Abstract
Neuropeptides and neurotrophins, stored in dense core vesicles (DCVs), are together the largest currently known group of chemical signals in the brain. Exocytosis of DCVs requires high-frequency or patterned stimulation, but the determinants to reach maximal fusion capacity and for efficient replenishment of released DCVs are unknown. Here, we systematically studied fusion of DCV with single vesicle resolution on different stimulation patterns in mammalian CNS neurons. We show that tetanic stimulation trains of 50-Hz action potential (AP) bursts maximized DCV fusion, with significantly fewer fusion event during later bursts of the train. This difference was omitted by introduction of interburst intervals but did not increase total DCV fusion. Interburst intervals as short as 5 s were sufficient to restore the fusion capacity. Theta burst stimulation (TBS) triggered less DCV fusion than tetanic stimulation, but a similar fusion efficiency per AP. Prepulse stimulation did not alter this. However, low-frequency stimulation (4 Hz) intermitted with fast ripple stimulation (200 APs at 200 Hz) produced substantial DCV fusion, albeit not as much as tetanic stimulation. Finally, individual fusion events had longer durations with more intense stimulation. These data indicate that trains of 50-Hz AP stimulation patterns triggered DCV exocytosis most efficiently and more intense stimulation promotes longer DCV fusion pore openings.
SIGNIFICANCE STATEMENT Neuropeptides and neurotrophins modulate multiple regulatory functions of human body like reproduction, food intake or mood. They are packed into dense core vesicles (DCVs) that undergo calcium and action potential (AP) fusion with the plasma membrane. In order to study the fusion of DCVs in vitro, techniques like perfusion with buffer containing high concentration of potassium or electric field stimulation are needed to trigger the exocytosis of DCVs. Here, we studied the relationship between DCVs fusion properties and different electric field stimulation patterns. We used six different stimulation patterns and showed that trains of 50-Hz action potential bursts triggered DCV exocytosis most efficiently and more intense stimulation promotes longer DCV fusion pore openings.
Introduction
The information in the human brain is processed through the release of many chemical signals, that are released from two specialized organelles. Neurotransmitters are stored in synaptic vesicles (SVs), whereas neuropeptides, often together with small neurotransmitters, are stored inside dense core vesicles (DCVs). Despite similarities in the protein composition of the exocytotic machinery, it has been shown that different stimulation patterns trigger the release of neuropeptides and neurotransmitters to a different extent (Kreutzberger et al., 2019). In vivo, efficient release of the neuropeptides vasoactive intestinal peptide (VIP) and neuropeptide Y (NPY) from adrenal glands requires high repetitive stimulation. The release of noradrenaline, however, is sufficiently triggered already at low-frequency stimulation. This discrepancy in stimulus strength dependency between exocytosis of neurotransmitters and neuropeptides is known as the chemical frequency coding hypothesis (Andersson et al., 1982; Lundberg et al., 1986).
Further research confirmed that the release of peptides relies on high-frequency or patterned stimulation (Iverfeldt et al., 1989; Balkowiec and Katz, 2000; Persoon et al., 2018), while classic neurotransmitters are secreted already on low-frequency stimulation. In neuronal cultures, the most common stimulation paradigm to trigger DCV fusion is tetanic stimulation, e.g., hundreds of stimuli at 50 Hz (Hartmann et al., 2001; Persoon et al., 2018, 2019; Hoogstraaten et al., 2020; Moro et al., 2020). Surprisingly, tetanic stimulation leads to the fusion of only few percentage of vesicles (∼6%) of the total number of DCVs present in a single hippocampal neuron (Persoon et al., 2018). Moreover, this type of stimulation does not resemble the natural firing patterns in the human brain. Theta burst stimulation (TBS; 25 bursts of four APs at 100 Hz, delivered at 5 Hz) is considered a better alternative and has been shown to efficiently triggers the release of BDNF from primary sensory neuronal (Balkowiec and Katz, 2002). Effectiveness of TBS was also demonstrated for BDNF-positive DCVs released from dendrites but not axons (Matsuda et al., 2009). However, a systematic analysis of the most optimal stimulation patterns for DCV fusion is currently lacking.
Here, we compared the effectiveness of different stimulation patterns for DCV fusion in mammalian CNS neurons. The stimulation that showed the highest rate of fused DCV was strong tetanic stimulation (16 trains of 50 APs at 50 Hz). However, during this stimulation the gradual decrease of fusion capacity was observed; from the highest fusion rate during first few second to almost no release during last few seconds. Introduction of interburst intervals did not increase total DCV fusion, but even 5 s of interburst interval allowed the DCV pool to partially restore its fusogenic properties. TBS (10 trains of four APs at 100 Hz, delivered every 20 s) was as effective as tetanic stimulation, but its effectiveness depended on the number of action potentials (APs). Both low-frequency stimulation (80 APs at 4 Hz) and very fast ripple stimulation (200 APs at 200 Hz) induced fusion of DCVs, but at lower lever than tetanic stimulation. Lastly, we show that electrical stimulation affected the duration of DCV fusion events.
