Lit Seminar 150609

Fluorescent False Neurotransmitters Visualize Dopamine Release from Individual Presynaptic Terminals Niko G. Gubernator,1*† Hui Zhang,2,3* Roland G. W. Staal,2 Eugene V. Mosharov,2 Daniela Pereira,2 Minerva Yue,2 Vojtech Balsanek,1 Paul A. Vadola,1 Bipasha Mukherjee,5 Robert H. Edwards,5 David Sulzer,2,3,4‡ Dalibor Sames1‡ 1

Department of Chemistry, Columbia University, New York, NY 10027, USA.

2

Departments of Neurology, Psychiatry, and Pharmacology, Columbia University, New York, NY 10032, USA.

3

Departments of Psychiatry and Pharmacology, Columbia University, New York, NY 10032, USA.

4

Division of Molecular Therapeutics, New York Psychiatric Institute, New York, NY 10032, USA.

5

Departments of Neurology and Physiology, University of California School of Medicine, San Francisco, San Francisco, California 94143, USA. *These authors contributed equally to this work. †Present address: eMolecules, San Diego, CA, 92014, USA. ‡To whom correspondence should be addressed. E-mail: [email protected] (D. Sulzer); [email protected] (D. Sames).

The nervous system transmits signals between neurons via neurotransmitter release during synaptic vesicle fusion. To observe neurotransmitter uptake and release from individual presynaptic terminals directly, we designed fluorescent false neurotransmitters as substrates for the synaptic vesicle monoamine transporter. Using these probes to image dopamine release in the striatum, we made several observations pertinent to synaptic plasticity. We found that the fraction of synaptic vesicles releasing neurotransmitter per stimulus was dependent on the stimulus frequency. A kinetically distinct “reserve” synaptic vesicle population was not observed under these experimental conditions. A frequency-dependent heterogeneity of presynaptic terminals was revealed that was dependent in part on D2 dopamine receptors, indicating a mechanism for frequency-dependent coding of presynaptic selection. Decision making, memory, and learning require activation and modification of particular synapses. Synaptic transmission in turn requires neurotransmitter accumulation into synaptic vesicles followed by neurotransmitter release during synaptic vesicle fusion with the plasma membrane. Optical methods have been developed to observe synaptic vesicle membrane fusion (1–4), but there has been no means to observe neurotransmitter release from individual synapses in the brain.

We designed optical tracers of monoamine neurotransmitters, or fluorescent false neurotransmitters (FFNs), inspired by classic reports that tyramine, amphetamine, and other phenylethylamines can be taken up into secretory vesicles and discharged during exocytosis (5). FFNs were designed by targeting the neuronal vesicular monoamine transporter (VMAT2) that carries monoamine neurotransmitters from the cytoplasm into synaptic vesicles (6). VMAT2 is relatively nonspecific and transports cellular monoamines (e.g., dopamine, serotonin, norepinephrine) as well as synthetic amines (e.g., amphetamine, MDMA, MPP+) (7, 8). We predicted that bulkier fluorescent monoamines might also be substrates (9) and developed compound FFN511 (Fig. 1A, fig. S1, design criteria are in SOM) (10). FFN511 inhibited serotonin binding to VMAT2-containing membranes, providing an apparent IC50 of 1 µM, a value close to dopamine itself (7). In adrenal chromaffin cells, catecholamines are stored in large dense core vesicles (LDCVs) that possess the vesicular monoamine transporter 1 (VMAT1). FFN511 accumulated in a pattern consistent with LDCVs in cultured mouse chromaffin cells (Fig. 1B), and the accumulation was abolished by the lipophilic base chloroquine, which collapses the vesicle pH gradient (fig. S2) (11). Exposure to 350 nM FFN511 (30 min) had no effect on the quantal size of evoked catecholamine release (fig. S3), and total internal reflection

/ www.sciencexpress.org / 7 May 2009/ Page 1 / 10.1126/science.1172278

fluorescence microscopy (TIRFM) showed that FFN511 undergoes stimulation-dependent exocytosis from LDCVs (Fig. 1C, fig. S4, movie S1) (12). We then examined FFN511 accumulation and release in mouse brain using multiphoton microscopy. FFN511 incubation of acutely prepared slices from the striatum resulted in fluorescent puncta that correlated well with the size of axon terminals (~1 µm, Fig. 2). No label was observed in striatal cell bodies (chiefly medium spiny neurons, MSNs), indicating that accumulation into lysosomes or other acidic organelles was below detection limits. As expected, no labeling was observed in the corpus callosum (CC) which lacks monoamine terminals, and sparser distribution was observed in the cortex (Fig. 2A, Fig. 2G) and hippocampus than in the striatum. Extensive experimental evidence was accumulated to confirm that FFN511 labels dopamine presynaptic terminals in striatum, including the overlap of FFN511 signal with that of GFP using striatal slices prepared from transgenic mice expressing GFP under the control of the tyrosine hydroxylase (TH) promoter (13), (Fig. 2B); exclusion of FFN511 from GFP labled GABAergic striatopallidal MSNs using the BACD2R-EGFP transgenic mice (fig. S5 and movie S2); loss of FFN511 labeling by lesioning dopamine neurons by in vivo injection of the selective dopaminergic neurotoxin 6hydroxydopamine (6-OHDA) into the striatum of one hemisphere (14) (Fig. 2C); inhibition of FFN511 labeling by the VMAT2 inhibitors reserpine (Fig. 2D) and Ro 4-1284 (fig. S6), and an extensive overlap of FFN511 with the endocytic synaptic vesicle marker FM1-43 (Fig. 2E, fig. S7). Furthermore, amphetamine (20 µM, 20 min), which redistributes vesicular dopamine to the cytosol and induces dopamine release without synaptic vesicle fusion (15), caused a substantial loss of fluorescence in striatum (Fig. 2F) and the medial prefrontal cortex (mPFC) (Fig. 2G). Levels of FFN511 sufficient to label terminals in the striatum (10 µM, 30 min) did not significantly alter evoked dopamine release measured by cyclic voltammetry (a reduction of 7.5 ± 4%, mean ± SEM, n = 4 slices, P >0.5). FFN511 cannot be oxidized during electrochemical detection, and higher concentrations of FFN511 (40 µM) decreased the evoked dopamine release by 35.4 ± 1.4% (n = 3) presumably by displacing vesicular dopamine. The probe thus acts as an optical tracer of dopamine, and is sufficiently fluorescent to provide resolution of individual dopamine terminals at concentrations that do not interfere with normal catecholamine accumulation and transmission. We examined activity-dependent release of FFN511 from dopamine synaptic terminals using a “pulse-chase” protocol in which an acute striatal brain slice was labeled with FFN511 (10 µM, 30 min) and then stimulated with current applied by a bipolar electrode (Fig. 3A, movie S3) or high

