TwoPathways of Synaptic Vesicle Retrieval Revealed by Single ...
Supplementary Material: Two Pathways of Synaptic Vesicle Retrieval Revealed by Single Vesicle Imaging
Yongling Zhu*, Jian Xu, and Stephen F. Heinemann
Supplementary experimental procedures Transferrin uptake Efficiency of RNAi knockdown for CHC expression was estimated by transferrin uptakes. Neurons at DIV10-11 were transfected with either pCherry/CHC-RNAi or pCherry/control-RNAi plasmids. Transferrin uptake experiments were performed as previous report (Granseth et al., 2006; Royle et al., 2005) with slight modifications. Briefly, neurons of DIV14 were starved in Neurobasal medial without B27 for 1.5 hrs hours and were incubated with 50ug/ml transferrin conjugated to Alexa 488 (Molecular Probes) for 30 min at 37°C and were subsequently fixed. Transferrin uptake was measured by total Alxa488 fluorescence at the soma of cells co-expressing mCherry.
Data Analysis Fitting of SypHluorin fluorescence intensity data was carried out in custom Mathcad (Adept Scientific, Bethesda, MD). For time course of single vesicle fluorescence intensity presented in Figure 2, each trace was fitted by the least-squares estimation (LSE) to two equations, with all the parameters allowed to vary freely
F(t) = F0 - F1 * Φ(t-T) * (1-exp(-(t-T)/τ)
(1)
F(t) = F0 - F1 * Φ(t-T1) * (1-exp(-(t-T1)/ τ1) - F2 * Φ(t-T2) * (1-exp(-(t-T2)/ τ2)
(2)
where F0 is the initial fluorescence response of a single vesicle fusion, Φ(x) is a heaviside step function that returns 0 if x is negative, and 1 otherwise. Equation 1 is derived from the “one-phase decay” model, in which the fluorescence intensity decays in one phase
starting from T seconds after fusion, with a decay time constant of τ seconds, the total fluorescence intensity decrease in this step is F1. On the other hand, Equation 2 is derived from the “two-phase decay” model. In this model, the fluorescence intensity decays in two phases, with F1 and F2 representing the fluorescence intensity decrease in each phase. The first decay initiates at T1 seconds, with a decay time constant of τ1 seconds, and the second decay initiates at T2 seconds, with a decay time constant of τ2 seconds. After curve fitting, the sums of squares error (SSE) of each model were calculated. To compare the goodness of the fits, we calculated the ratio: SSE monophasic/ SSE biphasic. Also because equation with more parameters normally lead to a better fit, we decided to perform a “control fitting” to establish the minimal SSE monophasic/ SSE biphasic ratio that can be used as a standard for distinguishing biphasic and monophasic processes in our study. In this “control fitting”, we first simulated a single vesicle fusion followed by monophasic fluorescence decay with dwell time of 14s and decay time constant of 3s. The simulated fluorescence response (without noise) was added to the failure traces to produce the mimic monophasic endocytosis, which was then fitted by both monophasic decay and biphasic decay models. The control SSE monophasic/ SSE biphasic was estimated to be 1.05 ± 0.05 (mean ± SD). Based on this measurement, we set 1.15 (mean + 2SD) as the minimal SSE monophasic/ SSE biphasic. The range of two standard deviations from the mean in one direction corresponds to ~ 97.5% confidence interval. After curve fitting of experimental data with both models, a trace with a value of SSE monophasic/ SSE biphasic lager than 1.15 was considered as biphasic, otherwise, it was considered as monophasic. Curve fitting for multiple-vesicle traces were conducted in a similar way except that the minimal SSE monophasic/ SSE biphasic ratio was estimated by simulating with the
corresponding number of vesicle fusion events using a single exponential decay (dwell time of 0s, and decay time constant of 14s). For single vesicle fluorescence responses after photobleaching or clathrin RNAi treatments, we used biphasic decay model for the curve fitting to specifically examine the effect of treatments on each phase of the retrieval process. The quantal fluorescence of SypHluorin (Figure 1F) and FM 1-43 labeling (Figure 4A) were estimated by fitting the distribution of fluorescence intensity to the quantal model presented by Murthy and Stevens ((Murthy and Stevens, 1998) and Gandhi and Stevens (Gandhi and Stevens, 2003). The fitting was carried out in custom Mathcad by using the least-squares estimation. For action potential induced or spontaneous FM1-43 uptake, the distributions of fluorescence intensity were fitted to multiple Gaussians, and the peaks of fluorescence were compared with the quantal size. Data are presented as mean ± SEM, and statistical comparisons were made by using t test, unless otherwise stated.
