Fragile X syndrome (FXS) is an X chromosome-related genetic disorder that is associated with neurodevelopmental problems such as hyperactivity, intellectual disability, social problems, and impulsive behaviors. FXS is caused by a mutation of the fragile X mental retardation 1 gene (FMR1) located at Xq27.3. This mutation initiates an abnormal repetitive expansion of the trinucleotide CGG in the 5′ untranslated region of the FMR1 gene. This causes silencing of the FMR1 gene and results in the deficiency of its genetic product, FMRP1. Because FMRP contributes to the synaptogenesis and maintenance of synaptic function, brain development, and neural plasticity, lack of FMRP will cause neural mRNA alternations, increases neural protein synthesis, and imbalance in the excitatory/inhibitory synaptic transmission, which is one of the major phenotypes in FXS2. Several reports suggested a causal relationship between the level of FMRP and the clinical characteristics of patients with FXS. The emotional and cognitive phenotypes of FXS patients depend on the degree of methylation for FMR1 and its number of repeats. This means that the lower the level of FMRP, the more severe the symptoms will be2. Common physical features of FXS include a long face, macroorchidism, prominent ears and jaws, flat feet, and joint hypermobility. FXS patients have developmental problems such as mental retardation, intellectual disability, and language delay2. About 15% of patients have seizures during their childhood3 and about 20% have strabismus4.

Pharmacological and educational treatments for the management of behavioral and medical problems such as attention deficit hyperactivity disorder (ADHD)-like symptoms, aggressive behaviors, anxiety, and hyperarousal of FXS patients were performed4. In previous clinical studies, a positive response to treatment was defined as the type of improvement in the behavioral symptoms for 6 months with no major side effects leading to discontinuance of the medication. The success rate of each medication was 53% for selective serotonin reuptake inhibitor (SSRI), 62% for alpha-adrenergic agonists, and 54% for antipsychotics. In addition, SSRIs are one of the most used forms of treatment for adults or females with FXS. This helped alleviate social anxiety and withdrawals, especially in females. Alpha-adrenergic agonists were effective in treating hyperactivity and over-arousal as well as in alleviating sleep problems in young children with FXS. The antipsychotics helped target the aggressive behavior as well as the irritability associated with FXS5. Abnormal signaling pathways activated by metabotropic glutamate receptor (mGluR) are considered plausible treatment targets for FXS6. Unfortunately, clinical trials using these approaches were not successful in requesting other target molecules to test for the treatment of the detrimental condition. The reduction of FMRP regulated the expression of individual proteins as well as their activity. Among these, increased expression of α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) receptors and regulation of activity of receptors or proteins that modulate glutamate signaling has been repeatedly suggested as a potential mechanism to treat FXS. MMP9, MAP1B, PSD95, CaMKII, Arc, STEP, PIKE, APP, potassium channel Kv 3.1b, and others are also among the synaptic proteins that correspond to the FMRP target mRNAs and elevated in the fmr1 knockout (KO) mice2. As an RNA binding protein, FMRP can bind and regulate its target mRNAs’ expression which results in the abnormal gene expression signature in FXS compared to normal control. Considering the importance of the level of FMRP protein in the modulation of clinical phenotypes of FXS and its role as a modulator of translation by binding to target mRNAs, we first hypothesized that drugs that have the potential to normalize the altered gene expression profile in fmr1 KO mice or FXS patients may alleviate the neurobehavioral deficits seen in fmr1 KO mice. These drugs are expected to target the brain and modulate aberrant neural activity, especially preferring molecules that modulate neural transmission. Based on this hypothesis and assumption, we identified a candidate molecule and further examined its therapeutic effects in the fmr1 KO mice model of FXS. Along with this hypothesis, recently, transcriptomic data have been analyzed to investigate disease-drug and disease-disease relationships7 in the drug discovery process. Transcriptomic data is composed of genome-wide gene expression profiles and it gives experimenters information about specific cells, tissues, or organisms' transcript levels under the experimental conditions8. The gene expression omnibus (GEO), which is administered by the National Center for Biotechnology Information (NCBI), is widely used to get functional genomic data publicly. By analyzing transcriptomics data, it is easier to understand the network of differentially expressed genes (DEG) from specific biological processes, diseases, or drugs. With regards to drug development, using DEG analysis can discover potential targets for disease therapeutics8. Connectivity map (CMap) is a method and database that helps to find molecules that induce or revert the gene expression signature of interest. CMap database has over 1.5 million gene expression profiles with around 5000 small molecule compounds-induced transcriptomic data sets. CMap utilizes a gene expression signature as an input and calculates the enrichment score. If a drug candidate molecule reverts the gene expression signature the score is negative. A positive score means a drug candidate molecule induces the gene signature. CMap output provides a list of candidate molecules' enrichment scores along with their statistical significance. CMap has been widely used to repurpose drugs, and discover new lead compounds in various diseases such as hair growth9, acute myelogenous leukemia2, smoking cessation2, cancers2.

