Probing Rotational Viscosity in Synaptic Vesicles

Abstract

The synaptic vesicle (SV) is a central organelle in neurotransmission, and previous studies have suggested that SV protein 2 (SV2) may be responsible for forming a gel-like matrix within the vesicle. Here we measured the steady-state rotational anisotropy of the fluorescent dye, Oregon Green, within individual SVs. By also measuring the fluorescence lifetime of Oregon Green in SVs, we determined the mean rotational viscosity to be 16.49 ± 0.12 cP for wild-type (WT) empty mice vesicles (i.e., with no neurotransmitters), 11.21 ± 0.12 cP for empty vesicles from SV2 knock-out mice, and 11.40 ± 0.65 cP for WT mice vesicles loaded with the neurotransmitter glutamate (Glu). This measurement shows that SV2 is an important determinant of viscosity within the vesicle lumen, and that the viscosity decreases when the vesicles are filled with Glu. The viscosities of both empty SV2 knock-out vesicles and Glu-loaded WT vesicles were significantly different from that of empty WT SVs (p < 0.05). This measurement represents the smallest enclosed volume in which rotational viscosity has been measured thus far.

Introduction

The fusion of neurotransmitter-containing synaptic vesicles (SVs) is the primary signaling event between neurons. Work over the last decade has identified most of the proteins associated with SVs, and studies aimed at determining where these proteins act in the cycle of vesicle formation, filling, targeting, fusion, and recycling are elucidating the molecular pathway that underlies synaptic transmission (1–3). As the next stage in understanding the functioning of SVs, we are interested in applying sensitive optical techniques to study individual SVs with single-molecule resolution. We believe the small size (only ∼40 nm in diameter) of SVs makes them potentially amenable to high-resolution, single-molecule experiments and thorough biophysical studies. Here, we report measurements of the rotational viscosity inside individual SVs.

It has been suggested that the interior of SVs may be gel-like, with a proteoglycan matrix that changes size upon neurotransmitter loading (4,5). Furthermore, it is thought that the SV protein SV2 (6), which is a 12-transmembrane-domain protein with heavily glycosylated intravesicular domains, is largely responsible for the gel-like interior of SVs. SV2 comes in three isoforms: A, B, and C (of which A and B are the most numerous). SV2A and SV2B are critical for proper SV function, and mice that lack these SV2 proteins (SV2A and SV2B double knock-outs (SV2 DKOs)) die within 3 weeks (7). SV2 resembles several well-known membrane transport proteins (1), is the binding target of the antiepileptic drug levetiracetam (8,9), and is the route of entry for botulinum toxin (10). By measuring the rotational diffusion rates of the small fluorescent molecule Oregon Green (OG) (11) within both wild-type (WT) and SV2 DKO SVs, we determined that SV2 is indeed an important determinant of viscosity in SVs.

We recently reported that the hydrodynamic diameters of SVs increase by as much as 25%, which corresponds to a doubling of the volume, when the vesicles are loaded with glutamate (Glu) (5). Given the large change in vesicle size that occurs upon loading with Glu, we also decided to gain a better understanding of how the luminal viscosity might differ between empty vesicles compared with Glu-loaded vesicles. In our studies, we focused on the neurotransmitter Glu because we estimated that ∼80% of SVs in the brain are glutamatergic (5). By measuring the rotational viscosity in empty and Glu-filled vesicles, we found that the presence of Glu resulted in a decrease in the luminal viscosity of SVs.

Through these two sets of studies, we sought to gain a better understanding of the local environment inside the SV, how tightly organized the internal matrix of the vesicle is, whether SV2 is indeed the key determinant of the luminal viscosity, and how the internal matrix might change as the vesicle becomes filled with neurotransmitters. This study provides biophysical insight into the luminal environment of the SV and the smallest subcellular organelle in which rotational viscosity has been probed.

