Research ResourcePharmacology

Subtle modifications to oxytocin produce ligands that retain potency and improved selectivity across species

See allHide authors and affiliations

Science Signaling  05 Dec 2017:
Vol. 10, Issue 508, eaan3398
DOI: 10.1126/scisignal.aan3398

A more selective oxytocin receptor agonist

Oxytocin is clinically used to induce labor, and there is interest in using this peptide to treat social disorders. However, oxytocin triggers adverse cardiovascular side effects because it activates the vasopressin receptor and the oxytocin receptor. Muttenthaler et al. generated ligands based on oxytocin with subtle modifications, yielding a lead compound that was more selective for the oxytocin receptor than for the vasopressin receptors. It reduced social fear in mice and induced contractile activity in human myometrial strips without affecting cultured cardiomyocytes. Given the cross-talk between oxytocin, vasopressin, and their receptors, this compound will also be helpful in identifying effects that are solely mediated by the oxytocin receptor.


Oxytocin and vasopressin mediate various physiological functions that are important for osmoregulation, reproduction, cardiovascular function, social behavior, memory, and learning through four G protein–coupled receptors that are also implicated in high-profile disorders. Targeting these receptors is challenging because of the difficulty in obtaining ligands that retain selectivity across rodents and humans for translational studies. We identified a selective and more stable oxytocin receptor (OTR) agonist by subtly modifying the pharmacophore framework of human oxytocin and vasopressin. [Se-Se]-oxytocin-OH displayed similar potency to oxytocin but improved selectivity for OTR, an effect that was retained in mice. Centrally infused [Se-Se]-oxytocin-OH potently reversed social fear in mice, confirming that this action was mediated by OTR and not by V1a or V1b vasopressin receptors. In addition, [Se-Se]-oxytocin-OH produced a more regular contraction pattern than did oxytocin in a preclinical labor induction and augmentation model using myometrial strips from cesarean sections. [Se-Se]-oxytocin-OH had no activity in human cardiomyocytes, indicating a potentially improved safety profile and therapeutic window compared to those of clinically used oxytocin. In conclusion, [Se-Se]-oxytocin-OH is a novel probe for validating OTR as a therapeutic target in various biological systems and is a promising new lead for therapeutic development. Our medicinal chemistry approach may also be applicable to other peptidergic signaling systems with similar selectivity issues.


Oxytocin and vasopressin [also known as arginine vasopressin (AVP)] are closely related neurohypophysial neuropeptides that are mainly synthesized in the magnocellular and parvocellular neurons of the hypothalamus, transported in association with neurophysins to the posterior pituitary, and released into the systemic circulation after enzymatic cleavage in response to relevant physiological stimuli (1, 2). In the periphery, oxytocin is involved in uterine smooth muscle contraction during parturition, milk ejection during lactation, ejaculation, and pain (35), whereas centrally released oxytocin functions as a neurotransmitter or neuromodulator that promotes multiple behaviors (68) such as maternal care (9, 10), partnership bonding (8, 11, 12), social interactions (12), and stress and anxiety responses (1315), as largely determined in rodents. Intranasal administration of oxytocin elicits broad behavioral effects in humans, and its therapeutic potential for psychopathologies characterized by social or emotional dysfunctions is under clinical investigation (14, 1621). Vasopressin increases and decreases fluid balance and blood pressure in the periphery (2224). Centrally, vasopressin is implicated in learning and memory (2527), in various social behaviors including pair bonding and aggression in rodents (6, 8, 28, 29), and in stress- and anxiety-related behaviors (2830). The ubiquitous involvement of the oxytocin and vasopressin signaling system in diverse physiological functions reflects its ancient origin dating back at least 600 million years (8, 31). The oxytocin receptor (OTR) and vasopressin receptor are members of the G protein–coupled receptor (GPCR) family (5, 32) and are attractive targets in the treatment of various high-profile disorders including cancer, pain, autism, schizophrenia, anxiety, and reproductive and cardiovascular disorders (5, 8, 16, 33).

In humans and rodents, oxytocin and vasopressin act through the oxytocin receptor (OTR) and the three AVP receptors (AVPRs; vasopressor V1aR, pituitary V1bR, and antidiuretic V2R). Oxytocin and vasopressin are structurally similar nonapeptides that differ only by two amino acids at positions 3 and 8 (Fig. 1A). Two cysteine residues in positions 1 and 6 form the cyclic part of the molecules followed by a three-residue amidated C-terminal tail. Their chemical similarity and the high sequence homology of the extracellular binding domains of OTR and AVPRs (~80%) lead to substantial cross-talk, with oxytocin able to activate the AVPRs and vasopressin the OTR (34, 35). Specific receptor functionality is thus not controlled by ligand selectivity but by cell-specific variations in receptor abundance, controlled release, receptor oligomerization, rapid clearance, and specific enzymatic degradation (36). High OTR and AVPR homology and overlapping distribution constitute a major hurdle in the development of selective receptor agonists, antagonists, and therapeutic candidates (37, 38). The identification of novel drug leads is further complicated by substantial species differences, such that rodent and human selectivity typically do not overlap, thereby restricting clinical translation (3739). For example, clinically used oxytocin and vasopressin analogs including desmopressin (40, 41), carbetocin (42), and atosiban (37, 43, 44) are receptor subtype–selective in rats but not in humans. This difference is mainly due to rapid biodegradation, renal clearance, and a limited administration window during which they can be used clinically without side effects (40, 44, 45). Despite these limitations, oxytocin remains the ligand of choice in the clinic to induce and progress labor (46). The lack of a complete set of selective receptor agonists and antagonists further limits our ability to characterize the physiological responses for each subtype receptor and their relevance in disease. We therefore commenced a program to overcome these limitations and to produce more selective ligands for this fundamental signaling system (31, 4751). Here, we demonstrated that small modifications to the structural framework of the pharmacophore of the endogenous and pharmacologically unselective neuropeptides could be used to tune selectivity and generate analogs with an improved selectivity profile that was conserved across mouse and human, thereby facilitating translational studies.

Fig. 1 Overview of the pharmacophore framework modifications to oxytocin and vasopressin.

(A) NMR structure of oxytocin (OT), with framework residues marked in red that were the focus of this study. The table provides an overview of all synthesized peptides with details on their modifications. U, selenocysteine; d, deamino (N terminus); Δ, deletion of residue; bold, modification; P, position; term, terminus; AVP, arginine vasopressin. (B) Synthesis of selenocysteine building blocks for Boc-SPPS to enable sulfur to selenium replacements. (C and D) Functional screen of oxytocin analogs 1 to 10 (C) and vasopressin analogs 11 to 16 (D) at the human OT receptor (hOTR), hV1aR, and hV1bR using the fluorescent imaging plate reader (FLIPR) Ca2+ signaling assay and at the hV2R by measuring cyclic adenosine monophosphate (cAMP) accumulation. Taller bars in graphs indicate loss of function at that particular receptor. The last row of beige bars illustrates how modifications affected potency compared to oxytocin at the hOTR and vasopressin at the hV2R (black). (E) Functional screen of oxytocin analogs 1 to 7, vasopressin, and dAVP at the hOTR, hV1aR, and hV1bR as assessed by measuring second-messenger inositol 1-phosphate (IP1) accumulation and at the hV2R as assessed by measuring second-messenger cAMP accumulation. CNS (central nervous system) and PNS (peripheral nervous system) indicate where these receptors can be found in humans. Exact EC50 values are shown in tables S2 and S3.