Materials and Methods
Primary cell culture
All animals used in the experiments were bred according to institutional and Dutch Animal Ethical Committee regulations (DEC-FGA 11-03). Embryos were killed at 18 d (E18). Hippocampi were dissected in HBSS (Sigma) supplemented with 10 mm HEPES (Life Technology) and digested with 0.25% Trypsin (Life Technology) for 20 min at 37°C. Then hippocampi were washed two times with HBSS (Sigma) followed by one wash with DMEM. Dissociated hippocampi were then triturated with a glass Pasteur pipette and resuspended in Neurobasal supplemented with 2% B-27, 18 mm HEPES, 0.25% Glutamax, and 0.1% penicillin/streptomycin (Life Technology). Hippocampal neurons were counted and plated on previously prepared rat astrocyte micro-island cultures on 18mm coverslips in 12-well plate, at a density of 1500 neurons/well. Rat astrocyte micro-islands were generated as previously described (Mennerick et al., 1995; Wierda et al., 2007).
Constructs and virus infection
Constructs used to visualize DCVs were made by introducing a single point mutation (M153R) within pHluorin sequence in previously described pr(Syn)NPY-pHluorin construct or pr(Syn)BDNF-pHluorin (Farina et al., 2015; Persoon et al., 2018). Single point mutation within pHluorin (M153R) was introduced in order to increase fluoresce signal (Morimoto et al., 2011). All constructs were cloned in lentiviral backbone and used to produce viral particles (Naldini et al., 1996). Primary hippocampal neurons were infected at day in vitro (DIV)9–DIV11 and imaged 6 d after infection.
Live cell imaging
Live imaging recordings were performed on a Zeiss 200M inverted microscope with 40× oil objective (NA = 1.3). Cells on coverslips were placed into the imaging chamber and perfused with Tyrode's (119 mm NaCl, 2.5 mm KCl, 2 mm CaCl2·2H2O, 2 mm MgCl2·6H2O, 25 mm HEPES, and 30 mm glucose·H2O, pH 7.4, 280 mOsm). For the recordings of calcium influx, cells were incubated with 1 μm Fluo-5-AM for 7–10 min. Electric field stimulation was applied by two platinum parallel electrodes with stimulus isolator (WPI A-385) controlled by a Master8 (AMPI), delivering 1-ms 30-mA pulses. Patterns of the simulations provided above in Results. Time lapse recordings were acquired using Metamorph v6.3 software (Molecular Devices) and EM-CCD camera. Time lapse recordings were acquired at 2 Hz with the exposure time of 250 ms for the detection of DCV fusion events or 50 ms for calcium imaging with Fluo5-AM. Every recording contained 30 or 10 s of baseline image acquisition prior to the stimulation. The dequenching of intracellular pHluorin was obtained by application of Tyrode's solution with 50 mm NH4Cl. All live cell imaging experiments were performed at room temperature (21–24°C).
Image analysis
The analysis of DCV fusion events was performed in Fiji using a semi-automatic DCV_pHluorin plug in previously published work (Moro et al., 2021b). Events, detected in a three by three-pixel region of interest (ROI), with sudden and sharp fluorescence increase were classified as fusion events. All selected events were then loaded into a custom-build MATLAB script (Persoon et al., 2019; Moro et al., 2020). Changes in fluorescence intensity (F/F0) were measured by dividing fluorescence gain (F) by baseline fluorescence obtained by averaging the 10 first frames (F0). Events with a fluorescence increase over two standard deviations above F0 and where maximum F/F0 was reached within two frames (500 ms), were considered as DCV fusion.
Statistics
The statistical evaluation was performed by testing sample normal distribution with Shapiro test. For normalized sample set the significance level was measured by performing Student's t test, for non-normalized groups Mann–Whitney test or Kruskal with Dunn post hoc analysis was used.
Results
Introduction of interburst intervals allows replenishment of a fusogenic properties of DCV Pool without increasing total number of DCV fusion events
To determine the effectiveness of different stimulation patterns on DCV fusion, we expressed the DCV fusion reporter NPY-pHluorin in single mouse hippocampal neurons cultured on glia micro-islands (Moro et al., 2021b). Phluorin is a pH-sensitive GFP, which is quenched in the acidic lumen of DCVs and becomes fluorescent on exposure to the extracellular environment (Fig. 1A). We previously showed 90% colocalization between NPY-pHluorin and endogenously expressed DCV-cargo Chromogranin B and overexpression of NPY-pHluorin in hippocampal neurons did not affect the total DCV pool or single DCV intensity (Persoon et al., 2018). The “fused DCV fraction” was calculated by dividing the number of fused DCVs during stimulation over the remaining number of DCVs, as measured by NH4Cl application, which visualizes DCVs by dequenching the pHluorin signal.
