Characterizing chloride-dependent acidification in brain Clathrin-coated vesicles

Abstract

Endocytic organelles maintain their acidic pH using the V-type ATPase proton pump. However, proton accumulation across the membrane generates a voltage and requires the movement of an additional ion, known as a counterion, to dissipate charge buildup. The role of counterion movement in endosomes is not clear, but a sub-population of early endosomes, Clathrin-coated vesicles (CCVs), has previously been shown to use external chloride (Cl⁻) to allow V-ATPase-dependent acidification. We aimed to determine the identity and function of this presumed Cl⁻ transporting protein. Our sample of highly enriched bovine brain CCVs exhibited V-type ATPase facilitated acidification in the presence of external Cl⁻, independent of the monovalent cations present. While unsuccessful at identifying the mechanism of anion transport, we used glutamate-facilitated acidification, density gradients, and mass spectrometry to show that most brain CCVs are synaptic vesicles, complementing results from earlier studies that argued similarity only on the basis on protein content. The source of Cl⁻-dependent acidification in brain CCVs may be vGLUT1, a synaptic vesicle glutamate transporter with known Cl⁻ permeability, although CCVs in other tissues are likely to utilize different proteins to facilitate acidification.

Introduction

Cells utilize the endosomal pathway to perform vital functions including protein sorting, trafficking, and cell signaling. The initial step of the pathway involves endocytic vesicles, largely originating from Clathrin-coated vesicles (CCVs), merging with early endosomes. Early endosomes mature into late endosomes, which ultimately mature into lysosomes, the terminal organelle of the endocytic pathway. These organelles progressively become more acidic and each compartment strictly maintains an internal pH essential for its function. Acidic luminal pH is generated and maintained by the V-type ATPase (V-ATPase), a multi-subunit membrane protein that utilizes the free energy of ATP hydrolysis to pump protons (H⁺) against the electrochemical gradient and into the organelle (Forgac 2007). The action of the V-ATPase is electrogenic: H⁺-pumping generates a voltage across the membrane and an additional ion, known as a counterion, must move to dissipate the charge buildup and allow additional proton pumping (Harikumar & Reeves 1983; Ohkuma et al. 1983). The counterion could be an anion moving into the organelle lumen, a cation exiting the lumen, or a combination of both. Although various organelles appear to utilize the same V-ATPase for luminal acidification, the compartments all maintain different internal pH values. Investigating the counterion requirements of organelles could provide useful not only for understanding the ions and proteins involved, but could also hint at how organelles sense and maintain their specific internal pH, a question that has persisted in the field for over 40 years. One possibility is that the regulation and movement of counterions helps set the internal pH levels, although the identity of the counterion(s) used in different organelles is not well understood. Substantial experimental data generated from endosomal and lysosomal studies suggest that chloride (Cl⁻) is an important counterion for acidification of organelles in the endocytic pathway (Mindell 2012; Novarino et al. 2010; Ohkuma et al. 1982), but there is still debate over the identity of the counterion(s). Counterion movement in endosomes is particularly challenging to study, largely due to their extremely heterogeneous composition and the difficulty of isolating a consistent, distinct population. To study counterions in a well-defined endosomal system, we evaluated Clathrin-coated vesicles (CCVs), a sub-population of early endosomes.

Previously, Xie, et al. observed that isolated brain CCVs required external Cl⁻ to acidify (Xie et al. 1983). Subsequent studies by Van Dyke and colleagues confirmed this Cl⁻-dependent acidification activity in both bovine brain and rat liver CCVs (Vandyke et al. 1985; Vandyke et al. 1984). Expanding on their previous work, Xie et al. solubilized, column-separated, and reconstituted CCV protein fractions into proteoliposomes, identifying a specific fraction that facilitated Cl⁻ movement across the membrane (Xie et al. 1989; Xie & Stone 1986). Co-reconstitution of the Cl⁻ transporting fraction with a V-ATPase containing fraction yielded proteoliposomes that acidify upon addition of ATP in the presence of Cl⁻, similar to the activity observed in CCVs (Xie et al. 1989). However, this protein was not identified and details of its role in acidification are unknown.

A clear candidate group of proteins that might provide the endosomal Cl⁻ permeability are the ClC proteins are a family of Cl⁻ transporters and channels. The ClCs are composed of two distinct subclasses: Cl⁻-conducting ion channels localized to the plasma membrane (ClC-1, -2, -K_a, - K_b) and 2Cl⁻/H⁺ antiporters (ClC-3-7) primarily residing in organelles along the endocytic pathway (Jentsch 2008). ClC-3, -4, and -5 are localized to endosomes and transport vesicles, and each contain sorting motifs that allow them to, at least transiently, associate with Clathrin (Stauber & Jentsch 2010). ClC-5 is localized to early endosomes and has been shown to facilitate Cl⁻ movement in endosomes to assist in acidification (Gunther et al. 2003; Novarino et al. 2010). However, it has a restricted tissue distribution and is found primarily in epithelia, most notably in kidney and intestine (Steinmeyer et al. 1995). ClC-4 is broadly expressed and localized to endosomes, but details of ClC-4’s physiology and mechanism remain unclear (Jentsch 2015; Mohammad-Panah et al. 2003). ClC-3 is reported to be present in synaptic vesicles of neurons, synaptic-like microvesicles of endocrine cells, and the endosomal compartments of most tissues, where it has been suggested to play a role for acidification (Hara-Chikuma et al. 2005; Jentsch 2015; Maritzen et al. 2008). However, no reports have directly established that a ClC protein is involved in facilitating CCV acidification. An additional CCV counterion pathway candidate is the cystic fibrosis transmembrane conductance regulator (CFTR), a cAMP-regulated Cl⁻ channel expressed primarily in epithelial cells. It has also been suggested to be involved in facilitating endosomal pH, but recent studies indicate that it is unlikely to play a role in endosomal acidification (Haggie & Verkman 2009).

