All synthetic peptides were synthesized at SCRUM Inc. (Tokyo, Japan) using Fmoc-based solid-phase peptide synthesis, followed by reverse phase-high performance liquid chromatography (RP-HPLC) purification. Peptide purity was ascertained by analytical RP-HPLC and structural assignment was performed by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS). All analytical data of peptides in this report are presented in Supplementary Table S1.

The following briefly describes the synthesis method of KS-133. After synthesis and purification of side chain-protected linear peptide-linked resin, it was dissolved in dichloromethane (DCM) and then mixed with a 2% hydrazine solution for 10 min to deprotect Dde for the side chain of Lys and ODmab for the side chain of Asp. EDC/HOAt was further dissolved in DCM to activate the side chain carboxy group of Asp for amide bond formation with the side chain amide group of Lys. After incubation overnight at room temperature, the resin was washed and then treated with trifluoroacetic acid (TFA) to deprotect the thiol group and excise the monocyclic peptide from the resin. The peptide was purified by RP-HPLC using a SunFire C18 column (Waters Co., Milford, MA, United States). The fraction that contained the product was then collected and lyophilized to produce a side chain-cyclized and -deprotected peptide, Ac-Cys-Pro-Pro-Tyr-Leu-Pro-_c(Lys-Tyr-Leu-Cys-Asp)-Leu-Ile-NH₂ (amide bond cyclization with side chains of Lys⁷ and Asp¹¹), as a white powder with a mass spectrum of (M–H)⁻ 1561.265 (Calc 1,560.9), purity of 96.14%, and elution time on RP-HPLC (flow rate 1 ml/min) of 11.567 min under linear density gradient elution conditions (A/B = 80/20–10/90 for 20 min using 0.1% TFA in water as eluent A and 0.1% TFA in acetonitrile as eluent B). The peptide was dissolved in 0.1 M NH₄HCO₃ (pH 8.0) and reacted for 24–36 h at room temperature to the link side chains of Cys¹/Cys¹⁰ by a disulfide bond. The peptide was purified by RP-HPLC using the SunFire C18 column. The fraction with the product was collected and lyophilized to obtain bicyclic peptide KS-133, Ac-_c[Cys-Pro-Pro-Tyr-Leu-Pro-_c(Lys-Tyr-Leu-Cys]-Asp)-Leu-Ile-NH₂ (disulfide bond cyclization with side chains of Cys¹ and Cys¹⁰, and amide bond cyclization with side chains of Lys⁷ and Asp¹¹) as a white powder whose mass spectrum was (M–H)⁻ 1,558.705 (Calc 1,558.8) with a purity of 100.00% and elution time on RP-HPLC (flow rate 1 ml/min) of 10.617 min under linear density gradient elution conditions (A/B = 80/20–10/90 for 20 min using 0.1% TFA in water as eluent A and 0.1% TFA in acetonitrile as eluent B.

Evaluation of antagonist activities of peptides against VIP-VIPR1, VIP-VIPR2, and PACAP-PAC1 signaling pathways and agonistic activity of KS-133 at VIPR2 was carried out by Eurofins DiscoverX (Fremont, CA, United States). Intracellular calcium mobilization was monitored using a calcium-sensitive dye loaded in CHO-K1 cells expressing human VIPR1 (ITEM 86-0030P-2243AN), VIPR2 (ITEM 86-0030P-2244AN), and PAC1 (ITEM 86-0030P-2066AN). The cells were preincubated with antagonistic peptides for 30 min at room temperature in the dark. After addition of the ligand at EC₈₀, calcium mobilization was immediately monitored by FLIPR Tetra for 2 min. Cyclic AMP production in CHO-K1 cells expressing human VIPR2 was evaluated using HitHunter^® cAMP assays (antagonist mode: ITEM 86-0007P-2362AN; agonist mode: ITEM 86-0007P-2362AG). The cells were preincubated with antagonistic peptides for 30 min at room temperature. After addition of the ligand at EC₈₀, cells were further incubated for 30 min at room temperature. Cells were lysed and the cAMP content of the cell lysate was determined by the β-galactosidase-based enzyme fragment complementation assay (Eglen, 2002). IC₅₀ values were estimated by the following equation: IC₅₀ = 10^[Log(A/B) × (50–D)/(C–D) + Log(B)], where A is the concentration at >50% inhibition, B is the concentration at <50% inhibition, C is the inhibition rate at concertation A, and D is the inhibition rate at concertation B.