Introduction of interburst intervals allows replenishment of a fusogenic properties of DCV pool without increasing total number of DCV fusion events. A, Graphical representation of NPY-pHluorin positive DCVs during fusion with the plasma membrane and typical example of a single isolated hippocampal neuron with DCV fusion events marked in green. The electrical stimulation paradigm is depicted on top. Each neuron was subjected to one stimulation pattern selected in randomized fashion. Blue bars represent the 16 bursts of 50 APs at 50 Hz, green rectangle marks superfusion of Tyrode's containing 50 mm NH4+ to dequench pHluorin. B–D, Median number of DCV fusion events per time point per neuron. Above graphical representation of each stimulation paradigm used in the experiment, 16 bursts high-frequency stimulation, each burst consisting of 50 APs delivered at 50 Hz with or without 30-s interburst interval. E, Cumulative plot of median fused DCVs divided by the remaining pool of DCVs. Inset, Coefficient of variation of fused DCVs normalized to the remaining DCV pool for each stimulation pattern. F, Number of fused DCVs normalized to the remaining DCV pool per neuron for each stimulation pattern. G, H, Cumulative plot of median number of fused DCVs normalized to the remaining DCV pool per neuron for each train within different stimulation patterns. I, Number of fused DCVs normalized to the remaining DCV pool per neuron. J, Cumulative plot of median number of fused DCVs normalized to the remaining DCV pool per neuron. Kruskal–Wallis test and Dunn test with “holm” adjustment or Mann–Whitney test. *p < 0.05, **p < 0.01, ***p < 0.001, ns p > 0.05. Box plots depict data distribution with a central line marking the median value and cross marking mean value, the notch represents 95% confidence interval for median value. n = number of neurons (number of independent experiments).
A single hippocampal neuron fuses on average ∼6% of its DCVs on high-frequency repetitive stimulation (16 bursts of 50 APs at 50 Hz) with most fusion events occurring during the first two bursts (Persoon et al., 2018). The decline in the number of fusion events during stimulation may be caused by a gradual depletion of the releasable DCV pool. In cultured neural lobes of the hypothalamus, the effectiveness of vasopressin release depends on the duration of interbust intervals, with the most optimal duration of 20–30 s (Cazalis et al., 1985). Therefore, we hypothesized that the introduction of interburst intervals promotes replenishment of the fusion capacity of neurons (how many DCVs are fusing on stimulation as a fraction of the total amount of DCVs) and increases the total number of DCV fusion events. To test this, we split the 16-burst stimulation paradigm into two trains, each containing eight bursts of 50 APs at 50 Hz (2×8), or four trains, each containing four bursts of 50 APs at 50 Hz (4×4), separated by 30-s intervals (Fig. 1B–D). Every neuron was challenged with one stimulation pattern selected in a randomized approach. For all stimulation patterns, DCV fusion was synchronized to the onset of stimulation with very little fusion before stimulation and during the interburst intervals (Fig. 1B–E). The total fused fraction, calculated as number of fused DCVs normalized to the remaining DCV pool, during different stimulation patterns ranged between 3–6%, as observed before (Persoon et al., 2018). No significant differences between 16×, 2×8, or 4×4 stimulation patterns were observed. Introduction of interburst intervals unmasked substantial heterogeneity in DCV fusion efficiency among hippocampal neurons (coefficient of variation for fused DCVs normalized to the remaining DCV pool was 1.05, 1.16, and 1.55 for 16×, 2×8, and 4×4, respectively). While some neurons responded to interval stimulation with higher total release, others showed lower, leading to substantially increased variation, without effects on the average released fraction (Fig. 1E,F). Overall, the introduction of one of multiple interburst intervals during tetanic stimulation did not increase DCV fusion.
The fusion on 16× stimulation decayed over time (Fig. 1B,G), as observed before (Persoon et al., 2018). When comparing DCV fusion during the first half of the 16× stimulation (bursts 1–8) to the second half (bursts 9–16), a 2.5-fold reduction in fusing DCVs was observed (Fig. 1G). The same comparison for stimulation with a 30-s interval (two trains of eight bursts) revealed no significant difference between the first eight and the second eight bursts (Fig. 1G). Similar results were obtained when comparing sets of four bursts in the 16× stimulation to each of the 4×4 stimulations with 30-s intervals (Fig. 1H). The most prominent decrease in DCV released fraction was detected during the fourth train of the 16× stimulation (bursts 13–16; Fig. 1H) with a much smaller decrease when 30-s intervals were introduced (Fig. 1H, right panel). Hence, introduction of one or multiple 30-s interburst intervals is enough to partially restore fusogenic properties of DCV pool.