Identifying a CCV Cl⁻ transporting protein is made more complicated when considering that the composition and acidification of CCVs may be tissue dependent (Vandyke et al. 1985; Vandyke et al. 1984; Xie et al. 1983). While both liver and brain CCVs display Cl⁻ facilitated acidification, a majority of brain vesicles are thought to be primarily composed of synaptic vesicles (SVs) on the basis of membrane protein content similarities (Maycox et al. 1992). SVs, while also formed via Clathrin-mediated endocytosis, do not traffic cargo like CCVs but instead store neurotransmitters for use in neuronal signaling. SVs also contain a Cl⁻ permeability, suggested to be mediated by either ClC-3 or the SV-specific glutamate transporter, vGLUT1 (Preobraschenski et al. 2014; Schenck et al. 2009; Stobrawa et al. 2001). However, recently vGLUT1 has been definitively shown to conduct Cl⁻ and, if a large overlap exists between brain CCVs and SVs, could be contributing to the Cl⁻-dependent acidification observed in CCVs (Preobraschenski et al. 2014; Schenck et al. 2009). Still, because liver CCVs also exhibit Cl⁻ dependent acidification, a tissue with no notable vGLUT1 expression (Morimoto et al. 2003), the identity of the Cl⁻ transporter responsible for CCV acidification in brain and liver is still unresolved.

We sought to reproduce the Cl⁻-dependent acidification results published by Xie et al. (1983) and extend these studies by identifying the responsible protein, hoping to take advantage of dramatic improvements in methods for protein identification, like mass spectroscopy, in the intervening decades. This approach is advantageous in that it makes no assumption as to the identity of the CCV Cl⁻ transporting protein, although several members of the ClC family are likely candidates. We successfully isolated brain CCVs and observed robust, Cl⁻-dependent acidification, but could not further characterize the protein due to technical challenges. However, our brain CCVs clearly contained SVs, as has been previously suggested (Maycox et al. 1992), and as indicated by detection of two synaptic proteins, raising the possibility that the observed Cl⁻ conductance is identical to that in SVs. We used both immunoprecipitation and density gradient based approaches to attempt isolation of a population of acidifying CCVs that contained no SV proteins, but could not obtain a subpopulation that satisfied those conditions. Thus, employing a functional approach that complements previous work based on protein composition, we too conclude that a majority of CCVs are SVs, confirming conclusions made by Maycox and colleagues (Maycox et al. 1992). Mass spectroscopy of CCVs reveal substantial levels of vGLUT1, strongly suggesting that this protein contributes to the observed CCV Cl⁻-dependent acidification in brain. However, the Cl⁻ permeable protein in liver remains unidentified.

Materials and methods

Materials

Fresh bovine brains were obtained from J.W. Treuth & Sons (Baltimore, MD, USA). All chemicals were purchased from Sigma, unless designated otherwise. Leupeptin hemisulfate and pepstatin A were from MP Biomedicals. Tubes and rotors were from Beckman-Coulter. Dounce homogenizers were from Wheaton. Detergents were purchased from Avanti or Anatrace. Lipids were purchased from Avanti. NuPage Novex 4–12% Bis-Tris protein gels (NP0322BOX), PageRuler Plus prestained protein ladder (10–250 kDa) (26619), Pierce BCA protein assay kit (23225), and Protein G Dynabeads (10003D) were purchased from Thermo Scientific.

Antibodies

Antibodies, with the dilution used for Western blots in parenthesis: Santa Cruz: Clathrin heavy chain, sc-12734 (1:2000); Clathrin light chain, sc-12735 (1:1000); Lamp1, sc-17768 (1:3333); Synaptophysin 1, sc-55507 (1:3333). Thermo: HRP goat α-mouse, #32430 (1:10000); HRP goat α-rabbit, #32460 (1:100000). BD transduction labs: Hsp60, #611562 (1:10000). Aves labs: Synaptotagmin1, STG (1:3000); HRP goat α-chicken, H-1004 (1:10000). For Westerns, PDVF blots were blocked with 5% milk and primary/secondary blocking steps were performed with 2.5–5% milk (Carnation).

Clathrin-coated vesicle isolation

Clathrin-coated vesicles (CCVs) were isolated based on the methods of Nandi, et al. (Nandi et al. 1982). Fresh bovine brains were obtained from a local slaughterhouse and placed on ice upon arrival (all solutions and brains were kept at 4°C or on ice, unless otherwise specified). 4–5 bovine brains were defatted and the meninges were removed. Brains were placed in a sieve and washed with 3.5 L of PBS (pH 7.0). Total brain mass was typically 800–1000 g. Brains were roughly minced and the sample homogenized with an equal volume of Buffer A (ex: 900 g of brain requires 900 mL of buffer) in a Waring blender with three 10-sec bursts at maximum speed. Buffer A was composed of 100 mM MES, 1 mM EGTA, 0.5 mM MgCl₂, pH 6.75, and contained the following protease inhibitors added right before using: AEBSF (0.02 mg/ml), leupeptin (0.0005 mg/ml), and pepstatin (0.001 mg/ml). The homogenate was centrifuged for 57 min at 22,000 × g. Avoiding the pellet, the supernatant was collected, filtered through gauze/cheesecloth to remove particulates, and then spun at 100,000 × g for 1 hr to obtain a crude vesicle pellet. Pellets were resuspended in Buffer A with a Dounce homogenizer (size ‘A’ dounce). The suspension was centrifuged at 10,000 × g for 10 min to eliminate aggregated material. The high-speed spin (now using 140,000 × g), resuspension, and low speed spin sequence was repeated 2 additional times. The resulting supernatant was then layered over a single step gradient of an 8% sucrose-deuterium buffer (containing buffer A salts) and centrifuged at 80,000 × g for 2 hrs at 17°C on a swinging bucket rotor (Beckman SW28). The pellet was washed and resuspended in buffer A. The solution was spun at 10,000 × g for 10 min to clear aggregate material and the supernatant was collected. This CCV sample was either stored at 4°C for several days or supplemented with 200 mM of sucrose, aliquoted, flash-frozen, and stored at −80°C for a minimum of 3 months with no loss of function. To scale up the protocol, 12–15 brains were processed, sample was stored overnight at 4°C after resuspending the pellet from the first 100,000 × g spin, and the protocol was finished the following day.