Rationale for subtle modifications

Traditional medicinal chemistry strategies that modify single or multiple functional groups of oxytocin and vasopressin commonly yield selectivity improvements that are not retained across species (37, 38). On the basis of structural data for oxytocin (4, 48, 5254) and earlier studies (37, 50, 5557), we knew that oxytocin and vasopressin binding and receptor activation are sensitive to even minor modifications. For example, ring reduction by one sulfur atom or disulfide bond replacement by a dicarba bridge (−CH2−CH2−) causes complete loss of activity (56). N-terminal deamination of oxytocin analogs often leads to improved OTR activation, which has led to the generation of the slightly more stable and hydrophobic oxytocin superagonist deamino-oxytocin (dOT) and the clinically used ligands carbetocin, desmopressin, and atosiban (37, 55, 57). Furthermore, changing Gly9 to Val9 in oxytocin and vasopressin switches these agonists to antagonists. In addition, [Gly9,Val]-oxytocin not only acts as an antagonist at the human V1aR (hV1aR) but also no longer binds to the hV1bR and the hV2R at concentrations up to 10 μM (50). For vasopressin, replacement of Gly9 by a carboxylic acid markedly decreases its vasopressor (V1aR) and antidiuretic (V2R) activity in rats (58). We also included replacement of the disulfide bond with a diselenide bond, which has been described as a structurally isosteric disulfide bond mimic that can increase potency and selectivity due to the slightly more hydrophobic character of the diselenide bond (56, 59). Moreover, the lower redox potential of the diselenide bond correlates to higher stability in a reducing environment and can therefore be exploited to improve peptide half-life (56, 5961). We therefore introduced subtle modifications to the pharmacophore framework of oxytocin and vasopressin (Fig. 1A and table S1) to identify those that yield selectivity gains that are retained across species.

Peptide synthesis

Two building blocks [Boc-l-Sec(Meb)-OH and dSec(Meb)-OH] were synthesized to generate the desired sulfur or selenium modifications (Fig. 1, A and B, and table S1). To acquire Boc-l-Sec(Meb)-OH in sufficient quantities, we optimized various protocols (6264) to obtain 28.7 g of pure Boc-l-Sec(Meb)-OH in only three steps (62% overall yield in a 4-day procedure; Fig. 1B). We also devised a new synthetic strategy for dSec(Meb)-OH, yielding the building block in a two-step synthesis with an overall yield of 12% (Fig. 1B). With these building blocks in hand, all of the oxytocin and vasopressin analogs 1 to 16 (Fig. 1A and table S1) were synthesized by Boc-SPPS (tert-butyloxycarbonyl solid-phase peptide synthesis), folded in 0.1 M NH4HCO3, and purified by reversed-phase high-performance liquid chromatography (RP-HPLC) (59, 65).

Functional response of novel oxytocin and vasopressin ligands

A fluorescent imaging plate reader (FLIPR) Ca2+ mobilization assay (for OTR, V1aR, and V1bR) and a second-messenger cyclic adenosine monophosphate (cAMP) assay (for V2R) showed that oxytocin activated all four human receptors. Half-maximal effective concentration (EC50) at the hV1aR and hV1bR was 10- to 300-fold higher than that at the hOTR (EC50: hOTR ≈ hV2R < hV1bR < hV1aR) (Fig. 1C and table S2). Deletion of the N-terminal amino group (dOT) increased the ability of the ligand to activate all of the AVPRs, suggesting an improved fit of dOT in the hydrophobic transmembrane binding pocket (47) across all four receptors. Replacement of the disulfide bond with the isosteric yet slightly more hydrophobic diselenide bond ([Se-Se]-OT and d[Se-Se]-OT) was well tolerated and did not lead to marked changes in potency. In particular, d[Se-Se]-OT showed the least change in potency and selectivity compared to oxytocin. Exchange of the C-terminal amide to acid in oxytocin (OT-OH) resulted in weaker potency at all receptors, and truncations of the C terminus of oxytocin (compounds 8, 9, and 10) reduced the potency at all receptors with complete loss of activity at the hV2R (up to 10 μM). Compounds 6, 7, 9, and 10 all showed substantial drops in interaction with AVPRs while retaining good potency at the hOTR, thereby indicating that the C terminus of oxytocin is a good target for improving selectivity for the hOTR against the three AVPRs.

In particular, the combination of diselenide replacement and C-terminal amide to acid change yielded [Se-Se]-OT-OH (compound 6) with an improved selectivity profile for OTR compared to oxytocin (Fig. 1C and table S2). [Se-Se]-OT-OH displayed low nanomolar potency (7.3 nM) and partial agonism (Emax = 52%) at the hOTR, no activation of the hV1aR (>10 μM), a 600-fold higher EC50 at the hV1bR, and a 15-fold higher EC50 at the hV2R (109 nM). N-terminal deamination of [Se-Se]-OT-OH yielded d[Se-Se]-OT-OH, which did not have improved potency at the OTR but at all three AVPRs, thus having a less interesting selectivity profile than [Se-Se]-OT-OH (Fig. 1C and table S2). d[Se-Se]-OT-OH was also a partial agonist at the hOTR (Emax = 59%) and a full agonist at the AVPRs.

Subtle modifications to vasopressin also modulated potency and selectivity without, however, yielding a clear lead compound (Fig. 1D and table S2). Vasopressin activated all four receptors, including the hOTR (EC50: hV2R > hV1bR > hV1aR > hOTR). Deamination of the N terminus [deamino-AVP (dAVP)] did not affect receptor activation, and the resulting ligand had potencies similar to that of vasopressin. The disulfide-to-diselenide exchange ([Se-Se]-AVP and d[Se-Se]-AVP) was well tolerated and did not substantially change the potency or selectivity profile. Deletion of Gly9 in dAVP and dAVP-OH resulted in substantial drops in potency at all four receptors.

Oxytocin analogs 1 to 7, vasopressin, and dAVP were further tested using the homogeneous time-resolved fluorescence (HTRF) inositol 1-phosphate (IP1) assay (Fig. 1E and table S3) to allow comparison with the downstream intracellular Ca2+ changes measured using the FLIPR assay. The HTRF-IP1 assay correlated well with the FLIPR assay (table S4), and the potency and selectivity trends, particularly with the lead compound [Se-Se]-OT-OH, were confirmed. [Se-Se]-OT-OH had an EC50 that was only 2.6-fold less potent than that of oxytocin and no activity at the hV1aR and hV1bR at concentrations up to 10 μM (table S3).