potassium (fig. S8). Bipolar stimulation at 1, 4 and 20 Hz each evoked exponential destaining (Fig. 3B), with a mean half-time (t1/2) of destaining of 330, 257, and 114 sec respectively, while negligible destaining occurred in the absence of stimulation or when stimuli were applied but calcium channels were blocked by 200 µM cadmium (Fig. 3B). For each stimulus frequency and at all times during the protracted stimuli, the destaining was well described by a single exponential. Thus, under these experimental conditions, neurotransmitter accumulation-competent presynaptic terminals do not display a population of kinetically distinct “reserve” synaptic vesicles. This approach provides a means to address the longstanding question of the fractional neurotransmitter released from terminals during stimulation-dependent exocytosis. The data suggest that 0.03% to 0.21% of dopamine synaptic vesicles fused per stimulus, depending on the stimulation frequency (Fig. 3C). The apparent low probability of release is consistent with cyclic voltammetry recordings that indicate that stimulation-dependent release of dopamine represents a very small fraction of that released with amphetamine (16). There was a depression at more rapid stimulation, with 2-fold less destaining per pulse at 4 Hz and 6-fold less destaining per pulse at 20 Hz than at 1 Hz (Fig. 3C). If dopamine terminals are mostly releasing neurotransmitter by full fusion of the synaptic vesicle (17), these measurements indicate the fraction of transmitter accumulation-competent synaptic vesicles fused per stimulus, and revealed a stimulationdependent form of presynaptic plasticity. The level of fractional destaining was enhanced with higher extracellular calcium (table S1), suggesting that frequency-dependent effects may be related to an activity-dependent decrease of calcium entry per stimulus, although other possibilities such as altered fusion mechanisms, depletion of a readily releasable pool or decreased signal propagation through axons cannot be ruled out. We next analyzed the heterogeneity within large ensembles of dopamine terminals in the striatum. The distribution of individual terminal activities showed a rightward skew at all stimulus frequencies when individual terminal half-times were displayed as histograms (Fig. 3D, P < 0.005, different from normal distribution by KolmogorovSmirnov (KS) test for all three frequencies). In normal probability plots, in which normally distributed data are linear, multimodal distributions can be seen that deviate further from normality with increased stimulus frequency (Fig. 3E; 1 Hz, D= 0.0628 ; 4 Hz D= 0.1160; 20 Hz, D= 0.1321; D statistics were obtained with the KS test and indicate the deviation from normality). All distributions were different from normal distribution and from each other (P < 0.005, KS test). The highly skewed distribution at 20 Hz indicates multiple populations of presynaptic terminals,

/ www.sciencexpress.org / 7 May 2009/ Page 2 / 10.1126/science.1172278

including the most active population in which the probability of terminal content release per stimulus, Pterminal, is ~0.2%, and the slowest population with Pterminal ~ 0.005%. The spatial distribution of presynaptic activity appeared to be complex, with very active and inactive terminals often nearby (Fig. 3F). There was no relationship between initial fluorescent intensity and terminal destaining rates (fig. S9), suggesting that the heterogeneity in terminal activity was unrelated to the number of functional synaptic vesicles. To examine whether terminal heterogeneity may be due to variation in the depolarization of the terminals, we stimulated the same slice preparation at two frequencies and measured FFN511 destaining, first at 1 Hz and then at 20 Hz (Fig 3G, H). The distribution of half-time values obtained from the same set of presynaptic terminals is well fit by a normal distribution at 1 Hz (blue, D= 0.062, P >0.1, KS test) and is in an apparent multiple population distribution at 20 Hz that is not fit by a normal distribution (red, D = 0.194, P < 0.0001, KS test). This confirms that presynaptic terminal heterogeneity is stimulation frequency-dependent, although we cannot exclude the possibility that this is due to differences in axonal action potential propagation. Differential activity of individual glutamate presynaptic terminals has been observed using either postsynaptic electrophysiological measurements (18, 19) or optical imaging indicating membrane fusion (4); FFN511 reveals heterogeneity of dopamine release from individual terminals in the striatum and shows that this is a dynamic phenomenon dependent on stimulation frequency. Cyclic voltammetry recordings in the striatum indicate that dopamine D2R activation inhibits the evoked release of dopamine in a frequency-dependent manner (20), and we thus examined how the frequency-dependent dopamine terminal heterogeneity was affected by the D2R antagonist, sulpiride (10 µM, 20 min) (Fig. 4). There was no significant change in the mean values of terminal kinetics at 1 and 4 Hz, although sulpiride accelerated release in the slow population of terminals more than one standard deviation slower than the expected mean (Fig. 4A and B). At 20 Hz, however, sulpiride increased the mean t1/2 by more than 50% (mean t1/2 = 106.2 ± 4.0 sec for control vs. 161.0±4.7 sec for sulpiride), slowing the destaining kinetics of > 65% of the terminals (Fig. 4C). Remarkably, the distributions of the destaining kinetics were closer to normal in the presence of sulpiride (D = 0.0445 at 1 Hz with sulpiride vs. D = 0.0628 with 1 Hz control; D = 0.0729 at 4 Hz with sulpiride vs. D = 0.1160 with 4 Hz control; D = 0.1033 at 20 Hz with sulpiride vs. D = 0.1321 with 20 Hz control by KS normality test). Except for 1 Hz with sulpiride, all distributions deviated from normality (P < 0.005) and from the corresponding controls (P < 0.001), confirming that the emergence of multiple populations was dependent in large part on D2R activation. Consistently, cyclic voltammetry recordings confirm that sulpiride