To test the relationship between interburst interval duration and fusion capacity of neurons, we used the 2×8 stimulation paradigm with 5- or 15-s interburst intervals (Fig. 1J,K). The fused fraction of DCVs on 2×8 stimulation with 5- or 15-s interburst interval was similar and led to the fusion of 4–6% of the remaining DCV pool (Fig. 1I). For each stimulation, a trend toward a 1–1.5% decrease in DCV fusion during the second train after interburst interval was detected (not significant; Fig. 1J,K).
In conclusion, the introduction of interburst intervals during strong repetitive stimulation did not affect the total number of DCV fusion events, but was sufficient to largely restore the fusion capacity during stimulation bursts, already after a 5-s interburst interval.
Theta burst stimulation triggers DCV fusion with similar efficiency per action potential as 50-Hz tetanic stimulation
To test DCV fusion efficiency upon more physiologically relevant stimulation patterns, we used theta burst stimulation (TBS). TBS is known to mimic in vivo neuronal activity and induce long-term potentiation (LTP) in hippocampal neurons (Otto et al., 1991). This type of high-frequency stimulation also efficiently induces the release of BDNF from rat hippocampal neurons (Balkowiec and Katz, 2002). TBS consisted of 10 trains of four APs at 100 Hz, delivered every 20 s. DCV fusion upon TBS was compared to 16× burst stimulation (16× 50 APs at 50 Hz; Fig. 2B). Upon TBS, 22% of neurons remained silent (did not fuse a single DCV), whereas 16× burst stimulation efficiently triggered fusion in all neurons (Fig. 2A). The data derived from nonsilent neurons were analyzed further.
Theta burst stimulation triggers DCV fusion with similar efficiency per action potential as 50-Hz tetanic stimulation. Neurons overexpressing NPY-pHluroin in (A–G). A, Percentage of silent cells during TBS or 16 × 50 AP at 50 Hz (16×) stimulation. B, Cumulative plot of median of fused DCVs normalized to the remaining DCV pool per neuron. Blue bars represent bursts of TBS stimulation, each burst consists of 10 trains of four APs at 100 Hz delivered every 200 ms, gray bars represent the 16 bursts of 50 APs at 50 Hz. Above graphical representation of 16× stimulation pattern. C, Cumulative plot of median of fused DCVs normalized to the remaining DCV pool per neuron for each train within TBS. Blue area represents 10 trains of four APs at 100 Hz delivered every 200 ms. Above graphical representation of TBS stimulation pattern. D, Number of DCVs normalized to the remaining DCV pool per neuron. E, Number of released DCVs per neuron. F, Number of released DCVs divided by number of APs used in stimulation (6×: 800 APs, TBS: 120 APs) per neuron. G, Percentage of persistent events per neuron. Neurons overexpressing BNDF-pHluroin (H–N). H, Percentage of silent cells during TBS or 16 × 50 AP at 50 Hz (16×) stimulation. I, Cumulative plot of median of fused DCVs normalized to the remaining DCV pool per neuron. Blue bars represent bursts of TBS stimulation, each burst consists of 10 trains of four APs at 100 Hz delivered every 200 ms, gray bars represent the 16 bursts of 50 APs at 50 Hz. J, Cumulative plot of median of fused DCVs normalized to the remaining DCV pool per neuron for each train within TBS. Blue shaded area represents 10 trains of four APs at 100 Hz delivered every 200 ms. K, Number of DCVs normalized to the remaining DCV pool per neuron. L, Number of released DCVs per neuron. M, Number of released DCVs divided by number of APs used in stimulation (16×: 800 APs, TBS: 120 APs) per neuron. N, Percentage of persistent events per neuron. Each neuron overexpressing NPY-pHluorin or BDNF-pHluorin (depicted in the figure) was subjected to one stimulation pattern (16× or TBS) selected in randomized fashion. Mann–Whitney test, *p < 0.05, **p < 0.01, ***p < 0.001, ns p > 0.05. Box plots depict data distribution with a central line marking the median value and cross marking mean value, the notch represents 95% confidence interval for median value. n = number of neurons (number of independent experiments).
Despite the fact that TBS is known to induce LTP (Otto et al., 1991), facilitation of DCV fusion was not observed among 3 subsequent trains of TBS (Fig. 2B,C). The fused fraction upon TBS (∼0.5%) was significantly lower compared to 16× burst stimulation (∼3.3%; Fig. 2D). A similar magnitude of reduction was observed for the total number of DCVs fused per neuron (Fig. 2E). Because the two high-frequency stimulations consist of different numbers of APs (TBS: 120 APs, 16×: 800 AP), we calculated the stimulation efficiency by dividing the number of fused DCVs per AP. Contrary to the fused fraction and the total number of fused DCVs, the stimulation efficiency did not show significant changes between TBS and 16×. During TBS and 16× stimulation neurons released 0.09 and 0.05 DCVs per AP, respectively (Fig. 2F). These data show that TBS triggers fusion of fewer DCVs than strong tetanic stimulation. However, when normalized to the number of APs the stimulation efficiency is similar for both patterns.
