Synapsin Utilization Differs among Functional Classes of Synapses on Thalamocortical Cells

Electrophysiology.

Mice with double knock-out mutation of the synapsin I and II genes and wild-type mice with the same genetic background were used in this study. Animals were prepared according to the guidelines and with the approval of the Animal Care Committee in Norway. Animals, 17–20 d old, were killed by rapid decapitation. A block of the brain containing the thalamus was dissected out and cooled to below 5°C in artificial CSF (ACSF) containing the following (in mm): 125 NaCl, 2.5 KCl, 2 CaCl₂, 1 MgCl₂, 25 NaHCO₃, 1.25 NaH₂PO₄, and 25 d-glucose (bubbled with 5% CO₂/95% O₂). Brain slices of LGN were cut from the block at a thickness of 250 μm, in a semi-parasagittal plane as described previously (Turner and Salt, 1998), using a vibroslicer (Sigmann Electronics, Hüffenhardt, Germany). The slices were stored submerged in ACSF at room temperature (∼24°C) until recording and were used for recording within 5 h.

During experiments, slices were kept submerged in a small chamber (volume, ∼1.5 ml). The chamber was perfused with ACSF at the rate of 3 ml/min at a temperature of 28 ± 1°C. Picrotoxin (80–100 μm; Sigma-Aldrich, St. Louis, MO) was added to the perfusion solution to block GABA_A-receptor-mediated synaptic currents, and 50 μm d-APV (Tocris Cookson, Avonmouth, UK) was added during experiments with train stimulus protocols to prevent induction of NMDA-receptor-mediated forms of synaptic plasticity. Cells were visualized with differential interference contrast using infrared light. Whole-cell voltage-clamp recordings were obtained with glass electrodes filled with the following (in mm): 80 Cs₂SO₄, 8 NaCl, 0.5 MgCl₂, 0.5 CaCl₂, 3 Na₂ATP, 10 HEPES, 5 EGTA, and 0.1 D-600 (blocks L-type Ca²⁺ channels; Sigma-Aldrich). The pH was adjusted to 7.2 with CsOH. Cs₂SO₄ was added to block GABA_B currents and to make the cells electronically more compact by blocking K⁺ channels. The access resistance (11 ± 3 MΩ) was continuously monitored during the experiments, and it varied <15% during an experimental series. A maximum of three recordings were done in each slice. Postsynaptic currents were evoked by electrical stimulation of the retinogeniculate and corticogeniculate fibers with bipolar electrodes, using 100- to 300-μs-long current pulses of 5–100 μA (Turner and Salt, 1998). The stimulus artifacts are truncated in the figures. The recorded electrical signals were low-pass filtered at 3 kHz and sampled at 10 kHz with a Multiclamp 700A amplifier and pClamp software (Molecular Devices, Palo Alto, CA). The amplitude of the postsynaptic currents was calculated as the difference between the peak synaptic current and the baseline holding current or the decaying current of a preceding synaptic event. Off-line analysis was done with Igor Pro (WaveMetrics, Lake Oswego, OR) in combination with the NeuroMatic tool packet (ThinkRandom) and custom-made functions.

Morphology.

The mice were anesthetized with Nembutal, perfused transcardially for 2 min with ACSF, and subsequently for 20 min with a fixating solution containing 4% paraformaldehyde and 1% glutaraldehyde dissolved in 0.1 m phosphate buffer at pH 7.4. A block of the brain was dissected out and postfixed for 2–3 h. The LGN was cut in coronal sections of 100 μm using a vibroslicer (Sigmann Electronics). The sections were treated with 1% NaBH₄ to terminate the cross-linking actions of the fixatives and stored at 4°C in PBS (0.01 m phosphate buffer and 9% NaCl, pH 7.4), including 0.05% sodium azide to prevent microbial growth.

