CLC3 IS A CRITICAL REGULATOR OF THE CELL CYCLE IN NORMAL AND MALIGNANT GLIAL CELLS

Abstract

While most brain cells are post-mitotic, small populations of progenitor or stem cells can divide throughout life. These cells are believed to be the most likely source for primary brain malignancies including gliomas. Such tumors share many common features with non-malignant glial cells but owing to their insidious growth form cancers that are typically incurable. In studying the growth regulation of these tumors, we recently discovered that glioma cell division is preceded by a cytoplasmic condensation that we called pre-mitotic condensation (PMC). PMC represents an obligatory step in cell replication and is linked to chromatin condensation. If perturbed, the time required to complete a division is significantly prolonged. We now show that PMC is a feature shared more commonly among normal and malignant cells and that the reduction of cell volume is accomplished by Cl⁻ efflux through ClC3 Cl⁻ channels. Patch-clamp electrophysiology demonstrated a significant upregulation of chloride currents at M phase of the cell cycle. Colocalization studies and co-immunoprecipitation experiments showed the channel on the plasma membrane and at the mitotic spindle. In order to demonstrate a mechanistic role for ClC3 in PMC we knocked down ClC3 expression using shRNA constructs. This resulted in a significant reduction of chloride currents at M-phase that was associated with a decrease in the rate of PMC and a similar impairment of DNA condensation. This data suggests that PMC is an integral part of cell division and is dependent on ClC3 channel function.

INTRODUCTION

In contrast to most other mammalian cell types, adult neural cells have traditionally been thought of as non-regenerative cells. Over the past 20 years, however, our understanding of the reality, importance and consequences of post-developmental neural cell proliferation has undergone considerable revision. Adult neurogenesis is now widely accepted to have implications for most areas of neuroscience including synaptic integration, neurodegeneration, injury and epilepsy. In turn, neurogenesis is dependent upon a resident population of cells in the brain which can be induced to proliferate in response to various stimuli. This has both positive and negative implications, and the most apparent pathology in which neural cell proliferation is a central mechanism of the disease process is brain tumors. Originally believed to be derived from astrocytes, there is now convincing genetic and functional evidence that gliomas, the predominant primary brain tumors, originate from neural stem cells (NSCs) (Oliver and Wechsler-Reya, 2004; Fan et al., 2007; Sanai et al., 2005), suggesting that these cell types may share common mechanisms of cell cycle regulation.

In dividing cells, the cell cycle is normally tightly regulated by a series of events which catalyze the transitions between G1, S, G2, and M phase. Ultimately, a successful cell division is dependent on the cell’s ability to distribute its genetic content between two daughter cells. This occurs along with the partitioning of the cytoplasmic content and the cell membrane. Both of these events require considerable structural changes, and adaptations in a cell’s biophysical characteristics may enhance its ability to effectively complete this process. Surprisingly little is known, however, about the biophysical aspects of neural cell proliferation. Although a number of studies have implicated ion channels in cell cycle progression in other cells, mechanistic causal relationships have been harder to establish. We recently reported an unexpected channel mediated decrease in cytoplasmic volume, termed pre-mitotic condensation (PMC), which occurs as glioma cells retract processes, round up and progress through mitosis. We found that this process is linked to chromatin condensation and is energetically driven by the release of chloride through ion channels (Habela and Sontheimer, 2007). As ion movement through channels is dictated by their electrochemical gradient, this outward movement suggests the accumulation of intracellular chloride ([Cl⁻]_in) well above the concentration observed in post mitotic neurons and glial cells. Interestingly, this is analogous to the elevated [Cl⁻]_in that is observed in NSCs during development and in the adult (Kuner and Augustine, 2000; Kakazu et al., 1999).

In this study, we show that PMC is a normal part of cell division in both normal and malignant glial cells and identify ClC3 as the Cl⁻ efflux pathway in this process. ClC3 localizes to the plasma membrane and mitotic spindle and co-immunoprecipitates with tubulin. When ClC3 expression was knocked down using shRNA constructs, a significant reduction of chloride currents at M-phase associated with a decrease in the rate of PMC and a similar impairment of DNA condensation resulted. This demonstrates a novel, unexpected and significant role for ClC3 at division in proliferating astrocytes and glioma cells.