Each peptide (10 mM, 1 μl) was incubated with rat plasma (20 μl) prepared in-house. Immediately after mixing with plasma and after 24 h of incubation at 37°C, 80% acetonitrile (200 μl) was added to extract VIpep-3 and acetonitrile (200 μl) was added to extract KS-132, KS-133(monocyclic), and KS-133. The mixture was stored for 10 min on ice and then centrifuged at 20,000 × g for 10 min at 4°C. The supernatant was recovered and directly applied to RP-HPLC to determine the remaining amount of unmodified peptides in the sample.

Experimental procedures that involved animals and their care were conducted in compliance with the Guide for the Care and Use of Laboratory Animals (National Research Council, 1996) and ARRIVE guidelines (Kilkenny et al., 2010; McGrath and Lilley, 2015). All animal experiments were approved by the Committee of Research Facilities for Laboratory Animal Science of Hiroshima University (#A20-115) and the Animal Care and Use Committee of the Graduate School of Pharmaceutical Sciences, Osaka University (#30-3-2). Pregnant ICR (CD1) mice at 16 days of gestation and male C57BL/6J mice at 7 weeks of age were purchased from Japan SLC Inc. (Shizuoka, Japan). Mice were housed individually in plastic cages under a standard 12-h light/dark cycle (lights on 08:00 h) at a constant temperature of 22 ± 1°C. The animals had ad libitum access to food and water. Pregnant females were monitored for the parturition date that was considered as postnatal day (P) 0.

For western blot analysis, ICR neonatal mice at P12 (7–10 g body weight) were subcutaneously (s.c.) injected with KS-133 (1 nmol/g) and Ro 25-1553 (0.4 nmol/g, Peptide Institute, Inc. Osaka, Japan), a selective VIPR2 agonist (Gourlet et al., 1997; Harmar et al., 2012), and adult C57BL/6J mice at 8 weeks of age (20–30 g body weight) were intranasally (i.n.) injected with KS-133 (20 nmol/mouse) and BAY 55-9837 (20 µg/mouse, Tocris Bioscience, Bristol, United Kingdom), another selective VIPR2 agonist (Tsutsumi et al., 2002; Tian et al., 2019). The dosages of these drugs were determined in a preliminary experiment using the change in phosphorylation of cAMP-response element-binding protein (CREB) as an index. Ro 25-1553 was dissolved in phosphate-buffered saline (PBS) and BAY 55-9837 was dissolved in distilled water. KS-133 was dissolved in saline with 1% dimethyl sulfoxide (DMSO; #13445-74, Nacalai Tesque, Inc., Kyoto, Japan) for s.c. injection or saline with 5% DMSO and 5 mg/ml poly-L-arginine (#P7762, Sigma-Aldrich, St. Louis, MO, United States) for i.n. injection. The volumes of the drug solutions were 5 ml/kg body weight for s.c. injection and 10 µl for i.n. injection.

For behavioral experiments, all litters were randomly divided into Ro 25-1553- and PBS-treated groups. From P1 to P14, mice were injected s.c. once daily with Ro 25-1553 at a dose of 0.07 nmol/g (Ago et al., 2015). Mice treated with PBS from P1 to P14 were used as the control. KS-133 or the vehicle was simultaneously injected with Ro 25-1553 or PBS. Animals were weaned at P21 and divided by gender at P28. We used male mice exclusively to minimize any potential variability due to sex-specific effects on behavioral performance. All groups were derived from at least four different litters to preclude possible differences in individual maternal behaviors as a mitigating factor in any subsequent long-lasting changes induced in the offspring. Behavioral analyses of mice were carried out at 2−3 months of age. Experimenters were blinded to the treatment while testing.

