i+1 alongone strand of a β–hairpin. Therefore, a designed β–hairpinmimetic can be used as scaffold to anchor in the correctrelative positions the side-chains of the residues identified ascrucial to the p53-HDM2 interaction, namely Phe19, Trp23,and possibly also Leu26. Optimization of the series led to thepreparation of derivative 80 (Fig. 13) that showed animproved profile with respect to the initial analogues.Evaluation for p53/HDM2 inhibitory activity by a solutionphase competition assay, performed with a surface plasmonresonance biosensor, proved that compound 80 exhibitedgood affinity values for HDM2. Moreover, the crystalstructure of the complex 80/HDM2 confirmed the β–hairpinconformation of the bound ligand, as well as the successfulmimicry of the location of the hot-spot residues Phe/Trp/Leu,in relation to the binding of p53 [217]. It has to be noticedthe improvement on activity achieved by the substitution ofthe Trp residue by a (6-Cl)Trp, on the basis of the workpreviously described by García-Echeverria [216].

Other type of scaffold also used as α–helix mimetic arepeptoids, oligomers composed of N-substituted glycinebuilding blocks. The side-chain of each amino acid inpeptides is formally shifted by one position from the C-α to


the amino group nitrogen in peptoids. Comparison of thepeptide chain with the peptoid chain shows that the directionof the peptide bond in peptoids should be reversed (retro-sequence) in order to provide the same relative arrangementof side-chain residues, R groups, and carbonyl groups (Fig.14). Peptoids are achiral, and helical structures are favoredwhen bulky N-alkyl side-chains are present. Interestingly,these helices are not stabilized by hydrogen bonds, and thehelicity depends on side-chain chirality and oligomer length[218]. The chiral peptoid helix is similar to a type-Ipolyproline helix, in which the amide bonds are cis, and thecarbonyl groups point the oxygens towards the N-terminus,causing the electrostatic dipole moment to be the reverse ofstandard α–helices. Because the helical conformation isdictated by steric constraints, and not by hydrogen bonds, asin regular proteins, it is able to persist both in aqueous andnon-polar solvents, as well as under a broad range of pH andtemperature conditions.

The group of Appella has very recently described apeptoid scaffold that binds to HDM2, and in consequence iscapable of inhibiting the HDM2-p53 interaction [219]. Thus,starting from a peptoid derivative of ten glycine subunits,and after several structural modifications in order to improveaqueous solubility and binding affinity, compound 81 (Fig.

14) was obtained as the derivative that showed higheraffinity for HDM2 in the series. In order to reproduce therelative position of key residues Phe19, Trp23 and Leu26 in atype-I polyproline helix of the peptoid skeleton two mainquestions had to be addressed. The first difference is that thepeptoid helix is more tightly bound than that of a peptide,only 3 residues per turn compared with 3.6. This problemwas resolved by deleting one position between the Phe andTrp residues. The other relevant structural differencebetween the two helices is that the side-chains project out atdifferent angles. This problem was obviated by elongation inan extra methylene group both the Phe and the Trp side-chains. Surprisingly, the best affinity results of the serieswere shown by an achiral compound (81), suggesting thepossibility that non-helical conformations adopted bypeptoids can also be important for binding to proteins.

Helical β-peptides, composed of β-amino acids, havealso been used as prototypes for PPI disruption. β-Aminoacids are characterized by the presence of an additionalcarbon atom between the amino and the carboxy groups. β-Peptides consist of linear chains of β-amino acids, and are inprinciple resistant to peptidases, as they are non-naturalfolding oligomers, or foldamers. This has promoted their useto modify the pharmacodynamic properties of peptide drug

Fig. (13). Linear peptides and β–hairpin derivatives as mimetics of the α−helical protein epitope of p53.

N

HN

N

O

O

O

N

N

NNH

N

O

O

O

O

H

H

H

H

O

NH

O

NH

O

HN

Cl

NH

HO2C

CO2H

CO2 H

HN

O

NH

NH

O

AcHN

O

O

O

HN

NH

O

CO2 H

O

HN

NH

NH2

O

O

HN

Cl

POH

OO

SCH3

HN

80

L-Pro

79

D-Pro


candidates, in order to improve their metabolic stability. β-Peptides can fold into stable helical secondary structures.Side-chains may be attached either to Cα or to Cβ, or to bothof them, and this fact considerably influences the secondarystructure of the corresponding oligomer. Thus, acyclicmonosubstituted residues tend to fold into 14-helices, or10/12 helices if patterned as alternating β2/β3 residues.Cyclopentyl and cyclohexyl ring constraints promoteformation of a 12–helix and a 14–helix respectively.Therefore, the choice of the appropriate substitution patternwill determine the preferred helical secondary structure.Helix formation occurs already in β-peptide hexamers, whilein the case of α-amino acids usually more than 10-12 aminoacid residues are required to form a stable helix [220].

Schepartz and co-workers realized that the side-chainslocated at the i, i+4, i+7 positions on an α–helix superpose

with those at positions i, i+3, i+6 on a 14–helix, and on thebasis of that, decided to present a short α-helical functionalepitope on a well-folded 14–he

current topics in medicinal chemistry, 33-62 33 modulation ... · current topics in medicinal...

Documents