The Ca2+ concentration–response curves of [Se-Se]-OT-OH at the hOTR, hV1aR, and hV1bR (representative curves in Fig. 2A; raw FLIPR data in fig. S1, A and B) highlight its partial agonism at the hOTR, its receptor selectivity, and its weak antagonism at the hV1aR [17% inhibition; half-maximal inhibitory concentration (IC50), 132 nM] (Fig. 2A; fig. S1, C and D; and table S2). The representative cAMP concentration–response curves of [Se-Se]-OT-OH at the hV2R showed that [Se-Se]-OT-OH was a full agonist at this receptor but not as potent as oxytocin and vasopressin (Fig. 2B). The IP1 concentration–response curves of [Se-Se]-OT-OH at the hOTR, hV1aR, and hV1bR showed that the same selectivity, potency, and partial agonism trends were also observed using second-messenger IP1 measurements (Fig. 2C).

Fig. 2 Pharmacological characterization of [Se-Se]-OT-OH at all four human oxytocin and vasopressin receptors.

(A) Representative Ca2+ concentration–response curves of [Se-Se]-OT-OH at the hOTR, hV1aR, and hV1bR. (B) Representative cAMP concentration–response curves of oxytocin, vasopressin, and [Se-Se]-OT-OH at the hV2R. (C) Representative IP1 concentration–response curves of [Se-Se]-OT-OH at the hOTR, hV1aR, and hV1bR. (D) Representative radioligand concentration-displacement curves for [Se-Se]-OT-OH at the hOTR, hV1aR, hV1bR, and hV2R. All curves were normalized to percentage of response or displacement of the control ligand (oxytocin for OTR and vasopressin for AVPRs). Data in (A) to (D) are means ± SEM of results obtained from at least n = 3 separate experiments, each performed in triplicate.

Binding analysis at the hOTR, hV1aR, hV1bR, and hV2R

Binding data for the oxytocin analogs 1 to 7, vasopressin, and dAVP were obtained in radioligand displacement assays at all four human receptors (table S5). The experimental inhibition constants (Ki) correlated well with the functional data (tables S2 and S3). [Se-Se]-OT-OH displaced the 125I-V1a antagonist ([125I]phenylacetyl-d-Tyr(Me)-Phe-Gln-Asn-Arg-Pro-Arg-Tyr-NH2) at the hV1aR with a Ki of 57 nM, thereby confirming the weak antagonistic effects observed in the FLIPR assay (Fig. 2D and table S5).

Schild regression analysis for binding mode

To confirm competitive binding of [Se-Se]-OT-OH at the hOTR, we performed a series of Schild regression experiments with oxytocin and [Se-Se]-OT-OH against the hOTR/Gq antagonist atosiban using the HTRF-IP1 assay. Increasing concentrations of atosiban shifted the concentration-response curves of oxytocin (fig. S2A) and [Se-Se]-OT-OH (fig. S2B) to the right, resulting in a linear Schild plot that is representative of competitive binding at the hOTR.

Binding and functional analysis of [Se-Se]-OT-OH and d[Se-Se]-OT-OH at murine OTR, V1aR, V1bR, and V2R

[Se-Se]-OT-OH retained its functional selectivity for OTR in mice and was inactive at murine V1aR (mV1aR) and mV1bR concentrations up to 10 μM (Fig. 3A, fig. S3A, and table S3). [Se-Se]-OT-OH activated mOTR with a potency similar to that of oxytocin (table S3) (66). [Se-Se]-OT-OH activated mV2R with an EC50 that was ~300-fold larger than that of vasopressin (67). d[Se-Se]-OT-OH also retained its selectivity profile in mice and was again less selective than [Se-Se]-OT-OH because it activated all four murine receptor subtypes (fig. S3B and table S3). Both compounds were partial agonists at the mOTR and full agonists at the mV2R. The binding data of both compounds (table S5) correlated well with the functional data (table S3), including low-affinity binding of [Se-Se]-OT-OH at the mV1bR fitting with the absence of activation at the mV1bR and binding to the mV1aR confirming mV1aR antagonism. [Se-Se]-OT-OH displayed biphasic binding to mV1aR and mV2R (fig. S4A and table S5), and d[Se-Se]-OT-OH displayed biphasic binding to mV2R (fig. S4B and table S5).

Fig. 3 Comparison of selectivity, stability, and ability to augment myometrial strip contractions between [Se-Se]-OT-OH and oxytocin.

(A) Functional selectivity profile of [Se-Se]-OT-OH and OT over all four human (h) and murine (m) oxytocin and vasopressin receptors. OTR, V1aR, and V1bR activity was measured by IP1 accumulation. V2R activity was measured by cAMP accumulation. (B) Metabolic stability of [Se-Se]-OT-OH (t1/2 = 25 hours) compared to oxytocin (t1/2 = 12 hours) in human serum. Data in (A) and (B) are means ± SEM of results obtained from at least n = 3 separate experiments, each performed in triplicate. (C and D) Representative contraction pattern for human myometrial strips exposed to increasing doses of oxytocin (C) or [Se-Se]-OT-OH (D). n = 5 women for [Se-Se]-OT-OH, and n = 8 women for oxytocin.

Human serum stability assay

[Se-Se]-OT-OH had a ~2-fold longer half-life of 25 hours compared to oxytocin, which had a half-life of 12 hours (Fig. 3B).

Human myometrial cell and strip contractility assays as proxies for induction and augmentation of labor

To determine whether the partial agonism of [Se-Se]-OT-OH and d[Se-Se]-OT-OH in the pharmacological studies would affect their physiological activity compared to the full agonist oxytocin, we tested the ability of compounds 1, 6, and 7 to induce contraction by activating the hOTR in myometrial cells immortalized with the human telomerase reverse transcriptase (hTERT-HM) (68) in a collagen gel contractility assay (69, 70). Both [Se-Se]-OT-OH and d[Se-Se]-OT-OH induced a contractile response comparable to oxytocin (fig. S5). [Se-Se]-OT-OH and d[Se-Se]-OT-OH increased contractility by 7.4 and 9% respectively, compared to 9% for oxytocin. We then confirmed these findings in contractility studies that used strips of human myometrium from cesarean sections. [Se-Se]-OT-OH and oxytocin increased contraction amplitude in a concentration-dependent manner (Fig. 3, C and D, and fig. S6) and had similar potencies (fig. S6). The contraction profile induced by [Se-Se]-OT-OH was, however, more phasic and frequent compared to a more tonic-like activity induced by oxytocin at the concentrations used (Fig. 3, C and D).

Behavioral efficacy in a mouse model of social fear

[Se-Se]-OT-OH was tested in the social fear conditioning (SFC) paradigm (71), which specifically generates social fear (as measured by reduced social investigation 24 hours after fear conditioning) in mice without any confounding behavioral alterations. [Se-Se]-OT-OH reduced social fear and facilitated social fear extinction 10 min after its intracerebroventricular infusion compared with vehicle-treated social fear–conditioned mice (Fig. 4A). This effect was similar to that of synthetic oxytocin and [Thr4,Gly7]-OT, a highly selective mOTR agonist (66) that also potently reduced social fear (Fig. 4A). All unconditioned control mice treated with either vehicle, [Se-Se]-OT-OH, oxytocin, or [Thr4,Gly7]-OT showed similar extents of social preference behavior as reflected by active exploration of conspecifics. Independently of subsequent treatment, all social fear–conditioned and unconditioned groups showed similar investigation of a nonsocial stimulus (small empty cage), indicating similar amounts of nonsocial anxiety. During the social fear recall (Fig. 4B), all groups of mice, irrespective of their treatment, showed high social investigation, which indicated reduced social fear and successful social fear extinction.

