Chiral interaction in Gly-capped N-terminal motif of 3(10)-helix and domino-type induction in helix sense

Chiral interaction of helical peptide with chiral molecule, and concomitant induction in its helix sense have been demonstrated in optically inactive nonapeptide (1) possessing Gly at its N-terminus: H-Gly-(Delta(Z)Phe-Aib)(4)-OCH(3) (1: Delta(Z)Phe = Z-dehydrophenylalanine; Aib = alpha-aminoisobutyric acid). Spectroscopic measurements [mainly nuclear magnetic resonance (NMR) and circular diochroism (CD)] as well as theoretical simulation have been carried out for that purpose. Peptide 1 in the 3(10)-helix tends to adopt preferentially a right-handed screw sense by chiral Boc-L-amino acid (Boc: t-butoxycarbonyl). Induction in the helix sense through the noncovalent chiral domino effect should be derived primarily from the complex supported by the three-point coordination on the N-terminal sequence. Thus the 3(10)-helical terminus consisting of only alpha-amino acid residues enables chiral recognition of the Boc-amino acid molecule, leading to modulation of the original chain asymmetry. Dynamics in the helix-sense induction also have been discussed on the basis of a low-temperature NMR study. Furthermore, the inversion of induced helix sense has been achieved through solvent effects.     

Noncovalent chiral domino effect on one-handed helix of nonapeptide containing a midpoint L-residue

We here clarify whether noncovalent chiral domino effect characterized by the terminal interaction of a helical peptide with a chiral small molecule can alter the helical stability of N-deprotected peptides containing an L-residue covalently incorporated into the inner position. Two nonapeptides consisting of the midpoint L-leucine (1) or L-phenylalanine (2) and the achiral helix-forming residues were employed. NMR and IR spectroscopy and energy calculation indicated that both peptides adopt a 3(10)-helical conformation in chloroform. They strongly preferred a right-handed screw sense because of the presence of the midpoint L-residue. These original right-handed screw senses were retained on addition of chiral Boc-amino acid, but their helical stabilities clearly depended on its added chirality. Here, Boc-L-amino acid stabilizes the original right-handed helix, whereas the corresponding Boc-D-amino acid tends to less stabilize or destabilize it. This tendency was not observed for the corresponding N-Boc-protected peptides 1 and 2, strongly suggesting that the N-terminal amino group is required for controlling the stabilization of the original right-handed helix. Therefore, noncovalent chiral domino effect in peptides 1 and 2 can contribute even to the helical stability of a chiral peptide prevailing one-handed helix strongly through the midpoint L-residue. In addition, the N-terminal moiety of a 3(10)-helical peptide was found to generate chiral discrimination in complexation process with racemic additives.     

Helix propensities of the amino acids measured in alanine-based peptides without helix-stabilizing side-chain interactions.

Helix propensities of the amino acids have been measured in alanine-based peptides in the absence of helix-stabilizing side-chain interactions. Fifty-eight peptides have been studied. A modified form of the Lifson-Roig theory for the helix-coil transition, which includes helix capping (Doig AJ, Chakrabartty A, Klingler TM, Baldwin RL, 1994, Biochemistry 33:3396-3403), was used to analyze the results. Substitutions were made at various positions of homologous helical peptides. Helix-capping interactions were found to contribute to helix stability, even when the substitution site was not at the end of the peptide. Analysis of our data with the original Lifson-Roig theory, which neglects capping effects, does not produce as good a fit to the experimental data as does analysis with the modified Lifson-Roig theory. At 0 degrees C, Ala is a strong helix former, Leu and Arg are helix-indifferent, and all other amino acids are helix breakers of varying severity. Because Ala has a small side chain that cannot interact significantly with other side chains, helix formation by Ala is stabilized predominantly by the backbone ("peptide H-bonds"). The implication for protein folding is that formation of peptide H-bonds can largely offset the unfavorable entropy change caused by fixing the peptide backbone. The helix propensities of most amino acids oppose folding; consequently, the majority of isolated helices derived from proteins are unstable, unless specific side-chain interactions stabilize them.     

The structural flexibility of the preferredoxin transit peptide.

