Introduction

Oxytocin (OT) is a nine-residue peptide that facilitates labor and lactation [1], and is associated with psychological issues such as trust, fear, anxiety, aggression, and pair bonding [2, 3]. In the early 1950s, du Vigneaud and his collaborators isolated [4] and synthesized OT [5], as well as determined the sequence as Cys1-Tyr2-Ile3-Gln4-Asn5-Cys6-Pro7-Leu8-Gly9-NH2, with a disulfide bond between Cys1 and Cys6 residues [6], forming a Cys1-Tyr2-Ile3-Gln4-Asn5-Cys6 peptide ring and a Pro7-Leu8-Gly9-NH2 tail. The structures of the ring and tail are of interest because they are implicated in divalent metal binding, which is known to increase the affinity of OT for its receptor [7]. With this interest, Bowers and coworkers used a combination of ion mobility spectrometry-mass spectrometry (IMS-MS) and theoretical methods to show that when Zn2+ binds to OT, it induces a structural change, and they proposed that this was key to the increased receptor affinity for the Zn2+–OT complex [8, 9]. Their calculations suggest two Zn2+–OT structures: one in which Zn2+ is octahedrally coordinated with backbone carbonyls associated with six residues (Tyr2, Ile3, Gln4, and Cys6 of the ring, as well as Leu8 and Gly9 of the tail); and another having an additional interaction with the amino terminus [9]. In both of these configurations, the Cys6–Pro7 bond adopted a trans-orientation, although the cis-isomer was not specifically investigated. Electron capture dissociation studies are in agreement with both the ring and tail residues coordinated by Zn [10].

Although, the prior IMS work did not explicitly consider the presence of cis-configured Pro7 in OT, nuclear magnetic resonance (NMR) studies indicate that a proportion of the Cys6- Pro7 bond in OT exists in the cis-configuration [11]; several recent studies from our laboratory show that peptides containing a proline residue often exhibit multiple conformations resulting from coexistence of proline in cis- and trans-configurations [12–16]. Because of this, and the biological importance of understanding the regulation of OT, in this present study we explicitly investigate the role of cis- and trans-forms of the Pro7 residue when OT binds to Zn2+. As shown below, we find that, in agreement with the NMR studies [11], the free peptide samples both cis- and trans-configured proline residues; upon binding the cis-isomer of proline must isomerize to the trans-isomer, such that the Zn2+-OT complex favors the trans-configuration, a final configuration that is consistent with the structure proposed from theory [9]. The cis→trans isomerization of the Pro7 residue, which is penultimate to the ring, appears to be important in regulating metal binding, and hence receptor binding.

Experimental

IMS-MS

IMS theory and experimental techniques have been discussed elsewhere [17–22]. In these studies, experiments were performed using a home-built ion mobility spectrometer coupled to a time-of-flight mass spectrometer. Positively charged ions are produced via electrospray ionization (ESI) [23] on a Triversa Nanomate (Advion). Ions are then accumulated in a Smith geometry ion funnel [24] and pulsed into a 1.8 m long drift tube. The drift tube is operated with ~3 Torr of He buffer gas at 300 K and a weak electric field (~10 V·cm−1). Ions exit the drift tube through a differentially pumped region and are analyzed in the time-of-flight mass spectrometer [25].

Electrospray Ionization for Sampling Solution-Phase Structure

As noted, the present studies utilize ESI to sample biomolecule conformations that exist in solution. A number of studies have shown that populations of peptide and protein conformations observed in the gas phase are related to the populations from solution and under gentle conditions can be monitored by IMS-MS [16, 26–28]. That is, as ions are transferred from solution to the gas phase, information about populations of different conformations from solution is preserved [26, 29, 30]. Silveira et al. implemented cryogenic IMS-MS to provide further insight into how gas phase ions are produced via ESI [30]. They and others [26] hypothesize that ions become kinetically trapped by the evaporative cooling process associated with ESI. Therefore, solution-like populations are retained for several milliseconds, which is on the time scale of MS and IMS experiments, making these techniques useful for structural characterization [31, 32].

