Introduction

SH3 domains are one of the most frequently occurring protein–protein interaction modules in eukaryotic cells, which are involved in a wide variety of cell processes (Cohen et al. 1995; Mayer 2001; Marchiani et al. 2009). These domains typically bind to proline-rich sequences containing the PxxP motif, which adopt a left-handed polyproline type II (PPII) helix conformation when they are bound to the SH3 domain (Feng et al. 1994). The position of a critical basic residue identifies the two major binding classes of peptide SH3 ligands, named class I or class II, respectively (Lim et al. 1994). Recently, different authors demonstrated the significant role of residues outside the core consensus in determining both affinity and specificity; this allows defining two distinct binding regions in the structure of many SH3 domains, referred to as Surface I and Surface II, respectively (Kim et al. 2008). Surface I has been extensively described and corresponds to the region which interacts with the PxxP motif (Li 2005; Zarrinpar et al. 2003). On the other hand, Surface II, referred to as the “specificity pocket” in some studies (Li 2005; Zarrinpar et al. 2003), is much broader and contains less conserved residues, located both in the RT or N-Src loops and in the strands c and d of the SH3 domain (Larson and Davidson 2000).

Previously, we demonstrated that the SH3 domain of murine cortactin (SH3m-cort) interacts with peptides derived from proline-rich region of the HPK1 kinase (Rubini et al. 2010). Using the near-UV CD analysis, we determined that the peptide reproducing the region 394–403 of HPK1, named P2, interacted with high affinity at the SH3m-cort domain.

PPII helix is a left-handed helix with all the amide bonds in the trans conformation. It has been well demonstrated that the stereoelectronic effects of the 4-substituent to the proline residue play a major role in determining the pucker of the proline ring and hence the cis/trans conformer ratio of the Xaa-Pro amide bond, influencing the PPII helical stability (Kuemin et al. 2010).

With the aim to enhance the tendency of the P2 peptide to adopt a PPII helix conformation in aqueous solution, with a consequent reduction of the entropic cost of the ligand–protein interaction, we synthesized a series of P2 analogues replacing Pro residues with either 4R-fluoroproline (FPro) or 4R-hydroxyproline (Hyp) (Table 1).

Table 1 Sequences of P2 proline-rich peptide analogues
Full size table

The interactions of the SH3m-cort domain with the aforementioned Pro-rich peptides were analyzed by CD spectroscopy, monitoring the CD changes of the Trp side-chain chromophore of the SH3 domain. The dissociation constants, K d, were determined by analyzing the CD data measured at 294 nm using a non-linear regression method (Siligardi et al. 2012; Siligardi and Hussain 1998). Additionally, CD thermal transitions of the P2 analogues were measured to correlate their propensity to adopt a PPII helix with their binding affinity for SH3. The comparative analysis of the binding properties of this set of closely related peptides confirms the complexity of the SH3 recruitment, supporting the role of additional interactions outside the region that interacts with the PxxP motif in the substrate recognition by the SH3m-cort domain as previously indicated (Rubini et al. 2010).

Experimental procedures

Peptide synthesis

Fmoc-protected amino acids and preloaded Wang resin were purchased from Calbiochem-Novabiochem (Läufelfingen, Switzerland). Fmoc-FPro-OH was obtained from Bachem (Weil am Rhein, D). HBTU, HOBt, DIEA and DMF were obtained from Iris Biotech (Marktredwitz, D), whereas TFFH was purchased from PerSeptive Biosystems (Foster City, CA). Peptides were assembled using the Fmoc/HBTU chemistry in 0.06 mmol scale by manual solid-phase synthesis. HBTU/HOBt activation employed a threefold molar excess (0.24 mmol) of Fmoc-amino acids in DMF solution for each coupling cycle unless otherwise stated. Coupling to the secondary amino group of 4-fluoroproline was performed using TFFH as coupling reagent (5 eq) in the presence of the carboxyl component (5 eq) and DIEA (10 eq). Coupling time was 40 min. Deprotection was performed with 20 % piperidine. Coupling yields were monitored on aliquots of peptide resin either by Kaiser test or by evaluation of Fmoc displacement (Wellings and Atherton 1997). Peptides were side chain-deprotected and removed from the resin by TFA treatment in the presence of 2.5 % TIS, 2.0 % anisole and 0.5 % water, and then precipitated by addition of diethyl ether.

