Sonic hedgehog (Shh) signaling is essential to the patterning of the embryonic neural tube, but its presence and function in the postmitotic differentiated neurons in the brain remain largely uncharacterized. We recently showed that Shh and its signaling components, Patched and Smoothened, are expressed in postnatal and adult hippocampal neurons. We have now examined whether Shh signaling has a function in these neurons. Using cultured hippocampal neurons as a model system, we found that presynaptic terminals become significantly larger in response to the application of Shh. Ultrastructural examination confirmed the enlarged presynaptic profiles and also revealed variable increases in the size of synaptic vesicles, with a resulting loss of uniformity. Furthermore, electrophysiological analyses showed significant increases in the frequency, but not the amplitude, of spontaneous miniature excitatory postsynaptic currents (mEPSCs) in response to Shh, providing functional evidence of the selective role of Shh in presynaptic terminals. Thus, we conclude that Shh signaling regulates the structure and functional properties of presynaptic terminals of hippocampal neurons.
The Shh signaling pathway is known for its multifunctional roles in embryonic development (Ingham and McMahon, 2001; Jiang and Hui, 2008; Varjosalo and Taipale, 2008) as well as adult tissue homeostasis (Beachy et al., 2004; Varjosalo and Taipale, 2008). In the central nervous system, the best-characterized function for Shh is its ability to stimulate the proliferation of cerebellar granule cells (Dahmane and Ruiz i Altaba, 1999; Wallace, 1999; Wechsler-Reya and Scott, 1999) and neural progenitor cells residing in specific areas of the brain (Lai et al., 2003; Palma et al., 2005; Breunig et al., 2008; Han et al., 2008). In addition to acting as a mitogen for cells with stem-cell properties, Shh signaling components remain expressed in differentiated (postmitotic) neurons, including hippocampal neurons (Traiffort et al., 1999; Sasaki et al., 2010; Petralia et al., 2011a; Petralia et al., 2011b). In this study, we investigated whether Shh signaling activity has a function in hippocampal neurons. We focused on the synapse of mature hippocampal neurons. Our findings reveal a new role for the Shh signaling pathway in regulating the structure and neurotransmitter release function of presynaptic terminals.
Results and Discussion
Shh signaling activity induces presynaptic differentiation
We added ShhN (supplementary material Fig. S1) or a Shh agonist designated SAG (Chen et al., 2002b) to cultured hippocampal neurons and examined the synapses of these neurons 2 days later using several synaptic markers, including a presynaptic active zone protein, Bassoon (tom Dieck et al., 1998), and two synaptic vesicle-specific proteins, Synapsin 1 (Micheva et al., 2010) and the zinc transporter ZnT3 (Palmiter et al., 1996; Grønborg et al., 2010). All three markers indicated that neurons that had been exposed to ShhN had significantly more presynaptic puncta than controls (Fig. 1A). A similar change was seen in the SAG-treated neurons (Fig. 1A). In addition, many presynaptic puncta were also enlarged (supplementary material Fig. S2).
Co-administering ShhN with a Shh antagonist cyclopamine (Taipale et al., 2000) completely prevented the ShhN-induced presynaptic puncta in these neurons (Fig. 1A; supplementary material Fig. S2), confirming that the presynaptic phenotype observed was a direct result of ShhN. Intriguingly, when neurons were treated with cyclopamine alone, none of the presynaptic markers indicated any obvious change (Fig. 1A; supplementary material Fig. S2). This finding was somewhat surprising because one would expect that, if endogenous Shh in these neurons is required for their synapse formation or maintenance, suppressing Shh pathway activity by blocking Smo should produce an opposite phenotype – a reduction or loss of synapses. One possibility is that Shh signaling transduction in neurons might operate via both canonical and non-canonical pathways (Jenkins, 2009), which would be reminiscent of the signaling transduction of the morphogen Wnt in neurons (Hall et al., 2000; Budnik and Salinas, 2011). If so, inhibiting Smo alone may not in and of itself eliminate Shh activity, and therefore, the cyclopamine-treated neurons may not exhibit readily detectable defects. An alternative or additional explanation for the lack of obvious alterations in the cyclopamine-treated neurons is that neurons employ a combination of multiple signaling pathways or molecular mechanisms to control synapse formation. This possibility seems probable because it has been found in studies of several synaptogenic molecules that reducing or silencing these molecules often generates milder than expected or even undetectable phenotypes [for examples, see Shen and Scheiffele and references therein (Shen and Scheiffele, 2010)]. Therefore, additional signaling mechanisms could compensate for the loss of Smo in the cyclopamine-treated neurons.
