Introduction

Oxytocin (OT) is a nine-amino acid metal-ion binding cyclic peptide (Fig. 1A) that serves as a neurotransmitter and hormone, and binds the oxytocin G protein-coupled receptor (OTR) [1]. OT is involved in social bonding [2], sexual attraction [3], maternal bonding [3] and childbirth [4]. Copper is an essential metal in nature, but is toxic in its unbound form, hence its binding and transport must be carefully regulated [5, 6]. OT binds copper ions (Cu2+), changing the conformation of the peptide and its affinity to the OTR [7,8,9].

Fig. 1
figure 1

a Schematic structure of OT and Ac-OT. b Proposed Cu2+ binding moieties in OT based on EPR (2) and by CID-MS (1,3,4,5). Black and white letters indicate copper-bound and unbound amino acids, respectively, as suggested by each model. Solvent ligands are marked in blue

Full size image

Copper ions have been shown to bind OT through the terminal amine and several backbone amides which are deprotonated as part of the complexation process. The mechanism of Cu2+ binding by OT has been studied by several methods such as affinity chromatography, mass spectroscopy (MS), and electrochemistry [7, 8, 10,11,12,13,14,15,16,17]. An electron paramagnetic resonance (EPR) and potentiometric measurement analysis of OT determined that [OT-Cu2+]− is a planar complex in which three amide hydrogens are deprotonated [11, 18]. The EPR study showed that Cys1, Tyr2, Ile3 and Gln4 are the OT amino acids involved in Cu2+-binding (Fig. 1B). Works using electrospray ionization mass spectrometry with collision-induced dissociation (MS-CID), suggested that the [OT-Cu2+]− complex was formed via the amide backbone of amino acids Cys1, Cys6, Leu8 and Gly9 (Fig. 1B) [7]. Additional studies that analyzed the [OT-Cu2+]− complex by MS-CID suggested yet a different binding mode in which the copper is complexed with either: Tyr2, Ile3, Gln4, and Asn5; Ile3, Gln4, Asn5 and Cys6; or Cys1, Tyr2, Cys6 and a solvent ligand (Fig. 1B) [19, 20]. Many of the above studies were performed in the gas phase after ionization and not in a biologically relevant aqueous solution. Furthermore, the disparity between the complexation modes calls for a more decisive analysis of this crucial interaction.

Crystallography and nuclear magnetic resonance (NMR) are the most common methods for determining peptide structures and may provide a way to define the exact amino acids involved in OT-Cu2+ complexation [21]. Surprisingly, the exact amino acids involved in OT-Cu2+ complexation were never determined by these direct structural methods. The NMR analysis of OT-Cu2+ is hindered by the paramagnetic properties and slow relaxation of Cu2+, which results in extreme broadening out of signals of nuclei that are within 8 Å of the Cu2+ ion [22, 23].

Many studies show that the effect of metal ion proximity on NMR signals can be utilized for structural analysis [23,24,25]. For example, lanthanides provide structural information since the shift in NMR signals correlates the distance from the complexed metal ion. Cu2+ proximity has shown to be a useful tool for structural analysis [26, 27]. Paramagnetic Cu2+ causes extreme line broadening due to the combined effects of paramagnetic relaxation enhancement (PRE) and pseudo contact shifts (PCS) [25, 28,29,30,31]. Cu2+ is nearly isotropic and has a PRE effect that erases peak signals up to a distance of 8 Å from its nucleus [22, 23, 32, 33]. While erasing peaks in NMR is usually a disadvantage, it can provide information on the proximity of the ion to specific moieties and of the entire complex structure. The PRE and PCS effects of Cu2+ have been used to determine protein-ion and fatty acid-ion structures by solution NMR [23, 34].

Solution-state NMR together with PRE information were used to analyze the binding mode and structure of [OT-Cu2+]− and to determine which amino acid moieties are involved in [OT-Cu2+]− complexation. The binding mechanism of the [OT-Cu2+]−complex was elucidated based on the PRE response of OT resonances during titration with Cu2+, and the structure was determined while in fast exchange with copper [24, 32].

Results and discussion

We first determined the structure of unbound OT by solution NMR to better evaluate the effect of copper complexation at later stages. The assignment and structure of unbound OT were determined in 5 mM acetate buffer at pH 6.75 to avoid pH-dependent changes in structure or chemical shift due to release of the amide protons upon subsequent Cu2+ binding. The ensemble was well-resolved where the low energy ensemble of 14 of the 50 calculated structures had backbone and heavy atom RMSD values of 0.72 and 1.22 Å, respectively (See details in ESI). Half of the low energy conformations were stabilized by a hydrogen bond between the Tyr2 amide proton and the Asn5 carbonyl oxygen (Assignment in Table S1, structures in Figure S3). When this ensemble was compared to a previously solved structure in pure water [35] (PDBid 2MGO) the lowest conformations of each ensemble had a backbone RMSD of 2.10 Å suggesting that the structures were dissimilar under the respective measured conditions.