Supplementary Figure 1. Fitting the Average Single Vesicle Response with Different Models. Black : Average trace from of 86 single vesicle responses in Figure 2B. Red: If fitted with a single exponential decay, the time constant is 19.6 ±0.7 s Blue: The fit with the model presented in a previous report (Granseth et al., 2006), where the reacidification time constant was set as 3s. The fit yields an estimation of endocytosis time constant = 10.5s.
Supplementary Figure 2. SypHluorin reporters have steady fluorescence after acidification and alkalization. (A) Example traces of SypHluorin fluorescence intensity as a function of time with extracellular pH alternating from 7.3 to 5.6, then back to 7.3. Switching extracellular pH from 7.3 to 5.6 led to the acidification of SypHluorin reporters on the surface membrane, and the acidification was relieved (alkalization) when pH was switched back to 7.3. The steady fluorescence under pH 5.6 and 7.3 suggest that SypHluorin reporters preserve steady fluorescence after acidification and alkalization. (B) Normalized fluorescence intensity change as a function of time, averaged from 110 individual traces.
Supplementary Figure 3. SypHluorin 1x and 2x share similar vesicular targeting property and endocytic pathways with SypHluorin (SypHluorin 4x). (A) Estimation of vesicle expression for SypHluorin 1x (sypH 1x), SypHluorin 2x (sypH 2x), SypHluorin (sypH 4x) and synaptopHluorin (spH) by NH4Cl alkalization (Sankaranarayanan et al., 2000) . (B) Vesicular fraction of SypHluorins and synaptopHluorin. Fractional vesicular expression was estimated as: 73.2% ± 1.0% for synaptopHluorin (n=140 synapses, from 6 experiments); 90.9% ± 0.6% for SypHluorin 1x (n=168 synapses, from 6 experiments); 90.8% ± 0.5% for SypHluorin 2x (n=138 synapses, from 6 experiments); and 90.4% ± 0.6% for SypHluorin 4x (n=138 synapses, from 6 experiments). No significant differences were observed among three SypHluorin sensors, but they were significantly higher than synaptopHluorin (*P < 0.05). Data are presented as mean ± SEM, and means are compared with the t test. (C) Example traces of fluorescence intensity as a function of times for single vesicles reported by SypHluorin 1x. Single action potential was delivered at 0s, as indicated by the arrow. (D) Histogram of dwell time of the fast (left panel) and slow (right panel) endocytosis for 97 vesicles reported by SypHluorin 1x. Smooth curves are the fits with Gaussian distribution, yielding a mean of 3.0 ± 1.6s (mean ± SD) for the fast endocytosis, and 16.6 ± 5.0s for the slow endocytosis. (E) Distribution of combined dwell times of both endocytosis shown in D. Smooth curve is the fit with two-component Gaussian distribution, yielding a mean of 3.0 ± 1.8s (mean ± SD) for the first component, and 16.3 ± 5.1s for the second component. (F) Example traces of fluorescence intensity as a function of time for single vesicles reported by SypHluorin 2x. (G) Dwell time distribution of the fast (left panel) and slow (right panel) endocytosis for 50 vesicles reported by SypHluorin 2x. Smooth curves are the fits with Gaussian distribution, yielding a mean of 2.4 ± 2.0s (mean ± SD) for the fast endocytosis, and 15.0 ± 3.8s for the slow endocytosis. (H) Distribution of combined dwell times of both endocytosis shown in G. Smooth curve is the fit with two-component Gaussian distribution, yielding a mean of 2.4 ± 1.6s (mean ± SD) for the first component, and 14.8 ± 4.3s for the second component. Error bars represent SEM
Supplementary Figure 4. Increasing pH buffer concentration slowed down the decay time constants of both Fast Endocytosis and Slow Endocytosis. (A) Example of single vesicle fluorescence intensity as a function of time when 10mM or 100 mM HEPES (pH=7.3) present in the external medium. The smooth curves are the fits with two distinct phases of exponential decay. (B) Time constants for both fast endocytosis and slow endocytosis were increased when buffer concentration increased from 10mM to 100mM, with τ increased from 2.6 ± 0.3 seconds to 5.1 ± 0.5 seconds for fast endocytosis and from 2.9 ± 0.4 seconds to 8.9 ± 0.6 seconds for slow endocytosis (n = 62 responses). Error bars represent SEM. ***p < 0.001, paired t test.
