Fragile X syndrome (FXS) is characterized by synaptic immaturity, cognitive impairment, and behavioral changes. The disorder is caused by transcriptional shutdown in neurons of the fragile X mental retardation 1 gene product, fragile X mental retardation protein. Fragile X mental retardation protein is a repressor of dendritic mRNA translation and its silencing leads to dysregulation of synaptically driven protein synthesis and impairments of intellect, cognition, and behavior, and FXS is a disorder that currently has no effective therapeutics. Here, young fragile X mice were treated with chronic bryostatin-1, a relatively selective protein kinase Cε activator, which induces synaptogenesis and synaptic maturation/repair. Chronic treatment with bryostatin-1 rescues young fragile X mice from the disorder phenotypes, including normalization of most FXS abnormalities in 1) hippocampal brain-derived neurotrophic factor expression, 2) postsynaptic density-95 levels, 3) transformation of immature dendritic spines to mature synapses, 4) densities of the presynaptic and postsynaptic membranes, and 5) spatial learning and memory. The therapeutic effects were achieved without downregulation of metabotropic glutamate receptor (mGluR) 5 in the hippocampus and are more dramatic than those of a late-onset treatment in adult fragile X mice. mGluR5 expression was in fact lower in fragile X mice and its expression was restored with the bryostatin-1 treatment. Our results show that synaptic and cognitive function of young FXS mice can be normalized through pharmacological treatment without downregulation of mGluR5 and that bryostatin-1–like agents may represent a novel class of drugs to treat fragile X mental retardation at a young age and in adults.
Fragile X syndrome (FXS), the most common form of inherited intellectual disability (Santoro et al., 2012), is characterized by synaptic immaturity, cognitive deficits (Koekkoek et al., 2005), and autistic-like behavior (Sabaratnam et al., 2003). The disorder is typically caused by an expansion of an untranslated CGG repeat in the 5′ untranslated region of the X-linked gene fragile X mental retardation 1 (FMR1; Verkerk et al., 1991). This triplet expansion leads to DNA methylation of FMR1 and transcriptional shutdown of the gene. The loss of the fragile X mental retardation protein (FMRP), an RNA-binding protein, causes dysregulation of the translation of dendritic mRNAs (Ashley et al., 1993; Darnell and Klann, 2013). FMRP represses translation probably by interacting with a specific subset of mRNAs, directly binding to the ribosome with high affinity, and thereby precluding the binding of tRNA and translation elongation factors on the ribosome (Chen et al., 2014). The key functional role of the binding of FMRP to RNA is supported by the evidence that a missense mutation in the KH2 domain (Ile304Asn) of human FMRP abolishes its binding to polyribosome and leads to a severe form of FXS (Siomi et al., 1994).
Signal processing at synapses is dramatically altered due to the lack of FMRP, resulting in an impaired ability in synaptogenesis, synaptic maturation, and synaptic plasticity to meet cognitive demands (Nelson and Alkon, 2015) and behavioral control. The most compelling is the evidence involving the metabotropic glutamate receptor (mGluR) 5 (Bhattacharya and Klann, 2012), based on the observation of an abnormally enhanced mGluR5-dependent long-term depression in fragile X mice (Huber et al., 2002). The leading mGluR theory of FXS indicates that overactive mGluR5 signaling, normally balanced by FMRP, underlies much of the brain pathology of FXS (Santoro et al., 2012; Hajós, 2014). Drug development for the treatment of FXS thus has been focused on achieving this balance by reducing mGluR5 hyperactivity in the brain. Consistently, genetic reduction of mGluR5 expression or pharmacologic mGluR5 antagonism has been reported to correct many FXS phenotypes in fragile X mice (Huber et al., 2002; Hajós, 2014).
We have found that chronic activation of protein kinase C (PKC) ε in the hippocampus can rescue synapses and spatial cognition, as well as other FXS phenotypes, in adult fragile X mice after the disorder has established (Sun et al., 2014). The downside of late-onset treatment is that the therapy may miss a critical period of youth, important in age-related sociocognitive and behavioral development. It is also not clear whether the observed therapeutic effects involve downregulation of the mGluR5 in the hippocampus. If the syndrome is a lasting consequence of brain development with deficit in synaptic maturation, it is possible that early intervention with agents that facilitate synaptic maturation could achieve a better outcome, unfolding full potential of therapeutics. This is highly achievable, considering the evidence that newborn screening for FXS is technically feasible (Tassone, 2014). In this study, we, therefore, evaluate therapeutic potential of a PKCε activator on synapses, cognitive function, and other FXS phenotypes, including mGluR5 expression in the hippocampus of young fragile X mice.