enhanced dopamine release by the second stimulus pulse at 20 Hz (2 single pulses with an interval of 0.05 sec, fig. S10A), and enhanced dopamine release by 2 pulses at 20 Hz but did not enhance dopamine released by 30 pulses at 20 Hz (fig. S10B). The phenomenon of frequency and pulsedependent reversal of receptor inhibition of dopamine overflow has been identified using nicotinic antagonists (21, 22). Thus the activity-dependent terminal heterogeneity is associated with receptor-mediated responses, and indicates the presence of frequency-dependent coding (23) that may determine how particular synapses are activated during decision making, habit formation, and learning. Spatial heterogeneity of dopamine release has been demonstrated by electrochemical recordings with a resolution of ~100 µm (24) and has been suggested to play an important role in the modulation of synaptic circuitry involved in motivation, reward and learning. Here we find that the activity of individual dopaminergic presynaptic terminals is modulated by neuronal activity and receptor activation. FFNs enable optical measurements of key presynaptic processes in the CNS, including accumulation of a vesicle transporter substrate and release by evoked activity or drugs such as amphetamine, at unprecedented spatial resolution. FFN511 is compatible with GFP-based tags, the FM1-43 endocytic marker, and other optical probes, which will allow construction of fine resolution maps of synaptic microcircuitry and presynaptic activity, particularly in regions such as the hippocampus and cortex where monoamine innervation can be too sparse for electrochemical recording. References and Notes 1. W. J. Betz, G. S. Bewick, Science 255, 200 (1992). 2. G. Miesenbock, D. A. De Angelis, J. E. Rothman, Nature 394, 192 (1998). 3. Q. Zhang, Y. Li, R. W. Tsien, Science 323, 1448 (2009). 4. V. N. Murthy, T. J. Sejnowski, C. F. Stevens, Neuron 18, 599 (1997). 5. I. J. Kopin, Annu Rev Pharmacol 8, 377 (1968). 6. A. Merickel, P. Rosandich, D. Peter, R. H. Edwards, J Biol Chem 270, 25798 (1995). 7. J. S. Partilla et al., J Pharmacol Exp Ther 319, 237 (2006). 8. Y. Liu et al., Cell 70, 539 (1992). 9. G. Chen, D. J. Yee, N. G. Gubernator, D. Sames, J Am Chem Soc 127, 4544 (2005). 10. Materials and methods are available as supporting material on Science online. 11. D. Sulzer, S. Rayport, Neuron 5, 797 (1990). 12. J. A. Steyer, H. Horstmann, W. Almers, Nature 388, 474 (1997). 13. K. Sawamoto et al., Proc Natl Acad Sci U S A 98, 6423 (2001). 14. T. F. Oo, R. Siman, R. E. Burke, Exp Neurol 175, 1 (2002).

/ www.sciencexpress.org / 7 May 2009/ Page 3 / 10.1126/science.1172278

15. D. Sulzer, M. S. Sonders, N. W. Poulsen, A. Galli, Prog Neurobiol 75, 406 (2005). 16. Y. Schmitz, C. J. Lee, C. Schmauss, F. Gonon, D. Sulzer, J Neurosci 21, 5916 (2001). 17. R. G. Staal, E. V. Mosharov, D. Sulzer, Nat Neurosci 7, 341 (2004). 18. N. A. Hessler, A. M. Shirke, R. Malinow, Nature 366, 569 (1993). 19. C. Rosenmund, J. D. Clements, G. L. Westbrook, Science 262, 754 (1993). 20. Y. Schmitz, C. Schmauss, D. Sulzer, J of Neurosci 15, 8002 (2002). 21. M. E. Rice, S. J. Cragg, Nat Neurosci 7, 583 (2004). 22. H. Zhang, D. Sulzer, Nat Neurosci 7, 581 (2004). 23. N. S. Bamford et al., Neuron 42, 653 (2004). 24. R. M. Wightman et al., Eur J Neurosci 26, 2046 (2007). 25. We thank Robert Burke for 6-OHDA injections and advice, Mark Sonders for discussion, Merek Siu for imaging analysis programming, and Jan Schmoranzer for technical support with the TIRF microscopy setup. Columbia University has applied for a patent on fluorescent false neurotransmitters. D. Sames thanks The G. Harold & Leila Y. Mathers Charitable Foundation and Columbia University’s Research Initiatives in Science and Engineering. D. Sulzer thanks NIDA, NIMH, and the Picower and Parkinson’s Disease Foundations. H. Zhang thanks NARSAD. R.H. Edwards thanks the Michael J. Fox Foundation, the National Parkinson Foundation, NIDA and NIMH. Supporting Online Material www.sciencemag.org/cgi/content/full/1172278/DC1 Materials and Methods Figs. S1 to S10 Table S1 Movies S1 to S3 References 13 February 2009; accepted 20 April 2009 Published online 7 May 2009; science.1172278 Include this information when citing this paper. Fig. 1. In mouse chromaffin cells, FFN511 is accumulated in large dense core vesicles (LDCVs) and is released by exocytosis. (A) Chemical structure of FFN511. (B) Multiphoton image of a chromaffin cell reveals a distribution of FFN511 consistent with LDCVs. Scale bar: 5 µm. (C) FFN511 exocytosis from a LDCV observed with TIRFM images acquired at 500 ms intervals. The upper row shows consecutive images of a single vesicle. Orthogonal section through this vesicle and its integrated intensity are in the middle and lower panels. The dotted line indicates stimulation by high potassium; the delay following

