Fragile X syndrome (FXS) is a trinucleotide repeat disorder associated with the loss of expression of fragile X mental retardation protein (FMRP) and subsequent intellectual disability, autism, hyperactivity and seizures (Hagerman and Hagerman, 2002). FMRP is an RNA binding protein that regulates the protein synthesis of numerous synaptic proteins, including amyloid-beta precursor protein (APP; Westmark and Malter, 2007; Lee et al., 2010). Altered levels of APP and its metabolites are found in Fmr1^KO mice, a mouse model that lacks expression of FMRP, as well as in subjects with autism and FXS (Westmark and Malter, 2007; Ray et al., 2011, 2016; Westmark et al., 2011). Genetic reduction of App in Fmr1^KO mice rescues behavioral, dendritic spine and electrophysiological phenotypes (Westmark et al., 2011, 2016). Furthermore, drugs that inhibit metabotropic glutamate receptor 5 (mGluR₅), which signals upstream of FMRP, reduce dendritic overexpression of APP in Fmr1^KO primary cultured neurons (Westmark et al., 2018) and seizures in mouse models that over-express human APP and amyloid-beta (Aβ; Westmark et al., 2009, 2010). Thus, APP is both a potential therapeutic target as well as a disease biomarker for FXS (Westmark, 2019). Glycogen synthase kinase 3 (GSK3) signals downstream of mGluR₅ and upstream of FMRP, and its activation affects all major hallmarks of Alzheimer’s disease and FXS (Figure 1). Herein, we sought to determine the efficacy of a novel GSK3 inhibitor developed by Angelini Pharma in attenuating in vivo (audiogenic-induced seizures, AGS) and in vitro (dendritic APP expression, spine length, ratio of mature dendritic spines to filopodia) phenotypes in Fmr1^KO mice.

Glycogen synthase kinase 3 is a constitutively active serine-threonine protein kinase that signals through numerous cellular signaling pathways including mGluR₅ (Portis et al., 2012). Phosphorylation of protein substrates by GSK-3 usually results in inhibition of those proteins. GSK3 has a bi-lobar structure including the N-terminus that binds to ATP and the C-terminus, or activation loop, that mediates the kinase activity. GSK3 activity is elevated in Fmr1^KO mouse brain where inhibition of mGluR₅ leads to inhibition of GSK3 indicating coordinated regulation of two key signaling proteins (Min et al., 2009). In 1996, Klein and Melton (1996) determined that the GSK3 inhibitor lithium rescues long-term memory in dfmr1 flies. Subsequent studies with lithium confirmed these findings in flies as well as demonstrated rescue of AGS, hyperactivity, social deficits, anxiety, learning and memory, cerebral protein synthesis, dendritic spine phenotypes, neurogenesis, and mGluR-LTD in Fmr1^KO mice (McBride et al., 2005; Min et al., 2009; Choi et al., 2010, 2011, 2016; Mines et al., 2010; Yuskaitis et al., 2010; Liu et al., 2011, 2012; Guo et al., 2012; Chen et al., 2013; King and Jope, 2013). Pharmaceutical GSK3 inhibitors also rescue a wide range of pathophysiological features in Fmr1^KO mice (Min et al., 2009; Luo et al., 2010; Mines et al., 2010; Guo et al., 2012; King and Jope, 2013; Chen et al., 2014; Franklin et al., 2014; King et al., 2014; Pardo et al., 2017; McCamphill et al., 2020). GSK3 inhibitors are under study for the treatment of numerous disorders, including Alzheimer’s disease, autism spectrum disorders and type 2 diabetes mellitus; and the preclinical data, in conjunction with a pilot clinical trial of lithium in FXS, suggest that GSK3 inhibitors are a potential treatment strategy for FXS (Berry-Kravis et al., 2008; Mines and Jope, 2011; Hagerman et al., 2012; Liu and Smith, 2014; Rizk et al., 2021).