We next assessed the types of synapses, and also compared presynaptic and postsynaptic terminals. For the glutamatergic synapses, we examined the presynaptic vesicular glutamate transporter (VGlut) and a postsynaptic density protein, PSD95. For the GABAergic synapses, we compared the presynaptic GABA transporter (VGAT) and a postsynaptic protein Gephyrin (Gphn). In both types of synapses, ShhN and SAG significantly increased the number of presynaptic terminals, but had little effect on the postsynaptic terminals (Fig. 1B; supplementary material Fig. S3). Consistently, double immunolabeling revealed that the proportion of presynaptic puncta that did not pair with visible postsynaptic puncta was greater in the ShhN-neurons than in the controls (Fig. 1B). Therefore, Shh activity affects the presynaptic terminals of both glutamatergic and GABAergic synapses.
Immunoblot analysis showed that the level of Bassoon was dramatically increased in the neurons treated with ShhN, SAG, or another Shh-agonist, Purmorphamine (Fig. 1C) (Sinha and Chen, 2006). Surprisingly, the levels of Synapsin1 and a second synaptic vesicle protein, Synaptophysin, were not affected by any of the treatments (Fig. 1C). The unchanged gross levels of synaptic vesicle proteins – contrary to the dramatic increase of the presynaptic terminal component Bassoon – indicate that Shh signaling may control formation of presynaptic terminals through selective de novo synthesis of terminal components, and the concomitant redistribution and clustering of pre-existing synaptic vesicle components.
As an additional approach to activate the Shh pathway, we ectopically expressed a constitutively active Smoothened mutant (SmoA) (Xie et al., 1998) in the neurons. We co-transfected SmoA with either Synaptophysin::EGFP (Syp::EGFP) or PSD95::EGFP to mark the presynaptic and postsynaptic terminals respectively. The expression of SmoA caused the enlargement of the Syp::EGFP-marked presynaptic boutons. Unlike the clearly spaced, small puncta seen in the control neurons, many of the Syp::EGFP boutons from the SmoA-expressing neurons became highly fused large clusters (Fig. 1D; supplementary material Fig. S4). This increase was similar, yet more evident, in older more mature neurons (Fig. 1D). Conversely, in both more and less mature neurons, the expression of SmoA had no obvious effect on the size of the postsynaptic PSD95-puncta (Fig. 1D). We were unable to compare the number of Syp::EGFP puncta because the clumps of clusters resulting from SmoA overexpression precluded a one-to-one identification of individual puncta (Fig. 1D; supplementary material Fig. S4). Exogenously introduced Gli2, which activates the downstream component of the Shh pathway (Roessler et al., 2005; Kim et al., 2009), mimicked the SmoA-induced effect (supplementary material Fig. S4). Thus, Shh signaling preferentially regulates the presynaptic terminals of hippocampal neurons.
Shh-induced presynaptic terminals are functional
Synaptic vesicle endocytosis is an essential process that determines the function of a synapse (Dittman and Ryan, 2009). We reasoned that if the Shh-induced presynaptic terminals were functional, synaptic vesicle endocytosis should take place, and that if these terminals were enlarged, they would take up more endocytic markers. We carried out FM dye uptake and antibody uptake assays. FM dye is a fluorescent styryl dye which partitions preferentially into lipid membranes, where it becomes intensely fluorescent (Betz and Bewick, 1992; Kay et al., 1999). We used a fixable fluorescent dye FM1-64 and measured its internalization by the neurons upon high K+-induced depolarization. The ShhN neurons took up more FM dye, as evidenced by the significantly increased number and size of FM clusters or puncta (Fig. 2A,B; supplementary material Fig. S5A). The same trend was seen in the SAG-treated neurons (Fig. 2B, and data not shown).