stimulation is typically observed in this preparation (movie S1). Fig. 2. FFN511 selectively labels dopamine terminals in live cortical-striatal acute slices. (A) Labeling by FFN511 in acute live cortical-striatal slice: abundant labeling in the striatum (STR), sparser labeling in cortex (CTX), and no label in corpus callosum (CC). Scale bar: 100 µm. (B) Overall pattern of FFN511 (red) and TH-GFP (green) fluorescent markers show extensive overlap (yellow) as expected for dopamine terminals. The GFP-label (green) is cytosolic and thus fills both terminals and axons, while FFN511 only labels terminals and is more punctate. Scale bar: 10 µm. (C) Nearly complete inhibition of striatal FFN511 labeling 21 days after unilateral 6-OHDA lesion. Striatal slices were examined by cyclic voltammetry to ensure complete loss of dopamine release in the lesion side. (D) FFN511 labeling was strongly inhibited by reserpine (20 µM). Scale bar for C and D: 10 µm. (E) Colocalization of FM 1-43 (red) and FFN511 (green) labeled terminals. Slices were sequentially loaded with 10 µM FM 1-43 and 5 µM FFN511. A representative image from four independent experiments depicts extensive colocalization of FFN511 and FM 1-43 labeled terminals. Scale bar: 4 µm. (F and G) Destaining of FFN511 from the striatum (F) and mPFC (G) by amphetamine. Left panels: before amphetamine; right panels: after 20 minutes of 20 µM amphetamine. Scale bar: 10 µm. Fig. 3. Frequency-dependent induction of multiple dopamine terminal populations. (A) Local stimulation at 4 Hz resulted in destaining from the terminals. Stimulation began at t = 0. Scale bar: 5 µm. (B) Destaining of FFN511 at 4 Hz is Ca2+and frequency-dependent. Controls received no stimulation (153 puncta from 3 slices). Destaining with cadmium chloride (200 µM) was identical to unstimulated controls (475 puncta from 5 slices). The destaining curves for each stimulation frequency were fit by a single exponential decay function and half-life (t1/2) values calculated as   0.693 (1 Hz: 765 puncta from 9 slices, 4 Hz: 410 puncta from 7 slices, 20 Hz: 416 puncta from 6 slices). (C) The dependence of mean fractional release parameter (f, i.e. destaining per stimulus) on stimulus frequency. (D) Histogram of t1/2 values of individual dopamine terminals stimulated at 1 Hz, 4 Hz, and 20 Hz. Bin size: 20 sec. (E) Normal probability plot of half-time values for each terminal at 1, 4 and 20 Hz. The deviation from normality was increased with stimulation frequency. (F) Spatial distribution of FFN511 destaining rates of individual puncta stimulated at 20 Hz are shown in false color. Scale bar: 10 µm. (G) Averaged destaining kinetics of 60 puncta from one region of a slice stimulated consecutively first at 1 Hz (blue) and then at 20 Hz (red). (H) Distribution of t1/2 values for the terminals in G; distribution at 1 Hz was well fit

/ www.sciencexpress.org / 7 May 2009/ Page 4 / 10.1126/science.1172278

by a normal distribution, while the distribution deviated from normality at 20 Hz (movie S3). Fig. 4. Effect of dopamine D2R antagonist sulpiride on presynaptic terminal populations is frequency dependent. (A) Effect of sulpiride on dopamine terminals stimulated at 1 Hz (765 puncta from 9 slices for control, and 901 puncta from 8 slices for sulpiride). Normal probability plot reveals that sulpiride accelerated transmitter release from the slowest terminals. (B) Effect of sulpiride on dopamine terminals stimulated at 4 Hz (519 puncta from 10 slices for control, and 520 puncta from 5 slices for sulpiride). The effect of sulpiride was similar to that at 1 Hz stimulation. (C) Effect of sulpiride on dopamine terminals stimulated at 20 Hz (416 puncta from 6 slices for control, and 508 puncta from 4 slices for sulpiride). Sulpiride inhibited release from >65% of the terminals, except for the most active population. At each stimulation frequency, sulpiride shifted the entire population of the t1/2 values closer to a normal distribution.

/ www.sciencexpress.org / 7 May 2009/ Page 5 / 10.1126/science.1172278

www.sciencemag.org/cgi/content/full/1172278/DC1

Supporting Online Material for

Fluorescent False Neurotransmitters Visualize Dopamine Release from Individual Presynaptic Terminals Niko G. Gubernator, Hui Zhang, Roland G. W. Staal, Eugene V. Mosharov, Daniela Pereira, Minerva Yue, Vojtech Balsanek, Paul A. Vadola, Bipasha Mukherjee, Robert H. Edwards, David Sulzer,* Dalibor Sames*

*To whom correspondence should be addressed. E-mail: [email protected] (D. Sulzer); [email protected] (D. Sames)

Published 7 May 2009 on Science Express DOI: 10.1126/science.1172278 This PDF file includes: Materials and Methods Figs. S1 to S10 Table S1 References Other Supporting Online Material for this manuscript includes the following: (available at www.sciencemag.org/cgi/content/full/1172278/DC1) Movies S1 to S3

Supporting Online Material For

Fluorescent False Neurotransmitters Visualize Dopamine Release From Individual Presynaptic Terminals

Niko G. Gubernator,1,2* Hui Zhang,3,4* Roland G. W. Staal,3 Eugene V. Mosharov,3 Daniela Pereira,3 Minerva Yue,3 Vojtech Balsanek,1 Paul A. Vadola,1 Bipasha Mukherjee,6 Robert H. Edwards,6 David Sulzer,3,4,5# Dalibor Sames1#

*

These authors contributed equally to this work.

