Many bioactive peptides have shown bioactivity by binding to various proteins, such as intracellular soluble proteins, membrane-binding proteins, and so on. (1) Such bioactive peptides have received much attention as promising chemicals for pharmaceuticals and food materials. Some high-affinity cytokine-like peptides are known to have cyclic structures. It has been reported that cyclic peptides have a high affinity for their target proteins because there is less entropy loss upon protein–peptide binding if the rigid structural peptide closely fits a target protein. (2)
One of the important problems to be solved for bioactive screening is the magnitude of the sheer number of peptide variants. A new expression screening method for bioactive peptides, such as the random peptide-integrated discovery (RaPID) display and split-intein circular ligation of peptides and proteins (SICLOPPS), has been proposed, and a large number of peptide libraries have been prepared. (3−5)
A typical method for establishing a chemically synthesized peptide library is a peptide array method established by Frank. (6) Linear peptides are simultaneously synthesized on an activated cellulose membrane in an array format. The synthesized peptide library can then be utilized for in situ binding assays with the proteins of interest. A methodology for the circularization of synthesized linear peptides has also been reported. (7) Compared with the above expression screening method, the library size for this peptide screening method is small (approximately 10 2–10 4) because peptide synthesis is performed via machine manufacturing. A peptide microarray for high-throughput binding detection on a solid surface has also been proposed. Usui et al. reported a peptide microarray method on the nanoliter scale using a robotic spotter for proteomic analyses and ligand screenings. (8) However, the synthesis of many peptide variants is a rate-limiting step in the production of chemically synthesized peptide arrays.
Recently, computationally associated peptide screening has been proposed for the assessment of the activities of various bioactive peptides, such as antimicrobial peptides, (9) antioxidative peptides, (10) and therapeutic peptides, (11) driven by the relationship analysis of their amino acid (AA) sequences and bioactivities. Peptide array screening is preferable to this method because quantitative activity data on all peptides can be gathered. It should be noted that the effect of AA residue substitution at any position can be investigated in detail because the sequence design is systematically planned.
In our group, bioactive peptide screening using a peptide array was performed. (12−15) It was confirmed that peptide activity could be improved via AA residue substitution. A cyclic peptide library was recently established and applied for the assay of NGR cyclic cell adhesion peptides. (16) Previously, we also established a free peptide library of linear peptides using a photocleavable linker. (17)
Oxytocin (OXT) is a natural cyclic nonapeptide with the sequence CYIQNCPLG, which generates a disulfide bond between the two Cys residues. OXT bioactivity is reported to have diverse roles mediated by specific oxytocin receptors (OXTRs). Its role during uterine contraction at childbirth and milk ejection from the mammary glands is well known. It is also known to regulate several physiological functions such as vascular and cardiac relaxation and interferes with salt and water balance. (18) OXTR is a G-protein-coupled receptor (GPCR) expressed in both the myometrium and endometrium of the uterus and the myoepithelial cells of the mammary glands. OXTR is also expressed in the central nervous system, and there are many studies on the relationship between OXT function as a neurotransmitter and social recognition, anxiety, autism, and so on. (19,20)
The OXT/OXTR system has also received attention as a pharmaceutical target for neuropsychiatric disorders, including schizophrenia and autism spectrum disorder (ASD). (19) So far, many researchers have investigated OXT substitution variants and OXT analogues for drug development or activity–structure analyses. There are a few reports on OXT variants generated via natural AA substitutions. Replacement of the asparagine residue at P5 to Ala, Gln, Ser, Orn, or Val has been reported to have a deleterious effect on its biological activity. (21) There have also been reports of OXT analogues using d-amino acids. For example, d-Cys analogues substituted at P6 showed high antagonist activity, while replacement of Tyr at P2 with an aromatic d-amino acid inhibited the uterotonic activity of OXT. Even now, studies on OXT analogues have been actively carried out. (22−24)
In the present study, we aimed to establish a free cyclic peptide library to systematically evaluate the agonist/antagonist activity of OXT variants. Therefore, we determined the quantitative activity of OXT variants substituted with one AA residue using a transforming growth factor-α (TGF-α) shedding assay and investigated the effect of the substituted residues on the relationship between the AA sequence and bioactivity. Consequently, no improvement in agonist activity was observed, but OXT variants with high antagonist activity were obtained. A free cyclic peptide library constructed via natural amino acid substitution is a powerful tool for screening bioactive peptides.
The cyclic peptides used in this study were synthesized on a cellulose membrane using a spot synthesis method. The reaction conditions for disulfide bond formation were established as described in our previous papers. (12,16) When we preliminarily synthesized a free cyclic OXT peptide and analyzed the released peptide via mass spectrometry, the peak of the intermolecular dimer was detected. To prevent intermolecular oxidation, the photolinker concentration in the peptide array synthesis was decreased. When the concentration was adjusted to 10% of the original concentration (25 mM), a single peak of cyclic OXT was obtained. It was concluded that intermolecular dimerization was eliminated and disulfide bonds were formed in all molecules.