Assessing Whether IMS Intensities are Due to Solution- or Gas-Phase Populations

While it is often assumed that IMS peak intensities for ions produced by ESI reflect populations found in solution [16, 27, 28], it is important to directly test this, especially for small peptide systems, where ion activation can produce IMS distributions that correspond to “gas-phase” distributions [33]. We test this by gently activating mobility-selected ions and measuring the resulting distributions of structures using an IMS-IMS approach, as described in detail elsewhere [22]. Briefly, IMS-IMS-MS experiments are performed by operating the drift tube as two independent drift regions. Ions are mobility selected in the mid-funnel and are collisionally activated when a high field is applied across a 0.3 cm region. Ions are activated by collisions with the buffer gas and are annealed. Changes in the IMS distributions are then monitored as the new populations are separated in the second drift region.

Determination of Experimental Cross-Section

It is often useful to plot drift time distributions on a cross section scale. Recorded drift times are converted to collisional cross sections (Ω) using [34]:

$$ \Omega =\frac{{\left(18\pi \right)}^{1/2}}{16}\frac{ze}{{\left({k}_{\mathrm{b}}T\right)}^{1/2}}{\left[\frac{1}{m_{\mathrm{I}}}+\frac{1}{m_{\mathrm{B}}}\right]}^{1/2}\frac{t_{\mathrm{D}}E}{L}\frac{760}{P}\frac{T}{273.2}\frac{1}{N}, $$
(1)

where ze is the charge of the ion, k b is Boltzmann’s constant, m l is the mass of the ion, m B is the mass of the buffer gas, t D is the time required to traverse the drift tube, E is the electric field, L is the drift tube length, T is the temperature, P is the pressure, and N is the neutral number density of the gas at STP.

Peptide Synthesis

Peptides (OT, and the alanine substituted sequence, Cys1-Tyr2-Ile3-Gln4-Asn5-Cys6-Ala7-Leu8-Gly9-NH2) were synthesized by Fmoc solid-phase synthesis on an Apex 396 peptide synthesizer (AAPPTec, Louisville, KY, USA) as described previously [13]. Fmoc side-chain protected amino acids and Wang-type polystyrene resins were used (Midwest Biotech, Fishers, IN, USA). Deprotections were performed with 20% piperidine in dimethylformamide, and 1,3-Diisopropylcarbodiimide/6-chloro-1-hydroxybenzotriazole was used as the coupling reagent. Peptides were cleaved from the resin with a solution containing trifluoroacetic acid:triisopropylsilane:methanol at a 18:1:1 ratio. Additionally, because the peptides contained cysteine residues, 5% 2,2′-(ethylenedioxy) diethanethiol was added. Peptides were washed, precipitated in diethyl ether, and the products were dried using a vacuum manifold.

Observation of [OT + Ca]2+ in the Mass Spectrum

In this study, ESI produces almost exclusively doubly-charged ions of OT (either [OT+2H]2+ or [OT+Zn]2+). All solutions are sprayed out of water or a mixture of water and methanol. Each sample contains 0.5 μM OT and Zn concentrations ranging from 0 to 6 μM. We note that, although extensive efforts to reduce Ca2+ in these experiments have been made, we observe [OT+Ca]2+ ions in the mass spectrum. Careful studies of the influence of this contaminant on Zn2+ binding have been carried out (see Supplemental Figure 1). The Ca2+-OT abundance does not appear to vary with introduction of Zn2+. Thus it appears to be a spectator in this system, and we do not consider it further.