#

To whom correspondence should be addressed. E-mail: [email protected],

[email protected].

This PDF includes: Materials and Methods Figure S1 to S10 and Legends Table S1 and Legend Movie S1 to S3 Legends References

1

I. MATERIALS AND METHODS

1. Design Criteria for Fluorescent False Neurotransmitters The chemical design of the novel probes loosely mimics the overall topology and physical properties of monoamine neurotransmitters. Specifically, the aminoethyl group was conserved while the aromatic system was expanded to engineer the desired fluorescent properties. In addition to fluorescence properties, it is paramount that the candidate compounds are mildly basic to provide the driving force for accumulation in the vesicles and sufficiently polar to disfavor passive transport to other acidic compartments such as lysosomes and acidic endosomes. Following these design guidelines, we developed compound FFN511.

2. Structural characterization of FFN511 (8-(2-Amino-ethyl)-2,3,5,6-tetrahydro1H,4H-11-oxa-3a-aza-benzo[de]anthracen-10-one) NMR 1H (300 MHz, CDCl3) δ ppm: 7.00 (s, 1H); 5.90 (s, 1H); 3.24 (m, 4H); 3.04 (m, 2H); 2.88 (t, 2H, J=6.5 Hz); 2.78 (m, 4H); 1.97 (m, 4H); 1.50 (bs, 2H). NMR 13C (75 MHz, CDCl3) δ ppm: 162.4; 154.1; 151.4; 145.8; 121.4; 118.0; 107.9; 107.7; 107.0; 49.9; 49.5; 41.2; 35.7; 27.7; 21.5; 20.6; 20.5. IR (NaCl, cm-1) 2939; 2845; 1702; 1612; 1553; 1519; 1428; 1378; 1312; 1182; 729. LRMS (APCI+): 285 (C17H21N2O2, M+H).

2

3. FFN511 Photophysical Properties Fluorescence measurements were acquired on a Jobin Yvon Fluorolog fluorescence spectrofluorometer (slits 3, HV 750) in 100 mM sodium phosphate pH 7 buffer or chloroform solution (Fig. S1). Measurements in striatal slice were visualized using a Zeiss LSM 510 NLO multiphoton laser scanning microscope with a Titanium-sapphire laser (excitation 760 nm/emission 480-520 nm) equipped with a 63 X 0.9 NA water immersion ultraviolet objective (Zeiss). In pH7 buffer the λem (max) is 501 nm, λex (max) is 406 nm. In chloroform the λem (max) is 447 nm, λex (max) is 391 nm while in slice the λem (max) is ~475 nm.

4. Uptake into primary adrenal chromaffin cells Primary adrenal chromaffin cells. The cultures were prepared as previously described (1). Briefly, adrenal glands from 3-6-month-old mice were dissected in ice-cold Hank's balanced salt saline (HBSS). The capsule and cortex of adrenal glands were removed and the remaining medullae cut in two pieces. After several washes with HBSS, the tissue was incubated with Ca2+-free collagenase IA (250 units/ml, Worthington) in HBSS for 30 min at 30oC with stirring. The digested tissue was rinsed three times and triturated gently in HBSS containing 1% heat-inactivated bovine serum albumin and 0.02% DNAse I. Dissociated cells were collected at 1000 x g for 2 min and resuspended in a culture medium comprised of DMEM, 10% fetal bovine serum, 2 mM glutamine, 50 units/ml penicillin, and 50 µg/ml streptomycin. The cell suspension (100 µl) was plated onto polyD-lysine and laminin coated 1 cm2 glass wells in 50 mm dishes and, after 1-2 hrs, the dishes were flooded with the culture medium. Cells were maintained in a 7% CO2

3

incubator at 37oC. All measurements were conducted between days 2-4 post-plating. The saline for cell incubation at room temperature contained 10 mM HEPES-NaOH (pH 7.2), 128 mM NaCl, 2 mM KCl, 2 mM MgCl2, 1.2 mM CaCl2, 1 mM NaH2PO4, 10 mM glucose. Drugs and Reagents. The reversible VMAT inhibitor 2-hydroxy-2-ethyl-3-isobutyl-9,10dimethoxy-1,2,3,4,6,7-hexahydrobenzo[a]quinolizin hydrochloride (Ro 4-1284) was a gift from Hoffman-La Roche (Nutley, NJ). Ro 4-1284 is a TBZ analogue distinguishable only by the pharmacokinetics and solubility from TBZ. All other compounds were obtained from Fisher Scientific (Springfield, NJ) or Sigma Chemical Co. (St. Louis, MO). Uptake and Inhibition Studies. For experiments with chloroquine, 350 nM FFN511 was added to cell cultures at room temperature. Time lapse microscopy of FFN511 fluorescence intensity of chromaffin cells was measured at 10 minute intervals. Upon visual inspection, only healthy chromaffin cells (e.g. circular, unbroken etc.) were selected. The vitality of cells was confirmed by the presence of normal quantal catecholamine secretion and was established before each labeling study by amperometry. FFN511 uptake rates varied widely between cells (from 4 F. arb. units/min to 30 F. arb. units/min) and one or two cells were analyzed per assay. Uptake of FFN511 was displayed as average mean intracellular fluorescence subtracting fluorescent signal from media nearby. Experiments were performed on an Axiovert 135 TV microscope and analyzed by Axiovision 4.6 software. Amperometry. Solutions used for amperometric recordings were as follows. The bath saline (pH 7.4) contained (in mM): 128 NaCl, 2 KCl, 1 NaH2PO4, 2 MgCl2, 1.2 CaCl2, 10