To assure the bioactivity of the cyclic OXT synthesized on the peptide array, its binding activity to OXTR was evaluated using a TGF-α shedding assay. The amount of AP-TGFα released based on G-protein signaling was observed in a dose-dependent manner. Thus, we confirmed the system construction to accurately evaluate the activity of cyclic peptides against GPCRs.
In addition, the detection range of the OXT concentration used in this evaluation system was 500 pM to 5 μM. Therefore, the peptide was eluted with 200 μL of Hank’s balanced salt solution (HBSS) buffer from the peptide array and then the solution diluted in five volumes was used for the assay.
First, we tried to find an OXT variant sequence with higher biological activity than CYIQNCPLG, the original OXT sequence. AA substitution at seven positions was planned, except for the two Cys residues at P1 and P6.
To reduce the number of substituted AA residues and increase the screening efficiency, the number of AA residues used for substitution was decreased from 19 to 8. Natural AAs were classified into eight groups based on their physical properties and the sizes of their side chains, and a representative AA from each group was selected for residue substitution. Ala (A), Leu (L), Phe (F), Ser (S), Asn (N), Asp (D), Lys (K), and Pro (P) were used for OXT variants as representative AAs in each group.
None of the OXT variants showed higher agonist activity than the original sequence. Among them, AA substitution at P2, P3, P5, P7, and P9 resulted in a significant decrease in agonist activity.
From the results shown, we further synthesized and validated different AA substitution variants only at positions 4 and 8. The results are shown. For the fourth residue, the Asn substitution variant in the same side-chain structure as the original Gln residue showed approximately 35% of the agonist activity of the original OXT. The Ser substitution variant, which has the same uncharged hydrophilic property as Gln, showed approximately 8% agonist activity of the original sequence. For the eighth residue, the Ile substitution variant, which retained the highest activity, showed 82% of the agonist activity of the original sequence, the Val substituent showed 55%, followed by 22% for Asn, and 18% for Ala. However, none of them showed higher agonist activity than the original sequence.
Next, we demonstrated the importance of the ring structure. In our previous report, we succeeded in forming heterodimers via intramolecular SS bonding using cysteine residues. For this purpose, we used Fmoc-Lys(ivDde)-OH to synthesize peptides with different main and side chains and introduced a disulfide bond into the interior of both peptides. (12) This method was used to synthesize a heterodimer that is cleaved at any peptide bond within its ring structure. Although all heterodimers were assayed, the results showed that none of the variants showed any activity (data not shown).
We were unable to obtain OXT variants with higher agonist activity than the original OXT. However, OXTR antagonists are required as pharmaceutical interventions to prevent premature births. Therefore, we also investigated the antagonist activity of OXT variants. For this purpose, an approximately 15 times larger amount of OXT variants than that in the agonist assay was used in the TGF-α shedding assay. In the experiment, 10 μL of HBSS solution containing the substituted variant was first added to a culture medium containing OXTR-expressing HEK293 cells, and subsequently, 10 μL of 10 nM OXT was added to evaluate the inhibitory effect of the substituted variant.
The results are shown. Of the 53 substituted variants, 18 variants showed a relative activity of less than 0.5. The variants with extremely low activity included three Phe substituents (Q4F, N5F, and P7F), and the observed inhibition was attributed to antagonist activity.
In contrast, 12 variants showed a relative activity of more than 2, which was the enhancement effect, not the inhibitory effect. Three variants, Q4N, L8A, and L8S, demonstrated at least tenfold greater activity than the control. The enhancement effect was observed in some variants substituted at the fourth residue and eighth residue. A high enhancement effect was predicted to be correlated with high activity in the agonist assay, as shown.
To define the relationship between agonist and antagonist assays, the activity of the variants listed is described. A moderate positive correlation was observed, and three groups containing variants with significantly different activity levels were defined. Group A consisted of variants showing significantly low activity in the antagonist assay (p< 0.01) and low activity in the agonist assay. These were Y2L, Q4F, N5F, and P7F, and they are defined as antagonists. Group B consisted of six variants with significantly high enhancement activity in the antagonist assay (p< 0.05) (L8S, L8A, L8N, L8K, Q4S, and Q4N), and group C consisted of three variants showing significantly high enhancement activity in the antagonist assay (p< 0.05) and low activity in the agonist assay (L3D, Q4D, and N5D). A major contributor to the high level of enhancement observed in the antagonist assays is the 15 times larger amount of OXT variants used in these assays as compared with the one used in agonist assays. However, the activity of L8A and L8S variants was significantly high even though a large amount of these variants was added. The observed high activity level may be explained by the additive or synergistic effect toward the agonist activity of OXT.
It should be noted that all variants identified in group C were Asp (D) substitutes. Other D substitutes, such as Y2D, I3D, Q4D, N5D, P7D, L8D, and G9D, exhibited no agonist activity with the singular exception of L8D; however, all D substitute