Crude peptides were purified by preparative reversed-phase HPLC using a Shimadzu LC-8 (Shimazdu, Kyoto, Japan) system with a Vydac 218TP1022, 10 μm, 250 × 22 mm column (Grace Davison Discovery Sciences, Deerfield, IL). The column was perfused at a flow rate of 12 mL/min with a mobile phase containing solvent A (0.05 % TFA in water) and a linear gradient from 15 to 35 % of solvent B (0.05 % TFA in acetonitrile/water, 9:1 by vol.) in 40 min. The fractions containing the desired product were collected and lyophilized to constant weight in the presence of 0.01 N HCl. Analytical HPLC analyses were performed on a Shimadzu LC-10 instrument, fitted with a Jupiter C18, 10 μm, 250 × 4.6 mm column (Phenomenex, Torrance, CA) using the described solvent system (solvents A and B), with a flow rate of 1 mL/min, and detection at 216 nm. All peptides showed less than 1 % impurities. Molecular weights of compounds were determined by ESI–MS on a Mariner (PerSeptive Biosystem) mass spectrometer instrument. The mass was assigned using a mixture of neurotensin, angiotensin and bradykinin, at a concentration of 1 pmol/μL, as external standard.

Expression and purification of GST-SH3m-cort

GST-SH3m-cort fusion protein was kindly provided by Chiara Rubini (Rubini et al. 2010). The GST-SH3m-cort concentration was determined by absorption spectroscopy (ε = 57,420 M−1 cm−1 at 280 nm).

Circular dichroism and GST-SH3m-cort titration

All measurements were obtained using a nitrogen flushed Jasco J-715 spectropolarimeters (Tokyo, Japan) and a 0.5 cm or a 1.0 cm quartz cells for the near-UV region and a 0.1 cm quartz cell for the far-UV region. Each titration was performed in aqueous buffer (20 mM Tris–HCl, pH 7.5, in presence of 1 mM dithiothreitol) at 25 °C by addition of small aliquots of peptide stock solution in the same buffer as previously described (Ruzza et al. 2006). The contribution of the phenylalanine aromatic side chain present in the ligand-peptides, at a given concentration, was subtracted from the CD spectra of the domain in complex with each peptide. Peptide concentration was determined by weight using a Mettler Toledo microbalance (Columbus, OH) model AT21 Comparator (sensitivity ± 1 μg). The dissociation constants K d of the different complexes were determined by analyzing the CD data at a single wavelength by non-linear regression analysis as previously described (Siligardi and Hussain 1998).

Far-UV CD spectra of proline-rich peptides were obtained in aqueous buffer (20 mM Tris–HCl, pH 7.5) by varying the temperature from 5° to 45 °C, and in n-propanol/20 mM Tris–HCl buffer, pH 7.5, (96:4, v/v) solution at room temperature.

Estimates of PPII helix content can be obtained using

$$ \% {\text{PPII}} = \frac{{[\Uptheta ]_{ \max } + 6 ,100}}{13,700} \times 100 $$

where [Θ]max is the molecular ellipticity at the characteristic maxima. The upper (100 %) and lower (0 %) limits were determined from CD spectra for the polyproline peptide in an 8.4 M guanidine hydrochloride solution and for a peptide model for completely disordered proteins, respectively (Kelly et al. 2001).

Results

Peptide design and synthesis

Our goal is to enhance the tendency of the P2 peptide to adopt a PPII helix conformation in aqueous solution, stabilizing the trans-conformers in Xaa-Pro amide bonds. This effect should decrease the entropic cost of the peptide-SH3m-cort interaction, increasing the affinity of peptide towards the SH3 domain. Proline derivatives with a substituent in the γ-position (C4) have proven to be useful for the tuning of the cis/trans conformer ratio (Renner et al. 2001; Kotch et al. 2008; Shoulders et al. 2006). Proline can adopt either a Cγ-exo ring pucker, in which Cγ is puckered toward the Cα proton, or a Cγ-endo ring pucker, in which Cγ is puckered toward the carbonyl group (Fig. 1). Conformational studies show that both the nature of the substituent at γ position and the absolute configuration at this center influence the pyrrolidine ring pucker and consequently the cis/trans conformer ratio of the Xaa-Pro amide bond. An exo pucker, and consequently a trans Xaa-Pro amide bond, is favored when an electron-withdrawing group is in the 4R position while an endo pucker is preferred when the substituent is in the 4S position. This has been attributed to the reduction of the bond order of the C–N linkage by the increased N-pyramidalization due to the electron-withdrawing group (Panasik et al. 1994; Eberhardt et al. 1996), and to an n → π* electrostatic interaction between the carbonyl groups of the i-1 and i residues present when the peptide bond adopts a trans conformation (Bretscher et al. 2001; Hinderaker and Raines 2003).