We performed an additional experiment by measuring the uptake of two different antibodies. One antibody was specific to the luminal domain of Synaptotagmin1, which has proven useful for assaying synaptic vesicle endocytosis at glutamatergic synapses (Matteoli et al., 1992; Kraszewski et al., 1995). The other antibody was to the luminal domain of VGAT (Martens et al., 2008), which can be used to examine the VGAT-containing vesicles at the GABAergic synapses. Fig. 2C–F and supplementary material Fig. S5A show that the ShhN-treated neurons took up substantially more of both Synaptotagmin1 and VGAT antibodies. Lastly, we found that the enlarged, Syp::EGFP-illuminated synapses in the SmoA neurons displayed correspondingly larger FM clusters (supplementary material Fig. S5B). These results therefore are in agreement with the results seen for immunolabeling of the synaptic markers (Fig. 1; supplementary material Figs S2–S4) showing that Shh signaling has a profound effect in the development of the presynaptic terminals of hippocampal neurons.
Shh signaling changes the ultrastructure of presynaptic terminals
To determine whether the larger presynaptic puncta revealed by the above light microscopic analyses were due to enlarged terminals or synaptic vesicles or both, electron microscopy was performed. Presynaptic terminal profiles were identified based on the presence of closely spaced synaptic vesicle clusters and a dark presynaptic active zone that juxtaposed with a well-defined synaptic cleft and postsynaptic density. Many ShhN-treated neurons had remarkably large presynaptic profiles, some of which had a noticeably elongated active zone (Fig. 3A–G). Size distribution analysis from 229 clearly traceable presynaptic profiles showed that ShhN caused neurons to grow substantially larger presynaptic profiles (Fig. 3H; supplementary material Fig. S6A). These larger presynaptic profiles with much longer active zone were in complete accord with the immunoblotting experiment (Fig. 1C), which demonstrated a marked increase in the level of the active zone protein Bassoon.
The presynaptic profiles of ShhN-neurons also harbored many visibly bigger synaptic vesicles, giving the vesicle clusters a less homogeneous appearance. This visual impression was confirmed by measurements of individual synaptic vesicles in all the 229 profiles (Fig. 3I; supplementary material Fig. S6B).
We also noticed a higher incidence of mitochondria present in the ShhN-presynaptic profiles (Fig. 3F,G; supplementary material Fig. S6C–E). In the 229 control presynaptic terminals analyzed, 56 of them had mitochondria, whereas in the 229 ShhN terminals analyzed, 72 of them had, or were even filled with mitochondria (supplementary material Fig. S6F). Thus, these analyses provide evidence at the subcellular scale that Shh signaling regulates the size and ultrastructure of presynaptic terminals and synaptic vesicles of hippocampal neurons.
Shh signaling changes the physiology of hippocampal neurons
Having observed the changes in the structure of presynaptic terminals and synaptic vesicles, we wished to explore the physiology of the ShhN-exposed neurons. Whole-cell patch-clamp recording was used to examine the intrinsic properties of these neurons. The ShhN-treated neurons did not exhibit a significant change in either resting membrane potential or capacitance, although their input resistance was increased (supplementary material Table S1). The ShhN-treated neurons displayed the same threshold for firing action potential (AP) as the control neurons (supplementary material Table S1), suggesting that ShhN treatment had no impact on the quality or localization of voltage-gated sodium channels. However, the ShhN neurons reached their AP threshold more easily than the control neurons did, requiring less injected current to fire a single AP (Fig. 4A). Moreover, we found that with increasing current amplitude steps, the AP frequency (Fig. 4B) and the cumulative number of APs (Fig. 4C) in ShhN-treated neurons were both significantly greater than in control neurons, confirming that ShhN increases neuronal excitability.