Materials and Methods

SV isolation and purification

WT SVs were collected from commercially stripped mouse brains (Pel-Freeze, Rogers, AR) or from SV2 DKO mouse brains graciously provided by the laboratory of S.M. Bajjalieh. The mouse brains were ground with liquid nitrogen in a Waring blender and homogenized in 50 mM HEPES, 2 mM EGTA, and 0.3 M sucrose at pH 7.4 with 20 strokes in a glass-Teflon homogenizer. The homogenate was spun in an ultracentrifuge (Beckman Coulter, Fullerton, CA) at 100,000 × g for 28 min to pellet cell debris. The supernatant containing SVs was removed and the pellet was discarded. The supernatant was further purified by layering onto a 1.5/0.6 M sucrose step gradient and spun at 260,000 × g for 72 min. The SVs were subsequently removed from the interface of the step sucrose gradients. The vesicles were separated into aliquots and flash-frozen in liquid nitrogen and stored at −80°C until use. All of the buffered solutions were made with Milli-Q water (Millipore, Billerica, MA).

Loading of SVs with OG

The details of loading OG 488 (Invitrogen, Carlsbad, CA) into SVs have been described elsewhere (11). Briefly, to load the SVs with OG, we first used a syringe to load the SVs into a microfluidic chip. After the SVs in solution adsorbed onto the coverslip glass surface (which formed the floor of the microchannel), the medium was replaced with a pH 5.1 citrate buffered solution containing 290 mM glycine, 10 mM citrate, and 5 μM OG (load buffer) for 5 min to load OG into the vesicles. The buffered OG solution was then flushed out using a pH 7.2 solution containing 10 mM HEPES and 290 mM glycine (clear buffer).

For the two-color overlay experiment, WT mouse SVs were labeled with anti-SV65 monoclonal primary antibody for 15 min at 20°C and then labeled with secondary antibody conjugated with the fluorescent probe Alexa Fluor 633 (Invitrogen) for an additional 15 min at 20°C. Excess antibody was removed using IgG conjugated agarose beads specific to the primary and secondary antibodies in two steps. The SVs were incubated for 45 min at 20°C with the IgG beads, and the beads were then separated via centrifugation at 1000 × g for 2 min. The supernatant containing fluorescently labeled SVs was removed, and the SVs were imaged in a polydimethylsiloxane (PDMS) well that was sealed to a glass coverslip after exposure to oxygen plasma.

Glu-loading assay

Loading of Glu into the SVs was verified by an NADPH fluorescence assay. The SVs were loaded with Glu by adding a solution of 10 mM HEPES, 4 mM KCl, 4 mM MgSO₄, 0.3 M sucrose, 1.0 mM Glu, and 1 mM ATP. After loading for 10 min, the SVs were pelleted by centrifugation at 125,000 × g for 30 min. The supernatant was removed and the pellet was repeatedly rinsed with clear buffer and then resuspended in 10 mM HEPES, 150 mM KCl, 0.30 M glucose, 1 mM NADP⁺, and 1.4 mg/mL L-glutamic dehydrogenase (Sigma Aldrich, St. Louis, MO). The SVs were ruptured in a sonication bath, and the presence of NADPH was detected by a Fluorolog 3 (Horiba, Edison, NJ) with excitation set to 360 nm.

Microscope setup

Imaging was carried out on an inverted TE-2000 microscope (Nikon, Melville, NY) equipped with a 100× magnification and 1.3 numerical aperture (NA) objective. The samples were illuminated with a Sapphire 488 nm laser (Coherent, Santa Clara, CA) directed through a polarizing beam splitter cube, and polarization was adjusted with the use of a zero-order λ/2 waveplate. The microscope was also equipped with a 500 long-pass emission filter and a 355/488/633/1064 polychroic beam splitter. The emitted light was separated by polarization using an in-house-built setup and imaged onto a Cascade 512b EMCCD camera (Photometrics, Tuscon, AZ). For two-color imaging, a 633 nm HeNe (Coherent) was also used, and total internal reflection fluorescence illumination was performed with a 60× magnification 1.45 NA objective (Nikon). Red fluorescence was collected with a 695/100 band-pass filter. All optics and optomechanics used were purchased from either Thorlabs (Newton, NJ) or Newport (Irvine, CA). All experiments were carried out at a temperature of 17–18°C.

Optical corrections

The G-factor (a term used to correct for the instrument's sensitivity difference between I_‖ and I_⊥) was reevaluated regularly, but no measurable changes were detected. Subtraction of the background and instrumental bias from the fluorescence images of SVs was carried out with the use of Metamorph software (Molecular Devices, Sunnyvale, CA).