Fig. 1
figure 1

Ring conformation of 4-substituted Pro derivatives. The ring conformational preferences depend on the stereochemistry and electronic effects of the substitution

Full size image

In the case of Hyp, the strong preference for a γ-exo pucker has been attributed also to the gauche effect (Wolfe 1972): in the γ-exo conformation, a gauche orientation of the 4-OH group and the pyrrolidine nitrogen, relative to the Cγ-Cδ bond axis, is possible (Taylor et al. 2005). Moreover, recent theoretical studies suggest that the driving force to adopt the γ-exo conformation is hyperconjugative interaction which involved the donation from the σ(Cβ-H)ax and σ(Cδ-H)ax bonding orbitals into the σ*(Cγ-O) antibonding orbital (Alabugin and Zeidan 2002; Improta et al. 2001).

The proline ring pucker constrains the φ, ψ, and ω main chain torsion angles, and thus, a selective control of the ring pucker provides the possibility of preorganizing the peptide backbone conformation and modulating the peptide stability (Zheng et al. 2010; DeRider et al. 2002).

Peptides were synthesized by the solid phase method on a Wang resin according standard Fmoc/HBTU protocol. To overcome low yields and subsequent side reactions, the coupling to the secondary amino group of 4R-fluoroproline residues was performed using TFFH as coupling reagent (Ruzza et al. 2006). Following the removal of the Fmoc group, side chain protecting groups were left on during cleavage from the resin by TFA standard treatment (see Experimental Section). Crude products were purified by preparative reversed-phase chromatography. The correct composition of the synthetic peptides was confirmed by ESI–MS analysis.

Peptide conformation in aqueous solution

CD spectroscopy is the technique of choice to detect PPII helices conformation in solution. The presence of a positive CD band at about 217 nm and a negative band between 200 and 210 nm are diagnostic of the PPII helix conformation (Siligardi and Drake 1995; Drake et al. 1988). The positive CD band is generally red-shifted towards 227 nm in proline-rich peptides due to the increased content of tertiary amide chromophore (Venugopal et al. 1994) and its intensity is proportional to the PPII helical content (Kelly et al. 2001). By contrast, truly unordered peptides, with a dynamic structure that samples all the available conformational space, exhibit a similar but somewhat distinct shape. Their hallmark is a stronger negative band at about 200 nm accompanied by a negative band at about 225–227 nm (Woody 1992). Thus, both shape and magnitude of the CD spectra differentiate the unordered polypeptides from those displaying a preference for a PPII-helical structure.

The conformation of P2 and its analogues in the temperature range from 5° to 45 °C was assessed by far-UV CD spectroscopy. As an example, the far-UV CD spectra of H3 registered at different temperature have been reported in Fig. 2. Upon heating, the positive CD band at about 228 nm, hallmark of the PPII helix, decreased, and the negative band at about 200 nm slightly shift towards a higher wavelength, according to the results described by different authors (see Woody 2010, for a compressive Review). An isodichroic point at about 210 nm was observed in temperature dependence CD spectra of H3, as well as in the CD spectra of other compounds (Figure 1S in Supplementary Material), indicative of an equilibrium between two forms: the PPII helix conformation and the irregular structure (the so-called extended). The position of positive maximum will vary slightly from the 228 nm, depending on the ratio of tertiary to secondary amides in each peptide, as well as upon the nature of other structure adopted.

Fig. 2
figure 2

Far-UV CD spectra of the H3 peptide (0.1 mM) in 20 mM Tris–HCl, pH 7.5, buffer solution as a function of temperature

Full size image

The plot of the molar ellipticity values of the positive band in the CD spectra versus temperature revealed that all peptides showed qualitatively similar trends (Fig. 3). In terms of absolute molar ellipticity intensity, however, the H3 analogue showed the highest values, and consequently the highest PPII helical content (Table 2). On the contrary, the parent P2 peptide displayed the lowest propensity to adopt a PPII helix conformation on the all range of tested temperature. H2 and F2 peptides were characterized by a very similar behavior to P2 peptide. The intensity of the positive band nm at 25 °C, and consequently the PPII helical content (Table 2), is seen to decrease in the order H3 > H5 ≈ F3 > F5 > F2 ≥ H2 ≈ P2.