Thin section electron microscopy

250 μg of CCVs were centrifuged for 30 minutes at 60,000 × g. Pellets were fixed with 4% glutaraldehyde in 0.1 N sodium cacodylate buffer at pH 7.4 at room temperature for 1 hr and then stored in buffer at 4°C. Fixed samples were washed in buffer, treated with 1% osmium tetroxide in cacodylate buffer for 1 hr on ice, washed and stained with 1% uranyl acetate in 0.1 N acetate buffer at pH 5.0 overnight at 4°C, dehydrated with a series of graded ethanol, and finally embedded in epoxy resins. Thin sections were cut at 70–90 nm, stained with uranyl acetate and lead citrate and examined on a JEOL 200 CX transmission electron microscope.

Monitoring CCV acidification

CCV total protein concentration was determined using a BCA assay (Thermo). CCV acidification experiments were performed in a FluoroMax-3 spectrophotometer. 75–90 μg of CCVs were added to a buffer containing 10 mM Hepes (pH 7.0 with NaOH), 10 mM MgSO₄, 100 mM cation/anion salt (ex: KCl), 200 mM sucrose, and 8 μM acridine orange. Similar to methods in Xie et al. (1983), acridine orange fluorescence was excited at 492 nm and emission measured at 540 nm. Reactions were initiated with 1 mM Mg-ATP (pH 7.0) and 1 μM FCCP (final concentrations). In those experiments were bafilomycin was present, CCVs were incubated in 2 μM bafilomycin for several minutes on ice prior to addition of ATP. Results were normalized to the first value measured after ATP addition and, because of slight variation in the activity of different CCV preps, only normalized to equivalent experiments performed on the same day.

CCV acidification was measured under synaptic vesicle acidification conditions, taken from Maycox et al (Maycox et al. 1988). CCVs were added to either a sucrose buffer (10 μM acridine orange, 320 mM sucrose, 4 mM KCl, 4 mM MgSO₄, and 10 mM Hepes-KOH, pH 7.4) or a KCl buffer (10 μM acridine orange, 150 mM KCl, 4 mM MgSO₄, and 10 mM Hepes-KOH, pH 7.4). 1.5 mM Mg-ATP, 150 mM KCl, 1 mM Kglutamate, and 1 μM FCCP (all final concentrations) were added when indicated.

CCV partial reconstitution

Solubilization and reconstitution methods were modified from Xie et al. (Xie et al. 1989). 1.0 mL of concentrated CCVs (10–15 mg/ml) were combined with 0.34 mL of 3.0 M Tris-Cl (pH 8.5) and incubated on ice for 30 min. The sample was diluted with 1.68 mL of H2O and centrifuged at 65,000 rpm for 1 hr (Beckman TLA 100.3 rotor). Pellets were dounce homogenized in 1.3 mLs of ice-cold 4.0% sodium cholate with 5.0 mM DTT and incubated on ice for 15 min. 1.7 mLs of ice-cold H₂O was added to the sample, incubated for 15 min, and the sample was centrifuged for 1 hr. Pellets were washed with 3.2 mLs of 5.0 mM DTT and centrifuged for 1 hr. Resulting pellets were dounce homogenized in 1.5 mls of buffer containing detergent (concentrations listed at end of paragraph), 10 mM DTT, 30mM Tris-Cl (pH 7.5), and 0.5 mM EDTA and incubated on ice for 45 min. 0.3 mLs of glycerol was added, the sample was mixed well, and then was centrifuged for 50 min. Levels of solubilized proteins in the supernatant were evaluated by comparison on SDS-PAGE.

Detergent concentrations screened, initially selected based on the CMC, were as followed: 1.0%: C9G, C10G, LDAO, cymal-6, cymal-7, cyclofos-7, and fos-choline-12. 1.8%: Deoxycholate. 3.0%: C8G, C8M, 0.5%: C₁₂E₉. After initial rounds of screening, the following detergents/concentrations were selected for continued optimization: 2.0% and 6.0% C9G, 2.0% cyclofos-6, 1.0% and 2.5% cyclofos-7, 1.5% and 4.0% fos-choline-12, 1.0% and 4.0% LDAO, and 1.8% and 4.0% deoxcholate. The most successful detergents were: 2.0% cyclofos-6, 1.0% cyclofos-7, 1.5% fos-choline-12, and 4.0% deoxycholate.

Reconstitution of the transporter was performed by combining 104 μl of buffer B (5.6% sodium cholate, 28% glycerol, 430 mM potassium gluconate), 20 μl of Buffer C (0.6 M DTT, 66 mM magnesium gluconate), and 120 μl of lipids (phosphatidylcholine, phosphatidylethanolamine, phosphatidylserine, & cholesterol at a mass ratio of 4.34: 2.76: 0.2: 2.7 and used at concentrations ranging from 12.5–25 μg/μl; lipids were suspended in 1% sodium cholate, 1 mM DTT, 10 mM Tris-Mes (pH 7.0)). Samples were incubated at room temperature for 5 minutes, frozen in dry ice-ethanol, thawed, and incubated at room temperature for 45 min. Samples were diluted with 40 mLs of room temperature dilution buffer (150 mM KCl, 50 mM Tris-Mes (pH 7.0), 3 mM magnesium gluconate), and centrifuged at 200,000 × g for 1 hr at 17°C.