Materials and Methods
Animals and Drug Treatment.
Two types of male mice (9 to 10 per group; at an age of close to 3 weeks when obtained; Jackson Laboratories, Bar Harbor, ME) were used: FVB.129P2-Pde6b+ Tyrc-ch Fmr1tm1Cgr/J (fragile X mice) and FVB.129P2-Pde6b+ Tyrc-ch/AntJ (age-matched wild-type controls). These mice do not suffer from blindness. They were housed in a temperature-controlled (20°C–24°C) room, allowed free access to food and water, and kept on a 12-hour light/dark cycle.
All mice were randomly assigned to different groups. Bryostatin-1 was administered (20 μg/m2, tail i.v., two doses per week for 6 weeks), starting at an age of close to 4 weeks (26 to 27 days). The treatment would catch the last week or so in brain development (a month in a mouse with a lifespan of 2 years is roughly equal to 3.3 years of brain development in humans; Sengupta, 2013) before FXS phenotypes are established in fragile X mice (Bhattacharya and Klann, 2012). Although synaptogenesis and synaptic maturation occur throughout a mammalian lifespan, mouse brain development is anatomically complete at approximately 5 weeks of age, when all FXS phenotypes are established in fragile X mice (Bhattacharya and Klann, 2012; Michalon et al., 2012). The age would roughly correspond to mid-adolescence (approximately 15 years of age) in humans in term of synaptic maturation in the prefrontal cortex (Huttenlocher and Dabholkar, 1997). The same behavioral training procedures were performed in all animal groups in this study, so that differences in observed results would indeed reflect effects of drug treatments. The dose was based on our preliminary dose-response studies that smaller doses were not effective against disorder-induced synaptic and cognitive impairments. Nontreated groups received the same volume of vehicle at the same frequency. Synaptic and memory functions and other phenotypic features were evaluated 9 days after the last dose.
Tissue Preparation for Confocal Microscopy.
Animals were deeply anesthetized with sodium pentobarbital (120 mg/kg, i.p.) and perfused through the heart with phosphate-buffered saline (PBS) for < 4 minutes at room temperature and then with 4% paraformaldehyde in PBS (approximately 40 ml) to remove any negative effects of hypothermia on dendritic spines (Kirov et al., 2004). Brains were then kept in 4% paraformaldehyde in PBS for 10 minutes at 4°C and stored in PBS at 4°C. Right dorsal hippocampi were sectioned with a vibratome (Leica VT1200S; Leica Microsystems, Wetzlar, Germany) into 200-μm thickness and kept in series in PBC at 4°C for three-dimensional reconstruction.
Three-Dimensional Reconstruction Analysis of Dendritic Spine Morphology on Dendritic Shafts.
The tips of glass electrodes, prepared for electrophysiological experiments, were immersed for 10 seconds in 5% (w/v) 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) in dichloromethane (Sigma-Aldrich, St. Louis, MO) and air dried at room temperature for 1 hour (twice). The tips of DiI-coated electrodes were inserted, broken, and left in the strata oriens of the CA1 area of hippocampal sections at 200-μm thickness (three to four electrode tips per slice). After maintenance in PBS at 4°C overnight, the hippocampal sections were resectioned to 7-μm thickness and kept in PBS at 4°C. In the case that DiI was not well diffused along dendritic shafts, the 7-μm thickness sections in PBS were changed from 4°C to room temperature. At each time of confocal scanning, only one section was freshly mounted on a silane-coated microscope slide (Silanated slides; KD Medical, Columbia, MD), using 4% paraformaldehyde in PBS as a mounting medium.
Dendrites were imaged by a Zeiss Axio Observer Z1 microscope equipped with a 710-confocal scanning microscopy (>510 nm/568 nm excitation/emission) using a 100× Plan-APO Chromat oil immersion objective (1.4 numerical aperture; Carl Zeiss, Jena, Germany). A series of randomized confocal images (1024 × 1024 pixels) were confocally scanned at every 0.145 μm. The image resolution was according to Nyquist sampling, and the pixel size was 48, 48, and 145 (x, y, and z in nanometers, respectively). Using the ImageJ plugin Deconvolution Laboratory (Biomedical Imaging Group, EPFL; Lausanne, Switzerland), stacks of confocal images were deconvoluted by the Tikhonov-Miller algorithm. A confocal stack of point spread function, used for deconvolution, was prepared from Fluoro-max dyed red aqueous fluorescent beads [63-nm diameter, 1% (w/v) stock solution; Thermo Fisher Scientific, Fremont, CA], further diluted to 1:10,000, smeared, and air dried on a microscopic slide coated with 0.01% poly(l-lysine).