Fig. 3
figure 3

Temperature dependence on the maxima in the CD spectra for the P2 peptide and its Hyp or FPro analogues in 20 mM Tris–HCl, pH 7.5, buffer solution

Full size image
Table 2 Positions and heights of maxima in CD spectra of peptides at 5 and 25 °C in 20 mM Tris–HCl, pH 7.5, estimated PPII helix content, and values of peptide binding affinity to GST-SH3m-cort domain
Full size table

The CD spectra of peptides were also registered in n-propanol/20 mM Tris–HCl buffer, pH 7.5, (96:4, v/v) solution, condition which favors the right-handed polyproline I helix (PPI), after a week’s incubation, due to the slow conversion of amide bond from trans to cis form. The shape of the CD spectra of the P2 analogues (data not shown) is lacking in the characteristic hallmark of the PPI helix (positive band at about 210–215 nm), but also of that of PPII helix, indicating that an electron-withdrawing group on the 4R position substantially disfavors the formation of PPI conformation (Horng and Raines 2006; Kümin et al. 2007).

Determination of the K d values for GST-SH3m-cort domain

The interaction occurring between the P2 analogues and GST-SH3m-cort fusion protein was analyzed using the near-UV CD spectroscopy. The interactions of the SH3 domain with ligands are mediated through the stacking of aromatic amino rings from residues like Trp, Tyr and Phe with the pyrrolidine rings of the prolines located in the binding motif. These interactions modify the environment of the aromatic ring, generating an excellent in situ molecular probe of the protein-peptide interaction. The binding of the peptides to GST-SH3m-cort fusion protein can be clearly determined by the spectral changes of the Trp side chain band at approximately 290 and 295 nm (Fig. 4). The values of the apparent dissociation constant K d are determined by the non-linear regression analysis of the ΔA values of the CD spectra at 294 nm with increasing SH3/peptide molar ratios (Table 2; Fig. 5).

Fig. 4
figure 4

Near-UV CD spectra of GST-SH3m-cort in the presence of increasing amount of H3 peptide. GST-SH3m-cort was 53.4 μM in 20 mM Tris–HCl buffer, pH 7.5; H3 peptide was 2.574 mM in 20 mM Tris–HCl buffer, pH 7.5. Spectra were recorded at 25 °C. The ΔA values were measured as a function of the increasing peptide/GST-SH3m-cort molar ratios (indicated)

Full size image
Fig. 5
figure 5

Determination of the K d values by CD spectroscopy analyses. The ΔA values at 294 nm, measured as a function of different SH3/peptide molar ratios, are plotted versus the peptide concentration

Full size image

The titration of the isolated GST protein with P2 does not induce any CD changes in both the near and far-UV regions (Figure 2S), thus confirming the absence of non-specific peptide-GST interaction. Moreover, since GST protein might undergo dimerization (Dourado et al. 2008), parallel titrations using either free SH3m-cort domain or GST-SH3m-cort fusion protein and a suitable proline-rich peptide were previously performed (Rubini et al. 2010). The K d values determined for the proline-rich peptide were very similar, supporting the validity of using the recombinant fusion protein instead of the free SH3m-cort domain in CD spectroscopy analysis.

Table 2 shows that P2 peptide interacts with the GST-SH3m-cort fusion protein with a very favorable K d value (2.5 ± 1.3 μM). The substitution of Hyp for Pro residues into both H2 and H3 peptides does not have a significant effect on the K d values, while H5 binds with an affinity about sixfold lower (K d = 13.2 ± 1.0 μM) than that P2.

Surprisingly, the replacement of Pro residues into the P2 sequence with FPro residues is a negative determinant in the binding process. Peptides containing the FPro residues are characterized by higher K d values with respect to P2 peptide (see Table 2).

Discussion

Many of the so far identified SH3 domain ligands are rich in Pro residues; this could be due to the fact that they readily adopt a PPII helix conformation in solution (Lim et al. 1994). PPII is highly extended with a helical pitch of 9.3 Å/turn and 3.0 residues/turn that seem to fit best steric and hydrogen bonding pattern of SH3 ligand surface, without a significant loss in conformational entropy.

Artificial induction of PPII helices has been reported using modified proline residues either by non-covalent stabilization (Tuchscherer et al. 2001; Ruzza et al. 2006; Jacquot et al. 2007; Flemer et al. 2008) or by ring closing metathesis macrocyclization (Liu et al. 2010, 2011).

The introduction of an electron-withdrawing group in the 4 position of the pyrrolidine ring of Pro residue influences both proline ring pucker and the cis/trans conformer ratio of the amide Xaa-Pro bond. The substitution of either Hyp or FPro for Pro residues into the P2 sequence has been used to stabilize the proline γ-exo pucker, and thus, the PPII helix conformation. In addition to the configuration of the γ-substituent, the ability of these residues to stabilize the PPII helix conformation is strictly related to their position into the peptide sequence.