Microfluidic chip

All anisotropy measurements were carried out in a microchannel (1 mm wide × 100 μm high × 25 mm long). To create the microchannel, we first made a chrome mask of the desired features and aligned the mask above a silicon wafer with a 100 μm layer of the negative photoresist SU-8 spin-coated onto its surface. The wafer was exposed to UV light, and after washing and development steps were completed, a silicon master with the desired features was left. A PDMS/catalyst mixture was then poured onto the master and allowed to harden in a 60°C oven for several hours. The PDMS mold was removed from the silicon master and bonded to a glass coverslip after exposure to oxygen plasma. Solutions and SVs were injected into the microchannel via a syringe. Before conducting the plasma-bonding step, we cleaned the glass coverslips by placing them in a boiling solution of 1:1:1 of water/ammonium hydroxide/30% hydrogen peroxide for 2 h followed by thorough rinsing with milli-Q water.

Measurements of fluorescence lifetime

We measured the fluorescence lifetime of OG inside the SVs using a 470 nm picosecond laser coupled to a Picoharp 300 TCSPC (PicoQuant, Berlin, Germany). SVs were loaded with OG and Glu, and then pelleted via centrifugation at 125,000 × g for 30 min. After the supernatant was removed, the pellet was thoroughly rinsed and then resuspended in a cuvette with clear buffer. The presence of intact SVs after this process was verified by microscopy. We observed resuspended SVs under epi-illumination prepared in this manner, and did not observe any noticeable OG leakage from the vesicles within our timescale of measurement.

Results and Discussion

Because of the extremely small volume enclosed within an SV (∼40 nm in diameter, which corresponds to a volume of ∼2 × 10⁻²⁰ L within the SV membrane), it is difficult to probe its internal environment by following a molecule's translational motion within this tightly confined space. Therefore, we decided to follow the rotational diffusion of molecules within this intravesicular volume using fluorescence polarization anisotropy (12). Here, a fluorophore's anisotropy, r, is a comparison of the difference between the intensity of light emitted that is parallel to the polarization of the excitation light (I_‖) and the intensity of emission perpendicular to the excitation polarization (I_⊥) divided by the total light intensity. Specifically,

$$r=\left(I_{\Vert }-I_{\mathit{\perp }}\times G\right)/\left(I_{\Vert }+\mathit{2}\times I_{\mathit{\perp }}\times G\right),$$

where G is a correction factor for instrumental bias. The ratio between measured anisotropy and the theoretical maximum anisotropy can be used to find a fluorophore's rotational correlation via the Perrin equation:

$$\left(r_0/r\right)=\mathit{1}+\tau /\theta ,$$

where r₀ is the maximum anisotropy of a stationary fluorophore, τ is the fluorescence lifetime, and θ is the rotational correlation time. The rotational correlation time is proportional to the viscosity of the medium as shown by

$$\theta =\left(\eta \mathit{V}\right)/\left(\mathit{kT}\right),$$

where η is the solvent viscosity, V is the hydrodynamic volume of the rotating unit, k is Boltzmann's constant, and T is the temperature (13).

Loading of SVs with OG

To measure the rotational viscosity inside SVs using fluorescence polarization anisotropy, we must be able to introduce fluorophores into the interior volume of the vesicles. Here, we take advantage of an observation we made previously (11), i.e., when the pH of the solution around vesicles is lowered to ∼5.1, OG can be loaded. OG loading is specific to this pH range and OG leaks out slowly at physiological pH. The mechanism of OG uptake into SVs is not clear, but several lines of evidence show that it is not nonspecifically adsorbed on the surface of the vesicles (11).

To verify that we were observing SVs and that they were indeed loaded with OG, we labeled SVs with primary monoclonal antibody against the membrane protein SV65 and a red fluorescent secondary antibody (Alexa633), which was spectrally distinct from the green emission of OG. The antibody-labeled SVs were then loaded with OG, after which we used two-color total internal reflection fluorescence microscopy to image the antibody-labeled (red) and OG-loaded (green) SVs, and to ensure that the OG indeed loaded selectively into the SVs as identified with the SV-specific antibody. Fig. 1 A shows representative two-color images in which it is evident that there is a good overlay between the antibody-labeled (red) and OG-loaded (green) SVs, which indicates a high specificity of OG loading into SVs. Because of its photostability and relative insensitivity to pH, OG is a robust fluorescence probe for the lumen of SVs.