Fig. 4 Social fear conditioning (SFC) mouse study.

(A and B) Unconditioned (SFC) and conditioned (SFC+) mice were intracerebroventricularly infused with either vehicle (2 μl of Ringer’s solution; n = 9 SFC mice; n = 10 SFC+ mice) or [Se-Se]-OT-OH (250 μM/2 μl; 500 pmol; n = 7 SFC mice; n = 7 SFC+ mice). Oxytocin (250 μM/2 μl; 500 pmol; n = 10 mice) and [Thr4,Gly7]-OT (250 μM/2 μl; 500 pmol; n = 4 mice) were only infused into SFC+ mice 10 min before extinction training. Percentage of mice that investigated three nonsocial stimuli (empty cage) and six social stimuli (cage with a conspecific) during social fear extinction (day 2; A) and six social stimuli during social fear extinction recall (day 3; B) is shown. Data are means ± SEM and were analyzed using two-way analysis of variance (ANOVA) for repeated measures. (A) #P < 0.05 SFC+/vehicle compared to SFC+/[Se-Se]-OT-OH; *P < 0.05 SFC+/vehicle compared to SFC+/OT, SFC+/[Thr4,Gly7]-OT, and SFC/vehicle; (B) $P < 0.05 SFC+/vehicle compared to SFC+/OT and SFC/vehicle.

Maximal Ca2+ concentrations in human cardiomyocytes as a proxy for assessing cardiovascular safety

V1aR is highly abundant in the vascular smooth muscle and heart (cardiomyocytes), and V1aR activation has been linked to cardiovascular risks (7276). As a proxy to assess cardiovascular risks, we measured maximal intracellular Ca2+ concentrations in human cardiomyocytes treated with the various ligands (Fig. 5). Vasopressin and oxytocin concentration-dependently caused an increase in maximal intracellular Ca2+ in human cardiomyocytes with EC50 values of 121 and 174 nM, respectively, which aligned well with our pharmacological results of hV1aR activation by oxytocin (tables S2 and S3). Consistent with its improved selectivity (OTR activation without V1aR activation), [Se-Se]-OT-OH did not affect the maximal intracellular Ca2+ concentration (up to 10 μM) in cardiomyocytes (Fig. 5).

Fig. 5 Effects of vasopressin, oxytocin, and [Se-Se]-OT-OH on maximal intracellular Ca2+ in human cardiomyocytes.

FLIPR assays showing the effect of vasopressin, oxytocin, and [Se-Se]-OT-OH on the maximal intracellular Ca2+ concentrations. Data are means ± SEM, n = 4 wells per data point.


Agonists and antagonists with good selectivity for the four oxytocin and vasopressin receptor subtypes are in short supply (37, 38, 77). Physiological function and dysfunction of the receptors are therefore studied by time- and cost-intensive knockout models, which have linked the subtypes to various disorders including autism, schizophrenia, epilepsy, stress, aggression, depression, anxiety, and pain (5, 8, 16, 33). Therapeutic development targeting these disorders has so far failed because of the challenge of developing subtype-selective ligands that retain their selectivity across species, thereby limiting translational studies with clinically relevant animal models (37, 38, 78).

Subtle pharmacophore framework modifications—A new strategy to retain selectivity across murine and human receptors

This structure-activity relationship study demonstrated that subtle modifications to the pharmacophore framework of the endogenous ligands oxytocin and vasopressin could tune potency and selectivity and yield ligands ([Se-Se]-OT-OH and d[Se-Se]-OT-OH) with an improved selectivity profile that was conserved across mouse and human (Fig. 3A and table S3). In particular, the replacement of the disulfide bond with the diselenide bond in combination with the C-terminal amide to acid change in oxytocin yielded [Se-Se]-OT-OH, the compound with the most pronounced selectivity gain for the OTR (Fig. 3A). Although an amide-to-acid modification can generally be considered a subtle modification, it seems that either the free lone pair of the nitrogen atom is involved in receptor recognition and activation or the negative charge of the acid contributes to activity loss at the AVPRs. Substitution of the disulfide bond by the diselenide bond slightly increases the bond length (+0.3 Å), torsion angle (Δ +11°), and hydrophobicity (59, 60). In particular, the change in torsion angle could play a part in the selectivity differences observed by pushing the ligand into a left-handed conformer, as observed in the crystal structure of dOT (54). Both modifications (diselenide bond and C-terminal acid) had to be present to improve selectivity.

The increased selectivity of [Se-Se]-OT-OH makes it an excellent probe to delineate OTR function in the central nervous system (CNS) in behavioral animal models (where V2R is not present) and a promising lead molecule for the clinic, which we further explored in this study. We recommend that an effective concentration of 50 nM or a 10-fold higher concentration than oxytocin (in comparative studies with oxytocin) is used to ensure OTR-mediated action while retaining OTR preference. The weak antagonism (17% inhibition of vasopressin at the hV1aR) does not affect the investigation of OTR function but needs to be taken into account either when binding studies are carried out ([Se-Se]-OT-OH binds to the hV1aR with a Ki of 56.8 nM) or when [Se-Se]-OT-OH is administered in the presence of V1aR ligands. [Se-Se]-OT-OH is a partial agonist for OTR; however, this did not affect biological function as suggested by our myometrial contractility study, which showed that OTR activation resulted in a full response similar to that induced by oxytocin (Fig. 3, C and D, and fig. S5). Partial agonists can be beneficial in treating chronic disorders such as pain and autism because they often do not trigger the development of adverse effects such as overstimulation, desensitization, adaptation, tolerance, and dependence as seen with full agonists.

General conclusions and guidelines for future oxytocin and vasopressin ligand design

N-terminal deamination of oxytocin ligands improves binding and potency at all four receptors, renders oxytocin slightly more hydrophobic, and improves its stability against proteases (55). Although N-terminal deamination of vasopressin did not affect potency and selectivity, we still recommend deamination because it offers enhanced proteolytic stability. Considering the advancements in the area of disulfide mimetics, we recommend the replacement of the disulfide bond by a nonreducible and therefore metabolically more stable thio- or selenoether bond (4, 56). Modification of the C terminus of oxytocin and vasopressin could yield novel antagonists, considering this study and the replacement of Gly9 in oxytocin and vasopressin by Val, which results in an agonist-to-antagonist switch at the hV1aR for both ligands (50). The FLIPR is a valuable instrument for primary screening of compounds that target oxytocin and vasopressin receptors because it rapidly provides agonist and antagonist information with high-throughput capabilities. Follow-up characterization of selected leads by ligand binding, second-messenger quantification, and arrestin recruitment can then be used to better understand ligand kinetics and secondary messenger signaling pathways. The structure-activity relationship study presented here assessed affinity and functional data (FLIPR and IP1) of the most commonly used control compounds oxytocin, vasopressin, dOT, and dAVP over all four receptor subtypes in a single study and could be a reference for future oxytocin and vasopressin ligand development.