Preclinical validation of novel GSK3 inhibitors is necessary to progress the most effective compounds into clinical trials (Roca and Campillo, 2020). Angelini Pharma identified novel inhibitors of GSK3 through a structure-based hit discovery process (Furlotti et al., 2015; Ombrato et al., 2015). The compounds were synthesized as described in the patent application WO2013124158A1 and evaluated for pharmacokinetic properties after intraperitoneal administration in C57BL/6J mice (Furlotti et al., 2015). Based on these data, this study compared the efficacy of the initial lead compound AFC03127 with the mGluR₅ inhibitor 2-Methyl-6-(phenylethynyl)pyridine hydrochloride (MPEP) and the GSK3 inhibitor SB216763 in Fmr1^KO mice with the goal of identifying a more efficacious compound indicated for FXS. SB216763 is a commercially available maleimide derivative with potent and selective cell permeable, ATP competitive inhibitor activity of GSK3 (Ki = 9 nM; Coghlan et al., 2000). AFC03127 is a 5-substituted-N-(piperidin-4-ylmethyl)-1H-indazole-3-carboxamide compound with reversible, linear, ATP-competitive GSK3 inhibitor activity (Ki = 15 nM; Furlotti et al., 2015). AFC03127 exhibits an IC₅₀ of 40 nM for GSK3α and an IC₅₀ of 18 nM for GSK3β in in vitro ATPase enzymatic assays as well as an AUC_o,inf of 370 ± 96 μg.min/mL, C_max of 2.1 ± 0.7 μg/mL, T_max of 15 min, t_1/2 of 203 ± 53 min, Vd of 6 ± 2 L/kg, and CI of 0.02 ± 0.01 L/min/kg by in vivo pharmacokinetic testing of plasma from mice dosed at 10 mg/kg I.P. with a C_max of 5.5 ± 0.4 ng/mg protein in brain (Furlotti et al., 2015). Additional preclinical GSK3 ATPase enzymatic assays indicate IC₅₀ values of 7.54 nM (SB216763, GSK3α), 21.3 nM (SB216763, GSK3β), 12 nM (AFC03127, GSK3α), and 12.3 nM (AFC032127, GSK3β; Angelini, unpublished data). AFC03127 is under development as a mood stabilizing drug and is expected to have superior activity to commercially available SB216763 based on high plasma stability and its potent ATP-competitive inhibitor activity (Furlotti et al., 2015; Capurro et al., 2020). AFC03127 is identified as 14i and AF3581, respectively, in the cited references. We hypothesize that AFC03127 will reduce AGS in Fmr1^KO mice as well as dendritic APP expression, spine length and the percent of filopodia in Fmr1^KO cells.

Inhibitors of GSK3 are under investigation for the treatment of numerous disorders including Alzheimer’s disease, type 2 diabetes mellitus and FXS. Herein, we compared the efficacy of a novel Angelini Pharma GSK3 inhibitor (AFC03127) with established inhibitors of mGluR₅ (MPEP) and GSK3 (SB216763; Coghlan et al., 2000; Cross et al., 2001; Ougolkov et al., 2005; Rancillac et al., 2006; Gurrieri et al., 2010) side-by-side in both in vivo and in vitro assays in Fmr1^KO mice and cells, respectively. In vivo testing included AGS susceptibility, the gold standard test for drug efficacy evaluation in Fmr1^KO mice. Both MPEP and SB216763 significantly reduced wild running and seizures in Fmr1^KO mice, but the novel Angelini Pharma GSK3 inhibitor did not (Figure 2). The lower dose of AFC03127 was selected based on Angelini behavior data in Fmr1^KO mice where chronic treatment for 10–14 days with AFC03127 at 10 mg/kg significantly reduced marble burying (Angelini, personal communication), total distance traveled in the open field and startle response (Porceddu et al., 2021), but an acute 30 mg/kg treatment that significantly reduced total distance traveled in the open field appeared to cause sedation (Angelini, personal communication). AF03127 was tested at 30 mg/kg in Fmr1^KO mice in the AGS assay and exhibited a 29% seizure rate (n = 17) compared to 32% after treatment with 10 mg/kg (n = 22; data not shown, testing was with a separate set of control mice). Male and female mice were tested for all treatments; sex-specific effects were not observed.

The in vitro assays included dose response assessments of SB216763 and AFC03127 efficacy in reducing dendritic APP expression in Fmr1^KO primary neurons. There was a 40% reduction in dendritic APP staining in response to 25 μM MPEP (Figure 3). Testing a 100,000-fold concentration range, there was a statistically non-significant dose response decrease in dendritic APP staining in response to SB216763 with equivalent staining to MPEP at the highest dose of 250 μM SB216763 (Figure 3). A bell-shaped curve was observed in response to AFC03127 treatment such that APP staining was equivalent or lower with 25 nM–25 μM AFC03127 compared to MPEP (Figure 3 and Supplementary Table 1). The lowest effective in vitro dose was 25 nM AFC03127. The highest dose of AFC03127 (250 μM) did not reduce APP levels likely because the drug visibly precipitated out of solution forming a fine suspension, which may have settled on the cells in a localized high concentration causing cell death as evidenced by the shorter dendrites and “hot spots” of stained clumped puncta (Supplementary Figure 1). A greater than 60% reduction in dendritic APP was observed with 250 nM AFC03127 compared to prior studies demonstrating a maximal reduction of 50% APP with 250 nM AFQ-056, an mGluR₅ inhibitor (Westmark et al., 2018).