Proteoliposome acidification was evaluated using the methods described in ‘monitoring CCV acidification’. Cl⁻ movement across the proteoliposome membrane was assessed by monitoring fluorescence changes in the Cl⁻ sensitive dye 6-Methoxy-N-(3-Sulfopropyl)Quinolinium (SPQ). Proteolipsomes were diluted into inside buffer (300 μM SPQ, 225 mM KCl, 20 mM Tris-Mes (pH 7.0), and 3 mM Mg gluconate), freeze-thawed 3x, and sonicated to change the internal buffer. Proteoliposomes were pelleted and resuspended twice to remove external dye and resuspended in inside buffer without SPQ. At the start of the experiment, proteoliposomes were added to outside buffer (25 mM Kgluconate, 2 mM Tris-Mes, 3 mM MgGluconate, and sucrose to match osmolarity of inside buffer) and the experiment initiated with 2 μM valinomycin.

Immunoprecipitation with dynabeads

50 μl of protein-G Dynabeads (Thermo) were resuspended in an antibody solution consisting of 9 μg of Clathrin X-22 (Thermo, MA1-065) in 200 μl of PBS + 0.02% Tween20 (PBS+T) and incubated for 15 minutes with rotation at room temperature. Antibody solution was removed from pelleted beads, which were then rinsed once with PBS+T. Beads were resuspended in 250 μl of 6 μg/μl CCVs in buffer A and incubated for 1 hr at room temperature (rotating). Beads were pelleted, supernatant removed, and resuspended in PBS. The wash step was repeated 2 additional times. Beads were moved to a fresh tube and sample was eluted by adding 26 μl of 50 mM glycine (pH 2.8) and rotating for 2 min. Once eluted, sample pH was neutralized with addition of 3 μl of 2.0 M Tris (pH 7.0).

Density centrifugation gradients

Nycodenz step gradient: gradient solutions, solubilized in either D₂O or H₂O, contained 1x buffer A, 8% sucrose, and one of the following concentrations of Nycodenz (w/v): 23%, 26%, 30%, or 34%. Each layer contained 2.2 mLs per step, bottom-loaded by adding steps of increasing density. 1 part CCVs in 1x buffer A and 250 mM sucrose were mixed with 2 parts of a 45% Nycodenz solution and layered underneath the gradient. The gradient was spun on a swinging bucket rotor (Beckman SW41) for 2 hrs at 95,000 × g at 4°C.

0.5 M and 0.7 M sucrose step gradient: CCVs were diluted to a total of 3 mLs with a solution of 320 mM sucrose and 4 mM Hepes and layered onto a 0.5 M and 0.7 M sucrose step gradient (both layers contained 10 mM Hepes (pH 7.0); 3 mLs per step). Gradient was spun for 1 hr on a Ti70.1 rotor (Beckman) for 38,000 rpm at 4°C.

Iodoxinal step gradient: Gradient solutions contained 1x buffer A, 0.24 M sucrose, and one of the following concentrations of Optiprep (50% w/v stock solution): 20%, 24%, 28%, or 32%. The gradient contained 2.5 mLs/step, layered by adding decreasingly density solutions with CCVs applied to the top. The gradient was spun for 100 min at 25,300 × g at 4°C (Beckman SW41).

To separate a spun gradient, 1 mL samples were fractionated either from the bottom with a peristaltic pump or manually with a pipette from the top of the gradient (pipetting from the top only performed when a loose, visible pellet was present at the bottom of the gradient).

Mass spectrometry of CCV samples

CCVs were treated with 0.5 M Tris (pH 7.0) for one hour on ice, pelleted at high speed, and run on a 4–12% Bis-Tris SDS-PAGE for either 2 min or 1 hour. The 1 hour sample was divided into 8 fractions. In-gel samples were digested with trypsin at 37°C for 18 hr. Tryptic peptides were extracted then desalted before injected into a nano-LC/MS/MS system where an Ultimate 3000 HPLC (Thermo-Dionex) was coupled to an Orbitrap Elite mass spectrometer (Thermo Scientific) via an Easy-Spray ion source (Thermo Scientific). Peptides were separated on a ES800 Easy-Spray column (75 μm inner diameter, 15 cm length, 3 μm C18 beads; Thermo Scientific) with a 25 min linear gradient of 2–27% mobile phase B (mobile phase A: 2% acetonitrile, 0.1% formic acid; mobile phase B: 98% acetonitrile, 0.1% formic acid). The HPLC flow rate was 300 nl/min. Thermo Scientific Orbitrap Elite mass spectrometer was operated in positive data-dependent LC-MS/MS mode. The resolution of the survey scan was set at 60k at m/z 400. The m/z range for MS scans was 300–2000. For MS/MS data acquisition, the minimum signal intensity required was 3e4, the top two most abundant ions were selected for product ion analysis, the isolation width was 1.9 m/z, and the dynamic exclusion window was 7 sec. Xcalibur RAW files were converted to peak list files in mgf format using Mascot Distiller (version 2.4.3.3). Database search was performed using Mascot Daemon (2.4.0) against NCBInr_Bovine database.

Results

Clathrin-coated vesicle isolation and characterization

We isolated bovine brain CCVs using a previously described method (Nandi et al. 1982). A Western blot containing samples extracted at the beginning and end of the preparation indicates that the Clathrin heavy chain was highly enriched in the final sample when compared to the initial brain supernatant (Fig. 1A), suggesting successful CCV isolation. Comparing immunoblotted CCVs to identical SDS-PAGE samples shows that the Clathrin heavy chain clearly corresponds to a band near 180 kDa, the known molecular weight for the heavy chain (Fig. 1A). Thin section electron microscopy (EM) images show frequent vesicles surrounded by a rod-like lattice with a diameter between 50–80 nm, characteristic of vesicles coated with Clathrin (Fig. 1B–C) (Cheng et al. 2007). In addition to coated vesicles, the sample also contained some large, non-coated vesicles (Fig. 1B), whose diameters ranged between 60–250 nm. While the extent of this contamination is difficult to quantitatively assess with EM, some contamination in a CCV prep is common (Pearse 1982; Vandyke et al. 1985). Likely contaminants of an organelle prep are other organelles due to their similar sizes and densities. Using Western blots to evaluate sample purity, we found that CCVs were highly enriched in Clathrin and showed no detectable signal for markers of lysosomes and mitochondria (Fig. 1A). Based on these data, we successfully isolated a highly enriched sample of CCVs with minimal contamination.