Dendritic spines were automatically detected and counted with the NeuronStudio (beta) program (Rodriguez et al., 2008; http://research.mssm.edu/cnic/tools-ns.html). Dendritic projections from dendritic shafts at 0.2–3 μm in length were classified as dendritic spines. Dendritic spines with a head-to-neck diameter ratio (neck ratio) greater than 1.1 were either considered as thin or mushroom-shaped spines (see below). Spines that did not meet the neck ratio value (head-to-neck diameter ratio of greater than 1.1) and had a length of spine to head diameter (thin ratio) above 2.5 were automatically classified as thin; otherwise, they were classified as stubby. Spines that met the neck ratio value and had a head diameter (mushroom size) of 600 nm or greater (Sorra and Harris, 2000) were automatically labeled as mushroom shaped; otherwise, they were labeled as stubby. Filopodia were identified manually (long profiles without enlarged heads) from the projection structures that was longer than dendritic spines and did not have enlarged heads.
Although histochemical levels of postsynaptic density (PSD)-95 or synaptophysin have been found to correspond well with those detected at the biochemical levels (Glantz et al., 2007; Horling et al., 2015), Western blots measure the total protein levels in the tissue and may have results that are different from histochemistry at a single structure or cell level, owing to protein expression or transport to a specific structure (Hongpaisan et al., 2013). Our study, however, focused on the structure changes, and we thus used immunohistochemistry to confirm the morphologic studies using electron microscopy (see below) and confocal microscopy of DiI staining.
Hippocampal sections at 200-μm thickness were further fixed in 4% paraformaldehyde in PBS for 2 to 3 days at 4°C and resectioned into 5-μm thickness. Sections at 5-μm thickness of the right dorsal hippocampus (one section every 400 μm; four serial sections from each animal) were processed as described below. Hippocampal slices were incubated free floating with Image-iTTM FX signal enhancer (Thermo Fisher Scientific/Life Technologies, Grand Island, NY) for 30 minutes at room temperature. Sections were incubated overnight at room temperature with the following primary antibodies (Hongpaisan and Alkon, 2007): mouse monoclonal anti-synaptophysin (1:2000; EMD Millipore, Billerica, MA), rabbit polyclonal anti-spinophilin (1:100; EMD Millipore), mouse monoclonal anti–growth-associated protein (GAP)-43/B-50 (1:2000; EMD Millipore), mouse monoclonal anti–PSD-95 (1:100; Santa Cruz Biotechnology, Santa Cruz, CA), mouse monoclonal anti–brain-derived neurotrophic factor (BDNF) (1:50; Santa Cruz Biotechnology), and/or rabbit monoclonal anti-mGluR5 (1:100; Abcam, Cambridge, MA). Tissue sections were then incubated with either Alexa Fluor 568 secondary antibody (1:200; Thermo Fisher Scientific) and/or a biotinylated secondary antibody (1:20; Vector Laboratories, Burlingame, CA) for 3 hours at room temperature and then with streptavidin-conjugated Alexa Fluor 488 (1:100; Thermo Fisher Scientific) for 3 hours at room temperature. Sections were mounted with VECTASHIELD mounting medium with 4′,6-diamidino-2-phenylindole to counterstain nuclei (Vector Laboratories).
The random area in the hippocampal CA1 stratum radiatum that appeared immediately after switching to a higher-magnification lens (63× Plan-APO Chromat oil immersion objectives; 1.4 numerical aperture) was imaged for appropriate fluorescence (e.g., Alexa Fluor 488 and/or 568). Confocal images of hippocampal sections were acquired in line-scan mode and with a pinhole of approximately 1.00 airy unit. Confocal images with similar levels of 4′,6-diamidino-2-phenylindole fluorescence intensity among experimental conditions were quantified with the ImageJ program (http://rsb.info.nih.gov/ij/; National Institutes of Health, Bethesda, MD), with control data being set at 100%.
Quantification for Immunostained Density of Pre- or Postsynaptic Structure.
Confocal images of the density of postsynaptic dendritic spines, presynaptic axonal terminals, postsynaptic membranes, or presynaptic membranes per 45 × 45 × 0.6 μm3 volume of the CA1 stratum radiatum were analyzed as previously described (Hongpaisan et al., 2011).