Figure 1.

(Color online) Two-color fluorescence imaging of SVs and optical setup for measuring fluorescence polarization anisotropy of individual vesicles. (A) Two-color imaging of SVs. Left column (panels 1, 4, and 7): SVs with green fluorescence from loaded OG. Middle column (panels 2, 5, and 8): SVs labeled with red fluorescent antibodies specific to the SV protein SV65. Right column (panels 3, 6, and 9): Overlays of the green and red images. (B) Schematic of the microscope setup used to determine the steady-state anisotropy of individual SVs; the fluorescent spots in the inset are individual SVs loaded with OG.

Fluorescence anisotropy measurements

Fig. 1 B shows the microscope setup we used to measure the fluorescence anisotropy of OG in SVs. Fig. 2 shows the measured anisotropy of OG in 752 SVs empty of neurotransmitters from WT mice (A), 794 empty SVs from SV2 DKO mice (B), and 781 SVs from WT mice loaded with the neurotransmitter Glu (C). The mean anisotropies of the SVs were 0.146 ± 0.003, 0.115 ± 0.002, and 0.119 ± 0.002 for empty WT, empty SV2 DKO, and Glu-loaded WT SVs, respectively (the uncertainties represent the standard error of the mean). The broad distribution (in terms of standard deviation, the values are ±0.074, ±0.066, and ±0.051 for WT SVs, empty SV2 DKO SVs, and Glu-loaded WT SVs, respectively) of measured anisotropies is likely caused by a combination of factors, such as the heterogeneity inherently present in the SVs, which came from a whole mouse brain preparation, or nanoscopic imperfections in the glass surface that could lead to local changes in the polarization of the light around each SV. However, there is still a statistically significant difference between the anisotropy of WT empty vesicles and the anisotropy of empty DKO or Glu-loaded WT SVs (p < 0.05). In a comparison of empty SV2 DKO and Glu-loaded SVs, the difference was not found to be statistically significant. Another concern is that the use of a high NA objective (NA = 1.3) can lead to uneven polarization across the field of view and also cause polarization mixing; however, we studied this issue and did not find it to be a problem (see next section).

Figure 2.

(Color online) Histograms of the measured anisotropies of three different sets of SVs. (A) Empty SVs from WT SVs have a mean anisotropy of 0.146 (n = 752). (B) Empty SVs from SV2 DKO mice have a mean anisotropy of 0.115 (n = 794). (C) WT SVs loaded with the neurotransmitter Glu have a mean anisotropy of 0.119 (n = 781).

Effect of high NA on polarization

To address concerns regarding the effect of high NA on fluorescence polarization measurements (14), we measured the anisotropy of OG in various glycerol/water solutions obtained with a 0.40 NA (20× magnification) objective and the 1.3 NA (100× magnification) objective used to measure the anisotropy of the SVs. The measured anisotropy between the two objectives was not significantly different (data not shown), so the effect of NA on polarization anisotropy was assumed to be small.

Assay for verifying loading of Glu into SVs

In the absence of ATP, Glu leaks out of SVs (5). As a result, the SVs we isolated and purified from whole brain were empty and devoid of Glu. However, incubating SVs in a loading buffer that contains ATP and Glu will refill the SVs with Glu (5). To verify that the Glu-loaded SVs were indeed refilled with neurotransmitter, we used a fluorescence assay for Glu:

$$\text{Glutamate}+{\text{NADP}}^{+}{}^{\stackrel{\mathrm{L-GDH}}{\to }}2\text{-oxoglutarate}+{\mathrm{NH}}_4^{+}+\text{NADPH}$$

Here, Glu dehydrogenase (GDH) converts Glu and NADP⁺ to 2-oxoglutarate and NADPH. Therefore, the presence of Glu can be verified by the conversion of NADP⁺ to NADPH, because fluorescence of NADP⁺ peaks at λ_em 400 nm, whereas NADPH has a fluorescence peak at 460 nm.