Comparison of the FLIPR and IP1 assays

Only three discrepancies with more than 50-fold difference were detected between the FLIPR and the IP1 assays (oxytocin and dOT at the hV1bR and d[Se-Se]-OT at the hV1aR; table S4), suggesting signaling bias for these ligands at these receptors. The characteristic Ca2+ transients elicited by activation of Gαq-coupled GPCRs such as OTR, V1aR, and V1bR originate from the production of IP3 downstream of phospholipase C activation, which induces the release of Ca2+ from intracellular stores through IP3 receptors, termination of receptor signaling, and subsequent uptake of increased intracellular Ca2+ into stores and the extracellular compartment through Ca2+ adenosine triphosphatases (79). In addition, Ca2+ signals can be magnified or inhibited through intracellular feedback mechanisms such as Ca2+-induced Ca2+ release, activation of downstream effectors that modulate intracellular Ca2+ amounts, or modification of Ca2+ signals through alternate G proteins such as Gαi, which is involved in OTR and vasopressin receptor signaling (5, 80). Thus, whereas IP3 production is a requisite step for initiation of a Ca2+ signal, other factors contribute to the magnitude and kinetics of the observed Ca2+ transients elicited by activation of GPCRs. In contrast, the IP1 assay indirectly quantifies the production of IP3 through accumulation of its breakdown product IP1. It is plausible that the observed differences arise because of mechanistic differences that contribute to increases in intracellular Ca2+ and accumulation of IP1. The cellular consequences of such differences remain unclear to date, although Ca2+ signaling is a physiologically relevant consequence of GPCR activation. The difference in signaling produced by these ligands indicates a potential signaling bias, a phenomenon that has been previously studied for this receptor class (43, 8082).

[Se-Se]-OT-OH modulating central actions: Reversal of social fear

Oxytocin is linked to multiple CNS disorders such as autism, bipolar disorders, schizophrenia, pain, anxiety, and depression (5, 8, 16, 33). Studies showing that intranasal oxytocin administration can modulate social behavior in humans have resulted in increased interest and clinical trials exploring the therapeutic potential of intranasal oxytocin for these disorders. The exact target pharmacology and underlying mechanisms of action are, however, still under debate because oxytocin can also signal through V1aR and V1bR (tables S2 to S5). Hence, receptor subtype–specific probes are critical because of the partly opposing behavioral effects of oxytocin and vasopressin, especially in the context of anxiety and stress regulation (14). [Se-Se]-OT-OH allows for a clear and simple distinction between the involvement of OTR compared to V1aR and V1bR in the CNS without the use of time- and cost-intensive knockout models or coapplication of species-selective V1a and V1bR antagonists. This is particularly valuable for therapeutic target validation and future drug development. Here, we chose to behaviorally test our novel probe in the SFC mouse model, which has clinical implication to treat social anxiety disorders (15, 71, 83), and to compare it to the commonly used mOTR-selective agonist [Thr4,Gly7]-OT, which does not retain its selectivity profile for human receptors. The use of [Se-Se]-OT-OH confirmed that this action is mediated through OTR and not through V1aR or V1bR, and [Se-Se]-OT-OH displayed similar effects to [Thr4,Gly7]-OT (Fig. 4, A and B).

[Se-Se]-OT-OH modulating peripheral action: Labor induction and augmentation

Selectivity against V1aR to reduce cardiovascular side effects is of particular importance for peripheral applications because V1aR is the primary receptor subtype in the vascular smooth muscle and heart (7276). Clinically, oxytocin is administered intravenously to induce labor, augment weak contractions in labor, and treat postpartum hemorrhage and intranasally to elicit lactation. Because of its activation of AVPRs, oxytocin administration has only a limited therapeutic window of use, and adverse effects reported in the mother include anaphylactic reaction, postpartum hemorrhage, cardiac arrhythmia, fatal afibrinogenemia, pelvic hematoma, subarachnoid hemorrhage, hypertensive episodes, rupture of the uterus, convulsions, coma, and death due to water intoxication. In the neonate, adverse effects reported include bradycardia, cardiac arrhythmia, jaundice, retinal hemorrhage, permanent CNS or brain damage, seizures, and death (46, 84, 85). Hence, the lack of AVPR activation in OTR agonists is needed to diminish side effects.

[Se-Se]-OT-OH may be such a candidate because it only minimally raised intracellular Ca2+ concentrations in human cardiomyocytes, in contrast to oxytocin, which displayed nanomolar activity (Fig. 5). V1aR is highly abundant in human cardiomyocytes where it mediates many of its cardiovascular side effects. Furthermore, [Se-Se]-OT-OH and oxytocin were equipotent in augmenting contractility in human myometrial preparations, although [Se-Se]-OT-OH displayed a more regular contraction pattern (Fig. 3D). By contrast, oxytocin reduced contraction frequency in a concentration-dependent fashion but increased the contraction duration, resulting in much greater total contractile activity (Fig. 3C), which can lead to uterine hyperstimulation and rupture, which are life-threatening events for mother and fetus (86, 87). [Se-Se]-OT-OH’s inactivity at V1aR, a receptor that is also abundant in the myometrium during pregnancy and is involved in contraction (88, 89), in combination with its partial agonism at the OTR, resulted in an improved contractility profile, suggesting that it would be a more controlled and safer alternative to oxytocin to induce and augment human labor. Thus, [Se-Se]-OT-OH not only reduced AVPR-related side effects but also offered additional advantages to aid labor induction and progression through a more regular increase of contraction strength without the risk of uterine hyperstimulation or rupture.

Human serum stability of [Se-Se]-OT-OH

Enzymatic stability is a key characteristic for many peptide drug candidates, and hence, it was important to characterize the stability of [Se-Se]-OT-OH compared to oxytocin. Diselenide bonds exhibit a lower redox potential than disulfide bonds, rendering such mimetics more difficult to reduce, a feature that can be exploited in improving the half-life of peptide drug candidates (4, 56, 5961). On the other hand, peptides with C-terminal amides are more stable than peptides with C-terminal acids. The improved stability of [Se-Se]-OT-OH in human serum (Fig. 3B) suggests that the impact of the more redox-stable diselenide bond outweighs the disadvantage of the C-terminal acid modification. However, the experimentally determined in vitro human serum half-life (which was in hours) of oxytocin and [Se-Se]-OT-OH does not reflect physiologically relevant in vivo half-life (which is in minutes) because it does not take renal clearance into account. Nonetheless, it can assess even small stability improvements, which could be important for treating CNS disorders through delivery methods [such as treatment of autism in children or treatment of migraines in adults (21)] that are not affected by renal clearance.