MATERIALS AND METHODS

Glioma Cell Culture and Pharmacology

D54-MG and U251–MG glioma cells (WHO IV, glioblastoma multiforme) were a gift from Dr. D. Bigner (Duke University, Durham, NC). The stable GFP cell line used for Figure 1A was derived from D54-MG cells (D54-MG-GFP) which were transfected with peGFP-N1 (Clontech, Mountain View, CA). The plasmid insertion was maintained with 0.25 mg/mL G418 disulfate salt (GIBCO). In all other experiments the stable lines were generated by transfection with a pGIPZ vector (see below).

Figure 1.

Pre-mitotic condensation occurs in glioma cells and spinal cord astrocytes. A, example glioma cell approaching division from top to bottom demonstrated in 3D (left) and 2D (right two panels). Error bar = 10 μm. B, Relationship between volume (open circles) and intensity measurements (closed squares) for cell in A. C – E, 2D time-lapse curves with cytoplasmic condensation demonstrated by closed squares and DNA condensation (Hoechst 33342) demonstrated by open circles. C, 2D time-lapse curves for D54 – MG – GFP glioma cells. Trace is the mean of n = 45 individual cell divisions. Independent experiments repeated 5 times. D, Example 2D-timelapse experiment for dye loaded U251-MG cells. n = 9. Similar results observed in 4 independent experiments. E, Example 2D time-lapse experiment for dye loaded cultured rat spinal cord astrocytes n = 9 dividing cells. Similar results observed in 3 independent experiments.

All glioma cells were maintained at 37°C, 10% CO₂ in 1:1 DMEM-F12 (Mediatech, Hendon, VA) with 7% FBS and 2 mM Gln and were used for experiments within the 20 passages. All stock concentrations of drugs used were dissolved in dimethyl sulfoxide (DMSO). Unless otherwise noted the DMSO stock was diluted 1:1000 to obtain the final working concentration. Drug treatment media was composed of control media supplemented with DMSO or drugs dissolved in the same. 5-nitro-2-(3-phynylpropylamino) benzoic acid (NPPB) was used at 200 μM for both 2D time-lapse experiments and for electrophysiology. 4,4′-Diisothiocyanatostillbene-2,2′-disulfonic acid disodium salt hydrate (DIDS) was used at 200μM. Unless otherwise stated, all reagents were purchased from Sigma (St. Louis, MO)

Astrocyte Cultures

Astrocytes were cultured from postnatal day 0 (P0) rat pups. Pups were anesthetized on ice and decapitated, the spinal cords were dissected into ice-cold serum-free EMEM (Gibco, Grand Island, NY) containing 20 mM glucose. Meninges were stripped and cords were minced and placed into an O₂ saturated papain solution (Worthington, Lakewood, NJ) for 20 min. The tissue was washed twice with spinal cord astrocyte media (EMEM supplemented with 10% fetal calf serum, 20 mM glucose, and penicillin/streptomycin) and then triturated. A high density of cells (1.0 × 10⁶ cells/mL) was plated in a small volume (50 μL) onto polyornithine- and laminin-coated 0.170 mm thick glass bottom 35mm dishes (Mattek, Ashland, MA). This allowed cells to adhere to the glass. After 4 hours, an additional 2 mL of media was added to each well. Cultures were maintained at 37°C, 5 % CO₂ and the media was changed every day for the first 3 days, then every fourth day thereafter.

3D Time-lapse Volume Determination

3D - Confocal images were acquired with a Hamatsu IEEE1394 Digital CCD camera (325–6, Sunayama-Cho, Hamamatsu City, Japan) mounted on an Olympus IX81 motorized inverted microscope equipped with an Olympus Disk Scanning Unit (DSU)(Olympus, 2 Corporate Center Drive, Melville, NY 11747–3157) and controlled by Slidebook software (Intelligent Imaging Innovations Inc., Denver Colorado). The microscope was housed in a temperature- and CO₂- controlled humidified incubator maintained at 37°C and 5% CO₂ for 3D-timelapse experiments. Image stacks were acquired with 0.5 μm steps throughout the depth of the cells for both GFP (excitation = 482 ± 17 nm, emission = 536 ± 20 nm) and Hoechst (excitation = 387 ± 11 nm, emission = 447 ± 30 nm). Stacks were acquired at 30-minute intervals through a 40x oil objective (numerical aperture = 1.3) and the camera was binned to minimized light exposure and phototoxicity. Image stacks were converted to a series of 16 bit TIF files and imported into Image Pro 3D Suite (Media Cybernetics INC, Bethesda, MD) where image volume measurements were made based on pixel intensity in the x–y plane throughout the z depth of the stack.