We next examined the synaptic properties of the ShhN neurons by measuring their spontaneous miniature excitatory postsynaptic currents (mEPSCs). Notably, the frequency of mEPSCs from the ShhN-treated neurons was approximately twice that of the control neurons (Fig. 4D,E), indicating increased neurotransmitter (glutamate) release probability. The amplitude of mEPSCs, on the other hand, was similar between the ShhN-treated neurons and the control neurons, as measured by average amplitude and 10–90% rise-time or 50% decay (Fig. 4F; supplementary material Table S1). The unchanged mEPSC amplitude might imply that enhanced Shh activity did not perturb the number of activated glutamate receptors at the postsynaptic terminals. Alternatively, given an increase in the size of the synaptic vesicles in the ShhN-treated presynaptic terminals, it is also possible that there was a compensatory change in the postsynaptic terminals. Thus, these experiments show that the physiological properties of the ShhN neurons correlate with the changes in the morphology and structure of their presynaptic terminals.
Taken together, we demonstrated that Shh signaling activity induces the growth of functional presynaptic terminals in hippocampal neurons. Shh is a mitogen for neural stem cells (Lai et al., 2003; Palma et al., 2005; Breunig et al., 2008; Han et al., 2008), but our work reveals that postmitotic hippocampal neurons are also regulated by Shh. While this work was under review, a paper by Harwell et al. was published that reported that Shh is involved in synapse formation of corticofugal projection neurons (Harwell et al., 2012). Our conclusion about the importance of Shh activity for regulating synapse development is fully consistent with that of Harwell et al., and the similar results obtained with different types of neurons using very different experimental approaches makes the conclusion even more meaningful and physiologically relevant. It is worth noting that the findings by Harwell et al. also support our previous observations of the dendritic source of Shh (Petralia et al., 2011b) in hippocampal neurons and the negligible Ptch and Smo on the axons (Petralia et al., 2011a; Petralia et al., 2012).
Our data suggest the involvement of Shh-induced upregulation of the presynaptic active zone protein Bassoon in the genesis of presynaptic terminals. However, when we analyzed the regulatory sequences for Bassoon (http://www.gene-regulation.com/cgi-bin/pub/programs/match/bin/match.cgi), we did not find any consensus Gli binding sites in the 10 kb upstream nor in the 100 bp downstream from the transcription start site. This is not totally surprising given that neuropilin 1, another neuronal protein, also lacks the Gli binding sites despite its upregulated expression in response to Shh pathway activation (Hillman et al., 2011). The Shh-induced Bassoon expression could be mediated by an intermediate transcription factor. Indeed, our analysis of Bassoon sequences found 11 potential binding sites for N-Myc, a transcription factor whose expression is increased upon Shh pathway activation (Kenney et al., 2003; Oliver et al., 2003).
The currently accepted model is that the primary cilium is essential for Shh signaling transduction to take place. Like other types of cells and other types of neurons, cultured hippocampal neurons contain the primary cilium (supplementary material Fig. S7). Our data presented here however do not address whether the primary cilium plays a role in the Shh-induced presynaptic growth. Nevertheless, a recent in vivo study demonstrated that deletion of the primary cilium disrupts synapse formation in adult-born dentate granule cells of the hippocampus (Kumamoto et al., 2012). This raises the possibility that the primary cilium might serve a functional role in the Shh signaling-mediated synapse development in hippocampal neurons in addition to that of the adult-born dentate granule cells.
Materials and Methods
All animal procedures were approved by the National Institute on Aging and National Institute on Deafness and Other Communication Disorders Animal Care and Use Committees, and complied with the National Institutes of Health Guide for Care and Use of Laboratory Animals.