Determination of Association Constants

The areas under each peak for both [OT+2H]2+ and [OT+Zn]2+ are integrated after normalization by total ion abundance. The areas of the unbound peaks can then be compared to the areas after addition of Zn2+ to determine the fraction bound. Association constants are determined by plotting the fraction of OT bound to Zn2+ as a function of the concentration of free Zn2+ in solution. The points are fit with a ligand binding, one site saturation model [35] using:

$$ f=\frac{B_{max}*\left[Zn\right]}{\left(\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{${K}_A$}\right.+\left[Zn\right]\right)}, $$
(2)

where f is the fraction of OT bound to Zn2+, B max is the maximum specific binding, [Zn] is the free Zn2+ concentration, and K A is the association constant.

Results and Discussion

Changes in Conformational Distribution of OT Conformations as a Function of Solution Composition

Figure 1 shows the collision cross section distributions of [OT+2H]2+ obtained from different water:methanol solutions. The distribution for [OT+2H]2+ shows two peaks, corresponding to populations of structures with collision cross sections of Ω = 236 and 243 Å2, respectively. In water, the Ω = 236 Å2 peak comprises ~40% of the distribution. As the amount of methanol in solution is increased, a shift in the distribution occurs, and at the final solution of 95:5 methanol:water, Ω = 236 Å2 ions are favored, comprising 83% of the distribution. Figure 1 also shows the distribution of conformations at pH 3 in water, which is dominated by Ω = 236 Å2 ions, comprising ~95% of the distribution. This distribution is interesting because the solution composition is identical to the one reported previously from NMR studies [11, 36], which found two populations corresponding to cis- and trans-proline configurations at relative abundances of ~10% and ~90%, respectively. The observed changes in the distribution of conformations as a function of solution composition are consistent with populations of conformations coming from solution. This agreement suggests that our Ω = 236 and 243 Å2 peaks correspond to trans- and cis-isomers, respectively; however, we test this assignment rigorously below.

Figure 1
figure 1

Collision cross section distributions for [OT+2H]2+ obtained from different water/methanol solutions. The top trace is the distribution from a pH 3 aqueous solution identical to that used in a NMR study [11]

Full size image

IMS-IMS-MS of OT

To further understand the relationship of the populations measured in Figure 1 as a function of solution composition, IMS-IMS-MS experiments were conducted. IMS-IMS-MS experiments for mobility selected [OT+2H]2+ in Figure 2 show that regardless of which solution is used to produce ions, collisional activation (at our highest energies) results in an identical “gas-phase” distribution. As discussed in detail previously [33], observation of identical populations, regardless of which peak is selected (or which solution is used to produce ions) is evidence for a gas-phase, quasi-equilibrium distribution established in the absence of solvent by collisional activation. This quasi-equilibrium distribution favors the Ω = 236 Å2 structure, and is clearly different from the distribution produced under gentle ESI conditions, in which the populations are imposed from solution. Figure 2 also shows studies using more gentle activation conditions (60 V) reveal partial conversion of structures, which is consistent with this interpretation. Hereafter, we assume that the distributions recorded from our gentle experimental conditions correspond to solution populations.

Figure 2
figure 2

Collision cross section distributions shown for [OT+2H]2+ from IMS-IMS-MS experiments. (a) Selection and activation of conformation with Ω = 236 Å2. (b) Selection and activation of conformation with Ω = 243 Å2. The bottom trace for each represents the ESI source distribution from a solution of 50:50 water:methanol with 0.1% FA. Selection and activation experiments were performed by increasing the voltages until dissociation occurs (120 V for OT)