2D Time-lapse imaging

Glioma cells were grown in # 1 German borosilicate 2 or 4 well chamber slides (Nunc). All imaging experiments were performed 24 to 48 hrs after plating for glioma cells and 3–5 days post-dissection for astrocytes. When nuclear condensation was accessed, Hoechst 33342 was loaded into cells at a concentration of 100 ng/mL in normal culture media for 25 minutes. After washing, cells were allowed 1–2 hours to recover prior to imaging. For experiments in which U251-MG cells and astrocytes were loaded with CellTracker^™ Green CMFDA (5-chloromethylfluorescein diacetate) (Molecular Probes, Eugene, Oregon), a 10μM solution of the dye was made up in loading media composed of serum/glutamine free media 1:1 DMEM/F12 warmed to 37°C. Cells were washed 3 times in loading media without CMFDA to remove serum and glutamine and then covered with the 10 μM CMFDA solution. After 45 minute incubation at 37°C the dye solution was removed and cells were washed once with loading media and 3 times with normal culture media. Imaging experiments were conducted 12–24 hours after loading to allow for recovery.

Time-lapse images were acquired with an Axiocam MRm camera mounted on an Axiovert 200M inverted microscope with a motorized stage and filter cube turret (Carl Zeiss, Thornwood, NJ). The microscope was housed in a temperature and CO₂ controlled humidified incubator maintained at 37°C and 5% CO₂ and cells were imaged in their respective normal culture media. Wide field fluorescence was generated from a metal halide X-cite 120 light source (EXFO Photonics Solutions Inc., Mississauga, Ontario, Canada) and reflected through a FITC (excitation = 480 ± 15 nm, emission = 535 ± 20 nm) filter cube for GFP or CMFDA fluorescence or a DAPI (excitation = 350 ± 25 nm, emission = 460 ± 25 nm) filter cube (Chroma Technologies, Burlington, VT) for Hoechst 33342 fluorescence. Neutral density filters were used to attenuate light and prevent phototoxity. 2D time-lapse experiments were conducted with a 20X Fluar air objective. Images were acquired at 5 to 20 minute intervals depending on cell type and duration of experiment. Intensity measurements for GFP and Hoechst 33342 were made from 16-bit TIF files in Image J (NIH Image, National Institutes of Health, USA) within regions of interest (ROI) selected to reflect the mean cytoplasmic or nuclear intensity, respectively. Measurements were reflective of mean pixel intensity within that ROI and therefore, not dependent on ROI size from image to image. Values were normalized by creating a ratio of the intensity at a given time to the intensity at the last imaged time point prior to division (t = 0).

Immunocytochemistry

D54-MG cells were plated on uncoated 12 mm round # 1 glass coverslips (Macalaster Bicknell, New Haven, CT) in 24 well plates at a density of approximately 10,000 cells per well. After 72 hours in culture, cells were fixed for 10 min at room temperature (25°C) with 4% paraformaldehyde in phosphate buffered saline (PBS), washed three times with PBS and then permeablized with 0.3% Triton X-100 in PBS + 5% normal goat serum (NGS). After the detergent was removed and the coverslips were washed, cells were covered with 5% NGS in PBS for 1 hour at room temp or overnight at 4 °C. Primary antibodies were diluted in PBS containing 1% NGS and incubated for 2–3 hours at room temperature followed by three washes with PBS and a 30 minute incubation with 5% NGS PBS to block nonspecific binding of the secondary antibodies. Secondary antibodies were diluted in 1% NGS PBS and incubated 1hour at room temperature followed by three PBS washes. Prior to mounting, cells were loaded with Hoechst 33342 at 100 ng/mL in PBS for 20 minutes. Coverslips were mounted onto 3″ by 1″ by 1mm slides (Fischer Scientific, Pittsburg, PA) with Gel-Mount aqueous mounting media (Sigma).