Fig 1. Characterizing the CCV preparation.

(A) Identical PDVF Western blots contain whole brain lysate (homogenate), clarified brain lysate (supernatant from whole brain lysate after 22,000 × g spin) (clarified lysate), and the final CCV sample (CCVs). Westerns were probed with antibodies against Clathrin, Hsp60, and Lamp1, proteins typically found in CCVs, mitochondria, and lysosomes (respectively). Western blots were overexposed to ensure purity. Included is the commassie-stained protein gel with a molecular weight ladder (M) included. (B) The electron microscopy image of the isolated CCVs displays the typical distribution of the sample, showing an enrichment of CCVs and some smooth vesicular contamination. (C) Magnification of the CCV sample shows CCVs with their typical lattice structure. An individual vesicle is shown in the blue box.

CCVs acidify in a Cl⁻-dependent manner

We characterized the Cl⁻-dependent acidification properties of the CCV sample using the membrane permeable, pH sensitive dye acridine orange. The dye is thought to become protonated and subsequently membrane-impermeable when in an acidic compartment, self-quenching as it accumulates, which then qualitatively indicates vesicular acidification (Palmgren 1991). Upon addition of ATP, CCVs in a bathing buffer with 100 mM KCl rapidly acidified to a stable level within 2–3 minutes (Fig. 2A), as seen by the decrease in fluorescence. Addition of FCCP, a H⁺ ionophore, returned fluorescence back to its initial intensity, demonstrating that fluorescence loss is both reversible and the result of a change in internal [H⁺]. CCVs incubated with bafilomycin, an inhibitor of the V-ATPase, displayed no acidification, confirming that the V-ATPase is responsible for generating the pH gradient. CCV acidification was stable, robust, and consistent, with very little variation between different CCV brain preparations. The CCV sample acidified to the same extent after being frozen for more than 3 months at −80°C and retained at least 50% acidification after being stored at 4°C for 7 days (data not shown).

Fig 2. Acidification of CCVs in the presence of various ions and inhibitors.

(A) A representative acidification trace of CCVs in a 10 mM Hepes buffer (pH 7.0) containing 100 mM KCl and 8 μM acridine orange. The experiment was initiated with addition 1 mM ATP (labeled) and a decrease in fluorescence emission indicates acidification. 1 μM FCCP, a H⁺ ionophore, was added where indicated. The experiment was performed in the presence or absence of 2 μM bafilomycin, an inhibitor of the V-type ATPase. The vertical axis depicts arbitrary fluorescent units measuring emission at 540 nm (A.U. 540) and the horizontal axis represents time in seconds. (B) In a similar experiment to (A), CCV acidification was tested in buffers where K⁺ is substituted for sodium, choline or lithium, but Cl⁻ constantly remains in the buffer. (C) Similar to (A), CCV acidification tested in buffers were K⁺ constantly remains in the external buffer but Cl⁻ is substituted for Br⁻, SO₄⁻, or gluconate. A Hepes buffer only control is included.

While CCVs in a buffer with KCl acidify, vesicles in a salt-free buffer display no measurable acidification (Fig. 2C). To determine the ionic requirements of CCV acidification, we tested the relative contributions of K⁺ and Cl⁻ by measuring activity in the absence of each. Acidification appeared unchanged when K⁺ was substituted with choline, sodium or lithium (Fig. 2B), indicating that acidification does not depend on the cation. In contrast to the robust acidification in KCl, substituting Cl⁻ with gluconate, phosphate, or sulfate yielded no detectable acidification, results very similar to the salt-free and +bafilomycin controls (Fig 2C). Substituting Cl⁻ with Br⁻ slightly enhanced the rate of acidification, unsurprising considering that most Cl⁻ channels/transporters tend to discriminate poorly between the two anions (Graves et al. 2008; Linsdell et al. 2000; Stauber et al. 2012). Therefore, we have confirmed the previous findings that brain CCV acidification requires external Cl⁻ and appears to be cation independent.

CCVs are not amenable to internal buffer exchange via freeze-thaw sonication methods

Given the robust CCV acidification, we sought to characterize these results in a more quantitative fashion. The acridine orange experiments cannot indicate the magnitude of the pH change, nor can it indicate initial or final pH values. The ability to control CCV internal buffer composition and load a membrane impermeable dye would allow for such measurements, and could designate if Cl⁻ is moved across the membrane via a channel or a coupled transporter. A typical method for modifying internal buffer composition is to freeze-thaw (FT) vesicles several times and then sonicate, which should produce unilamellar vesicles that have internally incorporated the external bathing buffer (Graves et al. 2008). We performed a 3x freeze-thaw and subsequent sonication (FTS) cycle on CCVs, finding they acidified to the same extent as unmodified vesicles when measured by acridine orange (data not shown). However, we detected no fluorescent signal when CCVs were FTS in a solution containing the membrane impermeable dye fluorescein, indicating no dye incorporation. We hypothesized that the Clathrin coat prevented the CCVs from breaking open during the freeze-thaw step and attempted to remove the coating protein by incubating vesicles in 0.5 M Tris (Keen et al. 1979). SDS-PAGE shows that a substantial portion of the Clathrin remained in the supernatant after treatment with Tris (confirmed with electron microscopy, data not shown), implying that many vesicles no longer retained their Clathrin coat (Fig. S1A). These data indicate the amenability of uncoated vesicles to structural modifications and we evaluated the functional impact of these treatments on the same group of vesicles. Increased modifications progressively decreased the vesicles’ ability to acidify (Fig. S1B) and, while this loss of function was substantial, enough activity remained to attempt internal dye incorporation. Unfortunately, no measurable dye incorporation occurred even after a wide range of dyes, buffers, and loading conditions were tested. The absence of dye may be attributed to a lack of tightly resealed vesicles, possibly due to a high concentration of membrane proteins, too little internal dye incorporation, or the inability of the Cl⁻ transporter to withstand the FTS conditions. The remaining acidification we observed after Tris-FTS may be from vesicles that retained their Clathrin coat during the harsh FTS step. Ultimately, we were unable to modify the internal buffer solution of the CCVs while retaining acidification activity.