Under anesthesia with pentobarbiturate, mice were perfused through the heart with PBS (5 ml) and 2% glutaraldehyde and 3% paraformaldehyde in PBS (80 ml). Brains were removed and stored in fixative at 4°C. Hippocampal slices were processed for Epon embedding, resectioned to 90-nm thickness, collected on a grid, and stained with uranyl acetate and lead citrate. Electron micrographs were taken of the middle of each section with a JEOL 1010 transmission electron microscope (JEOL Inc., Peabody, MA). Random sampling was achieved by orienting the hippocampal CA1 area under low-power magnification. The random area that immediately appeared after switching to a higher magnification (7000× magnification) was imaged with a charge-coupled device camera. Electron micrographs were quantified on the Preview program on a MacPro 4.1 computer with a 30-inch monitor (Apple, Cupertino, CA). Electron micrographs were collected within 14 × 14 × 0.09 μm3 of the CA1 stratum radiatum. We defined axospinous synapses as those located between dendritic spine structures that do not contain mitochondria and with axon boutons containing presynaptic vesicles.
Spatial Learning and Memory and Visible Platform Test.
A modified Morris water maze task (two trials per day for 8 days), a difficult task for revealing mild impairments, was used to evaluate spatial learning and memory.
Water maze training sessions began on the ninth day after the last dose of bryostatin-1, a time gap to separate potential acute effects from the chronic therapeutic impacts. The maze pool had a diameter of 114 cm and height of 60 cm and was filled with 40 cm H2O (22 ± 1°C), mixed with 200 ml nontoxic white tempera (BesTemp; Certified Color Corp., Orange, CA). Mice were trained to find a hidden platform, centered in one of the quadrants and submerged 2 cm below the water surface. At the start of each trial, mice were placed individually in the water facing the maze wall, using different starting positions each trial, and were allowed to swim until they found the platform, where they remained for 20 seconds. If a mouse failed to find the platform within 1.5 minutes, it would be guided there by the investigator, with 90 seconds scored. The swim path was recorded with a video-tracking system. After the training trials, a probe trial was given 24 hours after the last training trial, with the platform removed to assess memory retention for its location by the distance the mouse moved in the quadrants. The video-tracking system tracked the animal’s movements in each quadrant (1 minute).
The sensorimotor ability of mice was evaluated with a visible platform test. The platform was placed at a new location and was marked with a pole that protruded 9 inches above the water surface.
Statistical analysis was performed using analysis of variance (ANOVA), followed by the Newman–Keuls multiple comparisons test. P < 0.05 was considered statistically significant. All procedures were conducted according to the National Institutes of Health Animal Care and Use Committee guidelines and were approved by the Ethical Committee of West Virginia University.
Bryostatin-1 Improves Levels of PSD-95 and Metabotropic Glutamate Receptors in Young Fragile X Mice.
Significant differences among experiment groups for PSD-95 membranes, primarily accumulated in the postsynaptic membranes (F3,222 = 6.121, P < 0.001; Fig. 1, A and B), were observed. In wild-type mice, bryostain-1 significantly increased postsynaptic membranes (P < 0.01; Fig. 1B). Fragile X mice showed significant decreases in the densities of postsynaptic membranes (P < 0.05; Fig. 1B). Bryostatin-1 improved the formation of postsynaptic membranes (P < 0.001; Fig. 1B) in fragile X mice. The data using immunohistochemistry confirmed the results from electron microscopy analysis (see below).
Accumulation of mGluR5 into postsynaptic densities was examined with the colocalization of mGluR5 and PSD-95, using immunohistochemistry and confocal microscopy (Fig. 1A). Significant differences among the groups of wild-type mice (WC), wild-type mice with bryostatin-1 (WB), fragile X mice (TC), and fragile X mice with bryostatin-1 (TB) were observed for the density of mGluR5-containing postsynaptic membranes (colocalization area of mGluR5 and PSD-95; ANOVA, F1,187 = 7.814, P < 0.001; Fig. 1C) and mGluR5 level in postsynaptic membranes (mGluR5 fluorescence intensity in the colocalization area; F1,187 = 3.478, P < 0.01; Fig. 1D). Data with significant difference with ANOVA were subsequently analyzed with a two-tailed paired t test. Bryostatin-1 had no effect in wild-type mice, whereas decreases in the number of mGluR5-containing postsynaptic membranes (P < 0.01; Fig. 1C) and their mGluR5 level (P < 0.05; Fig. 1D) were significantly improved with bryostatin-1 in transgenic mice (P < 0.01; Fig. 1, C and D). Bryostatin-1 also enhanced the density of mGluR5-containing postsynaptic membranes in transgenic mice, compared with wild-type controls (Fig. 1C, compare TB with WC).