For immunocytochemistry, the following primary antibodies were used: polyclonal goat anti-synapsin Iab (1:100 dilutions; Santa Cruz Biotechnology, Santa Cruz, CA), and rabbit anti-synapsin IIab (0.1 μg/ml; StressGen Biotechnologies, Victoria, British Columbia, Canada). Before antibody incubation, the sections were rinsed in PBS and treated in 1% bovine serum albumin (BSA) dissolved in PBS for 30 min. The primary antibodies, the specificity of which had been confirmed by Western blotting of total brain extracts (data not shown), were diluted in PBS with 1% BSA and 0.05% sodium azide. The incubation lasted for 3 d at room temperature. Sections were rinsed in PBS and incubated for 2 h in biotinylated anti-goat or anti-rabbit secondary antibody (1:100, in PBS; Vector Laboratories, Burlingame, CA), followed by 2 h in HRP-conjugated avidin–biotin complex (ABC; Vector Laboratories). Immunoreactivity was visualized using 0.05% diaminobenzidine and 0.001% H₂O₂. For all antibodies, deletion of the primary antibody in the protocol eliminated the specific staining discernible at the electron microscopic level.

Immunostained sections were fixed with 1% OsO₄ for 1 h. The sections were briefly immersed in 50% ethanol, incubated in 4% uranyl acetate diluted in 50% ethanol for 1 h, and finally dehydrated in a series of ethanol. The sections were rinsed in acetone and saturated with Epon (Electron Microscopy Sciences, Fort Washington, PA) by first incubating them in equal amount of acetone and Epon and next in pure EPON. The sections were flat embedded between two pieces of Aclar sheet (Ted Pella, Redding, CA). The areas to be analyzed by electron microscopy were identified using camera lucida, encapsulated with Epon, and polymerized at 60°C for 2 d. Ultrathin sections of 80 nm were cut using an ultra microtome (Ultracut UC7; Leica, Nussloch, Germany). The cutting angle was oriented near parallel to the surface of the embedded tissue to maximize the area of the tissue–Epon interface within the sections. This approach permitted the most efficient detection of immunolabeled profiles because antibody penetration is limited to the surface layer of the tissue.

Ultrathin sections were examined in an electron microscope (JEOL-1010; Japan Electron Optic, Tokyo, Japan), and micrographs of regions with labeling were taken at 12,000–15,000× magnification. The negatives were scanned to digitize at 600 dots per inch (dpi) and examined at a final magnification of at least 60,000× using Image-Pro Plus 4.0 (Media Cybernetics, Silver Spring, MD). Additional photographs were captured using a 16 megapixel digital camera (SIA) at 10,000–15,000× magnifications. Retinal (RL), presumed cortical (RS), and inhibitory (F) terminals were identified by their specific morphology (Steriade et al., 1997). For vesicle density analysis, the cross-section area of each terminal (excluding the area occupied by mitochondrion or membrane-limited inclusions typically encountered in RL terminals) was measured using Image-Pro Plus, and the vesicles contained within the terminal were counted. To compute intervesicle distances, we determined the x–y coordinates of each vesicle within each terminal and calculated the intervesicle distances from these coordinate values. Using a spreadsheet algorithm, designed by Dr. Barry Condron (University of Virginia, Charlottesville, VA), we computed the intervesicle distances between every possible vesicle pairs within each terminal and extracted the shortest distance obtained for each vesicle. Pictures for illustrations were prepared by scanning the negatives at 2400 dpi to compose enlarged images at 400 dpi.

Statistical analyses.

Tests of statistical significance were done using t test or ANOVA with Bonferroni's correction (SSPS software; SPSS, Chicago, IL).

Electrophysiology

We used whole-cell voltage-clamp technique to record EPSCs evoked in thalamocortical cells by electrical stimulation of afferents from retina and visual cortex and compared transmission properties in synapsin I and II double knock-out mice with those in wild-type mice.

