Geranylgeranyltransferase I (GGT) is a prenyltransferase that mediates lipid modification of Rho small GTPases, such as Rho, Rac, and Cdc42, which are important for neuronal synaptogenesis. Although GGT is expressed in brain extensively, the function of GGT in central nerves system is largely unknown so far. We have previously demonstrated that GGT promotes the basal and neuronal activity and brain-derived neurotrophic factor (BDNF)-induced dendritic morphogenesis of cultured hippocampal neurons and cerebellar slices. This study is to explore the function and mechanism of GGT in neuronal synaptogenesis. We found that the protein level and activity of GGT gradually increased in rat hippocampus from P7 to P28 and subcellular located at synapse of neurons. The linear density of Synapsin 1 and post-synaptic density protein 95 increased by over-expression of GGT β, while reduced by inhibition or down-regulation of GGT. In addition, GGT and its known substrate Rac was activated by BDNF, which promotes synaptogenesis in cultured hippocampal neurons. Furthermore, BDNF-induced synaptogenesis was eliminated by GGT inhibition or down-regulation, as well as by non-prenylated Rac1 over-expression. Together, our data suggested that GGT mediates BDNF-induced neuronal synaptogenesis through Rac1 activation.
Synaptogenesis is a critical step for establishing neural circuits and is a complicated process of interaction between extracellular factors and intracellular signaling transduction(Waites et al. 2005; Shen and Cowan 2010; Tolias et al. 2011). Brain-derived neurotrophic factor (BDNF) is a well-known mediator of synapse formation and plasticity (Huang and Reichardt 2003; Luikart and Parada 2006; Lu et al. 2008; Yoshii and Constantine-Paton 2010). BDNF not only increases the number of dendritic synapses of hippocampal neurons(Ji et al. 2005; Yoshii and Constantine-Paton 2007; Zhou et al. 2008a),but also promotes presynaptic neurotransmitter release and neuron maturation(Shen et al. 2006; Cabezas and Buno 2011). Although it is reported that BDNF exerts its function on synaptogenesis through multiple signaling pathways, leading to cytoskeletal reorganization or gene expression required for synaptogenesis (Waites et al. 2005; McAllister 2007; Sheng and Hoogenraad 2007), the precise mechanisms that link the BDNF to F-actin remodeling in filopodial motility or synapse/spine formation remain unclear.
As key regulators of the actin cytoskeleton, members of the Rho-family GTPases, including Rac1, Cdc42, and RhoA, which play essential roles in orchestrating the development and remodeling of spines and synapses, are hotspots of research(Tada and Sheng 2006; Yoshihara et al. 2009). In cultured hippocampal neurons and slices, over-expressing constitutive active Rac l increases the density of dendritic spine, but over-expressing dominant-negative Rac1 decreases it (Nakayama et al. 2000; Tashiro and Yuste 2004; Wiens et al. 2005). Specifically, knockout Rac1 in excitatory neuron of forebrain not only decreases the density of dendritic spine but also affect hippocampal synaptic plasticity and spatial memory that depends on hippocampus(Haditsch et al. 2009). Since Rho GTPases are activated by guanine nucleotide exchange factors (GEFs) and inhibited by GTPase-activating proteins (Hall 1998; Schmidt and Hall 2002), many studies are focusing on finding different GEFs or GTPase-activating proteins. For example, it is reported that Vav2, a GEF of Rho small GTPases, which regulate Rho family GTPases and actin cytoskeletal dynamics, is activated by BDNF signaling (Shen and Cowan 2010). Similarly, Tiam1, another GEF of Rho small GTPases, mediates BDNF-induced Rac-GTP formation and chemotaxis in the developing cerebellum(Zhou et al. 2007), and Tiam1 also appears to mediate BDNF-induced neurite outgrowth in cultured cortical neurons (Miyamoto et al. 2006). However, before being activated by GEFs, GTPases need to be translocated from the cytosol to the membrane (Hancock et al. 1989; Philips et al. 1993), a prerequisite for its activation. This is achieved by prenylation, a reaction mainly catalyzed by farnesyltransferase (FT) or geranylgeranyltransferase I (GGT), which acts to add a lipid moiety to the cystine of C-terminal ‘CAAX’ box (Cys-aliphatic-aliphatic-X) of the GTPase (Zhang and Casey 1996). GGT includes two subunits: an α subunit (GGTα), the catalytic subunit, that is shared with FT and a distinct β subunit (GGTβ), which is responsible for substrate binding (Zhang and Casey 1996). It is reported that the basal expression level of GGTα is higher than that of GGTβ and GGTα also appear to be expressed in tissues that lack expression of the corresponding β subunits (Tsao and Waugh 1997; Maurer-Stroh et al. 2003).