In order to obtain insight into the structural flexibility of chloroplast targeting sequences, the Silene pratensis preferredoxin transit peptide was studied by circular dichroism and nuclear magnetic resonance spectroscopy. In water, the peptide is unstructured, with a minor propensity towards helix formation from Val-9 to Ser-12 and from Gly-30 to Ser-40. In 50% (v/v) trifluoroethanol, structurally independent N- and C-terminal helices are stabilized. The N-terminal helix appears to be amphipathic, with hydrophobic and hydroxylated amino acids on opposite sides. The C-terminal helix comprises amino acids Met-29-Gly-50 and is destabilized at Gly-39. No ordered tertiary structure was observed. The results are discussed in terms of protein import into chloroplasts, in which the possible interactions between the transit peptide and lipids are emphasized.     

External chirality-triggered helicity control promoted by introducing a beta-Ala residue into the N-terminus of chiral peptides

The noncovalent chiral domino effect (NCDE), defined as chiral interaction upon an N-terminus of a 3(10)-helical peptide, will provide a unique method for structural control of a peptide helix through the use of external chirality. On the other hand, the NCDE has not been considered to be effective for the helicity control of peptides strongly favoring a one-handed screw sense. We here aim to promote the NCDE on peptide helicity using two types of nonapeptides: H-beta-Ala-Delta(Z)Phe-Aib-Delta(Z)Phe-X-(Delta(Z)Phe-Aib)(2)-OCH(3) [Delta(Z)Phe = alpha,beta-didehydrophenylalanine, Aib = alpha-aminoisobutyric acid], where X as the single chirality is L-leucine (1) or L-phenylalanine (2). NMR, IR, and CD spectroscopy as well as energy calculation revealed that both peptides alone form a right-handed 3(10)-helix. The original CD amplitudes or signs in chloroform, irrespective of a strong screw-sense preference in the central chirality, responded sensitively to external chiral information. Namely added Boc-L-amino acid stabilized the original right-handed helix, while the corresponding d-isomer destabilized it or transformed it into a left-handed helix. These peptides were also shown to bind more favorably to an L-isomer from the racemate. Although similar helicity control was observed for analogous nonapeptides bearing an N-terminal Aib residue (Inai, Y.; et al. Biomacromolecules 2003, 4, 122), the present findings demonstrate that the N-terminal replacement by the beta-Ala residue significantly improves the previous NCDE to achieve more effective control of helicity. Semiempirical molecular orbital calculations on complexation of peptide 2 with Boc-(L or D)-Pro-OH reasonably explained the unique conformational change induced by external chirality.     

Determination of alpha-helix N1 energies after addition of N1, N2, and N3 preferences to helix/coil theory

Surveys of protein crystal structures have revealed that amino acids show unique structural preferences for the N1, N2, and N3 positions in the first turn of the alpha-helix. We have therefore extended helix-coil theory to include statistical weights for these locations. The helix content of a peptide in this model is a function of N-cap, C-cap, N1, N2, N3, C1, and helix interior (N4 to C2) preferences. The partition function for the system is calculated using a matrix incorporating the weights of the fourth residue in a hexamer of amino acids and is implemented using a FORTRAN program. We have applied the model to calculate the N1 preferences of Gln, Val, Ile, Ala, Met, Pro, Leu, Thr, Gly, Ser, and Asn, using our previous data on helix contents of peptides Ac-XAKAAAAKAAGY-CONH2. We find that Ala has the highest preference for the N1 position. Asn is the most unfavorable, destabilizing a helix at N1 by at least 1.4 kcal mol(-1) compared to Ala. The remaining amino acids all have similar preferences, 0.5 kcal mol(-1) less than Ala. Gln, Asn, and Ser, therefore, do not stabilize the helix when at N1.     

2H nuclear magnetic resonance of the gramicidin A backbone in a phospholipid bilayer.