4

glucose, 10 HEPES-KOH. Secretagogue solution was identical with the exception of 90 NaCl and 40 KCl. Secretagogue was applied by local perfusion through a pressurized glass micropipette (Picospritzer, General Valve Co., Fairfield, NJ) for 5 sec at ~10 μm from the cell. A 5 μm diameter carbon fiber electrode held at +700 mV was pressed against the cell surface and catecholamine oxidation was monitored as amperometric current spikes. The current was filtered using a 4-pole 5 kHz Bessel filter built into an Axopatch 200B amplifier (Axon Instruments, Foster City, CA) and sampled at 25 kHz (ITC-18, Instrutech, Great Neck, NY). After secretagogue application, the amperometric current was recorded for 60 sec and the data were analyzed using a locally written routine in IGOR Pro (WaveMetrics, Lake Oswego, OR) (2). The current was digitally filtered using a low-pass binomial 600 Hz filter and the root mean square of the noise (RMS noise) on the first derivative of the current (dI/dt) was measured in a segment of the trace that did not contain spikes. dI/dt was then used to detect amperometric events that were 4.5-fold larger than the RMS noise. To improve the quality of spike detection, the current was additionally filtered using low-pass binomial 150 Hz filter before taking dI/dt. For each amperometric spike, the following shape characteristics were determined: quantal size (Q, pC), amplitude (Imax, pA), duration at half-height (t1/2, ms), rise-time between 25% and 75% of Imax excluding the pre-spike foot (trise, ms), and the incline of the rising phase (slope, pA/ms). The falling phase of a spike was characterized by two time constants (τ1 and τ2, ms) of the double-exponential fit of the current between 25% of Imax and spike’s end. The number of catecholamine molecules released from individual vesicles (QN) was calculated as QN = Q/(nΔF), where F = 96,485 C/mole is Faraday’s constant and n = 2 is the number of electrons donated by each catecholamine molecule

5

(3). Spikes with Imax smaller than 3 pA and traces with fewer than 10 amperometric spikes were excluded from the analysis (Fig. S3).

5. FFN511’s VMAT2 IC50 value To measure FFN511’s affinity for VMAT2, we assayed its ability to inhibit uptake of 3Hserotonin by membranes from HEK cells stably expressing VMAT2 (4). Uptake by 50 µg extract from at least two different membrane preparations was carried out in triplicate using 2 mM ATP, 20 nM 30 Ci/mmol serotonin and varying concentrations of FFN511 for 2 minutes at 29oC.

6. Total internal reflection fluorescence microscopy (TIRFM) To investigate the ability of FFN511-labeled vesicles to undergo stimulation-dependent exocytosis, we employed live-cell imaging with evanescent field microscopy. Fluorescence was excited in a narrow 70-120 nm thick layer of cytosol immediately adjacent to the coverslip with a 488 nm laser beam to produce total internal reflection (TIR) at the glass/cell interface. TIRFM was performed on a Nikon TE2000 equipped with a 60X PlanApo N.A.1.45 objective. Images were acquired using a cooled CCD (ORCA I, C4742-95, Hamamatsu) controlled by MetaMorph software. Cells were treated with 7 μM FFN511 for 10 min at 37oC and rinsed twice. 300 images were taken from each studied cell at a rate of 2 frames per second (2.5 min total). At 60 sec, a 60 sec-long puff of 90 mM KCl was applied at ~10 µm from the cell through a pressurized glass micropipette (Picospritzer, General Valve Co., Fairfield, NJ). Detailed analysis of the kinetics of vesicle disappearance revealed that at the image acquisition rate of our

6

measurements (2 Hz), the fluorescence signal from a fusing vesicle typically reached the level of background within one frame, although similar to acridine orange studies (5), sometimes a diffuse ‘cloud’ of fluorescence was visible due to diffusion from the vesicle to the media (Fig. S4), indicating exocytotic release.

7. FFN511 in Striatal Slices Slice preparation. Two- to four-month old male C57BL/6 mice were obtained from The Jackson Laboratory (Bar Harbor, ME). All animal protocols were approved by the IACUC of Columbia University. Mice were decapitated without anaesthesia. Acute 250 µm thick coronal striatal brain slices (bregma, +1.54 to +0.62 mm) were cut on a vibratome and allowed to recover for at least 1 hr at room temperature in oxygenated artificial cerebrospinal fluid (ASCF, in mM): NaCl 125, KCl 2.5, NaHCO3 26, CaCl2 2.4, MgSO4 1.3, KH2PO4 0.3, glucose 10, HEPES 5; pH 7.3-7.4, 290-295 mOsm. FFN511 loading and destaining. FFN511 (10 µM in ACSF) was loaded into presynaptic terminals by a 30 min incubation at room temperature. The dye bound to extracellular tissue was removed by a 30 min incubation in ADVASEP-7 (CyDex, Overland Park, KS; 100 µM in ACSF) (6). The slice was then placed in a recording chamber and superfused (1ml/min) with ACSF. Slices were allowed to rest for at least 10 minutes in the chamber before imaging. For stimulation-dependent dopamine terminal destaining, stimuli at 1, 4 or 20 Hz (300 µs x 1 mA) were applied to the striatum locally by an Iso- Flex stimulus isolator triggered by a Master-8 pulse generator (AMPI, Jerusalem, Israel) using bipolar electrodes. To minimize the variation in the depolarization of release sites, we used a stimulation protocol applying 150% of the