We previously reported that BDNF activates GGT and in turn promotes membrane recruitment of Rac1 and increases dendritic arborization of cultured hippocampal neurons (Zhou et al. 2008b). GGT is also required for BDNF-induced cerebellum Purkinje cells morphogenesis in cultured cerebellar slices (Wu et al. 2010). At the neuromuscular junction, a synapse in periphery nerves system, agrin activates GGT through the muscle-specific receptor tyrosine kinase MuSK, leading to acetylcholine receptor clustering at the post-synaptic membrane (Luo et al. 2003). However, to our knowledge, whether GGT affect the synaptogenesis in central nerves system has not been reported until now. Furthermore, although many studies showed that both Rho GTPases and BDNF modulate synaptogenesis, respectively, whether the effect of BDNF on synaptogenesis was mediated by Rho GTPases is unknown.
In this study, we investigated the role of GGT in synaptogenesis by focusing on its regulation by factors that promote synaptogenesis. We found that the spatiotemporal expression pattern of GGT was closely related to synaptogenesis in vivo. Over-expression of GGTβ promoted the linear density of post-synaptic density protein 95 (PSD 95) and Synapsin 1, two major molecular markers of synaptogenesis, while down-regulation or inhibition of GGT decreased synaptogenesis. Treatment of cultured hippocampal neurons with BDNF increased GGT activity and membrane association of Rac, the known substrate of GGT. The effect of BDNF on synaptogenesis was attenuated by down-regulating GGT or Rac1 with the prenylation site deleted or mutated over-expression. Together, these observations reveal the critical role of GGT in synaptogenesis, and GGT mediates BDNF-induced neuronal synaptogenesis through Rac1 activation.
Materials and methods
Reagents, antibodies and plasmids
Recombinant human BDNF was from Peprotech. GGTi-2147 was from Calbiochem. K252a and GGPP were from sigma. Dansyl-GCVLL was from Biosynthesis and octyl-β-d-glucopyranoside was from Merck. CalPhos™ Mammalian Transfection Kit was from Clontech (Mountain View, CA) of USA. Antibodies used for immunostaining or immunoblotting were from Millipore (Billerica, MA) (rabbit anti-PSD95, mouse anti-Rac), BD bioscience (San Jose, CA) (mouse anti-Synapsin 1), Chemicon (Denver, Co) (mouse anti-β-actin), GGTβ (rabbit-anti-rat, kindly gifted by Dr ZG Luo at Institute of Neuroscience, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences). pKH3-GGTβ, pSUPER-GGTβ, GGTαWT, GGTαK164A, GGTαY200F, Rac1WT, Rac1C189S, and Rac1ΔC were kindly gifted by Dr ZG Luo (Zhou et al. 2008b; Wu et al. 2010).
Neuron culture, transfection, and treatment
Rat hippocampus was isolated from E18 Sprague–Dawley rats according to the protocol of Zhou et al. (Zhou et al. 2008b). Briefly, hippocampus were dissected, dissociated with 0.125% trypsin, triturated with medium containing 10% fetal bovine serum. The neurons were plated at a density of 100 000 cells on the poly-d-lysine coated coverslips in 24-well plate. For morphological analysis, neurons at day in vitro 5 (DIV5) were transfected with EYFP or together with testing constructs at a ratio of 1 : 3 with calcium phosphate precipitation method using CalPhos™ Mammalian Transfection Kit according to the instruction of manufacturer. Sometimes, neurons were treated with various reagents, for example, BDNF (50 ng/mL) or GGTi-2147 (2.5 μM) for 30 h, either alone or with combinations.