ClC3 was immunolabeled with either a commercially available rabbit polyclonal anti-ClC-3 antibody to the C- terminus of ClC3 at a final concentration of 1.5 μg/mL (Alomone Labs Ltd., Jerusalem, Israel) or a polyclonal rabbit antibody to the c-terminus referred to as 15K and kindly provided by Dr. William J. Hatton PhD (University of Nevada School of Medicine, Reno, NV). Final working concentration of the 15K antibody was 2.7 μg/mL. A FITC labeled monoclonal mouse antibody to α-tubulin (ab11303) was purchased from Abcam Inc. (Cambridge, MA) and diluted as suggested by the manufacturer. Alexa Fluor 488 and 546 labeled goat anti-mouse and goat anti rabbit secondary antibodies were purchased from Invitrogen (Carlsbad, CA). Semi-confocal fluorescent images were acquired with a 60x oil immersion lens (NA = 1.42) on the Olympus IX81 motorized inverted microscope described above. Alexa 546 was imaged with a TRITC filter set (excitation = 543 ± 22 nm, emission = 593 ± 20 nm), Alexa 488 with a FITC set (excitation = 482 ± 17 nm, emission = 536 ± 20 nm), and Hoechst 33342 was imaged with a Dapi filter set (excitation = 387 ± 5.5 nm, emission = 447 ± 30 nm) (Semrock Rochester, NY). Out of Focus light was removed by the DSU.

shRNA and Control Stable Cell Lines

In order to knockdown ClC3 expression in the D54-MG cell line, we obtained commercially available pGIPZ-lentiviral shRNAmir vectors containing either a nonsilencing scrambled sequence which does not match any known mammalian genes (NS) or one of three hairpin sequences targeting CLCN3 (Open Biosystems, Huntsville, AL). The hairpin sequences were as follows with underlined sequencing corresponding to region of the human CLC3 gene targeted: F3 = 5′-TGCTGTTGACAGTGAGCGCG GTGTT GGTACATATGATGATTAGTGAAGCCA CAGATGTAATCATCATATG TACCAACACCTTGCCTACTGCCTCGGA - 3′; A4 = 5′-TGCTGTTGACAGTGAGCG CGCCTGG AAGAGGTTAGCTATTTAGTG AAGCCACAGATGTAAATAGCTAACCTCTTCCAGGCTTGCCTACTGCCT CGGA3′; H8 = 5′-TGCTGTTGACAGTGAG CGCCCTACCTCTTTCCAA AGTATAT AGTGAAGCCACAGATGTATATA CTTTGGAAAGAGGTAGGATGCC TACTGCCTCGGA - 3′; NS = 5′-TGCTGTTGACAGTGAGCGATCTCGCTTGG GCGA GAG TAAGTAGTGAAGCC ACAGATGTACTTACTCTCGCCCAAGCGA GAGTG CCTA CTGCCTCGGA - 3′. D54-MG cells were transfected with the Amaxa Biosystems nucleofection technique as directed by the manufacturer. Briefly, cells were grown to ~80 % confluency in 10 cm dishes, trypsinized, pelleted, and resuspended in supplemented Nucleofector® Solution for a final cell density of 0.5–1.5 million cells per 100 μL of solution. The cell solution was then mixed with 1 μg of plasmid DNA/100 μL, transferred to an electroporation cuvette and subject to an electroporation protocol optimized for the D54-MG line. Transfected cells were transferred to 60 mm dishes containing pre-warmed media and incubated under normal culture conditions. Media was changed 24 hours post-transfection to remove cells which did not survive electroporation. To generate the stable lines, 10 ug/mL puromycin treatment was begun 96 hours after transfection. Media was changed regularly to remove cellular debris and dead cells throughout the selection period until colony formation was visibly evident (~ 2–3 weeks) and all cells expressed the vector encoded GFP. For the A4 cell line, all colonies were combined and amplified in the presence of puromycin. For the H8 and F3 lines individual colonies were transferred to separate dishes with a pipette tip and amplified in the continued presence of puromycin under otherwise normal culture conditions.

Immunoprecipitation

Whole cell lysates were diluted to equal concentrations (1 mg/mL). These samples were pre-cleared, with 75 μL of protein A beads (Roche Applied Science, Indianapolis, IN) if the precipitating antibody was generated in rabbit or protein G beads (Roche Applied Science) if the precipitating antibody was generated in mouse, for 1 hour. Agarose beads were equilibrated by washing and pelleting the beads on low speed three times in RIPA buffer (50 mM TrisCl, pH 7.5, 150 mM NaCl, 1% Nondet P-40 (NP-40). Antibodies were added to the pre-cleared lysates at 2–3 μg/mL and incubated overnight at 4°C. The next day, 60 μL of the appropriate agarose beads was added to each sample for an additional 2–4 hours at 4°C. Immunoprecipitates were pelleted at low speed and washed four times in RIPA buffer. To release the immunoprecipitates from the beads, 2X sample buffer (50 μL) was added and samples were boiled for 5 minutes. Blots were run as described below. Mouse and rabbit tubulin antibodies were obtained from Abcam and used according to manufacturer’s suggestions.