Supplementary Figure 5. Estimation of the minimal time required for FM1-43 binding to and washing-off from surface membrane. (A) Time course of FM 1-43 binding to the neuronal surface membrane. FM 1-43 (10 μM) was applied by using a VC-6 Perfusion Valve Control System (Warner Instruments). The fit with equation: F= 1-e-t/τ gave τ = 2.25 ± 0.01s (n = 118 ROI, from 8 experiments). Binding reached 99% of maximum at 7s. (B) Time course of FM 1-43 washing rate with (red) or without (black) 1 mM ADVASEP 7 (Kay et al., 1999) in the washing solution. Surface membrane was stained with 10 μM FM 1-43, followed by a dye-free wash by using a fast local perfusion system (VC-6M Mini-Valve System combined with SF-77B perfusion stepper system, Warner Instruments). Fits with single exponential decay yielded a time constant of 0.21 ± 0.01s for ADVASEP 7 wash (n = 234 ROI, from 9 experiments) and 2.73 ± 0.02s for control wash without ADVASEP 7 (n = 182 ROI, from 8 experiments). The remaining FM 1-43 binding after 1min of washing was 7.5% for ADVASEP 7 wash, and 19.0% for control wash.
Supplementary Figure 6. Relationship between exocytosis and endocytosis with the assumption that single action potential consistently triggers fusion of two vesicles. The number of endocytosed vesicles (estimated by FM 1-43 uptake during 35s) was plotted against the number of exocytosed vesicles (estimated by SypHluorin 4x response) under stimulation of 2, 5, 10, 20, 30 action potentials at 20Hz. The number of exocytosed vesicles was calculated by: Ftotal/ Fsingle vesicle, where Ftotal is the total fluorescence intensity, and Fsingle vesicle
is the fluorescence intensity of single vesicle. The total fluorescence intensities of
SypHluorin responses and FM 1-43 uptake were the same as those in Figure 4E. The single vesicle fluorescence in SypHluorin response was set as 98 units instead of 196.9 units in Figure 4E, assuming that the quantal fluorescence response estimated in Figure 1F represented two vesicle fusion events. The single vesicle fluorescence of FM 1-43 uptake was set as 402 units, the same value used in Figure 4E. Error bars represent SEM
Supplementary Figure 7. Photobleaching does not affect the functions of endocytosis machinery. (A) Estimate of surface fluorescence during photobleaching. Photobleaching was carried out continuously in normal external solution with pH=7.3. To measure the remaining surface fluorescence, a transient extracellular acidification (~2s) was applied every two minutes, and the amount of surface fluorescence was calculated by: Fsurface=FpH 7.3-FpH 5.6 (B) Surface fluorescence intensity as a function of time during photobleaching (n= 7 experiments, 93 boutons). When fitted with single exponential decay (smooth curve), the time constant is 3.5 ± 0.7s. (C) Comparison of fluorescence recovery in response to 40 action potentials at 20Hz before (filled circle, n = 62 traces) and after photobleaching (open circle, n = 62 traces). Under such a stimulation, the newly released vesicular proteins account for a major portion (~90%) of the total proteins participating in the following endocytosis. The similar kinetics before and after photobleaching suggests the photobleaching does not affect the rate of endocytosis. Error bars represent SEM.