The in vitro assays also included testing SB216763 and AFC03127 efficacy in rescuing dendritic spine length and density and percent filopodia in Fmr1^KO primary neurons at the optimal doses determined in the dendritic APP staining study. A single concentration of each drug was tested at 3 time points (5, 15, and 75 min). MPEP (25 μM), SB216763 (25 μM), and AFC03127 (250 nM) significantly reduced dendrite length compared to vehicle (Figure 4, Supplementary Figure 2, and Supplementary Table 2). The control Fmr1^KO cells exhibited an average spine length of 1.6 μm as expected, which is about two-fold longer than spines from wild type neurons (Westmark et al., 2011). With the 5-min treatment, MPEP and SB216763 reduced dendrite length to a greater extent than AFC03127. With the longer 15- and 75-min treatments, all drugs significantly reduced dendrite length compared to vehicle; however, the cells treated for longer times with GSK3 inhibitors were starting to die as evidenced by shrinking dendrites (data not shown). Similar to dendrite length, the percent filopodia were significantly reduced with MPEP, SB21673, and AFC03127 compared to vehicle (Figure 4). Spine density was significantly different by ANOVA for the 5-min treated samples (vehicle, MPEP, SB216763, and AFC03127), but post hoc tests did not show differences between either GSK3 inhibitor or the vehicle (Figure 4). Spine density was not significantly different by ANOVA for the 15- and 75-min treatments.

This study compared the efficacy of a novel Angelini Pharma GSK3 inhibitor AFC03127 with MPEP and SB216763 in the Fmr1^KO mouse model. The Fmr1^KO mice are the most widely employed animal model used for the study of FXS. They have good face validity in terms of exhibiting many of the physical and behavioral characteristics of human patients such as lower seizure threshold and abnormal dendritic spine morphology, even if they may be over predictive in terms of translating drugs to the clinic (Berry-Kravis et al., 2018). MPEP is a research grade mGluR₅ antagonist that reduces AGS, anxiety and dendritic spine protrusion phenotypes in Fmr1^KO mice (Yan et al., 2005; de Vrij et al., 2008). SB216763 is a potent maleimide compound with ATP-competitive inhibitor activity against both GSK3α and GSK3β (Coghlan et al., 2000; Cross et al., 2001; Ougolkov et al., 2005; Gurrieri et al., 2010). AFC03127 was compared to MPEP and SB216763 in both in vivo and in vitro assays. The assays under evaluation included AGS, dendritic APP expression and dendritic spine measurements. These assays were chosen because they were previously shown to be altered in the Fmr1^KO mouse and responsive to treatment. Specifically, genetic and pharmaceutical modulation of APP levels rescued AGS, spine length and the percent filopodia (Westmark et al., 2011, 2018).

SB216763 (30 mg/kg) rescued the AGS phenotype in Fmr1^KO mice as well as MPEP, which corroborates prior findings (Min et al., 2009). AFC03127 did not rescue the AGS phenotype nor did tideglusib (200 mg/kg oral gavage; data not shown), which is an irreversible, non-ATP competitive GSK-3β inhibitor (Dominguez et al., 2012). AFC03127 significantly reduced dendritic APP expression over a wide concentration range whereas SB216763 only approached statistical significance at the highest dose. Both SB216763 and AFC03127 significantly rescued spine length and the percent filopodia in cultured Fmr1^KO neurons. SB216763 and AFC03217 equally inhibit both GSK3α and GSK3β (Coghlan et al., 2000; Furlotti et al., 2015, Angelini, personal communication). Inhibition of GSK3α but not GSK3β corrected excessive protein synthesis, mGluR-LTD, AGS, sensory cortex hyperexcitability, learning and memory, and enhanced psychomimetic-induced hyperlocomotion in Fmr1^KO mice (McCamphill et al., 2020). GSK3 mediates phosphorylation of the cytoplasmic domain of APP and processing of APP to amyloid-beta (Aβ; Aplin et al., 1996; Phiel et al., 2003; Su et al., 2004). Chronic GSK3 inhibition with lithium (non-selective), BRD0705 (α-selective), or BRD3731 (β-selective) reduced phosphorylation of APP but did not affect total APP levels (McCamphill et al., 2020). GSK3β activity increases the number of thin spines whereas lack of GSK3β increases the number of stubby spines (Kondratiuk et al., 2017). SB216763 and AFC03217 are not paralog selective inhibitors of GSK3α, albeit paralog-specific effects may underlie the varied AGS and dendritic findings.