As a new candidate for FXS treatment, pirenperone was further validated for its potential therapeutic effects both in vitro and in vivo. This approach and study may offer chances for novel therapeutic development against FXS and can be expanded to other CNS disorders that also need effective treatment.


DEGs through transcriptome analysis

To make signatures for CMap analysis, representative FXS transcriptome data were analyzed (GSE7329). Lists of up-and down-regulated genes based on the 90th percentile value of mean difference calculation are shown as a heatmap (Fig. 1A). There were 17 genes for up-regulated and 25 genes for down-regulated. Some of the DEGs (e.g. ATRNL12, FMR12, LAMB12) were previously reported as issued genes related to neurodevelopmental diseases. DEGs were further analyzed by GeneMANIA to reveal the molecular functional relationship. Two genes in the subnetwork, GAREM1, and ACKR3, along with co-expressing genes of SPRY2 and TNFRSF11A are revealed to be involved in the positive regulation of ERK1 and ERK2 cascade function (Fig. 1B). Interestingly, it has been reported that abnormal ERK activation in tissues from Fmr1 KO mice and blood lymphocytes from FXS patients2. However, no significant functional enrichment was found with 42 DEGs when FDR is controlled at 0.2. The whole DEG list is uploaded as Supplement material (Supplementary Table S1).

Figure 1
figure 1

42 genes were selected as DEGs between FXS and healthy control from GSE7329 data. (A) Heatmap of DEGs. (B) Molecular relationship between DEGs and associated genes. DEGs are located at the inner circle, associated are at outer. The color of the edge represents co-expression (purple), shared protein domain (grey), genetic interaction (green), predicted association (orange).

Mapping DEGs to CMAP

BY applying DEGs to CMap, 56 drugs with p values lower than 0.05 were deducted and among them, 25 drugs with a negative enrichment score to change the level of DEG oppositely were selected (Table 1). Among 25 drugs, 11 drugs are approved drugs by the US FDA. Pirenperone, a 5-hydroxytryptamine receptor (HT) 2A antagonist, was selected as a potential candidate for FXS because of its antipsychotic effects2. It had a − 0.711 enrichment value with a p value of 0.00434 from the CMap analysis.

Table 1 Top chemicals correlated with up and down signatures in the connectivity map by using DEGs.

Pirenperone treatment reversed the expression signature of the Fmr1 gene in control rat cortical neuron

To validate the findings from the transcriptomic analysis, the expression level of the Fmr1 gene, one of the genes which regulate FMRP, was confirmed with qRT-PCR by treating pirenperone in the control rat cortical neuron cell (Fig. 2A,B). Other input genes were also used to confirm the reversal of signature in our study: four genes were downregulated (LAMB1, ANK3, RIMS3, and CHL1) and three genes were upregulated (ROBO1, SETBP1, and ATRNL1) in FXS. Pirenperone treatment at 20 μM elevates gene expression of originally downregulated genes in FXS (4.56, 2.04, 1.595, and 1.732 times at LAMB1, CHL1, ANK3, and RIMS3, respectively), where it marginally changes the gene expression of originally upregulated genes (Fig. 2C–I). The results suggest that the expression levels of genes in cortical neurons which were mutated and causative for FXS were counter-regulated by pirenperone. The co-expression gene network of seven genes are functionally enriched in epithelial cell proliferation (FDR q-value: 0.032), basement membrane (q-value: 0.09), extracellular matrix disassembly (q-value: 0.12) (Fig. 2J). Our findings in conjunction with the known mechanism of such inputs may suggest pirenperone possibly rescue FXS symptoms by affecting the synaptic coverage of neurons.