Partial reconstitution of CCVs using various detergents

An alternative approach to studying the properties of the Cl⁻ transporting protein in vesicles is to solubilize, isolate, and reconstitute the protein into proteoliposomes. Previously Xie, et al. solubilized uncoated vesicles in the detergent C₁₂E₉ and successfully reconstituted a partially purified Cl⁻ transporting fraction into proteoliposomes, but did not identify the protein (Xie et al. 1989). As a first step to investigate and characterize the protein, we omitted the complicated purification step and directly reconstituted solubilized proteins from uncoated vesicles into liposomes, a method previously used to reconstitute the V-ATPase from CCVs (Arai et al. 1989; Mulberg et al. 1991). We evaluated successful incorporation by assaying ATP-initiated acidification, which would occur if both the V-ATPase and the Cl⁻ transporting protein were properly reconstituted into proteoliposomes. With a second assay, we sought to monitor the functionality of the Cl⁻ transporting protein by loading the proteolipsomes with 225 mM KCl and the Cl⁻ sensitive dye 6-Methoxy-N-(3-Sulfopropyl)Quinolinium (SPQ), whose fluorescence decreases in the presence of Cl⁻. These vesicles were placed in an external buffer without Cl⁻ and we initiated the experiment with Valinomycin, an ionophore that allows K⁺ to freely cross the membrane. If a Cl⁻ transporter is present, K⁺ acts as a counterion to allow movement of the anion out of the lumen, resulting in an increase in the fluorescent signal. Neither the acridine orange nor the SPQ assays showed fluorescent changes, possibly due to a low concentration of reconstituted protein. Indeed, an SDS-PAGE of the proteoliposomes indicates little protein incorporation. To optimize reconstitution, we solubilized uncoated vesicles in various detergents and compared levels of solubilized protein with SDS-PAGE (Fig. S2). Cyclofos-6, cyclofos-7, fos-cholate-12, deoxycholate, LDAO, and C9G all appeared to solubilize more protein than C₁₂E₉, the previously used detergent (Fig. S2, cyclofos-6 not shown but was very similar to cyclofos-7). After optimizing different concentrations of these selected detergents, we determined that 2% cyclofos-6, 1% cyclofos-7, 1.5% fos-cholate-12, and 4% deoxycholate allowed for the most reconstituted protein. However, no SPQ fluorescence changes were detected after assaying for Cl⁻ transporting activity in these optimized proteoliposomes, again preventing us from characterizing/identifying the Cl⁻ transporting protein.

Examining CCV acidification using synaptic vesicle acidification conditions

Both brain and liver CCVs acidify in the presence of Cl⁻ (Vandyke et al. 1985; Vandyke et al. 1984; Xie et al. 1983), suggesting a general mechanism of CCV, and perhaps early endosome, acidification. However, experiments published by Maycox et al. concluded that brain CCVs are primarily composed of SVs based on the extremely similar membrane protein content of these two populations (Maycox et al. 1992). Their conclusions, while compelling, contain no functional comparisons of the two populations, nor did their work exclude the possibility of a small sub-population of non-SV CCVs. Our CCV sample, isolated using a somewhat different protocol than Maycox et al., appears to have a different protein distribution upon comparison of our SDS-PAGE results with theirs. Therefore, we investigated if our brain CCV sample also contained SVs by immunoblotting against two characteristic SV proteins, Synaptophysin-1 and Synaptotagmin-1 (Fig. 3A). The sample contained a substantial signal for each protein, suggesting that a significant portion of the CCVs could be SVs. We selected an additional property of SVs, their ability to acidify via the V-ATPase in the presence of glutamate and low concentrations of Cl⁻, to explore if there is functional overlap between the two vesicle populations. We tested CCV acidification using the same experimental parameters previously employed to test SV acidification (Maycox et al. 1988), where they found that adding glutamate to a low KCl solution increased acidification. Indeed, we found that addition of 1 mM glutamate to CCVs in 4 mM KCl caused a modest increase in the magnitude and rate of acidification, while addition of 150 mM KCl greatly enhanced both rate and magnitude (Fig. 3B). These results, extremely similar to those observed for SVs under identical conditions (Maycox et al. 1988), indicate that our CCVs can use glutamate as a counterion. Further, glutamate was unable to induce acidification when 0 mM KCl was present, but required at least 2 mM KCl to utilize the glutamate for acidification (Fig. 3C), again very similar to the properties of SVs (Maycox et al. 1988). The presence of additional Cl⁻ (4 mM KCl) did not substantially affect the rate of glutamate acidification. The striking similarities between SVs and brain CCVs acidification properties prompted the following questions: is the Cl⁻-dependent acidification a property of CCVs, SVs, or some combination of both? What percentage of the CCV sample was SVs? Could CCVs be separated from SVs and still retain acidification?

Fig 3. Evaluating synaptic vesicle-like properties of CCVs.