Each mouse was anesthetized with isoflurane and their brain was removed rapidly. The prefrontal cortex was dissected on ice, frozen on dry ice, and stored at −80°C until analysis. Tissue samples were homogenized at 4°C in N-PER™ Neuronal Protein Extraction Reagent (#87792, Thermo Fisher Scientific, MA) with a protease inhibitor cocktail (#25955-11, Nacalai Tesque) and phosphatase inhibitor cocktail (#4906845001, Sigma-Aldrich). The homogenate was incubated for 30 min on ice and then centrifuged at 14,000 × g for 15 min at 4°C and the resulting supernatant was collected. Forty-five (phospho-CREB and CREB) or three (β-actin) micrograms of protein was loaded onto a 4–20% precast gel (#4561096, Bio-Rad Laboratories, Inc., CA), separated by sodium dodecyl sulfate polyacrylamide electrophoresis, and then transferred electrophoretically to a hydrophobic polyvinylidene fluoride membrane. The blotted membranes were blocked with a blocking buffer (5% bovine serum albumin in Tris-buffered saline with 0.1% Tween-20) for 1 h at room temperature and then incubated with primary antibodies phospho-CREB (Ser133) (87G3) rabbit monoclonal antibody (1:500; #9198, Cell Signaling Technology, Danvers, MA), CREB (48H2) rabbit monoclonal antibody (1:1000; #9197, Cell Signaling Technology) at 4°C overnight, or β-actin mouse monoclonal antibody (1:5000; #A2228, Sigma-Aldrich) for 1 h at room temperature. The membranes were then incubated with horseradish peroxidase-conjugated anti-rabbit IgG (1:3000; #7074, Cell Signaling Technology) or anti-mouse IgG (1:5000; #ab6789, Abcam, Cambridge, MA) for 1 h at room temperature. The primary and secondary antibodies were diluted with Can Get Signal™ (#NKB-101T; TOYOBO Co., Ltd., Osaka, Japan) Solution 1 and 2, respectively. Immune complexes were visualized using ECL2 Western Blotting Detection Reagents (PerkinElmer, Inc., Waltham, MA). Densitometric analysis was carried out using the CS analyzer 4 software package (ATTO Co., Kyoto, Japan). Expression levels of phospho-CREB were normalized to total CREB and β-actin, and expression levels of CREB were normalized to β-actin. Data are presented as the fold change relative to the PBS/vehicle-treated control group.

The novel object recognition test was carried out in accordance with previous reports (Takuma et al., 2014; Hara et al., 2016). In brief, after habituation to the experimental box under dim light conditions (30 lx) for 3 consecutive days, the test mouse was allowed to freely explore two novel objects (A and B) placed in the box for 10 min. Twenty-four hours after the training session, the retention session was conducted. In the retention session, object B was replaced with novel object C and the mouse was allowed to move freely for 5 min in the same box. The exploration time for each object in the retention session was measured with a stopwatch. The discrimination index (%) was the difference between the exploration time for the novel object plus that for the familiar object divided by the total exploration time. This index was used to calculate values for recognition memory. This test was conducted between 9:00–15:00.

