Introduction

Aptamers are functional oligonucleotides, proteins, or peptides isolated from very large random sequence libraries by in vitro selection based on the affinity to a desired specific target. Protein and peptide aptamers were first selected using phage display (Smith 1985) and subsequently using in vitro translation systems, including ribosome and mRNA displays (Hanes and Pluckthum 1997; Nemoto et al. 1997; Roberts and Szostak 1997). Canonical components (i.e., four nucleotides and 20 amino acids) and non-canonical moieties have been used for aptamers. In particular, the number of non-canonical moieties in the selection library has increased over the last decade (Uzawa et al. 2013; Lane and Seeling 2014; Dodevski et al. 2015; Maini et al. 2016).

For example, photo-isomerizable (Liu et al. 2012; Jafari et al. 2014; Bellotto et al. 2014), inhibitor (Li and Roberts 2003; Wang et al. 2014a, b), catechol (Wilke et al. 2014) and environment-sensitive fluorescence groups (Wang et al. 2014a, b; Manandhar et al. 2015; Wang et al. 2016) have been incorporated into aptamers. These functional groups add new artificial or additional functions to aptamers, which inherently have molecular recognition functions. The introduction of a photo-isomerizable group facilitated control of binding by photo-irradiation (Liu et al. 2012; Jafari et al. 2014; Bellotto et al. 2014), whereas the addition of an aptamer sequence to an inhibitor increased binding affinity by increasing the number of interaction sites (Li and Roberts 2003; Wang et al. 2014a, b). The introduction of an environment-sensitive fluorescence group enabled a fluorescence change upon binding a target molecule (Wang et al. 2014a, b; Manandhar et al. 2015; Wang et al. 2016). This fluorogenic aptamer technology realizes universal analysis of arbitrary targets without bound/free separation.

We reported previously the identification of fluorogenic peptides that bind to calmodulin (CaM) through selection using a combination of ribosome display and misaminoacylated, biorthogonal tRNA (Wang et al. 2014a, b). A hydrophobic indicator, 7-nitro-2,1,3-benzoxadiazole (NBD), that emits fluorescence in a hydrophobic environment was incorporated into the library. In this report, the behavior of a peptide among candidate peptides was investigated in detail. However, in this study, we found that the addition of a common sequence, which was employed for ribosome display (Supplementary Fig. 1), makes the affinity of another candidate peptide as strong as the previously reported peptide. Therefore, the binding behaviors of these peptides with the addition of this common sequence were characterized by fluorescence spectroscopy, surface plasmon resonance (SPR) and nuclear magnetic resonance (NMR). The additional sequence added to a peptide was found to reflect the interaction between target and a peptide-ribosome-mRNA complex during ribosome display.

Materials and methods

Solubility

The synthesized peptide solutions were super-saturated using a microdialysis device overnight in buffer (50 mM Tris/acetate, 150 mM KCl, 5 mM CaCl2, pH 7.0). The saturated solutions were collected and centrifuged at 14,000×g for 20 min to obtain the supernatant for absorbance measurement. The maximum absorbance was then measured for NBD and the concentration was calculated using the extinction coefficient of NBD at 485 nm (ϵmax = –19,500 L/mol cm). The acidic, basic and hydrophobic sequence composition as percentages were calculated using the peptide property calculator LifeTein® online software program.

Surface plasmon resonance (SPR) measurements

SPR measurements were performed on a Biacore T100 instrument with a CM5 sensor chip (GE Healthcare Bio-Science, Uppsala, Sweden). Bovine CaM purchased from Sigma Aldrich (St Louis, MO, USA) was immobilized on the CM5 sensor chip surface according to previous reports (Lin et al. 1974; Crouch and Klee 1980).

After immobilization of CaM onto the sensor surface, affinity analysis was carried out for the peptides. All the experiments were performed at 25 °C with a constant flow of 30 µL/min. For the affinity analysis, various concentrations of peptide solutions in buffer with 0.05% Tween 20 and 5 mM CaCl2 were injected over the sensor chip containing the immobilized CaM. The sensor chip was regenerated by injection of 10 mM EGTA (pH 8.0) for 30 s, followed by buffer with 0.05% Tween 20 and 5 mM CaCl2 for 5 min to stabilize the baseline. This measurement was carried out only in the presence of Ca2+. All of the binding curves were collected by subtraction of the curve for a reference flow cell and fitted to a steady state affinity model using the Biacore T100 Evaluation Software (version 2.0.3).

Fluorescence measurements

Fluorescence measurements were performed excitation was at 488 nm and emission was measured from 500 to 600 nm: To determine the fluorescence intensity change, CaM from 0–16 µM was added to 2 μM MB4. Similarly, CaM from 0–30 µM was added to 5 µM of MC5 and MD9 peptides in buffer. Fluorescence intensity changes in the presence and absence of Ca2+ were measured. For CaM binding to the peptides, the concentration of CaM was increased until a saturation level was obtained. The relative fluorescence intensity changes (RFIC) of peptides with CaM were calculated at the maximum emission wavelength of 535 nm. RFIC were calculated as RFIC = (FX − F0)/F0, where FX and F0 are the fluorescence intensity of a peptide with and without CaM, respectively. The calculated RFIC were plotted using Graph Pad Prism 5 (Graph Pad Software Inc., CA, USA) and the dissociation constant (K d) values were estimated by non-linear regression analysis of the emission intensities at 535 nm.