Bryostatin-1 Prevents the Loss of BDNF in Young Fragile X Mice.
Immunohistochemistry of BDNF in the CA1 stratum radiatum, imaged with a confocal microscope, showed a significant difference among experimental groups (F1,120 = 27.484, P < 0.001; Fig. 1E). Bryostatin-1 did not increase BDNF expression in wild-type mice (Fig. 1F). However, in transgenic mice, the reduction of BDNF (P < 0.001) was significantly improved (P < 0.001) and was even elevated above wild-type levels (Fig. 1E, compare TB with WC; P < 0.05) with bryostatin-1.
Bryostatin-1 Increases Presynaptic Vesicle Concentration within Presynaptic Boutons.
Densities of presynaptic axonal boutons and their presynaptic vesicle content in the hippocampal CA1 stratum radiatum were evaluated with confocal immunohistochemistry (Fig. 2A). ANOVA showed no significant difference among experimental groups for presynaptic axonal bouton density, as determined by counting the grains of presynaptic vesicle protein synaptophysin (Fig. 2B). However, significant differences among experiment groups were seen for presynaptic vesicle concentration within the presynaptic axonal boutons, as determined with the fluorescent intensity of synaptophysin (F3,142 = 2.726, P < 0.05; Fig. 2C). Although bryostatin-1 had no effect on presynaptic bouton density in all experimental conditions (Fig. 2B), bryostatin-1 increased presynaptic vesicle concentration within the presynaptic boutons both in wild-type (P < 0.05) and fragile X (P < 0.01) mice (Fig. 2C).
Electron microscopy was further used to determine the concentration of presynaptic vesicles at a single axonal bouton level (Fig. 2E). There was a significant difference among experimental groups (F3,190 = 8.018, P < 0.001; Fig. 2D). Electron microscopy confirmed the results of confocal immunohistochemistry that no significant change in presynaptic vesicles level was observed in fragile X mice (Fig. 2D, compare TC with WC), and bryostyatin-1 significantly increased the presynaptic vesicle level in both wild-type and fragile X mice (Fig. 2D, WB versus WC, and TB versus TC).
Bryostatin-1 Increases Synaptic Densities.
Density of axodendritic synapses in the hippocampal CA1 stratum radiatum was determined with electron microscopy (Fig. 3A). There was a significant difference among experimental groups (F3,195 = 6.426, P < 0.001; Fig. 3B). In wild-type mice, bryostatin-1 significantly increased synaptic density (P < 0.05; Fig. 3B). In fragile X mice, synaptic loss (P < 0.05) was significantly reversed (P < 0.001) with bryostatin-1 treatment (Fig. 3B).
Changes in presynaptic membranes in the hippocampal CA1 stratum radiatum were further studied, using confocal immunohistochemistry (Fig. 3C). Significant differences among experiment groups for GAP-43, predominantly located in presynaptic membranes (F3,117 = 13.149, P < 0.001; Fig. 3D), were observed. In wild-type mice, bryostain-1 significantly increased presynaptic membranes (P < 0.05; Fig. 3D). Fragile X mice showed significant decreases in the densities of presynaptic membranes (P < 0.001; Fig. 3D). Bryostatin-1 improved the formation of presynaptic membranes (P < 0.05; Fig. 3D). The data using immunohistochemistry confirmed the results from electron microscopy (see below).
Bryostatin-1 Increases Mushroom-Shaped Dendritic Spine Formation.
Change in the dendritic spine number at a single apical dendrite level in the hippocampal CA1 pyramidal neuron was studied with DiI staining and imaged with a confocal microscopy (Fig. 3E). A significant difference among experimental groups for mushroom dendritic spines was seen (F3,142 = 7.186, P < 0.001; Fig. 3F).
In wild-type mice, bryostatin-1 significantly elevated the number of mushroom spines per 100-μm dendritic shaft above wild-type levels (P < 0.05; Fig. 3F, compare WB with WC), correlated with the enhancement of memory retention after water maze training. In fragile X mice, a significant decrease in mushroom dendritic spines (P < 0.01; Fig. 3F, compare TC with WC) was significantly improved with bryostatin-1 treatment (P < 0.001; Fig. 3F, compare TB with TC). The number of mushroom spines in fragile X mice treated with bryostatin-1 was not different from that in wild-type mice treated with bryostatin-1 (Fig. 3F, compare TB with WB).
Bryostatin-1 Improves the Maturation of Dendritic Spines in Young Fragile X Mice.