Single-pulse stimulations of optic tract fibers in the wild-type mice gave EPSCs in thalamocortical cells (n = 20) that had an average ± SEM amplitude of −862 ± 153 pA, a rise time (10–90%) of 0.48 ± 0.009 ms, and a decay time constant of 1.43 ± 0.007 ms. For cells (n = 20) in the double knock-out mice, the corresponding EPSCs had a mean ± SEM amplitude of −829 ± 213 pA, a rise time of 0.48 ± 0.015 ms, and a decay time constant of 1.45 ± 0.010 ms. We found no statistically significant difference between the two genotypes for any of these parameters (range of p values, 0.32–0.62). Single-pulse stimulations of cortical afferents in the wild-type mice elicited EPSCs in thalamocortical cells (n = 15) that had an average ± SEM amplitude of −48.4 ± 5.7 pA, a rise time of 1.62 ± 0.024 ms, and a decay time constant of 2.64 ± 0.052 ms. For cells (n = 14) in the double knock-out mice, the average ± SEM EPSC had an amplitude of −52.2 ± 9.3 pA, a rise time of 1.60 ± 0.027 ms, and a decay time constant of 2.57 ± 0.044 ms. For neither of these parameters did we observe any statistically significant differences between the wild-type and the double knock-out mice (range of p values, 0.85–0.89).

Repetitive stimulation of retinal afferents in normal rats and mice gives synaptic depression in thalamocortical cells (Turner and Salt, 1998; Chen et al., 2002; Kielland and Heggelund, 2002). We used both paired-pulse stimulation with interpulse intervals in the range 10–500 ms (Fig. 1A) and pulse-train stimulation with 300 pulses delivered at 10 Hz (Fig. 1B) and found synaptic depression in both double knock-out and wild-type mice with both types of stimuli. Interestingly, there was no statistically significant difference between the degrees of depression in the two genotypes (range of p values, 0.22–0.24).

We also tested whether there was any difference in posttetanic potentiation (PTP) at the retinogeniculate synapses between the two types of mice using a tetanic stimulus of 100 pulses delivered at 100 Hz. A possible potentiation was examined by evoking EPSCs at a frequency of 0.1 Hz for a period of 2.5 min after the tetanic stimulation. For both wild-type and double knock-out mice, the size of the EPSCs did not increase after the tetanic stimulation. Accordingly, at the retinogeniculate synapse, PTP was present in neither the wild-type nor the double knock-out mice (data not shown). In summary, these data indicate that gene inactivation of synapsin I and II have no effects on short-term plasticity in the retinogeniculate synapses.

Repetitive stimulation of corticogeniculate afferents in normal animals gives synaptic facilitation in thalamocortical cells (Wrobel and Lindström, 1990; Turner and Salt, 1998; von Krosigk et al., 1999). We found such facilitation in both double knock-out and wild-type mice. However, the strength of the facilitation differed between the two genotypes. Paired-pulse effects were studied with interpulse intervals in the range of 10–500 ms, as in the studies of the retinal input. In the wild-type mice, we found that the facilitation was most pronounced at the shortest interpulse interval with gradually less facilitation during prolongation of the intervals (Fig. 2A, filled spots). The time course did not show a simple exponential pattern, which suggests that different biological processes are involved (Granseth et al., 2002). In the double knock-out mice, the paired-pulse facilitation at the corticogeniculate synapse showed similar time course as in the wild-type mice, but the facilitation was considerably stronger in the double knock-out mice (Fig. 2A, open spots). The facilitation was ∼50% stronger at intervals in the range of 10–100 ms. At all of the interpulse intervals from 10 to 200 ms, there was significant differences between the two genotypes (p < 0.005), but, for the 200 and 500 ms interpulse intervals, the differences were not statistically significant.

Stimulation of the corticogeniculate fibers in wild-type mice with a train of 300 shocks at 10 Hz gave a large increase in the amplitude of the first EPSCs (Fig. 2B). The EPSCs continued to increase but less pronounced for the next 50–70 responses. Thereafter, the EPSCs gradually declined in size (Fig. 2B, filled spots). The synaptic responses reached maximal amplitudes after ∼5 s with EPSCs that were ∼2.7 times larger than the EPSCs recorded before the start of train stimulation. At the end of the train, the amplitudes of the EPSCs were ∼1.5 times larger than the pre-train responses. In the double knock-out mice, the same type of train stimulation gave a large increase in the amplitude of the second EPSC of the train. However, immediately thereafter, the synaptic responses began to decrease (Fig. 2B). Relative to the pre-train responses, there was an average increase of amplitude by a factor of 3.1 at the second EPSC, 2.5 at the third EPSC, and 2.2 at the fourth EPSC. For the rest of the train period, the EPSCs declined with a near-linear time course and reached pre-train level after ∼25 s (Fig. 2B, open spots). The difference obtained between the wild-type and the double knock-out mice in the pattern of EPSCs during train stimulation was statistically significant (p = 0.019).