The ability of OT to bind Cu2+ under these conditions was determined by titration. OT was titrated at 20 °C with aliquots of Cu2+ to give a 1/20th molar ratio to OT. NMR spectra were recorded after each titration step (Fig. 2). The NMR hydrogen peaks showed changes in chemical shift, in both broadening and a decrease in signal (Fig. 2 and Figure S4). These signal changes indicate binding to Cu2+ because: (1) The decrease in signal could stem from deprotonation as part of the Cu2+-binding mechanism; (2) The broadening can be attributed to the paramagnetic effect due to Cu2+ proximity; (3) The change in chemical shift may indicate a different chemical environment and/or a change in water exchange rate upon binding.

Fig. 2
figure 2

1D 1H-NMR spectra of OT titration with Cu2+ at 20 °C showing chemical shift broadening as a function of Cu2+ concentration. Fingerprint region with amide assignment in black and sidechain assignment in grey (CT is C-terminal amidation). Inset shows increasing broadening of the buffer acetate protons as a function of titration

Full size image

Several studies have shown that the OT terminal amine plays a key role in Cu2+-binding [11, 17]. An N-acetylated OT analog (Ac-OT) was synthesized and studied to determine the effect of blocking the terminal amine on Cu2+ binding. Ac-OT was titrated at 20 °C with aliquots of Cu2+ to give 1/20th of the molar ratio of the peptide. The NMR spectra (Fig. 3) showed there was no significant change in the intensity of any of the peptide peaks (Fig. 3). The acetate peak from the buffer displayed a marked change in chemical shift and broadened out as a function of the Cu2+ concentration, serving as an internal control for Cu2+. Furthermore, the increased degree of broadening of the buffer acetate peak in the Ac-OT titration relative to the OT titration indicated that the copper was not bound by the peptide but complexed the acetate instead. This indicates that Ac-OT does not bind Cu2+, proving that the N-terminus is essential for Cu2+ − OT binding.

Fig. 3
figure 3

1D 1H-NMR spectrum of N-acetylated OT titration with Cu2+ at 20 °C showing no effect of the Cu2+ on the chemical shift or broadening of OT, but broadening of the acetate buffer signal as a function of Cu2+ concentration. The intensities of the spectra are corrected for dilution

Full size image

The detailed mechanism of binding was elucidated by following the Cu2+ titration of OT by 2D NMR. The TOCSY spectrum at 20 °C (Fig. 4) showed resolved peaks (Fig. 4A). The degree of change in chemical shift and reduction in signal due to broadening out with increasing Cu2+ concentration was determined for each hydrogen in the peptide (e.g., Fig. 4B). The HN signal of Ile3 disappeared upon adding the first aliquot of Cu2+ and the HN signal of Tyr2 was not evident in the spectrum after 3/20th molar equivalents of Cu2+ (Fig. 4C). The HN of Gln4 was lost after 5/20th and that of Asn5 was still barely evident at the end of the titration. Amide signals of Cys1 and Cys6 had low intensity in the non-titrated spectrum and could not be used for mechanistic study. The HN signal of Leu8 showed significant change in chemical shift (Fig. 4B peak A), but neither it nor Gly9 HN completely broadened out. Throughout the titration, the Gln4 and Asn5 sidechain amide, and C-terminus amidation resonances were evident, whereas, Tyr2 aromatic signals disappeared after 4/20th molar equivalents of Cu2+. Non-exchangeable hydrogens in the aliphatic region of the spectrum (Fig. 4D) also disappeared due to close proximity to Cu2+, including Tyr2 α and β hydrogens that disappeared at the 3/20th aliquot, Ile3 and Pro7 at the 4/20th aliquot and Gln4 at the 5/20th aliquot, Asn5 was significantly reduced but still evident at the end of the titration, and Leu8 and Gly9 persisted throughout the titration. Stronger binding was seen in the parallel titration of OT with Cu2+ at 14 °C, as expected, and the spectral phenomena were the same (Figure S5).

Fig. 4
figure 4

2D 1H-NMR TOCSY spectra of OT titration with Cu2+ at 20 °C showing a fingerprint region; b enlarged peaks from fingerprint region to exemplify broadening out (e.g., C and D) and chemical shift changes (e.g., A, C and D) and as a function of Cu2+ concentration; c the backbone and sidechain amide and aromatic regions and d aliphatic region of the TOCSY spectra

Full size image

Hydrogens which were further away showed changes in chemical shift owing to the structural changes in the peptides that resulted from Cu2+-binding. Together, the results suggested a binding mode where the N-terminus nitrogen was both the first to bind and was a prerequisite for binding since there was no broadening when the Cys1 amine was acetylated (Fig. 3). Subsequently Tyr2, then Ile3 and finally Gln4 probably bind in that order, as that is the order of reduction of amide signal. The sidechains of Tyr2 and Ile3 were strongly affected by the paramagnetic broadening, indicating a close proximity to the Cu2+. The changes in chemical shift of Leu8 and Gly9 were probably due to changes in equilibrium structure due to binding.