Figure 2
figure 2

Upregulation of Fmr1 gene expression upon pirenperone treatment (A) FMR1 gene expression was significantly down-regulated in human FXS patients. Unpaired t-test. Control versus FXS, ***p < 0.001 (B) Fmr1 gene expression level was significantly up-regulated after pirenperone treatment at a dose of 20 μM in cortical neuron. (C) Gene expression level for ATRNL1 in human patients (left) and real-time PCR result in cultures (right). (D) Gene expression level for SETBP1 in human patients (left) and real-time PCR result in cultures (right). (E) Gene expression level for ROBO11 in human patients (left) and real-time PCR result in cultures (right). (F) Gene expression level for CHL1 in human patients (left) and real-time PCR result in cultures (right). (G) Gene expression level for LAMB1 in human patients (left) and real-time PCR result in cultures (right). (H) Gene expression level for RIMS3 in human patients (left) and real-time PCR result in cultures (right). (I) Gene expression level for ANK3 in human patients (left) and real-time PCR result in cultures (right). One-way ANOVA, post hoc test: Bonferroni’s multiple comparison test, Vehicle versus 20 μM, ***5 μM versus 20 μM, ###10 μM versus 20 μM, @@@p < 0.001, number of samples per group = 12. (J) Gene expression change after treatment of pirenperone. Color in the circle represents expression difference between the pirenperone and vehicle treatment (red: higher expression in pirenperone, grey means no difference). The thickness of edges represents the number of supporting evidence of a co-expressional relationship between genes.

In addition, pirenperone treatment significantly induced the FMRP expression in cortical neuron cultures (p value: 0.0152, Fig. 2B) demonstrating that pirenperone may be effective in the recovery of FMRP transcript levels in FXS.

The hyperactivity of Fmr1 KO mice was rescued by pirenperone treatment

Hyperactivity is one of the most significant features in Fmr1 KO mice2. To observe the hyperactivity of Fmr1 KO mice, the open field test was performed (Fig. 3A). Fmr1 KO mice showed hyperactivity with more distance moved and higher velocity than wild-type (WT) animals. After the injection of pirenperone, Fmr1 KO mice with pirenperone treatment showed decreased distance moved and velocity compared to Fmr1 KO mice (Fig. 3B). This result suggests that pirenperone can rescue the hyperactivity phenotype of FXS (Fig. 3C).

Figure 3
figure 3

Pirenperone treatment rescued the hyperactivity in the Fmr1 knockout animal. An open field test was performed at the age of 6 weeks. (A) After pirenperone treatment, Fmr1 KO mice showed similar distance-moved as WT. One-way ANOVA, post hoc test: Bonferroni’s multiple comparison test, WT versus KO + P (pirenperone), Fmr1 KO versus KO + P, **,##p < 0.01 (B) Pirenperone treatment, normalized higher moving velocity in Fmr1 KO mice to WT level. One-way ANOVA, post hoc test: Bonferroni’s multiple comparison test, WT versus KO + P, Fmr1 KO versus KO + P, **,##p < 0.01 (Number of samples per group = 6–7) (C) Tracking images of animals.

The spatial memory of Fmr1 KO mice was normalized by pirenperone treatment

Fmr1 KO mouse has deficits in learning and memory10. Y-maze was performed to confirm whether Fmr1 KO mice have lower spatial memory than WT mice. After the injection of pirenperone, Fmr1 KO mice with pirenperone treatment showed increased spontaneous alternation which suggests better spatial memory in pirenperone-treated mice than Fmr1 KO mice with saline (Fig. 4B). This result implies that pirenperone may improve spatial memory deficits in FXS. There was no significance in the total number of entries (Fig. 4A).