We loaded Glu into the SVs by incubating purified SVs with Glu in the buffer described in Materials and Methods. Fig. 3 shows the results of our fluorescence assay for Glu-loaded SVs. Without ATP, Glu does not load into SVs; therefore, we used SVs incubated in Glu loading buffer without ATP for the negative control. The reduced NADPH peak in the negative control shows that there was little Glu present in the vesicle without active transport. However, there seems to be a very small amount of NADPH fluorescence, likely because some extravesicular Glu that was present in high concentrations in the loading solution was not removed or adsorbed to the SVs. The positive control contained Glu diluted to a final concentration of 5 μM in a SV-free solution. When the Glu-loaded SVs were disrupted through sonication to test for the presence of Glu, their contents were diluted by several orders of magnitude by the extravesicular assay buffer. Because of this large dilution, the detected NADPH fluorescence was weak, although it still clearly demonstrated Glu loading into SVs.

Figure 3.

(Color online) Fluorescence assay to verify loading of Glu into SVs. In the presence of Glu, GDH converts NADP⁺ into NADPH, which has a broad fluorescent peak with a center wavelength of ∼450 nm. In this assay, the amount of Glu present in the sample is correlated with the intensity of the NADPH fluorescence. The red curve is the positive control, which had 5 μM of Glu in solution. The blue curve shows the presence of Glu in SVs loaded with Glu using a loading buffer that contained Glu and ATP. Because of the extremely small volume of the SVs, a large dilution of Glu occurred when the intravesicular Glu was released into solution after lysis of the SVs. The green curve is the negative control, which was identical to the procedure and solution used for the blue curve but without ATP, which is needed for active transport of Glu from the extravesicular solution into the vesicles during loading.

Measurements of the fluorescence lifetime of OG in SVs

To quantify the rotational viscosity of the vesicle lumen based on the fluorescence polarization anisotropy of OG and the Perrin equation, we had to also determine the fluorescence lifetime of OG. Because the fluorescence lifetime can change depending on the immediate environment experienced by the dye molecules, we first measured the fluorescence lifetime of OG after it had been loaded into the SVs. We carried out this experiment by loading the SVs with OG, as described above, and then separating the SVs from the free OG in solution by pelleting the SVs via centrifugation. We then thoroughly rinsed the OG-loaded SV pellet and resuspended it in a buffered solution before conducting the measurement. For Glu-loaded SVs, an additional Glu loading step followed the loading of SVs with OG.

Fig. 4 shows the measured fluorescence lifetimes of OG in WT empty SVs, SV2 DKO empty SVs, and WT Glu-loaded SVs. The fluorescence lifetime of OG did not vary greatly between the samples, which indicates that no significant amount of self-quenching or fluorescence depolarization (resulting from, e.g., homo-Förster resonance energy transfer) occurred after the OG was loaded into the SVs. Additionally, the fits to the fluorescence lifetimes were not improved by including additional lifetime components. This suggests that the OG loaded within the SVs was homogeneous, as OG molecules adsorbed to the membrane surface of SVs or stacked on top of one another should have quite different fluorescence lifetimes. The presence of only one resolved fluorescence lifetime, as well as results from previous experiments (11) and the specificity of OG loading at pH 5.1, indicates that the OG was freely rotating in the SV lumen instead of being absorbed to the SV membrane.

Figure 4.

(Color online) Fluorescence lifetimes of OG. Fluorescence decay curve of (A) free OG in solution, (B) OG in empty WT SVs, (C) OG in empty SV2 DKO SVs, and (D) OG in WT SVs filled with Glu. The additional peaks in panels C and D are from the instrument response function of the detector. They are more prominent here than in the other panels because of the low fluorescence signal from OG. Each graph also displays the χ²-values for the single-exponential lifetime fittings, and below each decay curve is a plot of the residuals from fitting the curves to a single-exponential decay function.

Rotational viscosity in SVs

To determine the rotational viscosity within SVs, we first had to obtain a calibration curve using the anisotropies and lifetimes of OG. To that end, we added OG to solutions containing progressive ratios of glycerol/water, and measured their anisotropies to extrapolate to the anisotropy of OG in a solution with infinite viscosity, that is, the equivalent of a stationary OG molecule as shown in the inset graph in Fig. 5. We found the maximum anisotropy of OG, r₀, to be 0.346 ± 0.003, which is close to previous measurements for different adducts of OG (15). Fig. 5 plots the theoretically calculated rotational viscosities and the experimentally measured values, which match well with the theoretically calculated viscosities as determined using a method developed by Cheng (16).