Prepulse stimulation does not potentiate DCV fusion induced by TBS
Previous studies showed that introduction of a short β pulse before TBS leads to an increased LTP magnitude (Grover et al., 2009). Here, we applied two types of prepulse stimulations 20 s before the start of TBS to investigate the effect of a prestimulation on DCV fusion during TBS. The prepulse consisted of 50 APs delivered at 10 Hz (TBS10) or 100 APs at 20 Hz (TBS20). Both prepulse patterns before TBS were compared with naive TBS.
The 20-Hz prepulse induced the fusion of 0.6–0.7% of the remaining DCV pool, which was significantly higher than on 10-Hz prepulse stimulation (0.4% of the total pool; Fig. 3A,C). Nonetheless, there was no significant increase in DCV fusion on TBS previously primed with 10 Hz or 20-Hz prepulse. As naive TBS, both prepulsed TBS paradigms induced DCV fusion in range of 0.5–1.5% of the remaining pool (Fig. 3B,D). Moreover, no potentiation of DCV fusion was observed within preprimed or naive TBS when subsequent bursts were compared (Fig. 3B), even when number of burst within naive TBS was increased to 8 (Fig. 3E,F). Hence, the introduction of prepulse stimulation did not potentiate DCV fusion on TBS stimulation.
Prepulse stimulation does not potentiate DCV fusion induced by TBS. A, Cumulative plot of median of fused DCVs normalized to the remaining DCV pool per neuron during prepulse. Shaded area depicts prepulse stimulation consisting of one train of 10 or 20 Hz for 5 s. B, Cumulative plot of median number of fused DCVs normalized to the remaining DCV pool per neuron during TBS stimulation. Blue bars represent bursts of TBS stimulation, each burst consists of 10 trains of four APs at 100 Hz delivered every 200 ms. C, Number of DCVs normalized to the remaining DCV pool per neurons for prepulse stimulations. D, Number of DCVs normalized to the remaining DCV pool per neurons for TBS stimulation. E, Cumulative plot of median number of released DCVs per neuron during eight bursts of TBS (TBS8). Purple bars represent bursts of TBS stimulation, each burst consists of 10 trains of four APs at 100 Hz delivered every 200 ms F, Number of released DCVs per neuron per train during eight bursts of TBS. Each neuron overexpressing NPY-pHluorin was subjected to one stimulation pattern selected in randomized fashion. Kruskal–Wallis test and Dunn test with “holm” adjustment, *p < 0.05, **p < 0.01, ***p < 0.001, ns p > 0.05. Box plots depict data distribution with a central line marking the median value and cross marking mean value, the notch represents 95% confidence interval for median value. n = number of neurons (number of independent experiments).
Ripple stimulation triggers DCV fusion more efficiently than low-frequency stimulation
Ripples are natural firing pattern detected in vivo characterized by spontaneous high-frequency events lasting for 40–100 ms, recorded mainly from hippocampal neurons (Buzsáki et al., 1983). Shakiryanova et al., showed that in Drosophila Ib motor neuron boutons, neuropeptide release during 3-Hz stimulation was potentiated after a burst of 70 Hz (Shakiryanova et al., 2005). We designed a similar experiment in single isolated mouse neuronal cultures by stimulating each neuron with three trains of 80 APs at 4 Hz delivered with 10-s interburst interval. Immediately after the first two 4-Hz trains, bursts of 200 APs at 200 Hz were applied (Fig. 4A). During 4-Hz stimulation, significant DCV fusion was observed, synchronized to the stimulation, with fusion of 0.07–0.15% of the remaining DCV pool (Fig. 4B). The ripple stimulation, on the other hand, was more effective and led to the fusion of 0.4–0.55% of the remaining pool (Fig. 4C). However, DCV fusion during 4-Hz stimulation after the “ripple” burst was not altered and showed no significant differences between each of 4-Hz trains (Fig. 4D). In conclusion, DCV fusion was observed during two biologically relevant stimulation patterns, but the ripples were more effective in triggering DCV fusion than intervals of low-frequency stimulation.
Ripple stimulation triggers DCV fusion more efficiently than low-frequency stimulation. A, The electrical stimulation paradigm, gray bars represent trains of 80 APs delivered at 4 Hz, red bars represent trains of 200 APs at 200 Hz. B, Cumulative plot of median number of fused DCVs normalized to the remaining DCV pool per neuron, gray and red shaded area represents 4- and 200-Hz stimulation, respectively, n = 41(4). C, Cumulative plot of median number of fused DCVs normalized to the remaining DCV pool per neuron for each 200-Hz burst. D, Cumulative plot of number of fused DCVs normalized to the remaining DCV pool per neuron for each 4-Hz stimuli. Each neuron overexpressing NPY-pHluorin was subjected to one stimulation pattern selected in randomized fashion. Kruskal–Wallis test, *p < 0.05, **p < 0.01, ***p < 0.001, ns p > 0.05. Box plots depict data distribution with a central line marking the median value and cross marking mean value, the notch represents 95% confidence interval for median value. n = number of neurons (number of independent experiments).