Figure 4
figure 4

Restored spatial memory and learning ability in Fmr1 knockout mice with pirenperone treatment. Y-maze was performed at the age of 6 weeks by measuring (A) total number entry for 8 min recording. (B) percent of spontaneous alternation by calculating [(number of alternation/total number of entry − 2) × 100]. One-way ANOVA, post hoc test: Bonferroni’s multiple comparison test, WT versus Fmr1 KO, **p < 0.01, Fmr1 KO versus KO + P (pirenperone), #p < 0.05 (Number of samples per group = 6–7).

The impulsivity behavior of Fmr1 KO mice was normalized by pirenperone treatment

Since impulsivity is a common feature of FXS patients2, we examined the impulsive behaviors of Fmr1 KO mice using an elevated plus maze (EPM)2. Staying time in the open arm is indicative of impulsivity and anxiety2. Fmr1 KO mice stayed more often in the open arms than WT and after pirenperone treatment, the impulsivity-related behavior was normalized to WT level (Fig. 5). The frequency of entering each arm was counted and Fmr1 KO mice entered closed arms more than normal fmr1 knockout mice (Fig. 5). These results suggest that the Fmr1 KO mice revealed higher impulsivity and it is rescued by pirenperone treatment.

Figure 5
figure 5

Level of impulsivity was normalized after pirenperone treatment. Elevated plus-maze was performed at the age of 6 weeks. (A) The duration of staying at both open and closed arms were measured and the percent time in open arms was calculated using the following formula [(duration in open arms)/(duration in both open and closed arms) × 100]. One-way ANOVA, post hoc test: Bonferroni’s multiple comparison test, WT versus Fmr1 KO, *p < 0.05, Fmr1 KO versus KO + P (pirenperone), #p < 0.05. (B) The frequency of entering each arm by the group (Number of samples per group = 6–7).


In our study, a novel potential drug candidate for FXS was determined through transcriptome analysis. Until recently, one of the most promising drug candidates for the treatment of FXS was group I metabotropic glutamate receptors (mGluR) modulators because mGluRs are regulated by the FMRP and plays important role in the pathophysiology of FXS2. Regulation of proteins that are related to glutamate signaling can also be a potential target to treat FXS2. Aside from these targets, we here discovered a potential new drug candidate pirenperone to treat FXS by CMap analysis of DEGs, which is helpful to understand the interaction among genes and biological processes8.

In our CMap analysis for the new drug candidates, the top 5 chemicals are thiamine, sc-560, hycanthone, biperiden, and myosmine. Among them, thiamine, sc-560, and myosmine were not approved for drugs, and hycanthone, a schistosomicide, targets multidrug resistance protein 111 and is shown to be an effective inhibitor of acetylcholinesterase (AChE) from Schistosoma mansoni, but is less potent against AChE from the mammalian origin12. Biperiden is a medication used to treat Parkinson's disease and certain drug-induced movement disorders2. It relieves muscle rigidity but is not used for neuropsychiatric problems. Thus, we selected pirenperone as a potential candidate for FXS because of its antipsychotic effects2. Pirenperone is a selective antagonist of the serotonin (5-hydroxytryptamine, 5-HT) receptor, specifically 5-HT2A2. It also has an affinity for 5-HT713 and dopamine receptors but has little effect on 5-HT1 binding14. Pirenperone is also reported to be liposoluble which means that it can penetrate the brain easily15. The drug is known as an antipsychotic and a tranquilizer that blocks the lysergic acid diethylamide (LSD) effects in animals16 implying that has no risk of side effects such as hallucinations.