Solid-state 2H NMR spectroscopy has been employed to study the channel conformation of gramicidin A (GA) in unoriented 1,2-dimyristoyl-sn-glycerol-3-phosphocholine (DMPC) multilayers. Quadrupolar echo spectra were obtained at 44 degrees C and 53 degrees C, from gramicidin A labels in which the proton attached to the alpha carbon of residue 3, 4, 5, 10, 12, or 14 was replaced with deuterium. Because of the nearly axially symmetric electric field gradient tensor, the quadrupolar splittings obtained from an unoriented multilamellar dispersion of DMPC and singly labeled GA directly yield unambiguous orientational constraints on the C-2H bonds. The average of the ratios of the quadrupolar splittings of the left-handed amino acids to those of the right-handed amino acids, (delta vQL/delta vQD), is expected to be 0.97 +/- 0.04 for a relaxed right-handed beta 6.3LD helix, while a ratio of 0.904 +/- 0.003 is expected for a left-handed beta LD6.3 helix. Since we have experimentally determined this ratio to be 1.01 +/- 0.04, we conclude that that the helix sense of the channel conformation of GA is right-handed. Assuming that the dominant motions are fast axial diffusion of the gramicidin molecule and reorientation of the diffusion axis with respect to the local bilayer normal, then the theoretical splittings may all be scaled down by a constant motional narrowing factor. In this case, a relaxed right-handed beta LD6.3 helix, whose axis of motional averaging is roughly along the presumed helix axis, gave the best fit to experimental results. The reasonably uniform correspondence between the splittings predicted by the relaxed right-handed beta LD6.3 helix and the observed splittings, for labels from both the inner and outer turn of GA, did not reflect a peptide backbone flexibility gradient, since an outer turn (i.e., the turn of the helix closest to the interface with water) with greater flexibility would show additional motional narrowing for labels located there.     

Characterization of a synthetic peptide corresponding to a receptor binding domain of mouse interferon gamma.

A receptor binding region of mouse interferon gamma (IFN gamma) has previously been localized to the N-terminal 39 amino acids of the molecule by use of synthetic peptides and monoclonal antibodies. In this report, a detailed analysis of the synthetic peptide corresponding to this region, IFN gamma (1-39), is presented. Circular dichroism (CD) spectroscopy indicated that the peptide has stable secondary structure under aqueous conditions and adopts a combination of alpha-helical and random structure. A peptide lacking two N-terminal amino acids, IFN gamma (3-39), had similar secondary structure and equivalent ability to compete for receptor binding, while peptides lacking four or more N-terminal residues had reduced alpha-helical structure and did not inhibit 125I-IFN gamma binding. Substitution of proline, a helix-destabilizing amino acid, for leucine (residue 8) of a predicted amphipathic alpha-helix (residues 3-12), IFN gamma (1-39) [Pro]8, resulted in a substantial reduction in the helical content of the peptide, supporting the presence of helical structure in this region. However, destabilization of the helix did not reduce the competitive ability of the peptide. A peptide lacking eight C-terminal residues, IFN gamma (1-31), did not block 125I-IFN gamma binding and had no detectable alpha-helical structure, suggesting a requirement of the predicted second alpha-helix (residues 20-34) for receptor interaction and helix stabilization. Substitution of phenylalanine for tyrosine at position 14, IFN gamma (1-39) [Phe]14, a central location of a predicted omega-loop structure, did not affect the secondary structure associated with the region yet resulted in a 30-fold increase in receptor competition.(ABSTRACT TRUNCATED AT 250 WORDS)     

Effects of hydrogen-bond deletion on peptide helices: structural characterization of depsipeptides containing lactic acid

The insertion of alpha-hydroxy acids into peptide chains provides a convenient means for investigating the effects of hydrogen bond deletion on polypeptide secondary structures. The crystal structures of three oligopeptides containing L-lactic acid (Lac) residue have been determined. Peptide 1, Boc-Val-Ala-Leu-Aib-Val-Lac-Leu-Aib-Val-Ala-Leu-OMe (Boc: tert-butyloxycarbonyl; Aib: alpha- aminoisobutyric acid; OMe: methyl ester), and peptide 2, Boc-Val-Ala-Leu-Aib-Val-Lac-Leu-Aib-Val-Leu-OMe, adopt completely helical conformations in the crystalline state with the Lac(6) residue comfortably accommodated in the center of a helix. The distance between the O atoms of Leu(3) CO group and the Lac(6) O (ester) in both the structures is 3.1-3.3 A. The NMR and CD studies of peptide 1 and its all-amide analogue 4, Boc-Val-Ala-Leu-Aib-Val-Ala-Leu-Aib-Val-Ala-Leu-OMe, provide firm evidence for a continuous helical conformation in solution in both the cases. In a 14-residue peptide 3, Boc-Val-Ala-Leu-Aib-Val-Ala-Leu-Val-Ala-Leu-Aib-Val-Lac-Leu-OMe, residues Val(1)-Leu(10) adopt a helical conformation. Aib(11) is the site of chiral reversal resulting in helix termination by formation of a Schellman motif. Residues 12-14 adopt nonhelical conformations. The loss of the hydrogen bond near the C-terminus appears to facilitate the chiral reversal at Aib(11). Published 2001 John Wiley & Sons, Inc. Biopolymers 59: 276-289, 2001     

The role of context on alpha-helix stabilization: host-guest analysis in a mixed background peptide model.