GGT Activity Assay
GGT activity was determined in 96-well microtiter plates as previously described with minor modifications (Zhou et al. 2008b). After being pre-treated with GGTi-2147 or dimethylsulfoxide or K252a for 60 min, neurons at DIV10 were treated with BDNF for 60 min and lysed. The cell lysates (50 μg protein) were mixed with reaction mixture (100 μL) containing 50 mM Tris-Cl (pH 7.5), 5 mM MgCl2, 50 μM ZnCl2, 20 mM KCl, 1 mM dithiothreitol, 0.2% octyl-β-d-glucopyranoside, 10 μM dansyl-GCVLL, and 10 μM GGPP and the activity of GGT was examined by measuring the value of fluorescence (excitation at 360 nm and emission at 460 nm) using a Synergy 2 (Bio-Tek, Winooski, VT) fluorescence measurement system.
Preparation of synaptosome and post-synaptic density fractions
Synaptosome and PSD fractions were prepared from rat hippocampus by sucrose gradient method as previously described (Zhou et al. 2008a). All purification steps were performed at 4°C. The hippocampus of P14 rats were rapidly isolated and homogenized in ice-cold Buffer A (5 mM HEPES pH7.4,1 mM MgCl2, 0.5 mM CaCl2, 1 mM NaF, 1 mM beta-glycerophosphate, 0.1 mM phenylmethlsulfonyl fluoride, 1 mM benzamidine, 0.1 mM pepstatin, 1 μg/mL aprotonin, 1 μg/mL leupeptin, and phosphatase inhibitors). The homogenized extract was spun at 1400 g for 10 min. The supernatant was saved and pellet (P1) was homogenized using Buffer A again. After centrifugation at 700 g, the supernatant was saved and pooled with previous supernatant. Pooled supernatant was centrifuged at 13 800 g for 10 min to collect the pellet (P2). The P2 was resuspended in Buffer B (320 mM sucrose, 6 mM Tris, pH 8.0, 1 mM NaF, 1 mM β-glycerophosphate, 0.1 mM phenylmethylsulfonyl fluoride, 1 μg/mL of aprotonin, 1 μg/mL of leupeptin, 1 mM benzamidine, and 0.1 mM pepstatin). The P2 suspension was then loaded onto a discontinuous sucrose gradient (0.85 M/1 M/1.2 M sucrose solution in 6 mM Tris, pH 8.0), followed by centrifugation for 2 h at 82 500 g. The fraction between 1 M and 1.2 M sucrose was collected and saved as the synaptosome. The synaptosome fraction was further adjusted to 4 mL with the Buffer B and mixed with equal volumes of Buffer C (6 mM Tris, pH 8.1, and 1% Triton X-100) for 15 min. The suspension was spun at 32 800 g for 20 min. The supernatant was discarded, and the resulting pellet was saved as the PSD fraction. The total lysates of hippocampus, synaptosome and PSD fractions were adjusted to an equal protein concentration (5 μg/μL) using Bio-Rad Protein Assay (Bio-Rad, Hercules, CA). Five independent experiments were performed.
Membrane protein extraction, pull-down assay, and western blot
At DIV10, cortical neurons were treated with BDNF or GGTi-2147 for 60 min and the membrane proteins were extracted using the Plasma Membrane Protein Extraction Kit (BioVision, Mountain View, CA, Catalog K268-50) according to manufacturer guidelines. For pull-down assays, cell lysates (0.5–1.0 mg protein/mL) were incubated with 2–10 μg glutathione-s-transferase (GST) fusion proteins immobilized on Sepharose beads for 2–4 h at 4°C. Bound proteins were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and immunoblotting were performed as described previously (Zhou et al. 2008b; Liang et al. 2010). Briefly, an equal amount of protein was separated on a 10% or 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred electrophoretically to the polyvinylidene difluoride membranes. The membrane was blocked with 3% bovine serum albumin and 0.1% Tween-20 in Tris-buffered saline for 1 h at 25°C. The blots were incubated with antibodies against-Rac, GGTα, GGTβ, Synapsin 1, PSD 95 or β-actin overnight at 4°C, followed with peroxidase-conjugated secondary antibodies for 2 h at 25°C. The bands were visualized with an enhanced chemiluminescence detection kit (BeyoECL Plus; Beyotime, Haimen, Jiangsu, China), and the result was analyzed using Image Pro Plus Software (Media Cybernetics, Inc. Rockville, MD). All fold changes in band densities were measured relative to the control groups. Western blot experiments were carried out in three biological replicates and average fold changes are reported.