In this study, the expression of the FXS causative gene Fmr1 as well as several other ASD-related genes such as LAMB1, CHL1, ANK3, and RIMS3 was transiently induced by pirenperone treatment in cortical neurons. LAMB1 gene drives the translation of laminin beta-1 protein, which is an extracellular matrix glycoprotein involved in several processes during embryonic development, such as laminar structure organization in the cerebral cortex, and are known stimulators of neurite outgrowth17. Many studies have found the possible involvement of LAMB1 mutation in the manifestation and severity of ASD2. CHL1 (cell adhesion L1-like) is a gene located in chromosome 3p26.3 that encodes a protein from the L1 family of cell adhesion molecules that are involved in CNS development and synaptic plasticity18. Mutations of this gene have been known to impair cognitive function19 with possible involvement in the development of ASD20,21 or other neurodevelopmental disorders22. The ANK3 gene-encoded AnkyrinG protein is involved in brain development and molecular signaling and is known to play an essential role in intellectual functioning23. In addition, the mutation of ANK3 has been linked to several neuropsychiatric conditions such as ASD24, ADHD, bipolar disorder2, and schizophrenia2, with the involvement of disrupted intellectual function23. RIMS3 is a member of the RIM protein family that serves as an essential component of the presynaptic mechanism for neurotransmitter release and synaptic vesicle fusion25. RIMS3 has been presented as a causal or contributing gene to ASD pathology2 and has also been linked to schizophrenia26. In this study, several up-regulated genes were not down-regulated in their expression level by pirenperone treatment on cortical neurons in this study, which includes ROBO1, SETBP1, and ATRNL1. While being associated with neuronal differentiation and axon guidance27,28, the gene Roundabout homolog 1 precursor (ROBO-1) is reported to be differentially expressed in some autistic individuals2. SETBP1 gene encodes for SET binding protein 129 which binds to the SET nuclear oncogene associated with DNA replication30. Recurrent de novo missense mutations of this gene are also reported to cause Schinzel-Giedion syndrome, characterized by severe mental retardation, characteristic facial features, and various congenital malformations29. A truncating point mutation in the SETBP1 gene was also identified in a simplex ASD case, as reported by O’Roak and his colleagues31. Highly conserved among species, the Attractin-like gene (ATRNL1) encodes for a single-pass transmembrane glycoprotein associated with cell adhesion and signaling events. De novo deletion of this gene was observed in a male patient with cognitive impairment, autism, and dysmorphic features32. Whether pirenperone can modulate the expression pattern of these genes in FXS brains remains to be determined. Especially, checking the protein expression level will provide important information on the regulatory effects of pirenperone on FMRP-dependent gene expression. In addition to the mRNA expression, the protein expression level is governed by several factors including turnover of the existing proteins, which we like to thoroughly investigate in the future studies.

The up-regulated mRNA level of the Fmr1 gene by pirenperone implies that pirenperone may exert its effect through the modulation of synapse formation, function, and neural plasticity by adjusting the level of FMRP expression. The RNA-binding protein FMRP is reported to be involved in the translation regulation of synaptic proteins. The decrease or loss of function of FMRP will lead to behavioral deficits and decreased cognitive functions in FXS patients33. This will also bring about other consequences such as altered GABA receptor subunits expression34 as well as increased long-term depression (LTD) of excitatory synapses35. This leads to an imbalance in the excitatory and inhibitory signaling which is one of the common features of FXS36,37. This implies that the pirenperone’s up-regulation of the Fmr1 gene leads to the increase in FMRP and might have regulated the ratio of the excitatory and inhibitory signaling thus restoring the balance and restoring some of the behavioral deficits. In this study, the most strongly affected gene is CHL1 whose expression has elevated 4.1 times after pirenperone treatment. CHL1 is called the cell adhesion molecule L1-like (CHL1 or CALL) gene, located on chromosome 3p26.3, and it is highly expressed in the central and peripheral nervous systems. The gene guides the growth of regenerating motor axons and regulates the synaptic coverage of motor neurons2. Several studies have also associated the duplication as well as the deletion of the CHL1 gene in humans with intellectual disabilities such as autism2. There have been recorded familial cases that presented deletion of 3p26.3 as well as the CHL1 the gene and was suggested that a reduction of about 50% of gene expression results to cognitive deficits2. Our finding in conjunction with the known mechanism of CHL1 may suggest the increase in the CHL1 gene expression after pirenperone treatment possibly rescues FXS symptoms by modulating synaptic development or increasing the synaptic coverage of neurons.

Pirenperone inhibited head twitches induced by 5-HT treatment and amphetamine-induced hyperactivity in SD rats2. Many clinical studies have already mentioned that 5-HT2A receptor antagonists are useful to regulate motor symptoms of Parkinson’s disease2. The cortical 5-HT2A receptors can affect the function of subcortical dopamine (DA) through modulating glutamatergic cortico-striatal circuits, therefore dysregulated locomotor activity could be normalized2.