The helix content of a series of peptides containing single substitutions of the 20 natural amino acids in a new designed host sequence, succinyl-YSEEEEKAKKAXAEEAEKKKK-NH2, has been determined using CD spectroscopy. This host is related to one previously studied, in which triple amino acid substitutions were introduced into a background of Glu-Lys blocks completely lacking alanine. The resulting free energies show that only Ala and Glu- prove to be helix stabilizing, while all other side chains are neutral or destabilizing. This agrees with results from studies of alanine-rich peptide modela, but not the previous Glu-Lys block oligomers in which Leu and Met also stabilize helix. The helix propensity scale derived from the previous block oligomers correlated well with the frequencies of occurrence of different side chains in helical sequences of proteins, whereas the values from the present series do not. The role of context in determining scales of helix propensity values is discussed, and the ability of algorithms designed to predict helix structure from sequence is compared.     

Chimeric Protein Engineering

Protein stability can be enhanced by the incorporation of non-natural amino acids and semi-rigid peptidomimetics to lower the entropic penalty upon protein folding through preorganization. An example is the incorporation of aminoisobutyric acid (Aib, α-methylalanine) into proteins to restrict the Φ and Ψ backbone angles adjacent to Aib to those associated with helix formation. Reverse-turn analogs were introduced into the sequences of HIV protease and ribonuclease A that enhanced their stability and retained their native enzymatic activity. In this work, a chimeric protein, design_4, was engineered, in silico, by replacing the C-terminal helix of full sequence design protein (FSD-1) with a semi-rigid helix mimetic. Residues 1–16 of FSD-1 was ligated in silico with the N-terminus of a phenylbipyridyl-based helix mimetic to form design_4. The designed chimeric protein was stable and maintained the designed fold in a 100-nanosecond molecular dynamics simulation at 280 K. Its β-hairpin adopted conformations that formed three additional hydrogen bonds. Compared to FSD-1, design_4 contained fewer peptide bonds and internal degrees of freedom; it should, therefore, be more resistant to proteolytic degradation and denaturation.     

The effect of mutations on peptide models of the DNA binding helix of p53: Evidence for a correlation between structure and tumorigenesis

The tumor suppresser protein p53 has been called the “guardian of the genome.” DNA damage induces p53 to either halt the cell cycle, allowing for repair, or initiate apoptosis. P53 is mutated in over 50% of human tumors and it has been proposed that many tumorigenic mutations are deleterious to p53 because they induce local unfolding. To explore this hypothesis, peptide models have been developed to study tumorigenic mutations in the H2 helix of the p53 core domain. This helix is rich with charged residues and is a key component of the DNA binding region. A 16‐residue peptide corresponding to the H2 wild‐type sequence extended with an Ala‐rich C‐terminus was synthesized and studied by ¹H‐nmr (500 MHz) and CD. The nmr studies demonstrate that this peptide adopts helical structure in solution. Six additional peptides corresponding to subtle tumorigenic mutations were synthesized and CD was used to assess the relative stability of these “mutant analogues.” All six mutations studied are destabilizing relative to the wild type, with ΔΔG values in the range of 0.26 to 1.35 kcal mol⁻¹. Surprisingly, substitution of Asp 281 with Ala resulted in a peptide with the greatest destabilization even though Ala possesses the largest helix propensity of the common 20 amino acids. Because this helix appears to be stabilized mainly by local electrostatics, we conclude that its structure is susceptible to even the most conservative mutations. These results provide support for the hypothesis that tumorigenic mutations induce local unfolding of p53. © 1999 John Wiley & Sons, Inc. Biopoly 49: 215–224, 1999     

2D NMR and structural model for a mitochondrial signal peptide bound to a micelle