Stimulation patterns influence DCV fusion events duration
In chromaffin cells, the properties of single DCV fusion events differ depending on the frequency of electric stimulation (Fulop et al., 2005). Here, by using the pHluorin dequenching assay, we calculated the duration of each fusion event on different stimulation patterns (Fig. 5A). First, we categorized events as persistent or transient. A transient event was defined as an event with sharp increase in fluorescence intensity that returned to baseline before the end of the stimulation; a persistent event was characterized by a sharp increase in fluorescence intensity that did not return to baseline before the end of the stimulation (Fig. 5B). All stimulations consisting of repetitive bursts of 50 APs at 50 Hz showed 8–10% of persistent events (16×, 2×8, 4×4), 200-Hz stimulation (200 APs at 200 Hz) had ∼4% of persistent events, whereas TBS (10 trains of four APs at 100 Hz, delivered every 20 s) and 4-Hz stimulation (80 APs at 4 Hz) showed no persistent events (Fig. 5B). Therefore, the percentage of persistent and transient events is influenced by applied stimulation pattern.
Stimulation patterns influence DCV fusion events duration. A, Graphical representation of NPY-pHluorin positive DCVs during fusion with the plasma membrane and typical DCV fusion events at different time points. Below F/F0 traces of single transient and persistent event in a and b, respectively. B, Percentage of persistent events per neuron for each stimulation. C, Duration of transient events per neuron. D, Histograms of durations of transient events represented as logarithmic transformation of probability density with each bin calculated as raw bin's number divided by the total number of counts and the bin width. Inset, Typical examples of F/F0 traces of intracellular calcium influx on different stimulation measured with Fluo5-AM. Traces were normalized to first 15 s of baseline. E, Correlation of events duration with duration of calcium elevation. Each neuron overexpressing NPY-pHluorin was subjected to one stimulation pattern selected in randomized fashion. Kruskal–Wallis test and Dunn test with, “holm” adjustment *p < 0.05, **p < 0.01, ***p < 0.001, ns p > 0.05. Box plots depict data distribution with a central line marking the median value and cross marking mean value, the notch represents 95% confidence interval for median value. n = number of neurons (number of independent experiments).
Next, we measured the duration of transient events; the start of the event was defined by first fluoresce peak and the end of the event was defined as moment when fluoresce came back to the baseline. The duration of transient events was affected by the stimulation pattern: 16× 50 APs and 2×8 50 APs at 50 Hz had a similar duration of 3 s, while 4×4 50 APs at 50-Hz stimulation showed a significant decrease to 2.5 s. Two other high-frequency stimulations (TBS and 200 Hz) also showed shortening of the event duration compared with 16×, with 2 and 1.75 s for TBS and 200 Hz, respectively. The shortest transient events were measured for 4-Hz stimulation (1.5 s; Fig. 5C,D).
Additionally, we measured correlation between the event duration and the duration of calcium elevation. The calcium elevation was visualized by stimulating neurons previously filled with calcium indicator Fluo5-AM. For the repetitive stimulation, the duration of calcium elevation was defined as one repeatable train of action potentials separated by interburst intervals (23.5, 11.5, 5.5, 1.2, 1, and 20 s for 16×, 2×8, 4×4, TBS, 200 Hz and 4-Hz stimulation, respectively; Fig. 5E). The Spearman's correlation showed positive, but weak correlation when all stimulations were considered (correlation = 0.24, p-value = 4.78 × 10−5). However, when only high-frequency stimulations were considered, the positive correlation increased (correlation = 0.43, p-value = 2.09 × 10−12). Taken together, these data show that the percentage and duration of transient events strongly depend on an applied stimulation pattern, with a 50% reduction in event duration during the lowest stimulation frequency tested (4 Hz).
Discussion
In this paper, we studied the relationship between different electrical stimulation patterns and DCV fusion. We showed that splitting high-frequency repetitive stimulation into short trains did not increase DCV fusion. The introduction of interburst intervals lasting 5–30 s was sufficient to partially restore the fusion capacity of the neurons (Fig. 1). The number of DCV fusion events on TBS was much lower compared with 16× 50 APs at 50 Hz. However, when the number of fusion events was normalized to the number of APs applied in the stimulation, both types of stimulation showed a similar effectivity (Fig. 2). Introduction of prepulse stimuli before TBS did not alter DCV exocytosis (Fig. 3). Furthermore, we showed that DCV fusion can be triggered also under lower, physiologically relevant stimulations like 4 Hz or fast ripples consisting of 200 APs at 200 Hz (Fig. 4). Stimulation patterns influence the properties of individual fusion events, promoting longer pore openings with more intense stimulation (Fig. 5).