Significance differences among experimental groups were found for immature spines or filopodia (F3,142 = 24.386, P < 0.001; Fig. 3G) and mature dendritic spines (mushroom plus thin plus stubby spines; F3,142 = 3.956, P < 0.01; Fig. 3H) were seen at a single apical dendritic shaft of the CA1 pyramidal neuron. In wild-type mice, bryostatin-1 treatment did not change the numbers of filopodia and mature dendritic spines (Fig. 3, G and H). In fragile X mice, an increased number of immature spines (P < 0.001; Fig. 2H) and a decreased number of mushroom plus thin plus stubby spines (P < 0.05; Fig. 3H) was observed compared with wild-type mice. These changes were reversed to the wild-type control level with bryostatin-1 treatment.
Changes in overall dendritic spine density (45 × 45 × 0.6 μm3 volume of stratum radiatum), a hippocampal part that correlated with the apical dendrites of CA1 pyramidal neurons, were studied with immunohistochemistry of the dendritic spine–specific protein spinophilin (Fig. 3I). A significant difference among experiment groups was seen (F3, 154 = 4.748, P < 0.01). Bryostatin-1 had no effect on dendritic spine density in wild-type mice but improved the loss of dendritic spine density in fragile X mice (P < 0.05; Fig. 3J). These data on the densities of dendritic spines per brain volume units were correlated with changes in the number of dendritic spines on dendritic spine shafts studied with DiI staining (compare Fig. 3, G and H).
Taken together, the results suggest that bryostatin-1 improves the maturation of dendritic spines from the immature spines in the young fragile X mice.
Bryostatin-1 Prevents Deficits in Spatial Learning and Memory of Young Fragile X Mice But Did Not Alter Sensorimotor Ability.
There were significant learning differences among the four groups (F3,560 = 58.528, P < 0.001; Fig. 4A), indicating different learning among the groups. Spatial learning was significantly impaired in fragile X mice (fragile X mice with vehicle versus wild-type with vehicle: F1,272 = 22.117, P < 0.001). Bryostatin-1 treatment significantly improved the learning performance of fragile X mice (fragile X mice with vehicle versus fragile X mice with bryostatin-1: F1,272 = 110.621, P < 0.001) to the level of the control mice with bryostatin-1 treatment (wild-type with bryostatin-1 versus fragile X mice with bryostatin-1: F1,272 = 0.669, P > 0.05), indicating that the chronic bryostatin-1 treatment not only improved spatial learning in the wild-type mice but also repaired the learning deficits of fragile X mice when tested at an adult age after an early 6-week treatment.
The results of the probe test (Fig. 4, B–E) were analyzed using the target quadrant ratio (dividing the target quadrant distance by the average of the nontarget quadrant values during the probe test; Fig. 4F). There were significant differences in the target quadrant ratio among the groups (F3,38 = 8.790, P < 0.001), indicating differences in the spatial memory among the groups. Detailed analysis reveals that the memory recall of the transgenic mice was impaired (fragile X mice with vehicle versus wild-type with vehicle: F1, 18 = 5.155, P < 0.05) and that bryostatin-1 treatment in the transgenic mice significantly improved the memory recall, compared with that of fragile X mice without the treatment (F1,17 = 21.145, P < 0.001), to the level of the control mice with bryostatin-1 (control mice with bryostatin-1 versus fragile X mice with bryostatin-1: F1,17 = 0.063, P > 0.05).
A visible platform test, determined after the probe test, revealed no significant differences among the groups (F3,35 = 0.179, P > 0.05; Fig. 4G), indicating that there were no significant group differences in sensorimotor ability and escape motivation of the mice among different groups. Therefore, the differences in learning and memory-recall performance among the groups cannot be attributed to the differences in their sensorimotor ability and escape motivation.
The main findings of this study of an early-onset treatment in young adult mice can be summarized as follows. First, the loss of FMR1 protein in fragile X mice suppresses BDNF expression, dendritic spine and synaptic maturation, PSD-95 and mGluR5 accumulation, and memory-dependent mushroom-shaped dendritic spine formation in the apical dendrites of the hippocampal CA1 neurons, resulting in spatial learning and memory deficits. Second, the synaptic immaturity and cognitive dysfunction, core features of FXS, and other FXS phenotypes can be rescued with an early treatment of bryotatin-1 in order to produce therapeutic effects that are better than those of a late-onset treatment (see below). Finally, therapeutic effects can be achieved without downregulation of mGluR5 in the hippocampus.