Dendritic morphological analysis was performed on mice after the novel object recognition test and samples were collected within 10 min of dissection. Golgi-Cox impregnation was performed using an FD Rapid GolgiStain™ Kit (#PK401, FD Neurotechnologies Inc., Columbia, MD) as described previously (Takuma et al., 2014; Hara et al., 2016; Ago et al., 2017). Briefly, mice were deeply anesthetized with isoflurane, and their brains were removed, rinsed with Milli-Q water, and immersed in impregnation solution composed of potassium dichromate, mercuric chloride and potassium chromate. The brains were stored at room temperature for 2 weeks and then transferred and stored in a cryoprotectant solution for 72 h in the dark. The impregnated brains were embedded in 3.5% agarose gel and cut with a vibratome (VT1000S; Leica Microsystems) at room temperature. Coronal sections of 100 μm in thickness were mounted on Gelatin-Coated Microscope Slides (#PO101, FD Neurotechnologies Inc.) and dried in air at room temperature in the dark for 24 h. After drying, the sections were rinsed with Milli-Q water, reacted in the working solution, and dehydrated with a 50, 75, 95, and 100% graded ethanol series. Finally, the sections were dewaxed in xylene and coverslipped using Mount Quick (Daido Sangyo, Saitama, Japan). Digitized images from the prefrontal cortex (+2.245 through +1.345 mm with respect to the bregma) (Dong, 2008) were obtained under an upright light microscope with a cooled CCD digital camera system (Axio Imager.M2/AxioCam MRc5; Carl Zeiss, Jena, Germany). A 20× lens was used to measure dendrites. Only fully impregnated neurons that displayed dendritic trees without obvious truncation and that were isolated from neighboring impregnated neurons were retained for analysis. Forty pyramidal neurons with the soma in layers II/III were selected in the prefrontal cortex from five mice per group. Morphologies of apical and basal dendrites were quantified in three dimensions using the Neurolucida neuron tracing system (MBF Bioscience, Williston, VT) with the experimenter blinded to the treatment. The total length and branch number of apical and basal dendrites were compared among treatments. To assess differences in the amount and location of dendritic material, Sholl analysis was performed using a NeuroExplorer (MBF Bioscience).

All data are expressed as the mean ± standard error of the mean (SEM). Data from western blots were analyzed using one-way analysis of variance (ANOVA), followed by the Tukey–Kramer post-hoc test. For behaviors, the total number of dendrites, and dendritic length, data were analyzed using two-way ANOVA, followed by the Tukey–Kramer test. For Sholl analysis, data were analyzed using two-way ANOVA with drug treatment as the intersubject factor and repeated measures with distance from the soma as the intrasubject factor. Statistical analyses were conducted using the software package StatView^® 5.0 for Windows (SAS Institute, Cary, NC). A value of p < 0.05 was considered statistically significant.

For the molecular design of VIpep-3 derivatives, it is favorable to have structural information of VIpep-3, Ac-_c(CPPYLPRRLC)TLLLRS-OH. However, unfortunately, the crystal structure of the VIpep-3/VIPR2 complex has not been obtained to date. Alternatively, three structural aspects, the extracellular domain of VIPR2²⁶⁻¹¹⁴ (PDB ID: 2X57), VIP¹⁻²⁸ (HSDAVFTDNYTRLRKQMAVKKYLNSILN-NH₂) (PDB ID: 2RRI) (Umetsu et al., 2011), and a structural model of the VIP/VIPR1 complex (Couvineau et al., 2012) have been reported. Using this information, we first predicted the interaction mechanism between VIP and VIPR2. As shown in Figure 1A, a deep and wide groove that comprised three pockets (P1–P3) was observed on VIPR2. When we superimposed the structure of VIP onto the groove, Val¹⁹, Leu²³, and Leu²⁷ of VIP were likely to form hydrophobic interactions with Val⁷¹, Val⁷³, and Pro⁷⁴ of VIPR2 (Figure 1B). Additionally, Ile²⁶ and Tyr²² of VIP were likely to interact with pocket-1 and pocket-2 of VIPR2, respectively (Figure 1B).

Next, we tried to use this predicted binding mode of VIP for the molecular design of VIpep-3 derivatives. We found that the sequence order pattern of aliphatic amino acids of VIpep-3 (Leu⁵, Leu⁹, Leu¹², and Leu¹³) was similar to that of VIP (Val¹⁹, Leu²³, Ile²⁶, and Leu²⁷) (Figure 1E). Therefore, we hypothesized that VIpep-3 had a similar secondary structure as VIP and then designed a VIP/VIpep-3 hybrid sequence, CPPYLPKYLCSLLN (underlined amino acids are from VIP and others are from VIpep-3). When we superimposed the structure onto the groove of VIPR2, Tyr⁴ was likely to interact with pocket-3 (Figure 1C) in addition to the predicted binding mode of VIP/VIPR2. A disulfide bond between Cys¹ and Cys¹⁰ could be formed by 180° reverse orientation of the backbone at the point of Cα³ (Figure 1D). This was supported by the fact that the reported VIpep-3 family has an extremely conserved Pro⁶ (Sakamoto et al., 2018), which is a 5-membered ring heterocycle amino acid without a hydrogen bond donor Nα and makes it possible to flip the main chain orientation by breaking the helix structure.