Regarding the 5-HT2A receptor antagonist as a target of FXS, research on the role of the modulation of 5-HT2A activity on behavioral features such as hyperactivity, memory formation, and impulsiveness can be found elsewhere. 5-HT2A receptor antagonists such as ketanserin, M100907, αMe-5HT, and MDL11939 enhanced the cognitive learning ability of Fmr1 KO and BTBR mice which are known as one of the most common ASD models2. Deficits in GluR1-dependent synaptic plasticity have been observed in Fmr1 KO mice2.

Decreased N-methyl-d-aspartic acid (NMDA) receptor-dependent long-term potentiation (LTP) is observed in Fmr1 KO mice because of impairments of the signal activation between Ras and PI3K/PKB that damage GluR1-dependent plasticity2. For the reinstatement of learning ability, synaptic GluR1 delivery is important2. 5-HT2A receptor regulates synaptic plasticity and glutamatergic synaptic transmission with an increase in GluR1 in pre-frontal cortex neurons and these neuronal mechanisms may underlie the modulation of long-term potentiation in Fmr1 KO mice2. Inhibiting 5-HT2A receptors by treatment with α-Me-5HT and MDL11939, activated Ras-PI3K/PKB signaling, GluR1-dependent synaptic plasticity, and increased the ability to learn in Fmr1 KO mice2.

Additionally, searching KEGG DB38 (, pirenperone (D05495) belongs to the chemical subclass of butyrophenone derivatives. Other components in this class include haloperidol (DG00885), melperone (DG00887), and pipamperone (DG00889) harboring dopamine D2 and 5HT2A antagonism property. Haloperidol, an antipsychotic, can be used for FXS in several respects: anxiety, agitation, and aggressive behaviors (Elizabeth Berry-Kravis’s presentation at the 13th international Fragile X conference). Since haloperidol is a dopamine D2 antagonist, it had a high affinity for D2 dopamine receptors, and moderate affinity for serotonin 5-HT2A, and 5-HT1D receptors in human brain tissue homogenates2. Melperone is used for the treatment of schizophrenia, sleep disorders, agitation, and mentally confused states by antagonist activity at D2 dopaminergic and 5HT2A serotonergic receptors39. Similarly, pipamperone can be used for the treatment of chronic psychoses and states of aggressiveness of various origins. To improve haloperidol's pharmacological effects, Janssen developed pipamperone, an agent whose pharmacological profile was distinct from haloperidol. Unlike all other known antipsychotic drugs at that time, pipamperone had significant anti-tryptamine activity40. Interestingly, when risperidone was created, Janssen suggested it was a more potent version of pipamperone that has a selective 5-HT2A, D1, and D4 antagonist.

Many researchers observed a low level of anxiety in Fmr1 KO mice41. Previous studies have reported that Fmr1 KO mice have higher rates of premature responses which means that its inhibitory controls are lacking42. In this study, Fmr1 KO mice revealed low anxiety, and probably high impulsivity in elevated plus maze tasks, which were rescued by pirenperone. The serotonin system has been heavily implicated in impulsivity studies in general. There are implications on the involvement of 5-HT2A polymorphisms in the G-1438A promoter43 regions as well as in the A-1438A allele of the 5-HT2A receptor gene44 as the reason behind the impulsive behavior. Furthermore, it has been reported that 5-HT2A receptor antagonism using the 5-HT2A receptor antagonist MDL100,907 attenuates impulsive behavior in mice45,46. As suggested by several pharmacological studies, the G-coupled 5-HT2A receptor has been reported to be responsible for impulsive activities and this means that selective 5-HT2A receptor antagonist blocks this inherent behavior47,48. Our current study adds to this growing data that suggests the role of the 5-HT2A antagonist in the attenuation of impulsive behavior.

Elevated blood serotonin levels or hyperserotonemia in ASD including Rett syndrome and FXS individuals have been observed in about 30% of total patients2. It is not clear which subtypes of serotonin receptors play important roles in the modulation of ASD phenotypes but the results from the present study suggest that 5-HT2A receptors should be given more attention.

