To induce LTD of PF-PC synaptic transmission, we applied conjunctive stimulation composed of PF-burst stimuli (5 stimuli at 100 Hz) and somatic depolarization (50 ms). Ninty conjunctive stimulations at 0.5 Hz for 3 min (Fig. 1a, top panel) induced LTD (Fig. 1b, c, top panels; d, 66.3 ± 4.8%, n = 9; 26–30 min), while PC somatic depolarization alone for 90 times did not alter EPSC amplitude (Fig. 1c, top panel; d, 99.4 ± 1.5%, n = 8; 26–30 min). This decrease in EPSC amplitude was significant (P < 0.001), and was consistent with the findings of a prior study [11]. Then we investigated the effects of a smaller number of conjunctive stimulations on PF-PC synaptic plasticity. Thirty times application of the conjunctive stimulation at 0.5 Hz induced LTP (Fig. 1a – c, bottom panels; d, 133.0 ± 5.4%, n = 11, P < 0.001;) instead of small LTD, while 30 times depolarization alone did not change EPSC amplitude (Fig. 1c, bottom panel; d, 99.8 ± 2.1%, n = 10). Additionally, neither 30 times nor 90 times stimulation of PF bust alone induced LTP or LTD (Fig. S1).
Two types of LTP at PF-PC synapses are known: one type is increase in neurotransmitter release from the presynaptic terminal [19], and this presynaptic type of LTP was induced by PF stimulation alone at a relatively high frequency (4–8 Hz). The other type of LTP was induced by PF stimulation alone at a relatively low frequency (1 Hz) [20], and was caused by exocytic insertion of AMPAR at the postsynaptic membrane of PC [21]. To investigate the which mechanism, presynaptic or postsynaptic, was underlying the LTP found in the present study, paired-pulse ratio (PPR), the relative ratio of the second EPSC amplitude to the first EPSC amplitude, was compared between before and after conjunctive stimulation. Thirty conjunctive stimuli, which caused LTP, did not significantly change PPR (PPFbefore 1.44 ± 0.10; PPFafter 1.43 ± 0.04, P > 0.949) before and after conjunctive stimuli (Fig. 2a upper, b left). Consistent with prior studies [20], PPR was not changed between before and after LTD induction by 90 conjunctive stimuli (1.34 ± 0.02 vs. 1.44 ± 0.08, P > 0.2) (Fig. 2a lower, b right). Thus, the no change in PPR indicated that neither of those LTP nor LTD caused presynaptic changes (Fig. 2) [20, 21] .
The no changes in PPR before and after LTP induction strongly suggest that this type of LTP may share the mechanisms of postsynaptic LTP. Since postsynaptic LTP depends on nitric oxide (NO) [20, 21], we attempted to block mediation via NO. A water-soluble type of NO scavenger, 2-(4-carboxypheny)-4,4,5,5,-tetramethilimidazoline-1-oxyl-3-oxide (carboxy-PTIO, 30 µM) was added to the perfusing ACSF. LTP was entirely blocked by the NO scavenger (Fig. 3a, b), suggesting that the LTP, induced by a small number (30 times at 0.5 Hz: Fig. 1c upper) of conjunctive stimulations, was similar to postsynaptic LTP. Importantly, under NO-free conditions, this 30-time conjunctive stimulation rather elicited a decrease in its EPSC-amplitude (Fig. 3a, b) (72.1 ± 3.7%, n = 8). To examine whether this decrease of EPSC shares some mechanisms with conventional LTD, which is known to depend on PKC activity [22, 23], we attempted to block PKC activity. Under NO-free conditions, the addition of a potential PKC inhibitor, Gö6976 (Gö), into the internal solution (0.3 µM) blocked LTD (100.8% ± 1.8%, n = 5), which indicated that this LTD-like phenomenon was also mediated by PKC activity. Thus, we called this LTD-like phenomenon under NO-free conditions “hidden LTD.” Together, these findings suggested that the LTP induced by 30-time conjunctive stimulation shared a common molecular mechanism with postsynaptic LTP, and hidden LTD also shared common properties with conventional LTD, at least partly.