Immunocytochemistry, image analysis, and quantification
At DIV10, neurons were washed twice with phosphate buffered saline (PBS) and fixed for 20 min at 25°C with 4% paraformaldehyde in PBS. The fixed cells were permeabilized and blocked with 0.3% Triton X-100, 10% goat serum in PBS for 1 h and then incubated overnight at 4°C in a humidified chamber with primary antibodies (PSD 95 and Synapsin 1) followed by the corresponding secondary antibodies (Alexa Fluor Life Technologies Corporation. Treble Ct, Maryland 633 goat anti-mouse antibody and Alexa Fluor 555 goat anti-rabbit antibody) for 1.5 h at 37°C.
Images were acquired on Olympus confocal laser scanning microscopy (Olympus FV10C-02, Olympus Corporation, Shinjuku-Ku, Japan) using a 60 × objective with sequential acquisition settings at 1024 × 1024 pixel resolution. A z series projection of each neuron was made using approximately 5 to 10 sections (1.0 μm per section) and the resultant stack was merged into a single image using a maximum projection. Synapse density was quantified using Zeiss LSM (Carl Zeiss Microscopy GmbH, Munchen, Germany) image and Image Pro Plus Software (Media Cybernetics, Inc. Rockville, MD). PSD 95 or Synapsin 1 clusters greater than 4 pixels in size and localized to the transfected neuron were counted. We observed at least three secondary dendritic segments totaling at least 150 μm of dendritic length per neuron, and the number of clusters was counted and the number of clusters per dendritic segment (density) was calculated. The average optical density (OD) of clusters was also quantified. Ten to fifteen transfected neurons were chosen randomly for quantification per experiment, and three independent experiments were done for each construct or treatment.
All data were obtained from several sets of cultures with its own control. In most cases, the mean value of control groups was set as 1.0. All values were normalized by the means of the control group. Data analysis was performed using one-way anova or independent-sample t test. Data are shown as mean ± SEM (p <0.05 considered significant difference).
Over-expression of GGTβ promotes synaptogenesis
We previously reported that the protein level of GGTα and mRNA level of GGTβ in cultured hippocampal neurons gradually increased and peaked at DIV 8 approximately (Zhou et al. 2008b). In addition, the protein level of GGTβ in developing cerebellum increased gradually from P0 to adult (Wu et al. 2010). Since the synapse develops quickly during this period (Zhou et al. 2004), this result points out that the expression of GGT might be closely related to the synaptogenesis. Then, at first, we examined the spatiotemporal expression pattern of GGT in developing hippocampus. We found that expression of GGTα and GGTβ gradually increased in developing rat hippocampus from post-natal day 7 to 28 (P7-P28) (Fig. 1a). At the same time, the GGT activity was also gradually increased during this time window (Fig. 1b). To elucidate the subcellular location of GGT, we tested whether GGTα and GGTβ was located at the synapse using the synaptosome and PSD fractionation assay. The PSD 95 and Synapsin 1 were used as markers for synapse and presynaptic fraction, respectively. PSD 95 is a neuronal PDZ protein that is associated with receptors and cytoskeletal elements at synapses, playing suggestive roles for synaptic development, stabilization, and plasticity (El-Husseini et al. 2000; Joseph et al. 2011). We found that GGTα and GGTβ were enriched in synaptosome and post-synaptic density fractions from P14 hippocampus (Fig. 1c). Furthermore, we transfected the cultured neurons with GGTβ together with YFP (to mark the morphology of the neuron) and costained with GGTβ and Synapsin 1, the presynaptic marker. As is shown in Fig. 1d, GGTβ was colocalized with Synapsin 1, indicating that GGT was subcellular located at the synapse of the neurons. The above results suggest that the spatiotemporal expression pattern of GGT was correlated with the synaptogenesis in vivo, suggesting that GGT might be involved in synapse formation.
Next, we examined the role of GGT in synaptogenesis by over-expressing GGTβ in cultured hippocampal neurons. These cells were transfected with a vector encoding HA-GGTβ together with an EYFP plasmid (to mark positive neurons) at DIV 5 and analyzed at DIV 10. We found that over-expression of GGTβ resulted in the formation of more complex dendritic arbors, in line with the previous report (Zhou et al. 2008b). To evaluate the effect of GGT on synaptogenesis, we used the clusters density of Synapsin 1 and PSD 95, two generally used presynaptic and post-synaptic markers, in secondary dendrites as the index of synaptogenesis. The value of GGTβ over-expression group was normalized to that of EYFP-transfected neurons, which was set as 1.0. Compared with cells transfected with EYFP alone, the clusters density of Synapsin 1 and PSD 95 increased by 22% and 18%, respectively, after GGTβ transfection (Fig. 1e–g), suggesting that over-expression of GGTβ promotes synaptogenesis.