Supplementary Figure 8. Inhibition of transferrin uptake by CHC RNAi. (A) Representative examples of transferrin uptake (green) in neurons expressing either control (top) or CHC siRNA (bottom), along with mCherry (red). (B) Quantitative analysis of transferrin uptake in neurons transfected with either CHC RNAi (n = 38) or control RNAi (n = 37). Error bars represent SEM. Asterisks denote significant difference between treatments (** p
Yongling Zhu*, Jian Xu, and Stephen F. Heinemann
Supplementary experimental procedures Transferrin uptake Efficiency of RNAi knockdown for CHC expression was estimated by transferrin uptakes. Neurons at DIV10-11 were transfected with either pCherry/CHC-RNAi or pCherry/control-RNAi plasmids. Transferrin uptake experiments were performed as previous report (Granseth et al., 2006; Royle et al., 2005) with slight modifications. Briefly, neurons of DIV14 were starved in Neurobasal medial without B27 for 1.5 hrs hours and were incubated with 50ug/ml transferrin conjugated to Alexa 488 (Molecular Probes) for 30 min at 37°C and were subsequently fixed. Transferrin uptake was measured by total Alxa488 fluorescence at the soma of cells co-expressing mCherry.
Data Analysis Fitting of SypHluorin fluorescence intensity data was carried out in custom Mathcad (Adept Scientific, Bethesda, MD). For time course of single vesicle fluorescence intensity presented in Figure 2, each trace was fitted by the least-squares estimation (LSE) to two equations, with all the parameters allowed to vary freely
F(t) = F0 - F1 * Φ(t-T) * (1-exp(-(t-T)/τ)
(1)
F(t) = F0 - F1 * Φ(t-T1) * (1-exp(-(t-T1)/ τ1) - F2 * Φ(t-T2) * (1-exp(-(t-T2)/ τ2)
(2)
where F0 is the initial fluorescence response of a single vesicle fusion, Φ(x) is a heaviside step function that returns 0 if x is negative, and 1 otherwise. Equation 1 is derived from the “one-phase decay” model, in which the fluorescence intensity decays in one phase
starting from T seconds after fusion, with a decay time constant of τ seconds, the total fluorescence intensity decrease in this step is F1. On the other hand, Equation 2 is derived from the “two-phase decay” model. In this model, the fluorescence intensity decays in two phases, with F1 and F2 representing the fluorescence intensity decrease in each phase. The first decay initiates at T1 seconds, with a decay time constant of τ1 seconds, and the second decay initiates at T2 seconds, with a decay time constant of τ2 seconds. After curve fitting, the sums of squares error (SSE) of each model were calculated. To compare the goodness of the fits, we calculated the ratio: SSE monophasic/ SSE biphasic. Also because equation with more parameters normally lead to a better fit, we decided to perform a “control fitting” to establish the minimal SSE monophasic/ SSE biphasic ratio that can be used as a standard for distinguishing biphasic and monophasic processes in our study. In this “control fitting”, we first simulated a single vesicle fusion followed by monophasic fluorescence decay with dwell time of 14s and decay time constant of 3s. The simulated fluorescence response (without noise) was added to the failure traces to produce the mimic monophasic endocytosis, which was then fitted by both monophasic decay and biphasic decay models. The control SSE monophasic/ SSE biphasic was estimated to be 1.05 ± 0.05 (mean ± SD). Based on this measurement, we set 1.15 (mean + 2SD) as the minimal SSE monophasic/ SSE biphasic. The range of two standard deviations from the mean in one direction corresponds to ~ 97.5% confidence interval. After curve fitting of experimental data with both models, a trace with a value of SSE monophasic/ SSE biphasic lager than 1.15 was considered as biphasic, otherwise, it was considered as monophasic. Curve fitting for multiple-vesicle traces were conducted in a similar way except that the minimal SSE monophasic/ SSE biphasic ratio was estimated by simulating with the
corresponding number of vesicle fusion events using a single exponential decay (dwell time of 0s, and decay time constant of 14s). For single vesicle fluorescence responses after photobleaching or clathrin RNAi treatments, we used biphasic decay model for the curve fitting to specifically examine the effect of treatments on each phase of the retrieval process. The quantal fluorescence of SypHluorin (Figure 1F) and FM 1-43 labeling (Figure 4A) were estimated by fitting the distribution of fluorescence intensity to the quantal model presented by Murthy and Stevens ((Murthy and Stevens, 1998) and Gandhi and Stevens (Gandhi and Stevens, 2003). The fitting was carried out in custom Mathcad by using the least-squares estimation. For action potential induced or spontaneous FM1-43 uptake, the distributions of fluorescence intensity were fitted to multiple Gaussians, and the peaks of fluorescence were compared with the quantal size. Data are presented as mean ± SEM, and statistical comparisons were made by using t test, unless otherwise stated.