Full size image

Pro7→Ala Substitution of OT

As noted above, the abundances of peaks measured for the 100:0 water:methanol solution at pH = 3 are consistent with those reported for the same solution composition by NMR, and suggest assignments of Ω = 236 and 243 Å2 are trans- and cis-isomers of Pro7, respectively. This assignment is rigorously tested by substituting an alanine for the proline to yield Cys1-Tyr2-Ile3-Gln4-Asn5-Cys6-Ala7-Leu8-Gly9-NH2, in which Cys6-Ala7 bond should favor the trans configuration [37], as illustrated by the data shown in Figure 3 that displays the collision cross sections for OT (Ala7) along with OT (Pro7). Note that the Pro7→Ala substitution yields a single conformation at 234 Å2, which makes up ~98% of the distribution. However, since alanine has a smaller intrinsic size parameter, the collision cross section has been corrected to account for the intrinsic size difference between proline and alanine [38, 39]. The resulting peak has a cross section of 236 Å2. This shows that the Pro7 residue in the more compact conformation is in the trans configuration. The loss in intensity for the Ω = 243 Å2 peak upon alanine substitution indicates that Pro7 in the elongated conformation is in the cis-configuration. Although proline is not penultimate to the N-terminus, it is penultimate to ring, consistent with the importance of penultimate prolines to the conformational heterogeneity of peptides [15, 40].

Figure 3
figure 3

Collision cross section distribution for OT (Pro7) along with the distribution of OT (Ala7). For OT (Ala7), the black trace corresponds to the collision cross section adjusted to correct for the size difference between alanine and proline, and the red trace is the cross section without correction

Full size image

Determination of Zn2+-OT Association Constants for trans- and cis-OT

Figure 4 shows the collision cross section distributions for [OT+2H]2+ and [OT+Zn]2+ as a function of Zn2+ concentration. Two populations are observed for unbound [OT+2H]2+. Solutions with 15 different concentrations of Zn2+, ranging from 0 to 6 μM, have been investigated. The concentration of OT is held constant at 0.5 μM. When no Zn2+ is added to solution, the relative abundances of the trans (Ω = 236 Å2) and cis (Ω = 243 Å2) configurations are ~40% and ~60%, respectively. As more Zn2+ is added to solution, the overall abundance of [OT+2H]2+ decreases and [OT+Zn]2+ is observed in the mass spectrum. Interestingly, as the Zn2+ concentration increases, the decrease observed for free OT occurs in a conformation-specific fashion as seen in the mobility distribution. The abundance of the trans-configuration decreases relative to the cis-configuration, and at the final concentration of Zn2+, the trans-configuration comprises ~12% of the total distribution of [OT+2H]2+. It should be noted that the collision cross section for the trans-configuration in the [OT+2H]2+ distribution and the single conformation for [OT+Zn]2+ are very similar, suggesting that the Pro7 bond in OT–Zn complex exists in the trans-configuration, which could be the reason that the trans-conformation for [OT+2H]2+ is favored by Zn2+, in agreement with previous modeling studies on OT–Zn complex [9]. Figure 4 also shows the Zn bound form of OT is comprised of only one peak with Ω = 235 Å2. These data suggest that zinc binds to the multiple OT conformations, and adds rigidity to the molecule, which is evident by one prominent peak in the collision cross section distribution. This further elucidates why zinc increases the affinity of OT to its receptor; it not only imposes a structural change to OT but it also gathers the multiple conformations of a dynamic peptide and adopts just one upon binding.

Figure 4
figure 4

Collision cross section distributions for [OT+2H]2+ and [OT+Zn]2+. [OT+2H]2+ distributions are displayed as the black trace and [OT+Zn]2+ distributions as the red trace. The concentration of Zn (μM) is displayed on the left side of each trace. Distributions are normalized by total abundance of [OT+2H]2+ and [OT+Zn]2+. OT-Zn complex structure is adapted from ref. [9]

Full size image

Differences in binding affinities for Zn2+ to the two prominent conformations for [OT+2H]2+ can be calculated because of the conformational-specific binding. Many studies have been performed to determine binding constants for various protein–ligand complexes with ESI-MS [41–46]. In this study, we use IMS-MS data to calculate KAs for Zn2+ bound to OT. Figure 5 shows the results from the fit of the titration data for both the cis- and trans-configurations of Pro7 in OT. To obtain this plot, we integrate the area under the peaks corresponding to each conformation in the mobility distribution for [OT+2H]2+ and [OT+Zn]2+. From this, we can relate the drop in abundance for each conformation as the concentration of Zn2+ increases, to the individual conformers’ contribution to the OT–Zn complex. From the data in Figure 5, we determine binding affinities for the trans- and cis-configurations with Zn2+ to be 1.43 ± 0.24 μM−1 and 0.42 ± 0.12 μM−1, respectively. Thus, the affinity of OT for Zn2+, when Pro7 exists as the trans-isomer, is more than three times greater than when Pro7 exists as the cis-isomer. This difference in affinity ultimately requires that the cis-configuration isomerizes to the trans-isomer to stabilize the Zn2+-bound structure. The final trans-Pro Zn2+-OT structure is consistent with the theoretical structures proposed by Bowers and coworkers [9].