GGT inhibition or down-regulation impedes synaptogenesis in rat hippocampal neurons
To further determine the role of GGT in synaptogenesis, we treated cultured hippocampal neurons with GGTi-2147, a potent and selective inhibitor of GGT. The IC50 value of GGTi-2147 for blocking the geranylgeranylation of Rap1A, another substrate of GGT, is over 60-fold lower than that required to disrupt the farnesylation of H-Ras, a substrate of FT, (IC50 = 500 nM for Rap1A vs. IC50 > 30 μM for H-Ras). We found that GGTi-2147 treatment caused a marked decrease in the clusters density of Synapsin 1 and PSD 95 (Fig. 2a–c). In addition, the optical density (OD) of clusters of GGTi-2147 treated neurons also decreased significantly (Fig. 2a, d, and e), compared with vehicle treatment group.
Next, we used vector-based small interference RNA (shRNA) against GGTβ, which down-regulates expression of endogenous GGTβ efficiently in primary neurons, compared with the scramble control group (Fig. 2f; Zhou et al. 2008b), to further examine the role of GGT in synaptogenesis. As is shown in Fig. 2f–j, we found that down-regulation of GGTβ caused a simpler hippocampal neuron morphology and decreased cluster density and OD of Synapsin 1 and PSD 95 significantly. In comparison to scramble control group, the cluster density of Synapsin 1 and PSD95 decreased by the 50% and 37% respectively, while the OD decreased by 49% and 34%, respectively. Thus, the above data suggest that GGT-mediated protein prenylation plays an important role in the synaptic morphogenesis of hippocampal neurons.
The activity of GGT and Rac are regulated by BDNF
From the above results, we can see GGT is sufficient and necessary for the neuronal synaptogenesis, the next question is what is the mechanism? It is reported that the activity of GGT was regulated by BDNF and neuronal activity, which are known to promote neuronal synaptogenesis, during dendritic morphogenesis period (DIV 5) (Zhou et al. 2008b). Thus, to address this question, we tested the possibility of BDNF regulating GGT activity during synaptogenesis period (DIV 10).
It is known that the function of GGT is to covalently couple a lipid moiety to the cystine of C-terminal ‘CAAX' box (Cys-aliphatic-aliphatic-X) of the small GTPases, promoting them to be translocated from the cytosol to the membrane for their activation (Hall 1998). A potential substrate of GGT is Rac, a small GTPase critical for synaptogenesis (Luo 2002; Sin et al. 2002). Therefore, we used membrane fractionation assay to detect the membrane level of Rac, as an indirect index of GGT activity of DIV 10 neurons. We found that the basal level of membrane bound Rac, hereafter referred to as Rac(m), decreased in GGTi-2147 treated neurons (Fig. 3a), suggesting that the basal level of Rac(m) depends on the GGT activity. In addition, the membrane level of Rac increased significantly after BDNF treatment for 60 min (Fig. 3a) and this elevation depended on the GGT activity since GGTi-2147 pre-treatment prevented the increase in the Rac(m) level induced by BDNF (Fig. 3a), suggesting that GGT activity is regulated by BDNF. Furthermore, we detected the GGT activity directly using dansyl-GCVLL as the substrate (Zhou et al. 2008b). We found that the GGT activity increased significantly after BDNF treatment and this effect was specific, as co-treatment with GGTi-2147 abolished it (Fig. 3b), in line with the results showed in Fig. 3a. Similarly, BDNF-induced Rac(m) level and the GGT activity increase were eliminated by pre-treatment with K252a, the inhibitor of BDNF receptor tropomyosin-related kinase B (TrkB) (Fig. 3c and d).
Furthermore, the Rac activity was determined by GST pull-down assay using GST-p21-binding domain of p21-activated kinase PAK1, which is generally used to isolate the active form of Rac (GTP-Rac) (del Pozo et al. 2000). We found that the level of GTP-Rac also increased by the BDNF treatment and this effect was also abolished by the presence of GGTi-2147 (Fig. 3e), in agreement with the changes in Rac (m). Together, these results support the notion that BDNF activates GGT, which in turn promotes membrane association and the activation of Rac.