7

maximal stimulation intensity determined by cyclic voltammetry. All experiments were performed at room temperature, except for (+)-amphetamine sulphate (AMPH, 20 µM) induced destaining experiments, which were performed at 36 ºC. Imaging and data analysis. Striatal terminals were visualized using a Zeiss LSM 510 NLO multiphoton laser scanning confocal microscope with a Titanium-sapphire laser (excitation 760 nm/emission 480-520 nm) equipped with a 63 X 0.9 NA water immersion ultraviolet objective (Zeiss). Images were captured in 8-bit, 73 x 73 µm regions of interest at 512 x 512 pixel resolution and acquired at 27-second intervals using Zeiss LSM 510 software. To compensate for z-axis shift, a z-series of 5 images, separated by 1 µm in the z-plane, was obtained for each period. Some images were acquired at 60second intervals with a z-series of 12 images. Images in each z-series were aligned and condensed with maximum transparency. The time projection was analyzed for changes in puncta fluorescence using Image J (Wayne Rosband, National Institutes of Health, Rockville, MD) and custom-written software in IDL (Research Systems, Boulder, CO) (7). The striatal region of interest (ROI) where fluorescent puncta were analyzed was within 300 µm of the stimulation bipolar electrodes. Fluorescent puncta 0.3 – 1.5 µm in diameter were identified. The criteria for including puncta were: 1) spherical shape, 2) fluorescence that is two standard deviations above the background, and 3) stimulation dependent destaining. The program aligned puncta in the x, y, and z plane by shifting each image in 3 dimensions based on the location of the peak of their cross correlograms with the first z-series. Images showing projections of maximal z-axis intensity were made of each stack and the intensity of FFN511 fluorescence for each punctum was measured over the time interval. To correct for minor changes in

8

background fluorescence (due to minor tissue bleaching), the background fluorescence of each image (<10%) was subtracted from the fluorescence intensity of individual puncta. The results were then normalized by the maximal fluorescence intensity of that punctum just before unloading. The halftime decay of intensity during destaining (t1/2) was determined graphically. The fractional release parameter ƒ, was calculated from ln (F1/F2)/ ΔAP (8), where F1 and F2 are the fluorescent intensities at t1 and t2, respectively and ΔAP is the number of action potentials delivered during that period. Imaging FM1-43 with FFN511. Mouse corticostriatal slices were loaded with 10 µM FM 1-43 by incubation with the dye for 10 min in regular ACSF followed by 10 min in 40 mM KCl ACSF also containing FM 1-43. The slices were allowed to recover for 15 min in ACSF and they were then incubated with 5 µM FFN511 for 30 min. ADVASEP7, at a concentration of 100 µM, was applied for 30 min to remove any non-specifically bound dye. Slices loaded with FM 1-43 and FFN511 were imaged by two-photon microscopy (Leica DM6000) at an excitation wavelength of 810 nm. FM 1-43 fluorescence was detected at 570-650 nm, while FFN511 was detected at 450-510 nm. Under these conditions, the FM 1-43 and the FFN511 fluorescent signals are lower than what is obtained at their optimal excitation and emission ranges, but can be simultaneously acquired in both channels with no significant crossover. Single Z-section images from slices belonging to four different mice of similar age were obtained using the LASAF software. Imaging FFN511 with GFP. FFN511 was excited at 760 nm, while TH-GFP was excited at 910 nm. Although FFN 511 and GFP have similar emission profiles (emission maximum ~475 nm for FFN511 and 510 nm for GFP in slices), the excitation

9

wavelengths are sufficiently different to allow selective imaging of each probe.

8. Cyclic voltammetry (CV). Cylinder carbon fiber electrodes of 5 µm diameter and 30 to 100 µm length were placed into the dorsal striatum. For CV, a triangular voltage wave (-400 to +900 mV at 280 V/s versus Ag/AgCl) was applied to the electrode every 100 ms. Current was recorded with an Axopatch 200B amplifier (Axon Instrument, Foster City, CA), with a low-pass Bessel Filter setting at 5 kHz, digitized at 20 kHz (ITC-18 board, Instrutech Corporation, Great Neck, New York). Triangular wave generation and data acquisition were controlled by a PC computer running a locally written IGOR program (E. Mosharov, Columbia University; WaveMetrics, Lake Oswego, OR). Striatal slices were electrically stimulated every 2 minutes with a single pulse stimulation by an IsoFlex stimulus isolator triggered by a Master-8 pulse generator (A.M.P.I., Jerusalem, Israel) using a bipolar stimulating electrode placed at a ~100 µm distance from the recording electrode. Background-subtracted cyclic voltammograms served to identify the released substance. The DA oxidation current was converted to concentration based upon a calibration of 5 µM DA in ACSF after the experiment.

10

II. SUPPLEMENTAL FIGURES AND LEGENDS

A

B Excitation Emission

2500000

1200000 Fluor, arb units

2000000 Fluor, arb units

Excitation Emission

1400000

1500000 1000000

1000000 800000 600000 400000

500000 200000 0

0 300

350

400 450 500 Wavelength (nm)

550

600

300

350

400

450

500

550

600

Wavelength (nm)

Fig. S1. FFN511 fluorescent excitation and emission spectra. (A) chloroform and (B) 100 mM sodium phosphate pH 7 buffer.

Fig. S2. Accumulation of FFN511 (350 nM) measured by epifluorescence microscopy of individual chromaffin cells, was linear from 10-60 min. The labeling was abolished by the lipophilic weak base, chloroquine (100 µM, arrow).

11

Quantal size (# molecules)

Imax (pA)

t1/2 (msec)

#events/cell

n cells recorded

%feet

FFN511

342,000 ± 35,000

15.0 ± 3.3

5.5 ± 1.0

35 ± 7

14

27

Control

291,000 ± 47,000

12.5 ± 1.4

4.9 ± 0.5

28± 3

30

25

Fig. S3. Average of the median values of quantal parameters from individual mouse chromaffin cells ± SEM. No parameters were significantly different by ANOVA, including additional parameters (e.g., slope) not shown.