The stimulation patterns used in this study were characterized by different number of APs and their frequency delivery. Therefore, for each stimulation we calculated the fusion efficiency described as number of fused DCVs per one AP. Strong tetanic stimulations consisting of trains of 50-Hz APs lead to the highest ratio of fused DCVs; however, the fusion efficiency for these stimulations was equally effective as for stimulations considered as more physiological like TBS, ripples (200 APs at 200 Hz) or low-frequency stimulation (80 APs at 4 Hz; Fig. 6). Nevertheless, the fusion efficiency was measured only for neurons that responded to the applied stimulation. During TBS stimulation 20% of neurons remained silent and did not released any DCVs (Fig. 2), therefore TBS stimulation is as effective as tetanic stimulation, but there is a subpopulation of hippocampal neurons that are not responsive to this type of stimulation. In order to maximize DCV release for hippocampal neurons strong tetanic stimulation consisting of trains of 50-Hz APs is needed.
The effectiveness of stimulation largely depends on the number of applied APs. A, Number of DCVs normalized to the remaining DCV pool per neurons. B, Number of released DCVs per neuron. C, Number of released DCVs divided by number of APs used in stimulation (16×, 2×8 and 4×4: 800 APs, TBS: 120 APs, 200 Hz: 200 APs, 4 Hz: 80 APs) per neuron. Each neuron overexpressing NPY-pHluorin was subjected to one stimulation pattern selected in randomized fashion. Kruskal–Wallis test. *p < 0.05, **p < 0.01, ***p < 0.001, ns p > 0.05. Box plots depict data distribution with a central line marking the median value and cross marking mean value, the notch represents 95% confidence interval for median value.
Our and others experiments showed that the fusion of DCVs during 16× 50 APs at 50 Hz stimulation rapidly decreases over time (Fig. 1; Persoon et al., 2018, 2019; Hoogstraaten et al., 2020; Moro et al., 2021a; Puntman et al., 2021). Because DCV exocytosis is calcium dependent (De Wit et al., 2009); this decrease during such prolonged stimulation may be because of the limited availability of calcium at the end of the stimulation. However, we previously showed that neurons treated with a calcium ionophore (ionomycin) released on average six times less DCVs than 16× 50 APs at 50 Hz. Moreover, calcium levels do not decline during 16× 50 APs at 50 Hz (Persoon et al., 2018). Therefore, calcium dynamics do not explain the decline of DCV fusion during strong tetanic stimulation.
Here, we show that the introduction of short interburst intervals largely replenished the fusion capacity of neurons, even 5 s of interburst intervals were sufficient to restore the fusion capacity (Fig. 1), which suggest that the factor that limits the fusion capacity can be restored within only few seconds. The potential limiting factor could be the availability of SNARE and accessory proteins at the fusion site. Most DCVs fuse at synapses (Farina et al., 2015; Persoon et al., 2018). high-frequency stimulation triggers redistribution and increase in mobility of many presynaptic proteins essential for DCV exocytosis such as Rab3a (Star et al., 2005), Munc18-1 (Cijsouw et al., 2014), and CAPS-1 (Farina et al., 2015). Moreover, synapses with higher CAPS-1 intensity show higher DCV fusion probability (Farina et al., 2015). Therefore, the distribution and dynamics of presynaptic proteins might be a factor causing a time-dependent decrease in DCV fusion during 16× high-frequency stimulation. Another explanation could be the organization of the DCV fusion site itself. Recently has been shown that dynamin proteins acutely control DCV exocytosis by limiting the fusion of DCVs that occurs during later stage of high-frequency tetanic stimulation (Moro et al., 2021a). We speculate that during these few seconds of interburst interval during tetanic stimulation the fusion site is fully restored with the required exocytic proteins to drive DCV fusion with the plasma membrane.
In this study, we stimulated single isolated hippocampal neurons with different stimulation paradigms that mimic the natural firing patterns of the brain. TBS is a physiologically relevant stimulation, previously shown to efficiently induce BDNF release from rats hippocampal neurons (Hartmann et al., 2001; Balkowiec and Katz, 2002). Our data show that DCV fusion, when normalized to the number of applied APs, is similar for TBS and 16× burst stimulation (Fig. 2), in line with the previous reports on BDNF release. However, we did not detect any potentiation of DCV fusion under TBS or TBS primed with prepulses (Fig. 3). This discrepancy cannot be explained by the fact that we used an NPY-based reporter for most of our experiments. NPY is known to attenuate LTP and inhibit release of glutamate from hippocampal slices (Whittaker et al., 1999; Sørensen et al., 2008). However, we showed that the fusion of DCVs labeled with BDNF-pHluorin also did not show potentiation (Fig. 2I,J). BDNF, in contrast to NPY, was shown to promote LTP (Rex et al., 2007). Therefore, we conclude that the cargo used in the current study did not have significant influence on the number of fused DCVs. The introduction of ripples within the low-frequency stimulation also did not lead to the potentiation of DCV fusion (Fig. 5). Moreover, the prolonged TBS, consisting of 8 trains instead of three trains, also showed no increase in DCV fusion when subsequent trains were compared (Fig. 3E,F). The possible explanations why none of the stimulation patterns induced detectable potentiation of DCV fusion was the short period of recordings that each stimulated neuron undergo.