GGT inhibition or down-regulation eliminates BDNF-induced synaptogenesis
Given the finding that BDNF could activate GGT, we further examined whether GGT loss of function affects BDNF-induced synaptogenesis. In line with previous studies (Ji et al. 2005; Yoshii and Constantine-Paton 2007; Zhou et al. 2008a), BDNF treatments caused an increase in the cluster density and O.D. of Synapsin 1 and PSD 95 in cultured hippocampal neurons at DIV10 (Fig. 4a–e). However, these effects of BDNF were abolished in the presence of GGTi-2147 (Fig. 4a–e). To further explore the role of GGT in BDNF-induced synapse formation, we suppressed the GGT expression using RNA interference approach together with or without BDNF treatment in neurons. As is shown in Fig. 4f–j, after being transfected into cultured hippocampal neurons, GGT β shRNA caused a decrease in the cluster density and OD of Synapsin 1 and PSD 95 (Fig. 4f–j), consistent with the previous results (Fig. 2f–j). Similar to the effect of GGTi-2147, down-regulation of GGTβ expression also abolished BDNF-induced effects on synaptogenesis (Fig. 4f–j).
It is reported that GGTα mutation of lysine 164 (K164) and tyrosine 200 (Y200) inhibited the prenylation activity by preventing the catalytic activity of prenyltransferase or the formation of the enzyme-substrate complex (Wu et al. 1999; Hightower et al. 2001; Luo et al. 2003). In our previous study (Zhou et al. 2008b), K164A (i.e., lysine to alanine, GGTαK164A) or Y200F (i.e., tyrosine to phenylalanine, GGTαY200F) was found to be associated with TrkB to a similar extent as the wild type GGTα (GGTαWT). These two mutated GGTα was used as dominant-negatives to compete with the endogenous GGT for TrkB and inhibit TrkB-mediated GGT activation and blocks the dendritic arborization induced by BDNF. In this study, we wonder whether K164A or Y200F could abolish the effect of BDNF on synaptogenesis. We transfected hippocampal neurons with empty vector or the vector encoding GGTαWT or mutated forms of GGTα (GGTαK164A or GGTαY200F), and then treated with or without BDNF. We found that over-expression of GGTαK164A or GGTαY200F not only impeded the basal synaptogenesis of hippocampal neurons but also prevented the effect of BDNF on synapse formation, compared with GGTαWT group (Fig. 4k–o). Thus, GGT is indeed crucial for basal and BDNF-dependent synaptogenesis.
Rac1 prenylation is required for BDNF-induced synaptogenesis
To further determine the role of GGT-mediated prenylation in BDNF-induced synaptogenesis, we used two mutated forms of Rac1 (Rac1ΔC and Rac1C189S) by deleting the C-terminal CAAX box or substituting the cystine with serine, respectively. These mutated forms of Rac1 might act as dominant-negatives by competing with endogenous Rac for GEFs such as Tiam1(Zhou et al. 2008b). After being transfected into cultured hippocampal neurons, Rac1WT increased the density and O.D. of PSD 95 and Synapsin 1, whereas Rac1ΔC or Rac1C189S had negative effect compared with neurons transfected with EYFP alone (Fig. 5a–e). In addition, over-expression of Rac1ΔC and Rac1C189S eliminated the promoting effect of GGT β over-expression on the neuronal synaptogenesis (Fig. 5a–e). Consistent with the previous data, over-expression of Rac1ΔC and Rac1C189S eliminated the promoting effect of BDNF-induced synaptogenesis (Fig. 6a–e). Therefore, Rac1 plays a very important role in synaptogenesis and the prenylation of Rac1 is required for this process. The above results showed that BDNF-induced neuronal synaptogenesis was mediated by GGT activation through promoting Rac membrane translocation and activation.