(A) Identical PDVF Western blots containing the enriched CCV sample were probed with antibodies against either Synaptotagmin or Synaptophysin. (B) CCV acidification was tested in the presence of glutamate using the acridine orange acidification assay with CCVs in a 4 mM KCl buffer. Marked are additions of 1 mM ATP, 1 mM glutamate (Glut), and 150 mM KCl (KCl) (final concentrations). (C) CCV acidification using the same conditions as (B) except the starting buffer contained 0, 2, or 4 mM KCl (black, blue, and red traces, respectively).

Attempts to separate CCVs and SVs using immunoprecipitation and density gradients

To determine if Cl⁻-dependent acidification is a property of brain CCVs and/or SVs, we attempted to isolate a subpopulation of acidifying vesicles from CCVs that did not contain SVs. We employed immunoprecipitation to isolate this subpopulation of CCVs. Using a Clathrin antibody as the bait, we evaluated if SVs were also extracted by monitoring the presence of Synaptophysin. Western blot analysis indicates we successfully immunoprecipitated a small amount of a Clathrin labeled sample, but Synaptophysin was also present (Fig. 4). We were unable to confirm the functionality of the isolated sample because neither the bead-bound sample nor glycine-eluted samples acidified when tested with the acridine orange assay, although exposure to the harsh, acidic glycine likely explains the lack of activity in the latter. To explore if targeting distinct locations of the Clathrin lattice or SVs themselves would pull-down a unique vesicle population, we performed immunoprecipitation with two additional Clathrin antibodies and the Synaptophysin antibody. However, SDS-PAGE showed no protein was pulled-down from the overall CCV sample. Thus, immunoprecipitation did not separate out a distinct subpopulation of non-SV vesicles from the overall CCV pool.

Fig 4. Immunoprecipitation of CCVs using a Clathrin antibody attached to protein-G dynabeads.

Protein-G Dynabeads, covalently attached to the X-22 Clathrin antibody, were incubated with CCVs and bound substrate was subsequently eluted. Shown are SDS-PAGE and Westerns with identical samples; the first Western was exposed to a Clathrin heavy chain antibody (α-Clathrin) and the second was exposed to the Synaptophysin antibody (α-syp). Lane 1 contains the unmodified CCV sample (starting CCVs), lane 2 is the CCV supernatant after Dynabead incubation (flow through), lane 3 contains a control of eluted antibody+Dynabeads (no CCV incubation) (control elution), and lane 4 contains the CCV sample eluted off the Dynabeads (CCV elution). Eluted antibody was visible in both Westerns and is marked.

If SVs and CCVs are different populations of vesicles, they might have different physical properties like density or mass. We employed density gradient centrifugation, a technique that separates heterogeneous samples based on the physical properties of mass, size, and/or density, to elucidate a difference between SVs and CCVs. Gradient compositions were inspired by previous methods used to isolate CCVs, SVs, or other similar organelles, and we selected sucrose, iodixonal, and/or Nycodenz as the primary gradient materials. After fractionating samples of a spun gradient, we evaluated protein distribution using SDS-PAGE, monitored the presence of Synaptophysin with Westerns, and measured acidification of select samples with the acridine orange assay. Gradients typically yielded one of three outcomes. The first, and most frequent, was a lack of separation of the proteins by the gradient and thus, suggesting a lack of distinction between the CCVs and SVs populations, such as with a Nycodenz step gradient (Fig. S3). The second was an asymmetrical distribution of several CCV proteins, implying a separation of the sample into subsets; but Synaptophysin was detected throughout these samples, as seen with the iodoxinal step gradient (Fig. 5A–B). The third was an asymmetric distribution of several CCV proteins with a gradient of Synaptophysin as well. However, the no/low Synaptophysin samples tended to have very little protein other than Clathrin and did not substantially acidify (Sucrose step gradient, Fig. 5C–D). Based on the reduction in the number of proteins, the non-acidifying CCVs are likely to be empty baskets, spontaneously formed Clathrin coats that lack an internal membrane, which are commonly found after CCV isolation (Keen et al. 1979). Therefore, we were unable to separate acidifying CCVs from SV markers, likely because that population is very small or too physically like SVs to extract. Our results agree with previous work from Maycox et al. arguing that the vast majority of CCVs are primarily composed of SVs (Maycox et al. 1992).

Fig 5. Separation of CCVs using density gradients.

(A) CCVs layered on top of a 20%, 24%, 28%, or 32% Iodixanol step gradient were spun and fractionated. The SDS-PAGE gel, with labeled molecular weight markers, contains fractions of CCVs, ordered by decreasing concentration of Iodixanol. The starting CCVs are the last sample. The concentration of the first sample fraction was so dilute it was not properly normalized. Below the SDS-PAGE is a Western blot of the same samples probed with an antibody against Synaptophysin. (B) Acidification of representative fractions from the gradient were tested using the same conditions described in Fig 3. (C) CCVs layered on top of a 0.5M and 0.7M sucrose step gradient were spun and fractionated. The SDS-PAGE gel, with labeled molecular weight markers, contains fractions of CCVs, ordered by increasingly high concentrations of sucrose. The final 2 samples are the resuspended gradient pellet and the starting CCVs. Below the SDS-PAGE is a Western blot of the same samples probed with an antibody against Synaptophysin. (D) Acidification of representative fractions from the gradient were tested under the same conditions described in Fig. 3.