The lowest effective dose of AFC03127 in our in vitro study was 1,000-fold lower than that for SB216763. We observed a bell-shaped dose response effect with AFC03127, in the reduction of dendritic APP levels, that could be due to a solubility issue or toxicity at higher concentrations of the drug. It is also feasible that there is a true bell-shaped dose response as both increased and decreased expression of Drosophila GSK3 impair larval crawling in the fly model of FXS (Jakubowski et al., 2012), suggesting that complex biological effects are associated with this enzyme. It is of interest that SB216763 but not AFC03127 attenuated AGS while AFC03127 but not SB216763 reduced dendritic APP expression and both drugs affected dendritic spine phenotypes. Recent work demonstrates that Fmr1 deletion in glutamatergic neurons in the inferior colliculus is necessary to cause AGS (Gonzalez et al., 2019). The effects of GSK3 inhibitors in the inferior colliculus remain to be determined.

Drug development is expensive. and it is difficult to attract pharmaceutical companies to the study of rare disorders. Thus, when drugs can be repurposed for multiple disorders, it benefits both industry and the affected families. It is important to identify and test novel GSK3 inhibitors as a potential therapeutic strategy for FXS. Lithium has been widely used in the clinic and demonstrated early efficacy in preclinical models of FXS; however, this drug can have off-target effects and issues with tolerability in humans, while cessation of treatment leads to reemergence of phenotypes in Fmr1^KO mice (King and Jope, 2013). Herein, we provide preclinical validation data comparing the efficacy of AFC03127 with MPEP and SB216763 in Fmr1^KO mice. A limitation of the work is that there is not dose response data for the seizure study due to the large number of mice required and the decision to screen multiple compounds versus multiple doses. The data contribute to a rapidly growing body of preclinical data regarding the efficacy of GSK3 inhibitors in the rescue of Fmr1^KO phenotypes. In addition, GSK3 activation affects all the major hallmarks of Alzheimer’s disease (Cai et al., 2012; Lauretti et al., 2020). Thus, the development and validation of novel GSK3 inhibitors may benefit multiple CNS disorders. It will be of interest to study clinically relevant behavioral alterations in mouse models of FXS and Alzheimer’s disease in response to AFC03127 in future studies, particularly considering the pivotal role of GSK3 in APP biology and the wide range of phenotypes rescued in Fmr1^KO mice in response to GSK3 inhibitors (Rizk et al., 2021; Figure 1).

In conclusion, inhibitors of GSK3 are under investigation for the treatment of a wide range of CNS disorders. Regarding FXS, GSK3 inhibitors have shown promise in fly, mouse and human models. Herein, we compared the efficacy of a novel inhibitor of GSK3 developed by Angelini Pharma, AFC03127, with SB216763. SB216763 was effective in attenuating seizures in vivo in a mouse model useful for the study of FXS. AFC03127 were effective at attenuating APP biomarker expression and both drugs were effective in attenuating dendritic spine phenotypes in in vitro assays. AFC03127 was more effective than MPEP or SB216763 in reducing dendritic APP expression. These findings support further preclinical study of this family of GSK3 inhibitors for the treatment of FXS, Alzheimer’s disease and other APP/Aβ-related disorders.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by University of Wisconsin Madison Institutional Animal Care and Use Committee (IACUC).

CW and PW conceived and designed the experiments, acquired data, and interpreted data. BG provided the test drugs. CW drafted the manuscript. CW, BG, PW, RO, CM, and FD edited and approved the manuscript. All authors agreed to be accountable for the work.

BG, RO, CM, and FD are employees of Angelini Pharma S.p.A. Angelini Pharma S.p.A. played no role in the design of the study or the collection and analysis of the data. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.