The difference between the two genotypes revealed during train stimulation suggests that the deletion of synapsin I and II can affect other types of synaptic plasticity besides facilitation. We therefore also tested for a difference with respect to PTP using the same stimulus protocol as used in the studies of the retinal input, i.e., 100 shocks at 100 Hz followed by test stimuli at a frequency of 0.1 Hz for 2.5 min (Fig. 3). In the wild-type mice, the mean amplitude of the EPSC at the first test stimulus increased by a factor of 3.9 relative to the pretetanic EPSCs (Fig. 3, filled spots). The test responses declined thereafter, with a time constant of 19.2 s. At the end of our recording period, the average EPSC was still 1.45 times the pre-train level. This demonstrated the presence of PTP at the corticogeniculate synapses in the normal mice. The double knock-out mice also showed PTP at corticogeniculate synapses, but the potentiation was significantly weaker (p = 0.003) (Fig. 3, open spots). At the first test stimulation, 10 s after the end of the tetanic stimulation, the amplitude of the mean EPSC had increased by a factor of 2.1 relative to the pretetanic EPSCs. The responses declined gradually with a time constant of 27.3 s, and reached pre-tetanic level after ∼1.5 min. Thus, in the late stage of the 10 Hz train and after the induction of PTP, we found weaker synaptic responses in the double knock-out mice compared with the wild-type mice at the corticogeniculate synapse, and these results are consistent with a putative role of synapsins in the regulation of availability of vesicles for the transmitter release process (Greengard et al., 1993; Hosaka et al., 1999).

Morphology

The physiological differences in synaptic plasticity between the double knock-out and the wild-type mice at the corticogeniculate synapses, together with the lack of differences at the retinogeniculate synapses, suggested differential utilization of synapsins at the two types of synapses. To analyze this further, we studied the distribution of synapsin I and II in the two types of terminals in wild-type mice using immunohistochemical labeling and electron microscopy. Retinal and cortical terminals can be distinguished at the ultrastructural level because of their characteristic appearance (Steriade et al., 1997). Retinal terminals have round vesicles, are large (RL terminals), contain pale mitochondria, and form asymmetric synapses. Cortical terminals have round vesicles but are small (RS terminals), contain no or a few dark-appearing mitochondria, and form asymmetric synapses. Terminals from inhibitory neurons have flattened vesicles (F terminals), dark mitochondria, and form symmetrical synapses.

We stained sections of LGN from the wild-type mice with antibodies specific for either synapsin Ia and Ib or synapsin IIa and IIb. Every micrograph used in our immunohistochemical analysis had at least several labeled profiles, confirming that antibody penetration did occur in the actual regions. Immunostaining of the double knock-out mice confirmed the specificity of the antibodies because sections from these animals were devoid of labeling for either type of antibody.

Labeling with antibodies toward synapsin I was examined in 44 micrographs from three mice, and all terminals in the micrographs were classified as far as possible. Within these micrographs, we encountered 89 labeled profiles of 209 recognized terminals. Labeling was found in 53 RS terminals (Fig. 4A,D) and 36 F terminals (Fig. 4B,D), whereas no labeling was found in RL terminals (Fig. 4D).

Labeling with antibodies toward synapsin II was examined in 32 micrographs from two mice. Within these micrographs, we encountered 64 labeled profiles of 171 recognized terminals. All of the labeled terminals were of the RS variety (Fig. 4C,D). Finally, the overall distribution of the different terminals, labeled and unlabeled, were 8, 60, and 32% for the RL, RS, and F types, respectively (Fig. 4D). These percentages are consistent with the expected distribution of input terminals in mice LGN based on what was described previously in detailed quantitative studies in cat (Montero, 1991; Erisir et al., 1998), confirming the validity of our identification criteria.

The lack of labeling of RL terminals could potentially be attributable to unsuccessful antibody penetration of these terminals. However, the simultaneous lack of staining of the RL terminals and positive staining of the RS terminals with antibodies against synapsin I and synapsin II observed in the same sections argues against this explanation. Moreover, the immunohistochemical data are consistent with our electrophysiological data.