Figure 5.

(Color online) Measurements of rotational viscosity. The blue line represents theoretically calculated rotational viscosities; the dots with error bars are experimental values we measured using our setup for OG in aqueous solution containing varying amounts of glycerol. WT empty vesicles, SV2 DKO empty vesicles, and WT Glu-loaded vesicles were placed on their respective positions on the curve based on their average anisotropies and measured fluorescence lifetimes. Inset: The Perrin plot used to extrapolate r₀, the absolute anisotropy of OG.

With this calibration curve, we were able to determine the mean viscosity within each of the three types of SVs we prepared. Empty SVs from WT mice had the highest viscosity, and both the empty SV2 DKO vesicles and WT SVs filled with Glu had significantly lower viscosities. Table 1 summarizes our data.

Table 1.

Data of merit

	WT	SV2 DKO	Glu-loaded	OG in water
Mean anisotropy	0.146 ± 0.003	0.115 ± 0.002	0.119 ± 0.002	0.0162 ± 0.0009
OG fluorescence lifetime (ns)	4.01 ± 0.02 (χ2 = 1.002)	3.99 ± 0.01 (χ2 = 1.100)	3.84 ± 0.05 (χ2 = 1.007)	3.85 ± 0.05 (χ2 = 1.209)
Viscosity (cP)	16.49 ± 0.12	11.21 ± 0.12	11.40 ± 0.65	1.08 ± 0.06
Rotational correlation time (ns)	2.93 ± 0.02	1.99 ± 0.02	2.01 ± 0.11	0.19 ± 0.01

WT: empty SVs from WT mice (n = 752); SV2 DKO: empty SVs from SV2 DKO mice (n = 794); Glu-loaded: SVs from WT mice filled with Glu (n = 781); OG in water: OG in water (n = 3). The SV anisotropy values reported in the table are the calculated standard error of the mean for each SV population. The standard deviations for WT, SV2 DKO, and Glu-loaded SVs are ±0.074, ±0.066, and ±0.051, respectively. Despite the large variations in each SV sample, there is a highly statistically significant difference between the measured viscosities of empty WT SVs and SV2 DKO SVs (t = 8.69, p < 0.0001) and between empty WT SVs and Glu-loaded SVs (t = 8.28, p < 0.0001).

The high viscosity (16.5 cP) measured inside WT empty vesicles reflects both the minuteness of the enclosed volume (∼2×10⁻²⁰ L) and the tight packing of the intravesicular volume by the luminal domains of SV proteins. In particular, the heavily glycosylated luminal domains of SV2 are important contributors to the viscous environment in SVs. Takamori et al. (2) estimated that on average, each SV contains 600 transmembrane domains, and we recently determined that each SV contains exactly five copies of SV2 (17), which represents a relatively small fraction of the proteins that are present on SVs. Yet, SVs lacking SV2 have a significantly lower viscosity of 11.2 cP, which confirms that SV2 is an important determinant of intravesicular viscosity.

We also found that empty vesicles had a higher viscosity on average than vesicles filled with Glu. Although this finding may at first seem unexpected because the intravesicular space should be more packed with molecules while filled with Glu, it is actually quite consistent with our recent observation that SVs loaded with Glu double their volume in comparison with empty SVs (5).

Conclusion

In conclusion, we note a significant difference in rotational viscosity between empty WT and DKO SVs, as well as between WT empty and Glu-loaded vesicles. Our result indicates that the heavily glycosylated intravesicular domains of SV2 are significant determinants of the rotational viscosity and local environment within the vesicle lumen. Additionally, our finding that empty vesicles are actually more viscous than Glu-filled vesicles supports our previous observation that SVs expand greatly in volume upon loading with neurotransmitters. Despite the small size (∼40 nm in diameter) and volume (∼2 × 10⁻²⁰ L) defined by single SVs, our results show that it is possible to determine the rotational viscosity within this tiny volume, which is the smallest enclosed volume in which rotational measurement has been measured thus far.