In conclusion, we showed that subtle modifications to the pharmacophore framework of oxytocin can lead to enhanced selectivity profiles that are conserved across species (mouse and human), thereby overcoming a substantial hurdle in ligand development. This proof-of-concept study provides a framework that could be applied to similar complex peptidergic signaling systems. The lead ligand derived from this study, [Se-Se]-OT-OH, is a valuable probe that can effectively delineate the physiological responses of OTR and is particularly suited for use in animal behavior studies looking at the central roles of OTR. Its functional selectivity and close structural similarity to oxytocin are important features that will help to clarify the contributions of the oxytocin signaling system in health and disease. The improved selectivity, stability, and safety profile of [Se-Se]-OT-OH could lead to superior alternatives to oxytocin in the clinic for peripheral indications such as labor induction and progression and postpartum hemorrhage, as well as for central indications such as autism, migraine, schizophrenia, anxiety, and stress, where clinical trials have shown that intranasally administered oxytocin is safe and therapeutically effective (20, 21, 9093). [Se-Se]-OT-OH has the potential to advance our understanding of the OTR as a therapeutic target in different animal models and represents a promising lead molecule for various high-profile OTR-related indications.


Selenocysteine building block synthesis

An optimized protocol with yield-improving modifications was developed on the basis of Chocat et al. (62), Tanaka et al. (63), and Oikawa et al. (64), enabling the synthesis of the Boc-l-Sec(Meb)-OH at a scale of 20 g with an overall yield of 62%. A new two-step strategy was devised for the synthesis of dSec(Meb)-OH with an overall yield of 12% (Fig. 1B).

Peptide synthesis

Peptides 1 to 16 were assembled manually by Boc-SPPS using the HBTU (2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate)–mediated in situ neutralization protocol (59, 65). The analogs were cleaved with hydrogen fluoride (94), purified by RP-HPLC, folded over 24 hours in 0.1 M NH4HCO3 buffer at pH 8.2 (1 mg/10 ml; ~100 μM), and purified by RP-HPLC to >95% purity. Peptide concentrations were determined on the basis of peak area detected at 214 nm by analytical RP-HPLC against oxytocin and vasopressin as standards with known peptide content established by amino acid analysis. Using the Beer-Lambert law, the peptide concentrations were calculated on the basis of absorptions of standards and samples using calculated extinction coefficients (9597).

Transfection and membrane preparation for FLIPR, HTRF-IP1, and radioligand displacement assays for the human receptors

hOTR, hV1aR, hV1bR, and hV2R complementary DNAs (cDNAs) were obtained from OriGene Technologies. [Tyrosyl-2,6-3H]-oxytocin (3H-OT) (46.3 Ci/mmol), [phenylalanyl-3,4,5-3H(N)]-AVP (3H-AVP) (2200 Ci/mmol), 125I-linear vasopressin hV1aR antagonist (2200 Ci/mmol), FlashBlue GPCR Scintillating Beads, TopSeal-A 96-well sealing film, and 384-well white OptiPlates were from PerkinElmer Life Sciences. The HTRF-IP1 assay kit was from Cisbio International.

COS-1 cells grown in Dulbecco’s modified Eagle’s medium (DMEM) and 5% fetal bovine serum (FBS) in 150-mm plates were transiently transfected with plasmid DNA (14.5 μg) encoding the hOTR, hV1aR, hV1bR, or hV2R using Lipofectamine 2000 (29 μl; Invitrogen). The cells were harvested 48 hours after transfection and homogenized using an Ultra-Turrax homogenizer (22,000 rpm) in assay buffer [50 mM tris-HCl, 10 mM MgCl2 (5 mM MgCl2 for oxytocin), and 0.1% bovine serum albumin (BSA) (pH 7.4)] with cOmplete protease inhibitor cocktail (Roche Diagnostics). The homogenate was centrifuged at 484g (2000 rpm) for 10 min, and the resulting supernatant was centrifuged at 23,665g (14,000 rpm) for 30 min. The pellet was resuspended in appropriate buffer without protease inhibitor containing 10% glycerol and stored at −80°C until assayed.

FLIPR assay for the hOTR, hV1aR, and hV1bR

COS-1 cells transfected with the hOTR, hV1aR, or hV1bR were plated 24 hours before the experiment at a density of 35,000 to 50,000 cells per well on black-walled 96-well imaging plates (Corning). Cells were loaded for 30 min at 37°C with Fluo-4-AM [4-(6-acetoxymethoxy-2,7-difluoro-3-oxo-9-xanthenyl)-4′-methyl-2,2′-(ethylenedioxy)dianiline-N,N,N′,N′-tetraacetic acid tetrakis(acetoxymethyl) ester] (4.8 μM) in physiological salt solution (PSS; 140 mM NaCl, 11.5 mM glucose, 5.9 mM KCl, 1.4 mM MgCl2, 1.2 mM NaH2PO4, 5 mM NaHCO3, 1.8 mM CaCl2, and 10 mM Hepes) containing 0.3% fatty acid–free BSA. To allow for complete dye de-esterification, cells were washed with PSS for 5 to 10 min before transferring them to a FLIPRTETRA (Molecular Devices) fluorescent plate reader. Ca2+ responses were measured using a cooled charge-coupled device (CCD) camera with excitation at 470 to 495 nM and emission at 515 to 575 nM. The baseline fluorescence was set to a minimum of 1000 arbitrary fluorescence units by adjusting camera gain and excitation intensity. Compounds were added as 3× concentrated stock solutions in PSS, with 10 baseline fluorescence readings before compound addition followed by fluorescence reading every second for 180 s.

Raw fluorescence data were converted to ΔF/F values by subtracting baseline fluorescence readings from subsequent time points and dividing the difference by baseline fluorescence values, as previously described (98). For the concentration-response curves, maximum ΔF/F values after the addition of compounds were plotted against agonist concentration and normalized to the response elicited by the native ligand (oxytocin for the hOTR and vasopressin for the hV1aR and hV1bR). A four-parameter Hill equation with a Hill coefficient of 1 was fitted to the data using GraphPad Prism (version 4.00).

cAMP assay for hV2R

The cDNA plasmid clones for hV2R were a gift from R. Schülein (FMP, Berlin). The hV2R sequence was inserted into pKaede-MN1 (MBL Life Science) using Eco RI and Hind III restriction sites to yield the wild-type receptor. The conditions for the propagation of human embryonic kidney (HEK) 293 cells and the creation of stably transfected cell lines were similar to those described previously (48, 99). Briefly, cells were transfected with CaPO4 transfection, and 1.8 × 106 of HEK293 cells per 10-cm dish were prepared. Twenty microliters of DNA, with a concentration of 1 μg/μl, was mixed with 480 μl of H2O + CaCl2 (430 μl of H2O + 50 μl of CaCl2) and added to 500 μl of Hepes-buffered saline solution. After 6 min, the solution was added to the DMEM high-glucose medium (PAA Laboratories), supplemented with l-glutamine and gentamicin, and the cells were incubated for 4 hours at 37°C. After a glycerol shock, DMEM was added again and the cells were put back at 37°C. In the following days, selection through the antibiotic Geneticin [geneticin G418-BC liquid (50 mg/ml; Biochrom)] took place.