Next, we examined whether this type of LTP was induced by physiological stimulation. We applied a conjunction of PF and CF stimulation under current clamp conditions using K+-based internal solution. Three hundred times conjunction of PF and CF stimulation at 1 Hz is known to induce LTD [12, 16]. When 60 times conjunction of PF and CF stimulation was applied, a small but significant increase in EPSC-amplitude (111.2 ± 3.5%, n = 8) was induced. This increase in EPSC amplitude was statistically significant compared to that of 60 times CF stimulation alone (94.5 ± 2.3%, n = 5, P < 0.01) (Fig. 4). Thus, physiological conjunctive stimulation consisting of PF and CF stimulation using a small number of stimuli could also induce LTP.
A number of reports have suggested the importance of Ca2+ concentration for cerebellar LTD and LTP induction [15, 21, 24]. Therefore, we conducted Ca imaging of PC dendritic region during the conjunctive repetitive stimulation. Representative examples of the Ca signals, in response to conjunctive stimulation to CF and PF applied 90 times at 0.5 Hz (Fig. 5b), are shown in Fig. 5c. These Ca signals consisted of a sharp peak followed by baseline responses (Fig. 5c, the rightmost panel). Since the conjunctive stimulation was given at 0.5 Hz and Ca signals were imaged at 0.1 s intervals, one peak response was observed every 20 frames. The average of the Ca signals obtained from 30 observation points from 10 different slices indicated that the peak signal intensities gradually emerged in the dendritic region of PCs and reached their maximum values at the end of the repetitive stimuli, while the baseline signal intensities showed an initial increase followed by a gradual decrease during the repetitive stimuli (Fig. 5d, left panel). The time courses of Ca2+ peak responses taken from 10 slices are indicated in the right panel of Fig. 5d (right panel, pink color denotes the mean). To compare the Ca concentration at the time point of 30- and 60-times stimulation, the latest 5 or 30 stimulations (26th to 30th vs. 86th to 90th or 1st to 30th vs. 61st to 90th, respectively), were selected. The averages of peak ΔF/F observed in the 26th to 30th and the 86th to 90th stimuli (solid pink bars, bottom of Fig. 5d, right panel) were not significantly different (10.6 ± 1.11 and 11.2 ± 0.899%, respectively, n = 10, P > 0.6, paired t test, Fig. 5e, left panel). On the other hand, the average of the peak responses observed in response to the first to 30th stimulus was significantly smaller than that observed in response to the 61st to 90th stimulus (empty bars, bottom of Fig. 5d, right panel) (7.16 ± 0.711 and 11.0 ± 0.887%, respectively; n = 10, P < 0.001, paired t test, Fig. 5e, right panel).
Finally, we investigated the Ca signals during PF or CF stimulation alone, which did not induce either LTP or LTD. The patch clamp experiment for stimulation of PF alone and CF alone showed no changes in EPSC (CF stimulation: Fig. 1 control experiment. PF stimulation: Fig. S1). CF stimulation alone had an early peak of response, which was around 45 to 50 s, while PF alone stimulation showed a gradual increase, which peaked at the end of the stimulation (Fig. 6a, b). The averages of peak ΔF/F between the 26th and 30th stimuli were significantly higher for the CF + PF stimulation group (10.6 ± 1.1%, n = 10, P < 0.01) and CF stimulation alone group (9.0 ± 1.6%, n = 10, P < 0.05) than for the PF stimulation alone group (2.8 ± 0.7%, n = 9), while the averages of peak ΔF/F between the 86th and 90th stimuli for the CF + PF group (11.2 ± 0.9%, n = 10 ) were significantly higher than those for the CF only group (3.6 ± 0.7%, n = 10, P < 0.01 ) and PF only group (5.7 ± 1.3%, n = 9, P < 0.05). For the average of the peak responses during the longer time period (for 60 s), average peak Ca signals were smaller in the PF-stimulation alone group (1st to 30th: 2.0 ± 0.5%, n = 9; 60th to 90th: 5.7 ± 1.3%, n = 9) than in the PF + CF stimulation group (1st to 30th: 7.2 ± 0.7%, n = 10, P < 0.05; 61st to 90th: 11.0 ± 0.9%, n = 10, P < 0.01), while average peak Ca2+ of the CF-stimulation alone group (4.2 ± 0.8%, n = 10, P < 0.01) was smaller than that of the PF + CF stimulation group (11.0 ± 0.9%, n = 10) only between the 61st and 90th response.