Figure 5
figure 5

KAs for the individual conformations of OT with Zn. The fraction bound of OT is plotted as a function of free Zn2+ concentration (μM). The solid curve is the fit for the trans configuration data and the dashed line is the fit for the cis configuration data

Full size image

There are two mechanisms to consider when discussing ligand binding: conformational selection [47, 48] and induced fit [49–51]. In conformational selection, a biomolecule may sample many structures along its energy landscape, and binding occurs through selection by the ligand to a preferred unbound conformation while maintaining similar structure. In the induced fit mechanism, the ligand binds to the unbound conformation and induces a specific structural change. From our scenario it is not hard to imagine both mechanisms involved in the binding process. Because the unbound trans-isomer has a similar cross section to that of the bound complex, the mechanism could be classified as conformational selection. However, we also observe the depletion of the cis-isomer when Zn is added to solution. In one case, a cis→trans isomerization could occur prior to Zn2+ binding, allowing Zn2+ to bind to the preferred structure, maintaining the conformational selection mechanism. However, there is also a scenario where Zn2+ could bind to the cis-isomer, inducing a cis→trans isomerization, which would then be considered under the induced fit mechanism. It is reasonable to assume that a combination of both exist in this system; however, our experimental approach cannot distinguish between the two possible mechanisms.

Cis→Trans Isomerization of Proline Regulates Zn2+ Binding

Ultimately, the studies presented above (and prior NMR studies [11]) show that the proline residue in OT gives rise to a conformational heterogeneity that varies substantially in different solution environments. Thus, in some environments where the cis-isomer is favored, it will be the case that cis→trans isomerization of proline regulates Zn2+ binding, and thus directly affects receptor affinity. This new finding suggests that mutations (e.g., OT sequences in which the proline residue is replaced with a trans-favoring residue such as alanine) may influence receptor binding and, hence, physiological efficacy of OT in environments where the trans form is not favored for proline. Studies that examine related sequences would be interesting.

Conclusions

IMS-MS studies find that OT adopts multiple conformations in solution due to cis-trans isomerization of the Cys6–Pro7 bond. Upon binding Zn2+ to OT, the cis-isomer converts to the trans-isomer; this differences leads to unique binding affinities for the trans- and cis-configurations with Zn2+ of KA = 1.43 ± 0.24 μM−1 and 0.42 ± 0.12 μM−1, respectively. While previous studies have qualitatively demonstrated conformation specific binding of a ligand to a biomolecule [52, 53], we used IMS-MS techniques to resolve and quantify differences in binding affinities that are associated with different structures. The structural transition that occurs in this system suggests that the proline residue regulates Zn2+ binding to OT in environments where the cis-isomer of proline is favored in the Zn2+-free OT form. This finding suggests that a more efficacious form of OT (at least in terms of metal- and presumably receptor-binding interactions in cis-isomer favoring environments) might be created by removing the ability of the peptide to form the cis-isomer; e.g., one can imagine introducing a mutation at the 7th residue in which the proline residue is replaced by amino acids, which favor trans-isomer configurations. We note that while the OT sequence is highly conserved across many types of organisms [1], other biologically active peptides (e.g., vasopressin and vasotocin) incorporate similar ring structures with a proline residue that is penultimate to the ring that may show similar behavior. We are currently investigating these systems.