The 19 amino acid signal peptide of rat liver aldehyde dehydrogenase, possessing a lysine substitution for an arginine and containing 3 extra amino acid residues at the C terminus, was studied by two-dimensional NMR in a dodecylphosphocholine micelle. In this membrane-like environment, the peptide contains two alpha-helical regions, both of which are amphiphilic, separated by a hinge region. The helix located closer to the C terminus is more stable than is the helix located near the N terminus. This suggests that the hydrophobic face of the C-terminal helix is buried within the hydrophobic region of the micelle. On the basis of these results a general model for protein translocation is presented in which the C-terminal amphiphilic helix of the signal region in the preprotein first binds to the mitochondrial membrane and then diffuses to the translocation receptor. The receptor then recognizes the N-terminal helix of the signal region, which is not anchored to the membrane. To explain how this signal peptide was imported into isolated mitochondria in the absence of energy or receptor protein [Pak, Y. K., & Weiner, H. (1990) J. Biol. Chem. 265, 14298-14307], a model for signal peptide translocation across a membrane barrier without the need for auxiliary membrane proteins is proposed. In this model the faces of the two helices fold upon each other, resulting in the mutual shielding of positively charged residues by the complementary hydrophilic face of the other amphiphilic helix.     

Enzyme-substrate interactions in the hydrolysis of peptide substrates by thermitase, subtilisin BPN', and proteinase K

Peptide substrates of the general structure acetyl-Alan (n = 2-5), acetyl-Pro-Ala-Pro-Phe-Alan-NH2 (n = 0-3), and acetyl-Pro-Ala-Pro-Phe-AA-NH2 (AA = various amino acids) were synthesized and used to investigate the enzyme-substrate interactions of the microbial serine proteases thermitase, subtilisin BPN', and proteinase K on the C-terminal side of the scissile bond. The elongation of the substrate peptide chain up to the second amino acid on the C-terminal side (P'2) enhances the hydrolysis rate of thermitase and subtilisin BPN', whereas for proteinase K an additional interaction with the third amino acid (P'3) is possible. The enzyme subsite S'1 specificity of the proteases investigated is very similar. With respect to kcat/Km values small amino acid residues such as Ala and Gly are favored in this position. Bulky residues such as Phe and Leu were hydrolyzed to a lower extent. Proline in P'1 abolishes the hydrolysis of the substrates. Enzyme-substrate interactions on the C-terminal side of the scissile bond appear to affect kcat more than Km for all three enzymes.     

Secondary structure and position of the cell-penetrating peptide transportan in SDS micelles as determined by NMR

Transportan is a 27-residue peptide (GWTLN SAGYL LGKIN LKALA ALAKK IL-amide) which has the ability to penetrate into living cells carrying a hydrophilic load. Transportan is a chimeric peptide constructed from the 12 N-terminal residues of galanin in the N-terminus with the 14-residue sequence of mastoparan in the C-terminus and a connecting lysine. Circular dichroism studies of transportan and mastoparan show that both peptides have close to random coil secondary structure in water. Sodium dodecyl sulfate (SDS) micelles induce 60% helix in transportan and 75% helix in mastoparan. The 600 MHz (1)H NMR studies of secondary structure in SDS micelles confirm the helix in mastoparan and show that in transportan the helix is localized to the mastoparan part. The less structured N-terminus of transportan has a secondary structure similar to that of the same sequence in galanin [Ohman, A., et al. (1998) Biochemistry 37, 9169-9178]. The position of mastoparan and transportan relative to the SDS micelle surface was studied by adding spin-labeled 5-doxyl- or 12-doxyl-stearic acid or Mn2+ to the peptide/micelle system. The combined results show that the peptides are for the most part buried in the SDS micelles. Only the C-terminal parts of both peptides and the central segment connecting the two parts of transportan are clearly surface exposed. For mastoparan, the secondary chemical shifts of the amide protons were found to vary periodically and display a pattern almost identical to those reported for mastoparan in phospholipid bicelles [Vold, R., et al. (1997) J. Biomol. NMR 9, 329-335], indicating similar structures and interactions in the two membrane-mimicking environments.     

Mutational analysis of bovine papillomavirus type 1 E5 peptide domains involved in induction of cellular DNA synthesis.