Cells were grown in six-well plates. The adenine nucleotide pool was metabolically labeled by incubating confluent monolayers with 3H-adenine (1 μCi per well) ([2,8-3H]-adenine; 27.2 Ci/mmol, PerkinElmer Life Sciences) for 16 hours as described previously (100). After preincubation, fresh medium that contained 100 μM RO201724 (a cell-permeable, selective inhibitor of cAMP-specific phosphodiesterase; Calbiochem) was added. After 4 hours, cAMP formation was stimulated by the hV2R agonist vasopressin (50 pM to 1 μM), oxytocin (60 pM to 1 μM), synthetic oxytocin analogs (700 pM to 10 μM), or forskolin (30 μM; a cell-permeable diterpenoid that has antihypertensive, positive inotropic, and adenylyl cyclase–activating properties; Sigma-Aldrich) for 20 min at 37°C. Assays were performed in at least three separate experiments in triplicate. The formation of 3H-cAMP was determined according to Bergmayr et al. (100). Potency (EC50) and efficacy (Emax) were calculated by fitting the data to a three-parameter logistic equation (Hill equation) using a Levenberg-Marquardt algorithm. EC50 values were presented as means ± SEM.

HTRF-IP1 assay for the human receptors

COS-1 cells were transiently transfected with plasmid DNA encoding the hOTR, hV1aR, or hV1bR using a Lipofectamine 2000/DNA ratio of 2 in DMEM. Assays measuring IP1 accumulation were performed 48 hours after transfection according to the manufacturer’s protocol. Briefly, cells were incubated with increasing concentrations of oxytocin and vasopressin analogs (10 pM to 10 μM) in stimulation buffer containing LiCl for 1 hour in 37°C and 5% CO2 in white 384-well OptiPlates. Cells were lysed by the addition of the HTRF reagents, the europium cryptate–labeled anti-IP1 antibody, and the d2-labeled IP1 analog and diluted in lysis buffer. The assays were incubated for 1 hour at room temperature. The emission signals at 590 and 665 nm were measured after excitation at 340 nm using the EnVision multilabel plate reader (PerkinElmer Life Sciences). Signal was presented as the HTRF ratio: F = [(F665nm/F590nm) × 104]. For concentration-response curves, the HTRF ratio values after the addition of compounds were plotted against the ligand concentration and normalized to the response elicited without ligand (negative control). Sigmoidal curves for the calculation of the EC50 values were fitted to individual data points by nonlinear regression with a Hill coefficient of 1, using the software package Prism (GraphPad Software).

Radioligand displacement assay for the human receptors

Competitive binding assays were performed using FlashBlue GPCR Scintillating Beads (PerkinElmer Life Sciences) as previously described (56). Reactions containing increasing concentrations of competing oxytocin and vasopressin analogs (10 pM to 10 μM); FlashBlue GPCR Scintillating Beads (100 μg for the hOTR and hV1aR; 200 μg for the hV1bR and hV2R); hOTR, hV1aR, hV1bR, and hV2R membrane preparations (5 μg of protein); and radioligand [3H-OT (2 nM) for the hOTR, 125I-V1a antagonist ([125I]phenylacetyl-d-Tyr(Me)-Phe-Gln-Asn-Arg-Pro-Arg-Tyr-NH2) (0.5 nM) for the hV1aR, and 3H-AVP (3 nM) for the hV1bR and hV2R] in assay buffer [50 mM tris-HCl, 10 mM MgCl2 (5 mM MgCl2 for oxytocin), and 0.1% BSA (pH 7.4)] were established in 96-well white polystyrene plates with clear flat bottoms in a total reaction volume of 80 μl. Radioligand binding was detected using a Wallac 1450 MicroBeta scintillation counter (PerkinElmer Life Sciences). Sigmoidal curves for the calculation of the IC50 values were fitted to individual data points by nonlinear regression with a Hill coefficient of −1, using the software package Prism (GraphPad Software).

Radioligand displacement assay for the murine receptors

Competitive binding assays were performed using 3H-OT and 3H-AVP (PerkinElmer Life Sciences) and increasing concentrations of the indicated peptides at 30°C on membranes prepared from HEK293 cells transfected with the mOTR, mV1aR, mV1bR, or mV2R as described previously (66). Compound affinities (Ki) were determined by means of competition experiments in which the unlabeled compound concentrations varied from 10 pM to 10 μM to displace 4 nM 3H-OT for the mOTR and 3H-AVP for the mV1aR, mV1bR, and mV2R. Ligand binding data and Ki were analyzed by means of nonlinear regression and binding-competitive fitting using Prism version 5 (GraphPad Software).

HTRF-IP1 assay and cAMP determination for the murine receptors

IP1 and cAMP accumulation were determined in HEK293 cells transiently transfected with DNA plasmids encoding the mOTR, mV1aR, mV1bR, or mV2R using HTRF assays (IP1 and cAMP assays, Cisbio International) as previously described (66). Sigmoidal curves for the calculation of the EC50 values were fitted by nonlinear regression log(agonist) versus response fitting, using Prism version 5 (GraphPad Software).

Stability assays

Peptides were added to human serum, and aliquots were taken and analyzed by RP-HPLC and liquid chromatography–mass spectrometry at 1-, 2-, 3-, 4-, 12-, 24-, and 48-hour time points as previously described (56).

SFC mouse model

SFC was performed as previously described (71). Sample sizes (>4 mice) were determined on the basis of previous studies (71, 83, 101). All experimental procedures were performed in accordance with the Guide for the Care and Use of Laboratory Animals of the Government of Oberpfalz and the guidelines of the National Institutes of Health. All male CD1 mice (8 to 10 weeks of age at the start of experiments; Charles River Laboratories) were group-housed under standard laboratory conditions (12-hour light/12-hour dark cycle, lights on at 06:00, 22°C, 60% humidity, and food and water ad libitum) in polycarbonate cages (16 cm × 22 cm × 14 cm). All experimental procedures were performed between 08:00 and 15:00. Surgeries were performed under isoflurane anesthesia, and extreme care was taken to minimize animal suffering. Analysis of the SFC paradigm was performed using GraphPad Prism version 6.0 (GraphPad Software).