As shown in Fig. 3, the introduction of Hyp residues at 2, −1 and −4 positions (peptide H3), involving the two canonical positions occupied by Pro residues in class II binding motif (2 and −1 positions, see Table 1), was found to promote the higher content of PPII helix either with respect to other positions or the presence of the FPro residues (Table 2). Indeed, the substitution of Hyp for Pro stabilizes the PPII helix more efficiently than the introduction of FPro analogue, although the hydroxyl group is less electronegative than the fluorine atom. This may be due to the capability of the hydroxyl group to perform additional non-covalent interaction between the γ-substituent of the Pro residue and the amide backbone, stabilizing the trans conformer in Xaa-Pro bond (Kuemin et al. 2010). Moreover, the capability of the hydroxyl group of Hyp to realize strong interactions with the aqueous solvent could efficiently stabilize the PPII helix conformation. Conformational studies have been suggested that peptide–solvent interaction is a major driving force in PPII helix formation (Shi et al. 2002; Rucker et al. 2003).

These results support our hypothesis that the introduction of a 4-(R)-electron-withdrawing group into proline residue belonging to a proline-rich peptide stabilizes the PPII helix conformation in aqueous solution. This is important when the PPII structure is the relevant biological binding conformation required to trigger the signal transduction mediated by ligand-binding interactions, particularly when peptides mimic a portion of protein structure where the intramolecular interactions, characteristic of the protein structure, are lost.

Unexpectedly, the induction of a stable PPII helix conformation in ligands does not improve the K d values of GST-SH3m-cort/peptide complexes (Table 2). The lack of improved affinity may be ascribed to two different factors: the unfavorable placement of the puckers of the Pro residue involved in the interactions with aromatic SH3 residue, and/or the difficulty to optimize interactions outside the classical Surface I binding site.

Previously we demonstrated that the Pro2 and Pro-1 residues into P2 sequence interacted with the SH3m-cort Tyr541 and Trp525 residues, respectively (Rubini et al. 2010). The pyrrolidine ring of the Pro-1 residue adopts almost an unpuckering conformation which can make an edge-face arrangement to form a strong aromatic-proline interaction (Bhattacharyya and Chakrabarti 2003) with the indole ring of Trp525. Thus a strongly preorganized exo ring pucker would push the Cγ atom and the pyrrolidine ring away from Trp525 weakening the aromatic-proline interaction, which cannot be compensated by the increase in the C–H·····π interaction due to the inductive effect imposed by the γ-substituent (F or OH). Recent studies using conformationally biased proline derivative showed that a preorganized ring pucker matching the native proline conformation would stabilize the protein structure; otherwise, a destabilizing effect would be observed when residues are involved in protein interaction (Shoulders et al. 2010; Zheng et al. 2010).

Additionally, our data also provide the evidence that the SH3m-cort domain recognizes proline-rich sequences, which contain basic residues outside the PxxPxK core. In particular, a Lys residue at the −5 position of the peptide-ligand is markedly effective in promoting the binding to the SH3m-cort domain by the interaction with Asp505 belonging to Surface II binding site (Rubini et al. 2010). The absence of increased affinity towards the SH3m-cort domain after the induction of a stable PPII helical conformation in the peptide-ligand suggests that a rigid scaffold is not suited for the ligand-binding surface, more likely due to a decrease of the peptide flexibility necessary for an optimal interaction with both Surface I and Surface II of the SH3m-cort domain.

Therefore, the binding affinity of the proline-rich peptides to the SH3m-cort domain is the result of a delicate balance of mutually compensating contributions that play a role in the binding energy: the PPII helix conformation that promotes aromatic-proline interactions, and the ionic interaction that involves the Lys-5 residue outside the classical class II binding motif.

Conclusions

An increasing body of evidence highlights the importance of SH3-mediated protein interactions in critical signaling processes that result in cancers and other pathological conditions. Therefore, several attempts have been performed aimed at generating drugs that can interfere with the SH3-mediated processes. However, the promiscuity and restrictions characterizing the SH3/ligand interactions are challenges for rational drug design. In this respect, the finding that the critical step in the SH3m-cort/ligand interaction involves both aromatic-proline interaction and ionic interaction outside classical class II binding motif allows the design of specific inhibitors of biological pathways, where cortactin-signaling has been demonstrated to be involved in malignancy.