Shh and related reagents
HEK 293 cells overexpressing Shh-N (N-terminal fragment of Shh) were kindly provided by James K. Chen (Stanford University). Shh-N-conditioned medium from the HEK 293 cells was prepared exactly as described (Chen et al., 2002b). Control medium was obtained from untransfected HEK 293 cells. Throughout this study, we use ShhN to refer to the Shh-N-conditioned medium. ShhN or control medium was used at 10%. The bioactivity of ShhN was validated using the Shh-Light2 assay (Taipale et al., 2000) (supplementary material Fig. S1A). The increased expression of Ptch (supplementary material Fig. S1B) (Rohatgi et al., 2007) and nuclear enrichment of Gli2 (data not shown) (Kim et al., 2009) confirmed the activity of ShhN in activating the Shh signaling pathway in cultured hippocampal neurons.
SAG, cyclopamine and purmorphamine were from Axxora (San Diego). SmoA plasmid was provided by Jeremy F. Reiter (University of California San Francisco). Gli2 plasmid was from Addgene (plasmid 17648) (Roessler et al. 2005).
Antibodies and other reagents
The antibodies to Bassoon (catalog no. 141002), ZnT3 (no. 197001), Gephyrin (no. 147011), Synaptotagmin1 (luminal; no. 105101), and VGAT (luminal; no. 131103) were from Synaptic Systems (Germany). Synapsin1 (no. AB1543P) and VGAT (no. AB5062P) antibodies were from Chemicon International. VGlut (no. AB5905) and GAD65 (no. MAB351) antibodies were from Millipore. PSD95 antibody was from Affinity BioReagents (no. MA1-046). Synaptophysin (no. S5768) and actin (no. 5441) antibodies were from Sigma.
The secondary antibodies for immunofluorescence labeling were species-appropriate IgG conjugate with either Alexa Fluor 488 or 568 (Molecular Probes). The secondary antibodies for immunoblots were peroxidase-conjugated species-appropriate IgG (Jackson ImmunoResearch Laboratories, Inc.).
Synaptophysin::EGFP was from Jane Sullivan (University of Washington). PSD95::EGFP was from David S. Bredt (Johnson & Johnson).
Cultured hippocampal neurons
Cultures of hippocampal neurons were prepared from embryonic day 18 rat brains as described previously (Mattson et al., 1989; Kaech and Banker, 2006; Bushlin et al., 2008). ShhN was added to neurons between 10 and 19 days in vitro (div) and experiments were performed 2–3 days later. Transfection was carried out using a calcium-phosphate-based kit (Clontech, catalog no. 631312) (Hering and Sheng, 2003; Jiang and Chen, 2006; Bushlin et al., 2008).
Immunocytochemistry, antibody uptake, fluorescence microscopy and image analysis
Immunofluorescence labeling was carried out as described previously (Bushlin et al., 2008).
For internalization of Synaptotagmin1 luminal antibody, the antibody (1:20) was incubated with neurons in 45 mM KCl-saline at 37°C for 5 min. Following washes in 0 mM Ca2+-saline, the neurons were fixed with 4% paraformaldehyde and permeabilized in 0.1% Triton X-100. After incubation with Alexa-Fluor-568-conjugated secondary antibody, the neurons were washed and mounted in ProLong antifade.
For internalization of VGAT lumenal antibody, neurons were incubated with VGAT-antibody conjugated with the fluorophore Oyster 550 (5 µg/ml) in 45 mM KCl-saline at 37°C for 5 min. The neurons were then fixed, washed, and mounted.
The labeled neurons were examined using a 40× or a 63× objective on a Zeiss LSM510 laser scanning confocal microscope. All images were acquired at a 1024×1024 pixel resolution and each image was an average of four scans. The confocal acquisition settings were kept the same for those samples when quantification was performed. The brightness and contrast of the images were minimally adjusted (in Adobe Photoshop 8.0) for those images presented. No additional digital image processing was performed.
We enforced rigorous standards for biological sampling. For light microscopy experiments, we performed more than five independent cultures for each experiment with four or five coverslips per culture. In each coverslip, we randomly selected neurons but carefully excluded those formed clusters with other neurons or glia and those displayed an unhealthy appearance of soma or exhibited frayed or fragmented processes.