It has been well established that FMRP mediates activity-dependent control of synaptic structure and function (Huber et al., 2002). Experimental studies suggest that the lack of FMRP leads to overactivity of mGluR5, a decreased GABAergic system (Olmos-Serrano et al., 2010), and elevated activity of glycogen synthase kinase 3β (Guo et al., 2012). Synaptic immaturity and cognitive dysfunction are core symptoms in FXS and learned-induced formation of dendritic spines is severely impaired in fragile X mice (Padmashri et al., 2013). Fragile X mice show an abundance of dense, immature dendritic spines (Scotto-Lomassese et al., 2011), as in FXS patients (Grossman et al., 2006). The hyperabundance of immature-looking lengthened dendritic spines could be the result of failed/delayed maturation (Cruz-Martín et al., 2010) and activity-dependent synaptic elimination (Pfeiffer et al., 2010). It has been reported that the left hippocampus in young male adult fragile X permutation carriers exhibits reduced structural connectome (Leow et al., 2014), consistent with evidence of a range of cognitive impairment, including spatial processing (Hocking et al., 2012; Wong et al., 2012). The deficit may also involve a loss of some dendritic channels (Routh et al., 2013) but can be rescued with chronic bryostatin-1 treatment at a young age, as shown in our study and previously in adults (Sun et al., 2014). The treatment in our study began at least 1 month earlier than our previous study (Sun et al., 2014), in which the same dose was administered at an adult age (2 months of age) for a longer treatment period (13 weeks). Histologic analyses were all performed at an adult age. Earlier treatment may lead to more favorable outcomes for several reasons. First, the therapeutic outcomes seem to be significantly better in the early-onset treatment group with the same doses of treatment, including the higher enhancement of BDNF expression and mushroom-shaped dendritic spine formation (Fig. 5, A and B), better synaptic maturation (Fig. 5, C and D), and greater improvements in performance in the water maze task (Fig. 5, E and F), suggesting that the brain network under the treatment has regained the capacity to meet cognitive demands. Second, an early intervention might facilitate a more “normal” development of sociocognitive skills and behaviors.
Our results of reduced mGluR5 labeling in the hippocampus of fragile X mice and increased labeling with bryostatin-1 treatment are somewhat unexpected. The data suggest that syndromic features of FXS might not be caused by an upregulated mGluR5 signaling pathway and synaptic and cognitive function can be rescued without downregulation of mGluR5 signaling. Desensitization of mGluR5, if induced (Gereau and Heinemann, 1998) with the treatment, and increased internalization would result in decreased surface expression of the receptor (Ko et al., 2012), a response not observed in this study. There is no evidence suggesting a direct association of FMRP with the metabotropic glutamate receptor 5 mRNA, although Lohith et al. (2013) revealed that both mGluR5 binding and protein expression were increased in the prefrontal cortex of FXS patients and carriers. In the fragile X mouse hippocampus, Western blot analysis was reported to show no difference in mGluR5 protein expression (Dölen et al., 2007), although a reduction was observed in the detergent-insoluble fraction of synaptic membranes isolated from the forebrain of fragile X mice (Giuffrida et al., 2005). There are also some observations of reduced expression of mGluR5 in fragile X mice (Giuffrida et al., 2005), but no difference in mGluR5 expression in total hippocampal homogenates has been reported by others (Huber et al., 2002). The lack of differences in mGluR5 expression (Huber et al., 2002; Giuffrida et al., 2005) between fragile X mice and controls but the presence of a reduced association of mGluR5 with PSD in fragile X mice might suggest an increased non-PSD association of mGluR5 labeling in fragile X mice. However, the evidence that supports the notion that mGluR5 overactivity reflects neuronal pathology in FXS seems very solid. Reducing mGluR5 expression or the use of mGluR5 inhibitors has been shown to correct a broad range of fragile X phenotypes in fragile X mice (Huber et al., 2002; Hajós, 2014). Although one would expect an increased colocalization of postsynaptic mGluR5 when it is overactive, our study does not rule out the possibility that the observed reduction in mGluR5 labeling in fragile X mice might partially reflect the reduced PSD labeling in the brain area. Nevertheless, our results indicate that the desired therapeutic effects can be achieved in fragile X mice without downregulation of mGluR5 in the hippocampus. The effective treatment actually improves mGluR5 labeling in the hippocampus. It is probably appropriate to mention here that while preclinical studies with mGluR5 antagonism appear promising, therapeutic values of mGluR5 inhibitors for FXS are still not clear. The mGluR5 inhibition exaggerates spine immaturity in fragile X mice (Cruz-Martín et al., 2010), an effect that would be considered opposite to the intended therapeutic outcomes. In addition, basimglurant and mavoglurant, two different and potent mGluR5 inhibitors, did not show therapeutic benefit in recent phase II clinical trials in FXS patients (Scharf et al., 2015).