Spectral analysis indicated the binding sequence. Since binding Cu2+ necessarily exchanges the backbone amide, the order of which the backbone amide signal was lost presumably indicated the order by which the amino acids bind Cu2+. Amino acids within 8 Å of the Cu2+ ion were paramagnetically broadened out.

The identities of the amino acids that bound Cu2+ were used together with the NMR-derived NOE distance restraints to solve the structure of the [OT-Cu2+]− complex at 20 °C while in equilibrium between a bound and free form. Structural COSY [36], TOCSY [37], using the MLEV-17 pulse scheme for the spin lock (150 ms) [36], and rotating frame overhauser effect spectroscopy (ROESY) experiments[36, 38, 39] were acquired under identical conditions (see Supplementary information), assigned [40] (Figure S6 and Table S1) and the NOE restraints were derived in the presence of 1/10th of a molar equivalent of Cu2+ to OT. The ROESY spectrum gave a total of 78 NOE interactions, comprising 54 intraresidual, 20 sequential and 4 longer range interactions. As stated, the signals for Cys1 and Cys6 had extremely low intensity in both the unbound OT and when in complex with Cu2+ and did not provide any structural information of the region. The structure was determined using XPLOR-NIH [41, 42] where covalent bonds to Cu2+ were introduced for the four determined ligand nitrogens, Cys1, Tyr2, Ile3 and Gln4 using known square planar geometry [11, 43]. The resulting ensemble of 50 structures had no violations of canonical geometry and backbone and heavy atom RMSD values of 1.29 and 2.28 Å, respectively. The 16 lowest energy conformations showed a stable structure with RMSD values of 0.80 and 1.38 Å, respectively. The presented 10 low energy structures (Fig. 5) had overall backbone and heavy atom RMSD values of 0.60 and 1.21 Å, respectively, and RMSD values in the OT ring region of residues 1–6 of 0.10 and 0.94 Å, respectively.

Fig. 5
figure 5

Low energy conformational space of OT-Cu2+ complex at 20 °C, showing atoms within 8 Å of the Cu2+ in purple

Full size image

The conformation was compared to that of unbound OT to ensure that we were not measuring the free fraction of the sample (Figure S6). The free structure was compared to the bound structure without constraints to the Cu2+ to ascertain that the NOE constraints led to a new structure and that the bound Cu2+ was not inducing all the structural changes (Figure S6). The RMSD of the ring residues 1–6 of OT in the lowest calculated structures of the unbound molecule to those of the OT-Cu2+ complex without introducing the Cu2+ bonds, was 2.19 Å, which is much larger than each individual RMSD (above and in Supplementary Information), strongly indicating that the structures are distinct.

The low-energy ensemble of the [OT-Cu2+]− complex (Fig. 5; PDBid 7OTD) shows a rigid ring area binding the Cu2+ and the Tyr2 aromatic ring within 3.5–5.8 Å of the metal ion center as has been seen in other systems [44,45,46,47]. The hydrogens that disappeared at the initial stage of the titration were within 5 Å of the Cu2+ explaining their early broadening out, and amino acids 1–7, which showed significant deviation of chemical shift upon titration, all resided within 8 Å of the Cu2+ in purple (Fig. 5). Asn 5 is above the plane of the bound copper with α, β and δ sidechain hydrogens at an average of 4.0, 5.5 and 7.9 Å, respectively. Leu8 and Gly9 are farther from the paramagnetic center and show changes in chemical shift associated with structural changes in OT upon binding.

This structure has a loss of resolution in the Cys–Cys bond region since the NMR signal is faint even in the non-bound spectrum before titration. Nonetheless, the titration analysis strongly correlates the structural results of the non-exchangeable hydrogens. Our proposed Cu2+-binding amino acids agree with the one reported by Bal et al. utilizing EPR [11]. Other reports, which utilized MS-CID, suggest different binding modes of OT [7, 13, 14, 19]. The variation between the models stem from differences in methodologies but may also hint at intricacies of Cu2+ transport by the peptide, which leads to several complexation modes.

Conclusion

This is the first time that the complex between OT and copper was determined in solution using NMR analysis. PRE of OT upon titration with Cu2+ was measured by NMR and used to map the binding amino acids that serve as copper ligands under these solution conditions and indicated the order of attachment during the binding process. The titration of OT and Ac-OT solutions with Cu2+ resulted in the following findings: (i) OT binding required the free Cys1 amino terminus for binding. (ii) Tyr2, Ile3 and Gln4 subsequently bound to the Cu2+ ion in that order. (iii) The conformational space of the [OT-Cu2+] complex showed the aromatic sidechain of Tyr2 within range of the Cu2+ to undergo a π-cation interaction. These findings give insight into the way copper binds OT in aqueous buffered media and hence may reflect on the nature of this important complex in the physiological environment. Therefore, peptidomimetics can be designed based on this structure and NMR can be used to study important complexes of small peptides with paramagnetic metal ions.