A 0

75

100

B -2s

-1.5s

-1s

-0.5s

0s

0.5s

1s

1.5s

Fig. S4. (A) Cultured rat chromaffin cells were incubated with 7µM FFN511 for 10 min, and secretion was then stimulated with 90 mM K+ starting at 60 s. Frames from the video at 0, 75 and 100 s are shown: in the second panel, a vesicle that does not fuse is circled in yellow, and one that fuses is shown in red. The blue circle indicates a vesicle that originally was not present and then either fused or moved away from the surface. The images are displayed as time series with the vesicles indicated by the colored arrows. Time scale bars are 30 sec; spatial scale bar is 1 µm. (B) A critique of TIRFM exocytosis experiments is that if the vesicle disappears as it moves away from the membrane, it could be mistakenly considered to have fused. We examined whether a vesicle could be caught “during the act” of exocytosis, when intermediate destaining could be observed. In this example of a single vesicle acquired as in (A), the fluorescence diffuses at 500 ms, which is thought to be evidence for exocytosis and is inconsistent with the vesicle moving away from the focal plane, which would not show expansion of the fluorescent area.

12

GFP

511

Merged

Fig. S5. FFN511 does not label GABAergic neurons and their neurites in the striatal brain slice. (A) Shown in green are the EGFP-labeled GABAergic striatopallidal neurons. The BAC-D2R-EGFP mice (bacterial artificial chromosome-transgenic mice that express GFP under the control of the dopamine D2 receptor promoter) were obtained from Gensat. (B) FFN511 labeling is shown in red. (C) Overlap of overall pattern of FFN511 with EGFP. FFN511 was excited at 760 nm, while EGFP was excited at 910 nm. Scale bar: 10 µm. Movie S2 shows 3-dimensional projection of this double label.

Control 6-OHDA lesion sideRO 4-1284

MSN

MSN

Fig. S6. FFN511 labeling was strongly inhibited by VMAT inhibitor, Ro 4-1284 (20 µM). Scale bar: 10 µm.

13

FM

511

Merged

Fig. S7. Colocalization of FM 1-43 (red) and FFN511 (green) labeled terminals. Mouse corticostriatal slices were sequentially loaded with 10 µM FM 1-43 and 5 µM FFN511 and simultaneously visualized. A representative image from four independent experiments depicts extensive colocalization of FFN511 and FM 1-43 labeled terminals. Scale bar: 4 µm.

control

After KCl

Fig. S8. Destaining of FFN511 labeling by high KCl. FFN 511 labeling is destained within 2 min application of 70 mM KCl ACSF. Scale bar: 10µm.

14

Fig. S9. Example of relationship between initial fluorescent intensity and terminal destaining. The t1/2 of terminal destaining (y) versus initial fluorescent (x) from 1 slice at 20 Hz is fit by linear regression, y=51.39+0.66x. The coefficient r2 is 0.014, suggesting that differences in the activity of dopamine terminals are unrelated to synaptic vesicle FFN accumulation. The analysis was performed in at least 3 slices at each frequency, and in no case was a relationship between initial fluorescent intensity and terminal destaining observed.

A

B

*

Control

*

Sulpiride

Fig. S10. Effect of sulpiride on evoked dopamine release. (A) Effect of sulpiride on paired pulse ratio. Sulpiride (10 µM, 20 min) increased release from the second pulse at 20 Hz (2 single pulses at an interval of 0.05 sec, paired pulse ratio is 0.133± 0.027 with sulpiride compared to 0.064± 0.012 as control, n=9, p < 0.05, t test), indicating D2R is effective at 20Hz. (B) Changes of evoked dopamine release after application of sulpiride. Sulpiride had no effect on dopamine release elicited by a single pulse (101 ± 4% of control, n=7, p>0.05), indicating no endogenous D2R tone in the slices. It slightly enhanced dopamine release elicited by 2 pulses at 20 Hz (114± 7%, n=6, p < 0.05), but it had no effect on evoked dopamine release elicited by 30 pulses at 20 Hz (96 ± 4%, n=7, p > 0.05). One way ANOVA (Dunnett) test. 15

III. Supplemental Table 1 [Ca2+]

t 1/2 (sec)

f

0 mM 0.5 mM 2.4 mM 10 mM

>5000 258 210 184

<0.01% 0.067% 0.082% 0.094%

Table S1. Dependence of fractional FFN511 release per stimulus (f) on extracellular Ca2+ levels in slices stimulated at 4 Hz. Between three to five slices with a total of 205-503 puncta were measured per group.

IV. Movie Legends Movie S1 Video micrograph from TIRF microscopy of FFN511 destaining during LDCV exocytosis from a chromaffin cell. Movie S2 Video micrograph of a 3-dimensional projection of FFN511 label and EGFP labeled striatopallidal medium spiny neurons. Movie S3 Video micrograph from 2-photon microscopy of FFN511 destaining during exocytosis in a striatal brain slice preparation.

V. SUPPORTING REFERENCES 1. E. V. Mosharov, L. W. Gong, B. Khanna, D. Sulzer, M. Lindau, J Neurosci 23, 58355845 (2003). 2. E. V. Mosharov, D. Sulzer, Nat Meth 2, 651-658 (2005). 3. D. Bruns, R. Jahn, Nature 377, 62-65 (1995). 4. J. P. Finn, R. H. Edwards, J Biol Chem 272, 16301-16307 (1997). 5. J. A. Steyer, H. Horstmann, W. Almers, Nature 388, 474-478 (1997). 6. A. R. Kay et al., Neuron 24, 8098-17 (1999). 7. S. S. Zakharenko et al., Nat Neurosci 4, 711-717 (2001). 8. J. S. Isaacson, B. Hille, Neuron 18, 143-152 (1997).

16