Mass spectrometry reveals CCV- and SV-specific proteins

We attempted to identify a possible Cl⁻-transporting protein using mass spectrometry (MS) analysis. Because Clathrin and its related adaptor proteins are such substantial portions of the protein sample, we uncoated the vesicles using Tris to increase the percentage of membrane proteins in the sample. We split the initial vesicle sample in half and evaluated it two different ways. First, we performed MS on the entire first half of the sample to evaluate the overall protein makeup of the vesicles. We ran the other half on SDS-PAGE to separate the sample based on mass, split it into 8 different samples, and performed MS on these individually to increase the likelihood of detecting low abundance proteins (Graph S1). In both the fractionated and total protein samples, we found many of the expected structural components of CCVs, including the Clathrin heavy and light chains, adaptor proteins, coating and uncoating proteins, components of the V-type ATPase, tubulin, and proteins related to microtubules. Additionally, we identified many neuronal specific proteins such as neurofilaments, synaptosomal associated proteins, and SV-specific SNAREs. In agreement with the CCV immunobotting against SV proteins (Fig. 3A), we also detected Synaptophysin-1 and Synaptotagmin-1. We examined the samples for Cl⁻ transporting proteins, finding vGLUT-1 in both the individual and overall samples. No other chloride transporting proteins were confidently identified. Interestingly, we did not detect ClC-3, a protein known to be present in SVs but rarely identified using MS, implying that even though MS methods for detecting membrane proteins are improving, they remain challenging to identify (Barrera & Robinson, 2011).

Discussion

Endosomal organelles require a consistent, highly regulated internal pH to function. They are acidified by the H⁺ pump V-ATPase, but H⁺ movement across the membrane is an electrogenic action that requires an ionic shunt or counterion to dissipate voltage accumulation. In this study, we examined the ions required to facilitate acidification in brain CCVs and explored their similarity to SVs. While Cl⁻ appears to function as the counterion in CCV acidification, the identity of the Cl⁻ transporting protein remains unclear (Vandyke et al. 1985; Vandyke et al. 1984; Xie et al. 1983). Our data are consistent with a previously proposed model showing that brain CCVs are primarily composed of SVs and thus the Cl⁻ transporting protein is most likely vGLUT1, a SV-specific glutamate transporter that also has been shown to transport Cl⁻ (Maycox et al. 1992; Preobraschenski et al. 2014). However, similar Cl⁻-dependent acidification has been observed in liver CCVs, and the identity/function of that Cl⁻ transporting protein remains unknown.

Cl⁻ has long been implicated as a counterion in endosomal acidification, so we sought to investigate its role by characterizing isolated CCVs in a well-defined ionic environment. Our sample of highly enriched CCVs relies on the V-ATPase and requires external Cl⁻ for acidification (Fig 1A, 2). The Cl⁻-dependence of acidification data agree nicely with the results from previous brain CCV experiments, though our experiments show even more pronounced Cl⁻-dependence than previous work (Vandyke et al. 1985; Xie et al. 1983). Similarly, our explorations revealed almost no effect of external cations on acidification, again similar overall to those previously observed in brain CCVs. Thus, our work even more clearly highlights that brain CCV acidification absolutely requires external Cl⁻ but is independent of the external cation.

Our efforts to further characterize the properties of the Cl⁻ transporting protein, either by replacing the internal buffer via freeze/thaw or by partial reconstitution detected no observable activity (Fig. S1, S2). The lack of transporter function was surprising, given the robust CCV acidification and reconstitution success reported by others (Arai et al. 1989; Mulberg et al. 1991; Xie et al. 1989). Possibly a step in our solubilization/reconstitution conditions prevented the Cl⁻ transporting protein from functioning or the final concentration of the transporter was too small to yield detectable Cl⁻ movement.

While CCVs in both brain and liver exhibit Cl⁻ dependent acidification, Maycox et al. (1992) showed that brain CCVs are primarily composed of recycling SVs based on membrane protein content similarities. However, they provided no functional data to support this overlap. Our CCV sample was isolated using a slightly different protocol than Maycox et al. and appeared to have a somewhat different SDS-PAGE protein pattern, leading us to question if our acidification was from CCVs, SVs or both (Maycox et al. 1992). Our CCV sample positively tested for SV proteins (Fig. 3A, Graph S1) and had similar acidification properties in the presence of glutamate (Fig. 3B–C) (Maycox et al. 1988). We tried to extract a population of non-SV containing CCVs using immunoprecipitation and density-gradient based approaches (Fig. 4, 5), again finding that SVs are prevalent throughout the acidifying CCV population. Thus, by taking a functional approach and using a separate set of criteria and experiments than Maycox and colleagues, we too have shown that CCVs and SVs have a remarkably similar composition based on acidification, the presence of Synaptophysin in most CCV samples, and mass spectroscopy results showing the presence of many SV-specific proteins.

The prevalence of SVs in our CCV sample argues that the protein facilitating the observed Cl⁻ dependent acidification is likely to be vGLUT1. This glutamate transporter, located predominately in SVs, has two anion binding sites (one for Cl⁻ and the second for glutamate) and an additional flexible cation binding site, which allows the transporter to rapidly acidify and balance the internal SV electrochemical gradient (Preobraschenski et al. 2014). vGLUT1 requires a small concentration of Cl⁻ to facilitate glutamate uptake, similar to the acidification properties that we observed in CCVs (Fig. 3B–C) (Maycox et al. 1988; Preobraschenski et al. 2014). However, vGLUT1 is a neuronally expressed protein and CCVs are ubiquitously present through the body. If both liver and brain CCVs need Cl⁻ to acidify, do they use the same protein or are the Cl⁻ proteins/counterion requirements organ specific? Alternatively, perhaps we were unable to isolate the subset of non-SV brain CCVs that use a different Cl⁻ transporting protein, potentially because of its extreme similarities to coated SVs or because they are a very small portion of the overall vesicle pool. Liver CCVs have been reported to have a slightly different counterion requirement profile; they do not require Cl⁻ for acidification, but its presence greatly enhances the extent of acidification (Vandyke et al. 1984). If the anion/cation utilization is tissue-dependent, then is Cl⁻ facilitated acidification necessary or just useful based on specific situations/cell type? Historically, CCVs have been isolated from liver and brain due to the ease of obtaining these larger organs. Exploration into the CCVs from other organs could provide insight into the role of Cl⁻ and counterions in these early endosomes.