Supplementary Figure 1. Fitting the Average Single Vesicle Response with Different Models. Black : Average trace from of 86 single vesicle responses in Figure 2B. Red: If fitted with a single exponential decay, the time constant is 19.6 ±0.7 s Blue: The fit with the model presented in a previous report (Granseth et al., 2006), where the reacidification time constant was set as 3s. The fit yields an estimation of endocytosis time constant = 10.5s.
Supplementary Figure 2. SypHluorin reporters have steady fluorescence after acidification and alkalization. (A) Example traces of SypHluorin fluorescence intensity as a function of time with extracellular pH alternating from 7.3 to 5.6, then back to 7.3. Switching extracellular pH from 7.3 to 5.6 led to the acidification of SypHluorin reporters on the surface membrane, and the acidification was relieved (alkalization) when pH was switched back to 7.3. The steady fluorescence under pH 5.6 and 7.3 suggest that SypHluorin reporters preserve steady fluorescence after acidification and alkalization. (B) Normalized fluorescence intensity change as a function of time, averaged from 110 individual traces.
Supplementary Figure 3. SypHluorin 1x and 2x share similar vesicular targeting property and endocytic pathways with SypHluorin (SypHluorin 4x). (A) Estimation of vesicle expression for SypHluorin 1x (sypH 1x), SypHluorin 2x (sypH 2x), SypHluorin (sypH 4x) and synaptopHluorin (spH) by NH4Cl alkalization (Sankaranarayanan et al., 2000) . (B) Vesicular fraction of SypHluorins and synaptopHluorin. Fractional vesicular expression was estimated as: 73.2% ± 1.0% for synaptopHluorin (n=140 synapses, from 6 experiments); 90.9% ± 0.6% for SypHluorin 1x (n=168 synapses, from 6 experiments); 90.8% ± 0.5% for SypHluorin 2x (n=138 synapses, from 6 experiments); and 90.4% ± 0.6% for SypHluorin 4x (n=138 synapses, from 6 experiments). No significant differences were observed among three SypHluorin sensors, but they were significantly higher than synaptopHluorin (*P < 0.05). Data are presented as mean ± SEM, and means are compared with the t test. (C) Example traces of fluorescence intensity as a function of times for single vesicles reported by SypHluorin 1x. Single action potential was delivered at 0s, as indicated by the arrow. (D) Histogram of dwell time of the fast (left panel) and slow (right panel) endocytosis for 97 vesicles reported by SypHluorin 1x. Smooth curves are the fits with Gaussian distribution, yielding a mean of 3.0 ± 1.6s (mean ± SD) for the fast endocytosis, and 16.6 ± 5.0s for the slow endocytosis. (E) Distribution of combined dwell times of both endocytosis shown in D. Smooth curve is the fit with two-component Gaussian distribution, yielding a mean of 3.0 ± 1.8s (mean ± SD) for the first component, and 16.3 ± 5.1s for the second component. (F) Example traces of fluorescence intensity as a function of time for single vesicles reported by SypHluorin 2x. (G) Dwell time distribution of the fast (left panel) and slow (right panel) endocytosis for 50 vesicles reported by SypHluorin 2x. Smooth curves are the fits with Gaussian distribution, yielding a mean of 2.4 ± 2.0s (mean ± SD) for the fast endocytosis, and 15.0 ± 3.8s for the slow endocytosis. (H) Distribution of combined dwell times of both endocytosis shown in G. Smooth curve is the fit with two-component Gaussian distribution, yielding a mean of 2.4 ± 1.6s (mean ± SD) for the first component, and 14.8 ± 4.3s for the second component. Error bars represent SEM
Supplementary Figure 4. Increasing pH buffer concentration slowed down the decay time constants of both Fast Endocytosis and Slow Endocytosis. (A) Example of single vesicle fluorescence intensity as a function of time when 10mM or 100 mM HEPES (pH=7.3) present in the external medium. The smooth curves are the fits with two distinct phases of exponential decay. (B) Time constants for both fast endocytosis and slow endocytosis were increased when buffer concentration increased from 10mM to 100mM, with τ increased from 2.6 ± 0.3 seconds to 5.1 ± 0.5 seconds for fast endocytosis and from 2.9 ± 0.4 seconds to 8.9 ± 0.6 seconds for slow endocytosis (n = 62 responses). Error bars represent SEM. ***p < 0.001, paired t test.