Many studies showed that synaptogenesis is a combined action of multiple factors, including external signals, such as BDNF, and a variety of intracellular mediators, such as Rho family small GTPases (Waites et al. 2005; Tolias et al. 2011), which lead to cytoskeletal reorganization or gene expression required for synaptogenesis. However, the mechanism that BDNF activated Rho family small GTPases and affected cytoskeletal reorganization and synapse/spine formation remains unclear. In this study, we showed that GGT played an important role in synaptogenesis and BDNF-induced synapse formation was mediated by GGT through Rac1 activation. Several findings support this conclusion. First, the expression and activity of GGT gradually increased from P7 to P28, a period that is known to be important for the synaptogenesis in vivo (Zhou et al. 2004) and subcellular located at the synapse of hippocampal neurons. Second, over-expression of GGTβ promoted synaptogenesis, while down-regulation or inhibition of GGT impeded synaptogenesis. Third, GGT and its known substrate Rac were activated by BDNF. Fourth, BDNF-induced synaptogenesis was not only abolished by GGT down-regulation or inhibition but also eliminated by un-prenylated Rac1 mutation over-expression.
GGT regulates the synaptogenesis of hippocampal neurons
It is known that, to be activated efficiently, many cytoplasmic signaling proteins need to be associated with the plasma membrane. This process is achieved by post-translational modifications, such as palmitoylation, myristoylation, or prenylation (el-Husseini Ael and Bredt 2002). Rho family GTPases, which play critical roles in neuronal synaptogenesis, were prenylated by GGT at the C-terminal CAAX box (Zhang and Casey 1996; Sinensky 2000). In addition to small GTPases, many other signaling proteins were also modified by prenyltransferases (Zhang and Casey 1996). For example, prenylation of Ca2+/calmodulin-dependent protein kinase CLICK-III/CaMKIγ plays an important role in its association with lipid raft and dendritogenesis (Takemoto-Kimura et al. 2007). Although GGT is extensively expressed in the brain (Yokoyama et al. 1991; Ericsson et al. 1993), its roles and mechanisms during development of central neurons is largely unknown. We previously reported that GGT not only regulates the basal neuronal dendritic growth but also mediates BDNF-induced dendritogenesis by interacting with its receptor TrkB (Zhou et al. 2008b; Wu et al. 2010). Since dendritic growth is closely related to synaptogenesis(Cline and Haas 2008), we deduced that GGT might be involved in neuronal synaptogensis. After having established the spatiotemporal relationship between expression of GGT and synaptogenesis (Fig. 1), we examined the role of GGT in synapse formation using gain-of function and loss of function approaches. In this study, we found that the linear density of Synapsin 1 and PSD 95 increased by over-expressing GGTβ (Fig. 1c–e). On the contrary, the linear density and OD of Synapsin 1 and PSD 95 decreased either by GGTi-2147 treatment, GGTβ shRNA transfection, or mutated forms of GGTα (GGTαK164A or GGTαY200F) over-expression (Figs 2 and 4). The above results showed that GGT regulates the linear density and optical density of PSD 95 and Synapsin 1 under basal conditions, suggesting that GGT is critical for synaptogenesis (Fig. 2).
One curious thing in our study is that over-expression of GGTαWT did not increase the linear density of Synapsin 1 and PSD 95 (Fig. 4), while over-expression of GGTβ did (Fig. 1). GGT is a heterodimer in nature, consisting of a distinct β subunit and an α subunit that is almost identical to FT (Seabra et al. 1991; Yokoyama et al. 1991; Moomaw and Casey 1992). It is reported that α subunit determines the enzyme activity of GGT, while the β subunit determines its functional specificity (Lane and Beese 2006). Since the basal expression level of GGTα is higher than that of GGTβ and GGTα also appear to be expressed in tissues that lack expression of the corresponding β subunits (Tsao and Waugh 1997; Maurer-Stroh et al. 2003), we think the result difference between GGTα and GGT β over-expression in synaptogenesis is because of that over-expression of GGTα could not obtain equivalent GGT β to form heterodimer to function, while over-expression of GGTβ could.