Consistent with our previous study in fragile X mice, chronic bryostatin-1 rescues other FXS phenotypes, such as PSD-95 (Zalfa et al., 2007; Tang and Alger, 2015) and BDNF levels in the hippocampus. PSD-95, a major synapse organizer, plays an important role in the stabilization of spines and synapses as well as in activity-regulated formation of PSD. BDNF is important for local protein synthesis and synaptic plasticity (Neumann et al., 2015). Although images of the BDNF fluorescence labeling were used to measure the BDNF levels in this study, the labeling intensity was found to reliably reflect the BDNF levels in the same brain areas determined with an immunochemical assay, such as the enzyme-linked immunosorbent assay, in our previous studies (Sun et al., 2014). Reduction of BDNF expression in fragile X mice induces cognitive deficits (Louhivuori et al., 2011; Uutela et al., 2012). Infusion of BDNF has been found to restore synaptic function in slices from fragile X mice (Lauterborn et al., 2007). The decrease in BDNF accumulation in the CA1 stratum radiatum was significantly prevented with both the early-onset (this study) and late-onset (Sun et al., 2014) bryostatin-1 treatment. However, early-onset bryostatin-1 treatment, but not late-onset bryostatin-1 treatment, also enhanced the BDNF level (higher than the control wild-type without bryostatin-1 treatment).
It is interesting that in the wild-type mice, the early-onset bryostatin-1 treatment significantly elevated the spatial memory retention (in this study) above wild-type levels, whereas the late-onset bryostatin-1 treatment did not enhance memory retention (Sun et al., 2014), with the same lag period to separate acute effects from the chronic therapeutic effects. However, we did not find a significant enhancement of PKCε-induced BDNF levels after early-onset bryostatin-1 treatment. This might implicate that other PKCε-dependent pathways might also be involved in synaptic plasticity. The direct interaction of PKCε with actin is important for synaptic function and neurite growth and synaptic formation (Prekeris et al., 1996). PKCε may also induce synaptogenesis directly by activating structural changes through its phosphorylation substrates, GAP-43, the myristoylated alanine-rich C kinase substrate, and adducin (Matsuoka et al., 1998).
Bryostatin-1, a highly potent and relatively specific PKCε activator with pharmacological profiles of synaptogenesis and synaptic maturation/repairing (Hongpaisan and Alkon, 2007; Hongpaisan et al., 2011; Sun et al., 2015), rescues synaptic and memory functions and other phenotypic features in young fragile X mice. At low concentrations (about 1 nm), its PKC activation is mostly on PKCε and less on PKCα. At higher concentrations, however, activities of other PKC isoforms might also be affected. At a lower dose (10 μg/m2, two doses per week for 3 weeks), bryostatin-1 alone was found to have no significant effects on spatial learning and memory in rats (Sun and Alkon, 2008). We previously observed that its chronic administration did not alter swimming speed of rodents in the swimming test (Sun and Alkon, 2008). Evidence is accumulating, supporting an essential role of some PKC isoforms in various phases and types of learning and memory (Alkon et al., 2005, 2007). Bryostatin-1–like agents enhance the synthesis of proteins required for memory processing and synaptic repair/synaptogenesis, reduce Aβ formation through activation of α-secretase and increase Aβ degradation via the endothelin-converting enzyme, and are antiapoptotic. Since intellectual ability, as well as retardation (Wang et al., 2012), involves multiple players in signal processing, these agents with multitargeting actions (Sun et al., 2015) may represent a more effective class of therapeutics than agents that target a single factor in this complex pathologic process (Vislay et al., 2013).
Conducted experiments: Sun, Hongpaisan.
Performed data analysis: Sun, Hongpaisan.
Wrote or contributed to the writing of the manuscript: Sun, Hongpaisan, Alkon.
- Received November 30, 2015.
- Accepted March 2, 2016.
Financial disclosure: The authors declare no conflict of interest.
- analysis of variance
- brain-derived neurotrophic factor
- 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate
- fragile X mental retardation protein
- fragile X syndrome
- growth-associated protein
- metabotropic glutamate receptor
- phosphate-buffered saline
- protein kinase C
- postsynaptic density
- fragile X mice
- fragile X mice with bryostatin-1
- wild-type mice with bryostatin-1
- wild-type mice
- Copyright © 2016 by The American Society for Pharmacology and Experimental Therapeutics