Our current data show that different stimulation patterns result in different ratios of persistent/transient events and lead to changes in event duration. Different stimulation patterns affect the kinetics of NPY-positive and BDNF-positive DCVs in a similar manner (Fig. 2G,N). We previously showed that DCVs that carry Semaphorin-pHluorin or NPY-pHluorin display different ratios of persistent and transient events. The differences in ratio of persistent/transient events between NPY and Semaphorin 3A were not caused by molecular weight of cargo, but by the interactions of specific cargoes with the extracellular matrix and/or the luminal matrix of vesicles (De Wit et al., 2009). Together, these results suggest that the kinetics of DCV fusion pores are influenced both by the intrinsic properties of cargo and by activity patterns (Table 1).
Summary of different DCV cargo properties
It was shown that Synaptotagmin 1 (Syt1) and Synaptotagmin 7 (Syt7) are redundant calcium sensors for DCV exocytosis but only Syt1 regulates DCV fusion event duration; depletion of Syt1 shortened event duration whereas overexpression prolonged event duration (van Westen et al., 2021). Syt1 is known to have lower affinity to calcium than Syt7 (Sugita et al., 2002). In all our stimulations calcium dynamics are strictly synchronized to the stimulation pattern. Tetanic stimulation leads to persistent influx of calcium throughout stimulation, but 4×4 (four trains, each consists of 50 APs at 50 Hz) and TBS contain interburst intervals when calcium levels return to baseline (Fig. 5D). Our data show a positive correlation between the event duration and the duration of calcium elevation; stimulation patterns with longer calcium influx showed longer fusion pore opening than stimulations with short peaks. The only exception was 4-Hz stimulation, which despite long period of calcium influx (20 s), showed the shortest event duration. However, 4-Hz stimulation, in contrast to other stimulations used in this study, is a low-frequency stimulation. The low-frequency stimulations typically lead to low or moderate calcium elevation. Therefore, we argue that the duration of calcium influx modulates the duration of DCV fusion pore opening but only when calcium levels reach a certain threshold, which can be achieved with high-frequency stimulations. We speculate, that high amplitude and duration of calcium influx is needed to recruit Syt1, which prolongs DCV fusion pore opening by increasing fusion pore stability.
These results, differ from observations made in chromaffin cells, where higher intracellular calcium concentrations lead to the shift of fusion pore kinetics to fast “kiss an run” mode (Alés et al., 1999). Exocytosis in chromaffin cells, similarly to neurons, depends on Syt1 and Syt7 calcium sensors (Schonn et al., 2008). However, the calcium concentration at DCV fusion sites in synapses is probably much higher than in chromaffin cells (Chow et al., 1994; Neher, 1998). Moreover, despite chromaffin cells and neurons sharing many of the same proteins involved in the exocytosis, several differences have been described. First, different dynamin paralogs are expressed in neurons and chromaffin cells; Dynamin 1 is crucial in neurons, whereas chromaffin cells depend on Dynamin 2 (Cao et al., 1998). The function of two dynamin paralogs, despite sharing ∼80% sequence identity, is only partially redundant. In chromaffin cells, Dynamin 2 regulates fusion pore expansion by interaction with the actin cytoskeleton (Anantharam et al., 2011). Second, chromaffin cells contain a unique submembrane accumulation of F-actin proposed to create an “actin barrier,” limiting (but probably also facilitating) the directed mobility and exocytosis of DCV (Trifaró et al., 1993; Steyer and Almers, 1999). In contrast, neuronal DCVs show a higher degree of mobility that is not sensitive to actin disruption (Silverman et al., 2005). Therefore, despite neurons and chromaffin cells sharing many key aspects of the DCV exocytosis, differences in protein composition might contribute to the observed differences in fusion pore kinetics.
Footnotes
This work was supported by the European Research Council (ERC) Advanced Grant 322966 of the European Union (to M.V.), COSYN (Comorbidity and Synapse Biology in Clinically Overlapping Psychiatric Disorders, Horizon 2020 Program of the European Union under RIA Grant Agreement 667301; to M.V.), and the JPND Neuron Cofund ERA-Net SNAREopathy (R.F.T.). We thank Joke Wortel for animal breeding, Robbert Zalm for cloning and producing viral particles, Desiree Schut for astrocyte culture and cell culture assistance, Ingrid Saarloos for assistance in protein chemistry.
The authors declare no competing financial interests.
- Correspondence should be addressed to Matthijs Verhage at matthijs{at}cncr.vu.nl or Ruud F. Toonen at ruud.toonen{at}cncr.vu.nl