We used two methods to analyze the number and size of synaptic puncta. For immunolabeled neurons, we used a customized Plugin in ImageJ that provided automatic measurements of puncta number (per unit area) and size with identical threshold settings. We reasoned that counting puncta number per unit area instead of per unit length was appropriate because some synapses were located on the soma. For transfected neurons, we used a different Plugin in ImageJ that was tailored to quantify puncta along a defined unit length of a neurite. The number of synaptic puncta is influenced by a number of factors including cell and neurite density. For this reason, we expressed the data for puncta number measurements as the percentage of control to minimize variation among cultures. The size of synaptic puncta varies (0.2–2 µm in diameter) depending on the synaptic markers and fluorophores used. Therefore, we expressed the data for size measurements also as the percentage of control.
Electron microscopy was performed as described (Petralia and Wenthold, 1999; Petralia et al., 2010). Briefly, neurons were fixed in 2% paraformaldehyde/2% glutaraldehyde in phosphate buffer at room temperature for 30 minutes. Following washes, they were postfixed in 1% osmium tetroxide in cacodylate buffer for 30 minutes, and dehydrated in an ethanol series (including 10 minutes in 1% uranyl acetate in 50% ethanol), and then in propylene oxide and embedded in Epon. The glass coverslip was removed with hydrofluoric acid and thin sections were stained with 0.03% lead citrate for 3 minutes, and were examined in a JEOL JSM-1010 electron microscope.
From two independent cultures, we photographed 215 randomly selected fields from the control and 192 from the ShhN-treated samples. Traceable presynaptic profiles from these micrographs were selected for further analyses by an observer (P.J.Y.) who was unaware of the identity of the groups until after measurements. The number of synapses analyzed is indicated in the legend for Fig. 3.
Whole-cell patch-clamp recordings were performed using an Axopatch 200B amplifier (Axon Instruments). The extracellular recording solution contained (in mM): NaCl, 124; KCl, 5; HEPES, 25; MgCl2, 1; CaCl2, 2; D-glucose, 30; (pH ∼7.35, adjusted to 300–315 mOsm). The intracellular pipette solution contained (in mM): K-gluconate, 120; NaCl, 5; EGTA, 0.1; HEPES, 10; KCl, 20; MgCl2, 4; phosphocreatine, 10; Mg-ATP, 4; Na-GTP, 0.3 (pH 7.3, adjusted to 290 mOsm). Membrane voltages were sampled at 10 kHz and filtered at 2 kHz using Pclamp 10 Software, and recordings were analyzed with Clampfit 10.2 software (Axon Instruments).
Intrinsic excitability protocols were adapted from Li et al. and Bolton et al. (Li et al., 1998; Bolton et al., 2000). Only cells exhibiting a resting membrane potential between −60 and −70 mV, access resistance <20 mΩ, and input resistance >150 mΩ, were used for analysis. Capacitance and series resistance were compensated for by >80%.
mEPSCs were recorded in the voltage-clamp whole-cell mode (Vh = −70 mV) in the presence of 20 µM Bicuculline methiodide (Tocris; Ellisville, MO, USA) and 1 µM TTX (Tocris). mEPSCs were measured over a period of 120 sec. Mini-detection software (MiniAnalysis Program, Synaptosoft, NJ) was used to analyze mEPSC frequency, amplitude, and kinetics. Detection parameters were established based on amplitude threshold (4×RMS), and area half-width. Noise levels varied between records, but thresholds were always between 8 and 10 pA.
Statistical significance was determined by the Student's t-test. All results were expressed as mean ± s.e.m.
We are grateful to James K. Chen for many helpful discussions and reagents. We are also grateful to Wayne Rasband for customized ImageJ Plugins that made our image analyses unbiased and efficient.
This work was supported by the Intramural Research Programs of the National Institute on Aging/National Institutes of Health; and the National Institute on Deafness and Other Communication Disorders/National Institutes of Health. Deposited in PMC for release after 12 months.