GGT activity is regulated by BDNF during synaptogenesis period
Prenylation was believed to be a constitutive modification of proteins (Sinensky 2000). However, recent studies suggest that prenylation of small GTPases of the Rho family is under tight regulation. For example, it is known recently that GGT activity can be strictly regulated by multiple factors in breast cancer cells or skeletal muscle cells (Chappell et al. 2000; Luo et al. 2003), but whether and how protein prenylation is regulated in neurons were largely unknown so far. We have recently reported that GGT activity is regulated by neuronal activity or BDNF in cultured neurons, as after being treated with high K+, bicuculline, or BDNF, the GGT activity markedly increased in cultured neurons (Zhou et al. 2008b; Zhou and Luo 2009). The activation effect of BDNF and neuronal depolarization on GGT activity was also observed in Purkinje cells of cultured cerebellar slices (Wu et al. 2010). GGT activity was also up-regulated under enrichmental environment, an in vivo neuronal activity model (Zhou et al. 2008b). The effect of BDNF on GGT activity was inhibited by the K252a or inhibition of GGT (Zhou et al. 2008b; Zhou and Luo 2009). In line with our previous reports, in this study, we also found that BDNF can active GGT, promote Rac1 to translocate to cell membrane during synaptogenesis stage (Fig. 3).
GGT mediates BDNF-induced synaptogenesis
As a member of neurotrophin family, BDNF is widely expressed in the vertebrate nervous system (Lin et al. 2011; Zhai et al. 2011). By binding to its surface receptor TrkB, BDNF plays critical roles in neuronal development, differentiation, synaptogenesis, synaptic plasticity, and neural protection from the harmful stimuli (Fu et al. 2011; Jeon et al. 2011). In present study, we also found the linear density and OD of Synapsin 1 and PSD 95 was significantly promoted by BDNF treatment, which is consistent with many other reports (Ji et al. 2005; Yoshii and Constantine-Paton 2007; Zhou et al. 2008a). In addition, BDNF-induced synapse formation was abolished by inhibition, down-regulation of GGT or GGTαK164A or GGTαY200F over-expression (Fig. 4). Since GGT could be activated by BDNF, the above result indicates that GGT mediates BDNF-induced synaptogenesis.
GGT prenylated the Rho family GTPases, which has been shown to be critical for neuronal morphogenesis (Redmond et al. 2002; Szebenyi et al. 2005; Fu et al. 2011), at the C-terminal CAAX box (Zhang and Casey 1996; Sinensky 2000). Thus, the CAAX box is necessary for Rho family GTPases membrane association and then activation (Hall 1998). Accordingly, mutation or deletion of the CAAX box abolished the prenylation modification (Zhou et al. 2008b). In this study, we found that Rac1ΔC or Rac1C189S, both un-prenylated form of Rac1, eliminated the increased linear density and O.D. of PSD 95 and Synapsin 1 induced by GGTβ over-expression and BDNF treatment (Figs 5 and 6). The PSD 95 and Synapsin 1 levels were roughly equivalent in cells transfected with Rac1WT and Rac1WT plus GGTβ (Fig. 5), suggesting that GGTβ and Rac1 were in the same signaling pathway. However, the PSD 95 and Synapsin 1 levels in Rac1WT plus BDNF group were higher than that of Rac1WT group, suggesting GGT independent signaling pathway (Fig. 6). Since Rac could be activated by its GEF, such as Tiam1, which is activated by BDNF (Miyamoto et al. 2006; Zhou et al. 2007), BDNF-Tiam1-Rac might be a potential signaling pathway to promote synaptogenesis independent of GGT. The above results indicated that Rac1 prenylation is required for BDNF-induced synaptogenesis, further suggesting that GGT-mediated prenylation is necessary for BDNF-induced synaptogenesis.
Taken together, our data suggested that GGT mediates BDNF-induced neuronal synaptogenesis through Rac1 activation. Because GGT has many other substrates besides GTPases (Zhang and Casey 1996), it is possible that GGT may also regulate synaptogenesis by its modification of other substrates. Furthermore, it is known that BDNF regulates synaptogenesis and synaptic plasticity (Zhou et al. 2008a) and that its receptor TrkB regulates synaptic plasticity through BDNF-dependent or independent manner (Shen and Cowan 2010). Given our findings that BDNF and its receptor regulate GGT activity, it shall be interesting to determine the role of GGT-mediated prenylation in regulating other neuronal functions, such as synaptic plasticity.
The study was supported by National Natural Science Foundation of China (No. 31070933); the grant from Xuzhou Medical College (No. 09KJZ18) and Natural Science Foundation of Jiangsu province (No. BK2011195); Program for New Century Excellent Talents in University (NCET-10-0181) and the grant from A Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD). We thank Chunmei Zhu at English Department of Xuzhou Medical College for English writing assistance. There is no conflict of interest.