The amino acid sequence of VIpep-3 is highly conserved in its family except for Arg⁷, Arg⁸, Thr¹¹, Leu¹⁴, Arg¹⁵, and Ser¹⁶ (Sakamoto et al., 2018). On the basis of the predicted binding mode, it appeared to be better to substitute Arg⁸ with Tyr. Because the three C-terminal residues Leu¹⁴, Arg¹⁵ and Ser¹⁶ were unlikely to be deeply involved in VIPR2-binding (Sakamoto et al., 2018), they were deleted. Consequently, the remaining amino acid-substitutable residues were Arg⁷ and Thr¹¹.

A strategy for VIpep-3 structure optimization is introduction of positively charged amino acid L-2,3-diaminopropionic acid (Dap) to the 7^th and 11^th positions to interact with negatively charged Glu³⁴ of VIPR2. KS-132, Ac-_c(Cys-Pro-Pro-Tyr-Leu-Pro-Dap-Tyr-Leu-Cys)-Dap-Leu-Ile-NH₂ (disulfide bond cyclization with side chains of Cys¹ and Cys¹⁰), was designed and synthesized (Figures 1E, 2).

Another strategy is bicyclization. In accordance with previous studies on VIP structure optimization, cyclization by bridging between the 21^st and 25^th positions was effective (Gourlet et al., 1997; Yung et al., 2003; Onoue et al., 2008; Giordanetto et al., 2013). For example, VIP analogues Ro 25-1553 (Gourlet et al., 1997) and Ro 25-1392 have been reported as VIPR2-selective agonists that have a cyclic structure by bridging between Lys²¹ and Asp²⁵ (Figure 1E). Therefore, we chose bicyclization as another strategy. KS-133, Ac-_c[Cys-Pro-Pro-Tyr-Leu-Pro-_c(Lys-Tyr-Leu-Cys]-Asp)-Leu-Ile-NH₂ (disulfide bond cyclization with side chains of Cys¹ and Cys¹⁰, and amide bond cyclization with side chains of Lys⁷ and Asp¹¹), was designed and synthesized (Figures 1E, 2). This bicyclization structure should stabilize active conformation of peptide.

First, the antagonistic activities of parental VIpep-3, KS-132, KS-133 (monocyclic) (no bridging between Lys⁷ and Asp¹¹), and KS-133 against human VIPR1, VIPR2, and PAC1 were evaluated by a calcium influx assay. After antagonistic peptide treatment for 30 min, ligand at a concentration of EC₈₀ was added to the cells and then calcium signaling was measured. All peptides antagonized the VIP-VIPR2 signaling pathway in a peptide concentration-dependent manner (Figure 3) but did not antagonize VIP-VIPR1 or PACAP-PAC1 signaling pathways up to 5 μM (Table 1). KS-132 and KS-133 had IC₅₀ values of 33.4 and 24.8 nM, respectively, which were stronger antagonistic activities than parental VIpep-3 (40.6 nM) (Table 1). This result suggested that both of the abovementioned two molecular design strategies were successful. Additionally, this result indicated that these derivatives had inherited the molecular properties required for VIPR2-selective binding activity from parental VIpep-3.

Next, antagonistic and agonistic activities of KS-133 at VIPR2 were evaluated by a cAMP assay. As shown in Figure 3, KS-133 antagonized VIP-VIPR2 signaling pathway in a peptide concentration-dependent manner. The IC₅₀ value was determined as 500 nM (Table 1). On the other hand, KS-133 did not exhibit VIPR2 agonist activity, even at high concentration of 5 μM (Figure 3 and Table 1).