Loss of FMRP expression leads to the hyperactivation of the extracellular-signal-regulated kinase (ERK) signaling pathways in both FXS and KO mice2. Stimulation of group I metabotropic glutamate receptors via 5-HT2A activates phospholipase C, splitting membrane phosphatidylinositol into inositol triphosphate (releasing intracellular Ca2+ from cytoplasmic stores) and diacylglycerol, a specific activator of protein kinase C (PKC). PKC activation triggers a cascade leading to MEK phosphorylation of ERK. Previous studies have reported 5-HT2A receptor function by phosphorylating ERK1 and ERK2. The 5-HT2A antagonist ketanserin inhibited the activation of the ERK signaling pathway thus activating the downstream signaling cascade2. Therefore, similar to kentaserin’s mechanism of action pirenperone treatment might be able to rescue the ERK hyperactivation in FXS models as well, but further studies would be necessary.

Even though there are still many questions to comprehend the mechanism of pirenperone as a potential drug candidate for FXS, we assume that the current study will provide new insights into discovering effective drug candidates for the disorder and also the possibility that 5-HT2A modulators can be drug targets for FXS.

Materials and methods


Fmr1 KO mice were obtained from Jackson Laboratory (Sacramento, CA, USA). Fmr1 KO mice were mated for littermates and the vaginal plug was confirmed the following day of mating. Embryo 0 (E0) was designated if the vaginal plug was checked and the day of birth was considered postnatal day 1 (P1). Dams were housed separately until their pups were weaned (P21). At P14, the genomic DNA was extracted from each mouse for the genotyping, and PCR was performed to confirm their genotype. For primary cortical neuron culture, pregnant Sprague–Dawley (SD) rats were used (Orient Bio, Seoul, Korea). Animals were maintained on a standard 12:12 h circadian cycle with consistent temperature (20–24 °C) and humidity (30–70%). All animals were able to take food and water freely.

Animal handling, caring, anesthesia, and drug administration were carried out according to the Principle of Laboratory Animal Care (NIH publication No. 85-23, revised 1985). All processes were approved by the Institutional Animal Care and Use Committee of Konkuk University (KUIACUC), Korea (KU18054).


Poly-d-lysine (0.1 mg/ml) was obtained from Sigma (St. Louis, MO, USA) and Neurobasal medium (NBM), dimethyl sulfoxide (DMSO), 1% penicillin/streptomycin, and 1% l-glutamine were from Gibco BRL (Grand Island, NY, USA). The 2% B27 supplement was from Invitrogen (Carlsbad, CA, USA). Dulbecco's phosphate-buffered saline (DPBS) was purchased from Welgene (Gyeongsan, Korea). For quantitative real-time PCR, ABI7500 (Applied Biosystems) and SYBR® Premix Ex Taq II (2×) (RR820A, Takara, Japan) were used. For RNA extraction, the Trizol solution was obtained from Ambion (Carlsbad, CA, USA). Pirenperone (Cat No. P126, > 97% purity) was purchased from Sigma Aldrich (St. Louis, MO, USA).

Data collection

Gene expression omnibus's (GEO) gene expression profile GSE7329 was selected to analyze DEG. The dataset contained Agilent Genome Human Microarrays data from human lymphoblastoid cells which has the following profiles; 5 autistic males with 15q11-q13 duplication, 8 autistic males with FXS, and 15 normal control males2. The normalized data were downloaded through the GEO database, administrated by NCBI.

Data processing and CMap analysis

The mean gene expression difference between the FXS group (n = 11) and a healthy control group (n = 15) was derived. Two filtering criteria are sequentially applied to define DEGs for CMap input. Firstly, we select genes where their gene expression difference belongs to the top 5% (highly expressed in FXSs) or bottom 5% (lowly expressed in FXSs). Secondly, we performed a t-test for gene expression differences to select genes with a cut-off of the p value less than 0.05. Then, we matched the probe ID of the Agilent platform to the Affymetrix U133A probe ID. This step is necessary as CMap supports only Affymetrix probe ID for the query. Finally, we obtained 42 genes as DEGs to be used as a signature pattern for CMap analysis.