Early gene E5 of bovine papillomavirus type 1 encodes a 44-amino-acid protein whose expression can transform immortalized mouse cell lines. We have previously reported that a chemically synthesized E5 peptide functions to induce cellular DNA synthesis upon microinjection into growth-arrested mouse cells. We further defined the two E5 domains essential for the full DNA synthesis induction activity by the analysis of E5 deletion and amino acid substitution mutant peptides. The first domain is the C-terminal 13-amino-acid core which is sufficient to activate DNA synthesis at high peptide concentration and contains two essential, highly conserved cysteine residues. The second domain is the 7-amino-acid hydrophobic sequence contiguous to the core domain which is sufficient to confer a 1,000-fold higher molar specific activity to the E5 peptide. A random hydrophobic sequence, but not charged amino acids, fulfills the function of the second domain.     

Biosynthesis of peptide neurotransmitters: studies on the formation of peptide amides

A high proportion of peptide transmitters and peptide hormones terminate their peptide chain in a C-terminal amide group which is essential for their biological activity. The specificity of an enzyme that catalyses the formation of the amide was investigated with the aid of synthetic peptide substrates. With peptides containing l-amino acids the enzyme exhibited an essential requirement for glycine in the C-terminal position; amidation did not take place with peptides that had leucine, alanine, glutamic acid, lysine or N-methylglycine at the C-terminus and a peptide extended by the attachment of lysine to the C-terminal glycine did not act as a substrate. Amidation did occur with a peptide containing C-terminal D-alanine but no reaction was detected with peptides having C-terminal, D-serine or D-leucine. In tripeptides with a neutral amino acid in the penultimate position, amidation, took place readily but the reaction was slower when this position was occupied by an acidic or a basic residue. A series of overlapping peptides with C-terminal glycine, based on partial sequences of calcitonin, underwent amidation at similar rates, indicating that the amidating enzyme recognizes only a limited sequence at the C-terminus of its substrates. The results provide evidence that the amidating enzyme has a highly compact substrate binding site.     

Molecular architecture of the voltage-dependent Na channel: functional evidence for alpha helices in the pore

The permeation pathway of the Na channel is formed by asymmetric loops (P segments) contributed by each of the four domains of the protein. In contrast to the analogous region of K channels, previously we (Yamagishi, T., M. Janecki, E. Marban, and G. Tomaselli. 1997. Biophys. J. 73:195-204) have shown that the P segments do not span the selectivity region, that is, they are accessible only from the extracellular surface. The portion of the P-segment NH(2)-terminal to the selectivity region is referred to as SS1. To explore further the topology and functional role of the SS1 region, 40 amino acids NH(2)-terminal to the selectivity ring (10 in each of the P segments) of the rat skeletal muscle Na channel were substituted by cysteine and expressed in tsA-201 cells. Selected mutants in each domain could be blocked with high affinity by externally applied Cd(2)+ and were resistant to tetrodotoxin as compared with the wild-type channel. None of the externally applied sulfhydryl-specific methanethiosulfonate reagents modified the current through any of the mutant channels. Both R395C and R750C altered ionic selectivity, producing significant increases in K(+) and NH(4)(+) currents. The pattern of side chain accessibility is consistent with a pore helix like that observed in the crystal structure of the bacterial K channel, KcsA. Structure prediction of the Na channel using the program PHDhtm suggests an alpha helix in the SS1 region of each domain channel. We conclude that each of the P segments undergoes a hairpin turn in the permeation pathway, such that amino acids on both sides of the putative selectivity filter line the outer mouth of the pore. Evolutionary conservation of the pore helix motif from bacterial K channels to mammalian Na channels identifies this structure as a critical feature in the architecture of ion selective pores.     

Structural basis for specific binding of the Gads SH3 domain to an RxxK motif-containing SLP-76 peptide: a novel mode of peptide recognition

The SH3 domain, which normally recognizes proline-rich sequences, has the potential to bind motifs with an RxxK consensus. To explore this novel specificity, we have determined the solution structure of the Gads T cell adaptor C-terminal SH3 domain in complex with an RSTK-containing peptide, representing its physiological binding site on the SLP-76 docking protein. The SLP-76 peptide engages four distinct binding pockets on the surface of the Gads SH3 domain and upon binding adopts a unique structure characterized by a right-handed 3(10) helix at the RSTK locus, in contrast to the left-handed polyproline type II helix formed by canonical proline-rich SH3 ligands. The structure, and supporting mutagenesis and peptide binding data, reveal a novel mode of ligand recognition by SH3 domains.