Supplementary Figure 5. Estimation of the minimal time required for FM1-43 binding to and washing-off from surface membrane. (A) Time course of FM 1-43 binding to the neuronal surface membrane. FM 1-43 (10 μM) was applied by using a VC-6 Perfusion Valve Control System (Warner Instruments). The fit with equation: F= 1-e-t/τ gave τ = 2.25 ± 0.01s (n = 118 ROI, from 8 experiments). Binding reached 99% of maximum at 7s. (B) Time course of FM 1-43 washing rate with (red) or without (black) 1 mM ADVASEP 7 (Kay et al., 1999) in the washing solution. Surface membrane was stained with 10 μM FM 1-43, followed by a dye-free wash by using a fast local perfusion system (VC-6M Mini-Valve System combined with SF-77B perfusion stepper system, Warner Instruments). Fits with single exponential decay yielded a time constant of 0.21 ± 0.01s for ADVASEP 7 wash (n = 234 ROI, from 9 experiments) and 2.73 ± 0.02s for control wash without ADVASEP 7 (n = 182 ROI, from 8 experiments). The remaining FM 1-43 binding after 1min of washing was 7.5% for ADVASEP 7 wash, and 19.0% for control wash.
Supplementary Figure 6. Relationship between exocytosis and endocytosis with the assumption that single action potential consistently triggers fusion of two vesicles. The number of endocytosed vesicles (estimated by FM 1-43 uptake during 35s) was plotted against the number of exocytosed vesicles (estimated by SypHluorin 4x response) under stimulation of 2, 5, 10, 20, 30 action potentials at 20Hz. The number of exocytosed vesicles was calculated by: Ftotal/ Fsingle vesicle, where Ftotal is the total fluorescence intensity, and Fsingle vesicle
is the fluorescence intensity of single vesicle. The total fluorescence intensities of
SypHluorin responses and FM 1-43 uptake were the same as those in Figure 4E. The single vesicle fluorescence in SypHluorin response was set as 98 units instead of 196.9 units in Figure 4E, assuming that the quantal fluorescence response estimated in Figure 1F represented two vesicle fusion events. The single vesicle fluorescence of FM 1-43 uptake was set as 402 units, the same value used in Figure 4E. Error bars represent SEM
Supplementary Figure 7. Photobleaching does not affect the functions of endocytosis machinery. (A) Estimate of surface fluorescence during photobleaching. Photobleaching was carried out continuously in normal external solution with pH=7.3. To measure the remaining surface fluorescence, a transient extracellular acidification (~2s) was applied every two minutes, and the amount of surface fluorescence was calculated by: Fsurface=FpH 7.3-FpH 5.6 (B) Surface fluorescence intensity as a function of time during photobleaching (n= 7 experiments, 93 boutons). When fitted with single exponential decay (smooth curve), the time constant is 3.5 ± 0.7s. (C) Comparison of fluorescence recovery in response to 40 action potentials at 20Hz before (filled circle, n = 62 traces) and after photobleaching (open circle, n = 62 traces). Under such a stimulation, the newly released vesicular proteins account for a major portion (~90%) of the total proteins participating in the following endocytosis. The similar kinetics before and after photobleaching suggests the photobleaching does not affect the rate of endocytosis. Error bars represent SEM.
Supplementary Figure 8. Inhibition of transferrin uptake by CHC RNAi. (A) Representative examples of transferrin uptake (green) in neurons expressing either control (top) or CHC siRNA (bottom), along with mCherry (red). (B) Quantitative analysis of transferrin uptake in neurons transfected with either CHC RNAi (n = 38) or control RNAi (n = 37). Error bars represent SEM. Asterisks denote significant difference between treatments (** p