On day 1 of the SFC paradigm, mice were transferred from their home cage into the conditioning chamber; after a 30-s adaptation period, an empty cage was placed in the conditioning chamber as a nonsocial stimulus that mice were allowed to investigate for 3 min before it was replaced by an identical cage containing an unfamiliar male conspecific (which represented the social stimulus). SFC mice were allowed to investigate the social stimulus for 3 min without receiving any foot shocks, whereas SFC+ mice were given a 1-s electric foot shock (0.7 mA) each time they investigated (sniffed) the social stimulus. Mice were returned to their home cage when no further social contact was made for 2 min. All mice investigated the nonsocial (empty cage) stimulus to a similar extent and received a similar number of foot shocks (table S6). On day 2, which consisted of social fear extinction training, mice were exposed to three nonsocial stimuli (empty cages) in their home cage to assess nonsocial investigation as a parameter of nonsocial fear and general anxiety-related behavior. Mice were then exposed to six unfamiliar social stimuli (six different male mice) to assess social investigation as a parameter of social fear. Mice received single acute intracerebroventricular infusions (35 to 45 s) of either vehicle (2 μl of sterile Ringer’s solution), oxytocin (2 μl of 250 μM; 500 pmol), [Se-Se]-OT-OH (2 μl of 250 μM; 500 pmol), or [Thr4,Gly7]-OT (2 μl of 250 μM; 500 pmol) 10 min before social fear extinction. On day 3, which assessed social fear extinction recall, mice were exposed in their home cage to six different unfamiliar social stimuli to investigate whether repeated exposure to social stimuli during extinction leads to a complete reversal of social fear.

Human myometrial cell contractility assay

Human uterine myometrial smooth muscle cells (hTERT-HM) were cultured, collagen gels were prepared, and the assays were performed as described previously (48, 70). Briefly, cells were cultured in DMEM/F-12/10% FBS (Invitrogen). Collagen gels were prepared from rat tail type 1 collagen (Sigma-Aldrich) to a final concentration of 1.5 mg/ml and seeded in 24-well culture dishes, with 150,000 hTERT-HM cells per well. Cells in collagen gels were allowed to equilibrate overnight in serum-free DMEM. The ligands of interest (1 μM concentration) were added to the serum-free medium, and the gels were released from the sides of the wells. FBS [10% (v/v)] was used as a positive control for contraction, and unstimulated cells were used as the negative control in all experiments. Gel images were captured over time using a FluorChem 8900 imager, and the area (in square centimeters) of the gels was measured using AlphaEaseFC software (Alpha Innotech Corporation). A decrease in gel area correlated with an increase in contractility.

Tissue bath myometrial contractility assays

Biopsies of myometrium were obtained from women undergoing term (39 to 40 weeks) prelabor elective cesarean section delivery at Liverpool Women’s Hospital, Liverpool, UK. All women gave written informed consent to participate, and the study was approved by the North West (Liverpool East) Research Ethics Committee (reference #10/H1002/49) and by the Research and Development director at Liverpool Women’s Hospital NHS Foundation Trust, Liverpool, UK. A full-thickness biopsy (~2 cm3) was cut from the upper lip of the lower uterine incision site and placed into Hanks’ balanced salt solution at 4°C (102). In the laboratory, strips of myometrium (1 mm × 2 mm × 5 mm) were dissected and placed between a force transducer and a fixed hook using aluminum clips. Strips were continually superfused with PSS [154 mM NaCl, 5.6 mM KCl, 1.2 mM MgSO4, 7.8 mM glucose, 10.9 mM Hepes, and 2.0 mM CaCl2 (pH 7.4)] maintained at 36°C. After stable spontaneous contractions developed, the strips were exposed to rising concentrations of oxytocin or [Se-Se]-OT-OH. Data were analyzed by measuring the amplitude of contraction using Origin Pro 9.0 software (OriginLab Corporation) as described previously (48, 103, 104). The effect of [Se-Se]-OT-OH or oxytocin was compared to the activity preceding the application of the first dose, and the data represent the change in activity compared to spontaneous control activity (100%). Concentration-response curves were fitted using nonlinear regression to calculate EC50 values.

Human cardiomyocyte assay

Wild-type WTC11 human-induced pluripotent stem cells (hiPSCs) were used in this study. Undifferentiated cells were maintained in mTeSR1 medium (STEMCELL Technologies). Standard cardiomyocyte-directed differentiation using a monolayer platform was performed with a modified protocol based on previous reports (105109). The differentiation setup was initiated by plating undifferentiated hiPSCs as single cells. The cultures were treated with 1 μM CHIR-99021 (Cayman Chemical) for 24 hours before reaching confluence. Cells were induced to differentiate (designated day 0) by replacing the culturing medium with RPMI 1640 medium (catalog number 11875-119, Thermo Fisher Scientific) containing 3 μM CHIR-99021, BSA (500 μg/ml), and ascorbic acid (213 μg/ml). On day 3, the medium was changed to RPMI 1640 medium with BSA and ascorbic acid containing 1 μM XAV-939 (Tocris Bioscience). On day 5, the medium was changed to RPMI 1640 containing ascorbic acid and BSA. On day 7, the medium was replaced with RPMI 1640 containing B-27 supplement with insulin (catalog number 17504-044, Thermo Fisher Scientific). Cells were harvested using trypsin, replated into a black-walled 384-well imaging plate coated with CellBIND (Sigma-Aldrich) at a density of 4 × 104 cells per well and cultured for 8 days before the experiments. Functional activity was assessed using a high-throughput FLIPRTETRA (Molecular Devices) FLIPR assay. The growth medium was removed and replaced with Calcium 4 (no-wash) dye (Molecular Devices) diluted in PSS [140 mM NaCl, 11.5 mM glucose, 5.9 mM KCl, 1.4 mM MgCl2, 1.2 mM NaH2PO4, 5 mM NaHCO3, 1.8 mM CaCl2, and 10 mM Hepes (pH 7.4)] according to the manufacturer’s instructions and incubated at 37°C for 30 min. Changes in intracellular Ca2+ amounts were measured using a cooled CCD camera (excitation, 470 to 495 nm; emission, 515 to 575 nm) with reads taken every 0.4 s for 240 s after the addition of compounds. The maximum intracellular Ca2+ amount was determined using ScreenWorks (version, Molecular Devices).

Statistical analysis

All statistical comparisons were made using GraphPad Prism software, and values were expressed as means ± SEM. Student’s t test and one-way analysis of variance (ANOVA) were used to determine statistical significance for the human myometrial cell contractility and organ bath assays. Two-way ANOVA was used to analyze the SFC data. P values <0.05 were considered significant.


Materials and Methods

Fig. S1. Representative raw FLIPR data.

Fig. S2. Schild plot analysis.

Fig. S3. Functional study of [Se-Se]-OT-OH and d[Se-Se]-OT-OH at the mOTR, mV1aR, mV1bR, and mV2R.

Fig. S4. Binding study of [Se-Se]-OT-OH and d[Se-Se]-OT-OH at the mOTR, mV1aR, mV1bR, and mV2R.

Fig. S5. Human myometrial cell contractility assay.

Fig. S6. Human myometrial strip contractility assay.

Table S1. Overview of the synthesized peptides including details on their modifications.

Table S2. Functional potencies for oxytocin and vasopressin analogs at the hOTR, hV1aR, hV1bR, and hV2R.

Table S3. Functional potencies for oxytocin and vasopressin analogs at the human and murine OTR, V1aR, V1bR, and V2R.

Table S4. Comparison of functional data from the FLIPR and HTRF-IP1 assays.

Table S5. Radioligand displacement data for oxytocin and vasopressin analogs at the human and murine OTR, V1aR, V1bR, and V2R.

Table S6. Overview of the number of electric foot shocks per mouse.

Reference (110)