Because synapsins presumably tether vesicles in the terminals, knock-out of the synapsins would be expected to reduce the density of vesicles (Rosahl et al., 1995; Takei et al., 1995; Gitler et al., 2004; Samigullin et al., 2004) and increase the distance between vesicles in the terminals. We studied these parameters to look for morphological changes induced by synapsin I and II gene inactivation. Qualitatively, we saw no pronounced differences in the density of vesicles between terminals from wild-type and double knock-out mice (Fig. 5A–C). However, careful quantitative analysis revealed significant differences for the RS terminals but not for the RL terminals, consistent with a differential expression of synapsins in the two types of terminals. For each terminal, we computed vesicle density by measuring the cross-section area, excluding mitochondrial areas, and counted the number of vesicles. We analyzed vesicle density in 87 RL and 71 RS terminals from five wild-type mice and 83 RL and 91 RS terminals from five double knock-out mice. In the RL terminals, the average ± SEM vesicle density was 104.7 ± 3.8 vesicles/μm² in the wild-type and 100.5 ± 3.8 vesicles/μm² in the double knock-out mice (Fig. 5D), and this difference was not statistically significant (p = 0.21). In contrast, in the RS terminals, the vesicle density differed significantly between the double knock-out mice and the wild-type mice. In the double knock-out mice, the average ± SEM density was 106.0 ± 3.7 vesicles/μm² and in the wild-type was 127.9 ± 4.1 vesicles/μm² (p = 0.001) (Fig. 5D). An important and surprising finding expressed by these values is that the density of vesicles in the RS terminals of the double knock-out mice was not different from the density in the RL terminals of both genotypes (Fig. 5D). This finding strongly supports the conclusion that synapsin I and II are not expressed in the RL terminals.

The reduced vesicle density in the RS terminals of the double knock-out mice could reflect a change in vesicle tethering, but, alternatively, it could simply reflect a change in number of vesicles or in terminal area (Fig. 6A, inset). We therefore compared the RS terminals in the double knock-out and the wild-type mice with respect to these parameters. The average ± SEM number of vesicles per RS terminal in the double knock-out was 27.2 ± 1.7 (n = 90) and in the wild-type mice was 28.2 ± 2.2 (n = 70), and the difference between these averages was not statistically significant (p = 0.35). However, the average ± SEM area of the RS terminals in the double knock-out mice (0.256 ± 0.012 μm²; n = 90) was larger (Fig. 6A) than the area in the wild-type mice (0.219 ± 0.012 μm²; n = 70), and this difference was statistically significant (p = 0.022). Accordingly, the increased area of the RS terminals in the double knock-out mice compared with the wild-type could account for the decreased vesicle density in these terminals. For the RL terminals, there was no difference between the two genotypes with respect to number of vesicles or terminal areas.

To address the question of vesicle tethering more thoroughly, we measured the distance between each vesicle and its closest neighbor in RL and RS terminals from four wild-type and four double knock-out mice. The ranges of vesicle numbers in each RS and RL terminal were 15–85 and 40–260, respectively. In RL terminals, the average ± SEM intervesicle distance was 54.5 ± 1.7 nm (n = 25) in the wild-type and 57.5 ± 1.4 nm (n = 27) in the double knock-out mice, and this difference was not significantly different (p = 0.18). In the RS terminals, on the contrary, the average intervesicle distance differed significantly between the two genotypes. In the RS terminals of the wild-type mice, the average ± SEM intervesicle distance was 46.4 ± 1.0 nm (n = 24) and in the double knock-out mice was 55.1 ± 1.1 nm (n = 27; p = 0.0001). As for vesicle density, the average intervesicle distance in the RS terminals of the double knock-out mice was similar to the average distance in the RL terminals of the double knock-out mice and the wild-type mice (Fig. 6B). The data on vesicle density and intervesicle distance strongly support the conclusion that the synapsins are present in the corticogeniculate terminals but not in the retinogeniculate terminals. Furthermore, these data are consistent with a function of synapsins in tethering of synaptic vesicles and of increasing the density within the vesicle clusters in the nerve terminals.