NMR measurements

For the NMR titration experiments of 15N-labelled human CaM with MB4, MC5 and MD9 peptide aptamers, a series of two-dimensional (2D) 1H–15N HSQC spectra were measured on 15N-labeled CaM (0.1 mM) at various protein:peptide ratios to monitor gradual changes in the chemical shifts or intensity of 1H–15N correlations. The 15N-labeled human CaM sample was purified using the protocols presented previously (Hayashi et al. 1998). 15N-Labelled CaM was dissolved in 50 mM Tris/HCl buffer (pH 7.5) containing 120 mM NaCl and 2.5 mM CaCl2 and 10% 2H2O. 2D 1H–15N HSQC spectra were acquired with eight transients and a total of 2048 (t 2, 1HN) × 256 (t 1, 15N) complex points. Initially, a reference spectrum was recorded in the absence of peptides. Then the sample was mixed with 20 mM peptide to achieve a protein:peptide molar ratio of 1:0.5, prior to the next measurement. More peptide solution was added incrementally to measure 2D 1H–15N HSQC spectra at protein:peptide molar ratios of 1:1, 1:2 and 1:3. The concentration of the protein was not adjusted after each titration point because of the small volume of the peptide solution added. Details about NMR measurement system and backbone assignments are in the supporting information.

Results and discussion

The selected peptides with five fixed residues, MAMQA, at the N-terminus, which derive from the restriction enzyme site during selection (Supplementary Fig. 1), were synthesized using the NBD labeled aminophenylalanine by solid phase peptide synthesis. We left the additional five amino acids in common, because the translational efficiency of a polypeptide largely depends on its N-terminal amino acids sequence. Ueno et al. (2012) proposed an elegant strategy to use a protease to generate the library whose N-terminus are randomized; however, the efficiency of protease activity would also depend on the library sequence.

Solubility

The solubility of the peptides in an aqueous buffer and the calculated hydrophobic percentage of each synthesized peptide are presented in Table 1. The solubility was dependent on the sequence. Although the hydrophobic percentages were the same for peptides MB4 and MC5, the percentage of acidic and basic amino acids differed between the two peptides.

Table 1 Synthesized selected peptide sequences with solubility
Full size table

SPR measurements

The binding properties of the peptide aptamers (MB4, MC5 and MD9) on immobilized CaM in the presence of Ca2+ were performed by SPR (Supplementary Figs. 2, 3, 4). Table 2 presents the affinity analysis of the peptides using a steady state affinity model with dissociation constants given. MB4 and MC5 have similar affinity to CaM, whereas the affinity of MD9 was slightly lower.

Table 2 Peptides with K d (µM) values
Full size table

Fluorescence measurements

The fluorescence of the three peptides in the presence of calcium ions was measured (Supplementary Fig. 5). The increase in fluorescence intensity indicates that the peptides bind in a hydrophobic pocket of CaM, that is exposed to Ca2+. The fluorescent intensity of MB4 was extremely low in the absence of CaM and the wavelength of maximum fluorescence was 539 nm. The wavelength shifted to 532 nm as the concentration of CaM increased. Conversely, in the absence of CaM both MC5 and MD9 showed weak fluorescence. As observed for MB4, the wavelength of maximum fluorescence intensity of MC5 shifted from 541 to 531 nm and that of MD9 from 542 to 531 nm as the concentration of CaM increased. Although the initial fluorescence of each peptide differed, the shift in maximum fluorescence intensity was to the blue, and the width was between 7 and 10 nm. The blue shift demonstrated the hydrophobic interaction of the NBD moiety with CaM.

The relative fluorescence intensity changes (RFIC) of peptide with CaM were at 535 nm. In the presence of Ca2+, the fluorescent signal of the MB4 peptide increased up to 16-fold with a concomitant increase in CaM to 10 µM (Fig. 1a). Although the fluorescence intensity of the corresponding peptide without the common sequence (B4 in the supporting information in Wang et al. 2014) was found to increase linearly, the presence of this common sequence drastically changed the behavior yielding a fluorescence increase for MB4, which is typically seen for specific binding. In the case of MC5 and MD9, the increase in fluorescence intensity was 3- and 3.5-fold, respectively (Fig. 1b, c). The K d between peptides and CaM were estimated from the results of Fig. 1, as shown in Table 2. The K d values of the three peptides were of the same order. Although the binding affinities of peptides are dependent on sequence, the micro-environmental change of NBD induced by the interaction with CaM was not dependent on the peptide sequence.

Fig. 1
figure 1

Data points fitted by the blue line represent the RFIC of Ca2+ bound CaM at different concentrations and the curve shows the best fit to non-linear regression analysis of emission intensities at 535 nm. The red line containing data points shows the RFIC in the absence of Ca2+ bound CaM

Full size image

NMR measurements

The binding mechanism of peptides MB4, MC5 and MD9 with CaM was characterized by a series of stepwise, multipoint NMR titration experiments using 15N-labeled human CaM (Fig. 2). Initially, a 2D 1H–15N HSQC was measured in the absence of peptides (zero concentration point) and then spectra at each peptide titration point were measured to identify the binding interface on CaM.