For probe ID mapping, we first map the Agilent probe ID to the official human gene symbol, then the human gene symbol is converted to Affymetrix probe ID by using the probe ID map table from For visualization of DEGs, a Multiple Experiment Viewer (MeV) was used. Statistical analyses were performed by R-3.4.3 ( For the interpretation of functional connections among selected DEGs, we used a web tool version of GeneMANIA (

CMap web interface provided by the Broad institute is used for our analysis ( Chemicals with negative enrichment scores meeting p value criteria less than 0.05 were selected. As we aim to find molecules to revert disease-related gene expression signatures, chemicals with negative enrichment scores were only considered. The whole analysis scheme is explained in Fig. 6.

Figure 6
figure 6

Schematic diagram of the transcriptome, CMap analysis, and therapeutic efficacy validation process.

Culture of rat cortical neurons

Primary cortical neurons were isolated from the embryonic day 18 (E18) brain of the SD rat as previously described2. Briefly, pregnant rats were anesthetized with 1% isoflurane, and the uterus was rapidly exposed to remove the fetus under sterile conditions. After the fetus was anesthetized and decapitated, the cortices were exposed and separated from brain tissue. By using mechanical trituration, cortices were divided into single cells and those cells were plated into poly-d-lysine pre-coated plates and incubated in NBM supplemented with 2% B27 supplement, 1% l-glutamine, and 1% penicillin–streptomycin in 2% CO2 incubator at 37 °C2. The density of cells was 6 × 105 cells per well.

Drug treatment (in vitro)

Pirenperone was diluted into 5, 10, and 20 μM with 0.2% DMSO and culture media for the treatment of cortical neuron cells. Treatment was conducted on the 8th day of in vitro culture of cortical neuron cells.

RNA extraction

After 24 h of the cortical neuron cell treatment of pirenperone, total RNA was isolated using Trizol reagent. The RNA extraction protocol was followed by the supplier's instructions. After the extraction, RNA concentration was confirmed by using NanoDrop ND-1000 Spectrometer (NanoDrop Technologies Inc., Montchanin, DE, USA). cDNA synthesis was achieved after normalization of all RNA concentrations into 1 μg.

Real-time qPCR

Quantitative real-time PCR was achieved with ABI7500 (Applied Biosystems) and results were analyzed by using the comparative threshold cycle (CT) method. Primer information is provided in Supplementary materials.

Drug treatment (in vivo)

50 μg/kg of pirenperone was diluted in 0.2% DMSO and saline for the vehicle was given via intraperitoneal (i.p.) injection to each mouse 30 min before the behavioral experiment.

Open field test

An exploratory hyperactivity test was performed in an open field box (40 × 40 × 30 cm). Each mouse was placed in the bottom center of an open field box and given 5 min of habituation followed by 20 min of behavior recording using EthoVision software (XT8.5)49. The velocity and the total distance moved were analyzed.

Elevated plus-maze

The elevated plus-maze was performed to test the impulsivity and anxiety-like behavior2. The maze was composed of two open and closed arms (67 × 7 cm). The shape of the whole maze was a plus symbol, each arm facing one another and connected through a neutral square-shaped space in the center. The maze was placed 55 cm above the floor. The mouse was put on the center (neutral area) and the test started. The mouse was allowed to move across arms freely for 8 min and the movement of the mouse was recorded by using EthoVision software (XT8.5). For each arm, the total time spent and the frequency of entry were analyzed.


Y-maze test was performed to assess the spatial working memory ability2. The maze has Y-shaped three arms (5 × 35 × 10 cm) and each corner has the same angles. Each mouse was put in the middle of the maze and was observed for their total entries and spontaneous alternations for 8 min. Spontaneous alternation was analyzed by using the following formula [total alternations/(total arm entries − 2) × 100]2.

Statistical analysis

Mean ± standard error of the mean (S.E.M) was used to express data for all behavior tests and qRT-PCR analysis. Unpaired t-test and one-way ANOVA followed by Bonferroni test were used to analyze for statistical significance. Statistical significance was considered when the p value was less than 0.05. All statistical analyses were calculated using GraphPad Prism version 5 software (

Ethics declarations

All methods were carried out in accordance with ARRIVE guidelines. All processes were approved by the Institutional Animal Care and Use Committee of Konkuk University (KUIACUC), Korea (KU18054).