The stability of peptides was evaluated in rat plasma. A peptide was incubated in rat plasma at 37°C for 0 and 24 h and the remaining amount of peptide was determined by RP-HPLC. By comparison with the peak area of the peptide, the remaining amount of the peptide after plasma incubation was estimated. As shown in Table 2, monocyclic peptides VIpep-3 and KS-132 were significantly degraded within 24 h. Conversely, bicyclic peptide KS-133 was highly stable for at least up to 24 h. Interestingly, KS-133(monocyclic) was also highly stable. KS-133(monocyclic) may form a pseudo bicyclic structure by ionic bond interaction between side chains of Lys and Asp.

KS-133, which had the strongest antagonistic activity in vitro and good resistance to protease degradation, was selected for in vivo evaluation. To determine whether KS-133 blocked VIPR2-mediated signaling in the brain, we first examined the effects of systemic administration of KS-133 on phosphorylation of CREB, a biomarker downstream of VIPR2, in neonatal and adult mice (Figure 4). Subcutaneous administration of selective VIPR2 agonist Ro 25-1553 (0.4 nmol/g) to ICR mice at P12, when VIPR2 is highly expressed in mouse brain (Waschek et al., 1996), significantly enhanced phosphorylation of CREB in the prefrontal cortex (Figure 4A). Simultaneous injection of KS-133 (1 nmol/g) suppressed the Ro 25-1553-induced increase in CREB phosphorylation. We also evaluated the effects of KS-133 in adult mice (Figure 4B). Intranasal administration of selective VIPR2 agonist BAY 55-9837 (20 µg/mouse) significantly increased phosphorylated CREB in the prefrontal cortex, which was blocked by coadministration of KS-133 (20 nmol/mouse) with BAY 55-9837.

Next, we investigated the pharmacological efficacy of KS-133 in a mouse model of psychiatric disorders on the basis of early postnatal activation of VIPR2 (Ago et al., 2015) (Figure 5). Repeated administration of Ro 25-1553 (0.07 nmol/g, s.c., once daily) during P1–14 significantly decreased the difference in time spent exploring each object, whereas PBS-treated control mice explored the novel object by a significant preference in the novel object recognition test (Figure 5A). Additionally, the discrimination index was significantly lower in mice treated with Ro 25-1553 than in control mice (Figure 5B). Chronic treatment with KS-133 (1 nmol/g) attenuated the decreases in the difference of time spent exploring each object and cognitive dysfunction. Two-way ANOVA of the discrimination index revealed significant main effects of Ro 25-1553 treatment (F _1,63 = 17.378, p < 0.0001) and KS-133 treatment (F _1,63 = 5.261, p < 0.05), and there was a significant interaction between Ro 25-1553 and KS-133 treatments (F _1,63 = 23.988, p < 0.0001).

Because a reduced length of basilar dendrites and a reduced dendritic number have been found in layer 3 in prefrontal cortical areas [Brodmann area (BA) 10, BA 11, and BA 46] and the anterior cingulate cortex (BA 32) of schizophrenic patients (Glantz and Lewis, 2000; Kalus et al., 2000; Broadbelt et al., 2002; Black et al., 2004; Konopaske et al., 2014), we examined dendritic morphology in the prefrontal cortex of mice treated with Ro 25-1553 and KS-133 after the novel object recognition test (Figure 6). Figure 6A shows Golgi-stained pyramidal neurons and representative tracings of soma and dendrites in the prefrontal cortex. Ro 25-1553 reduced the cell soma size, total branch number and length of apical and basal dendrites of prefrontal cortex neurons (Figure 6B). These effects were counteracted by simultaneous administration of KS-133 with Ro 25-1553. Sholl analysis also revealed that the amount of dendritic material distal to the soma in apical (treatment × distance interaction: F _{19, 1482} = 1.957, p < 0.01) and basal (F _{19, 1482} = 1.750, p < 0.05) dendrites was decreased in Ro 25-1553-treated mice, which indicated a reduction in dendritic complexity. KS-133 prevented the morphological abnormalities in both apical (F _{19, 1482} = 3.261, p < 0.001) and basal (F _{19, 1482} = 3.877, p < 0.001) dendrites in the prefrontal cortex.