Fig. 2
figure 2

A 2D 1H–15N HSQC spectrum. Mapping of the affected residues by addition of the MC5 peptide on CaM in a ribbon diagram. Large chemical shift changes to resonances induced by the addition of MC5 are shown in red. Ala129, Asn138 and Glu140–Met145 are represented as stick models. Ca2+ ions are shown as magenta spheres

Full size image

Titration of 15N-labeled CaM with MB4 peptide gave rise to chemical shift changes to a large number of resonances. We observed the disappearance and emergence of signals, indicating that the binding of CaM to MB4 was on the slow exchange timescale. The disappearance and emergence of resonances were observed at the first titration point (protein:peptide molar ratio of 1:0.5). A gradual increase in MB4 peptide concentration to saturate CaM protein (protein:peptide molar ratio of 1:3) gave rise to a large amount of precipitant and the change in chemical shifts of resonances was not saturated. Therefore, CaM and MB4 were mixed under diluted conditions (0.01 mM CaM with 0.03 mM MB4), and the sample was gradually concentrated using an ultrafiltration membrane. As a result, we observed saturation of chemical shift changes to resonances at a CaM:MB4 molar ratio of 1:3. NMR titration of 15N-labeled CaM with MC5 and MD9 peptides also gave the disappearance and emergence of signals, indicating that binding of CaM with these two peptides is also on the slow exchange timescale. The disappearance and emergence of resonances were observed at the first titration point (protein:peptide molar ratio of 1:0.5) and was almost saturated at a protein:peptide molar ratio of 1:3 (MC5) and 1:4 (MD9).

Due to the timescale of the interaction, it was difficult to follow chemical shift changes to resonances during the peptide titration experiments with assignment information for apo-CaM. Therefore, assignment of the backbone 1HN, 13Cα, 13C′, 15N and side-chain 13Cβ resonances of 13C/15N-uniformly labeled CaM in the presence of the peptides was performed. By analyzing the spin–spin connectivity in the 3D triple-resonance NMR spectra, 67% of the backbone resonances (100/149 amino acids) were assigned in the presence of the MC5 peptide at the protein:peptide molar ratio of 1:3. In the case of MB4 and MD9 peptides, no assignment information of the backbone resonances of CaM was completed because the quality of the 3D NMR spectra was poor. This was due to the low solubility of the peptides, which caused field inhomogeneity and thus poor quality 3D spectra; although, peptide solubility was enhanced by the addition of the five residues at the N-terminus.

Large chemical shift changes induced by the addition of the MC5 peptide (Δave > 80 Hz) were found for residues Ser18, Phe20, Ile28, Val36, Met37, Phe65, Met73–Met77, Glu85, Asp94, Lys95, Leu106, Arg107, Met110, Thr111, Lys116, Leu117, Val122, Ile126, Arg127, Ala129, Asn138 and Glu140–Met145. These residues form a solvent exposed cluster. The residues affected by the addition of the MC5 peptide were mapped to the crystal structure of Rattus rattus CaM (PDB ID:3CLN). The mapping result shows that MC5 binding to CaM affects many regions of CaM; however, the changes to Ala129, Asn138 and between Glu140–Met145 were particularly large and clustered (Fig. 3a). This result indicates that residues Ala129, Asn138 and Glu140–Met145 furnish the primary binding surface for the MC5 peptide (Fig. 3b), and that the binding surface for MC5 is located in the C-terminal domain of CaM, which correlates with the results observed for the C5 peptide (Wang et al. 2014). For MB4 and MD9 binding, the NMR spectra obtained were compared with the NMR data for MC5, and this overlaying of spectral data showed that the changes in chemical shifts of resonances yielded a similar pattern, indicating that these three peptides bind a similar region located in the C-terminal domain of CaM.

Fig. 3
figure 3

a Chemical shift changes to resonances representing CaM upon interaction with MC5. The chemical shift changes in both 1H and 15N were normalized by the equation [(∆15NHz)2 + (∆1HHz)2]1/2. b The expected primary binding surface for the MC5 peptide

Full size image

CaM captures various kinds of molecules and targets are not restricted to proteins. Nevertheless, physical properties of the target ligands are conserved with most showing basic amphiphilicity (Hayashi and Titani 2010) and in the many cases CaM recognizes these ligand properties using a similar core region. Therefore, the similar resonance shift patterns of the 2D 1H–15N HSQC spectra obtained from the titration experiments of CaM with target peptides suggests that CaM uses a similar binding region as shown for other ligands, even though these three peptides do not display explicit basic amphiphilicity.

Conclusion

Three major in vitro selected peptide sequences were synthesized by solid phase peptide synthesis. The interaction of the peptides with a target molecule was investigated by SPR, fluorescence and NMR spectroscopy. We found that the addition of the common sequence, which was employed for ribosome display, makes the affinity of another candidate peptide as strong as the previously reported peptide.