Western Blot Protocol

Cultured cells were lysed using RIPA buffer (50 mM TrisCl, pH 7.5, 150 mM NaCl, 1% Nondet P-40 (NP-40), 0.5% sodium deoxycholate, 1% sodium dodecylsulfate) for 30 min supplemented with a protease and phosphatase inhibitor cocktail obtained from Sigma (St. Louis, MO). Lysates were sonicated for 10 seconds and the homogenates were centrifuged for 5 min at 12,000 g at 4°C. Protein quantification was performed on the supernatant using a DC protein assay kit from Bio-Rad (Hercules, CA). Protein was heated to 60°C in 2X buffer (100 mM Tris pH 6.8, 4% SDS, 0.2% bromophenol blue, 20% glycerol, 200 mM dithithreitol) for 15 min. Equal amounts of protein were loaded into each lane of a 7.5% gradient pre-cast SDS-gel (Bio-Rad). Proteins were separated at 120 V constant. Gels were transferred onto PVDF paper (Millipore, Bedford, MA) at 200 mA constant for 2 hours at room temperature and these membranes were blocked in blocking buffer (10% milk in TBST, Tris-Buffered Saline Tween-20) for 1 hour by room temperature. Two ClC-3 antibodies were used to probe the blots-Alomone ClC3, (Jeruselum, Israel) and 14K, kindly provided by Dr. William J. Hatton PhD, (University of Nevada School of Medicine, Reno, NV). The membranes were incubated in primary antibody overnight at 4°C. The following day the membranes were washed three times for 15 min in TBST and then incubated with horseradish peroxidase-conjugated secondary antibody for 60 min at room temperature. Blots were again washed three times for 10 min and developed with Super Signal West Fempto enhanced chemiluminesence (Thermoscientific, Rockford, IL) using a Kodak Image Station 4000 MM (New Haven, CT). Secondary horseradish peroxidase-conjugated antibodies were purchased from Santa Cruz, Biotechnology, Inc (Santa Cruz, CA). ClC3 immunoreactivity was normalized to actin (Sigma), the Na⁺/K⁺ ATPase (Chemicon, Tumecula, CA) or GAPDH (Abcam).

Electrophysiology

Whole-cell patch-clamp recordings were acquired from glioma cells 72 hours after plating using standard whole cell patch clamp techniques (Hamill et al., 1981). Patch pipettes were pulled from thin walled borosilicate glass (World Precision Instruments (TW150F-4), Sarasota, FL) with an upright puller (Narishige Instruments (PP-830) Tokyo, Japan) to a resistance of 3–5 MΩ. An Axopatch 200A amplifier (Axon Instruments, Foster City, CA) was used and currents were low-pass filtered at 1 kHz then digitalized online at 10 kHz using a Digidata 1322A (Axon Instruments). Data was acquired and stored with the use of pClamp 8 (Axon Instruments). Whole cell capacitance and series resistance was measured from the amplifier, and compensation was adjusted to 80%. Standard pipette KCl pipette solution contained (in mM): 140 KCl. 1 MgCl2, 10 EGTA, 10 Hepes Sodium salt pH adjusted to 7.2 with tris base. Osmolarity was adjusted to 302mOsm. Normal extracellular NaCl Bath solution contained (in mM): 130 NaCl, 5 KCl, 1 CaCl2, 10.5 Glucose, 32.5 Hepes acid. pH was adjusted to 7.4 and osmolarity ranged from 306–312 mOsm/kg.

Statistical Analysis

Current subtractions and current-voltage relationships from electrophysiological recordings were determined using Clampfit (Axon Instruments, Foster City, CA). For all experiments raw data was analyzed and plotted using Origin 6.0 software (Microcal Software, INC, Northampton, MA). Appropriate statistical tests were chosen according to the type of data analyzed and performed using Graphpad Instat (Graphpad Software INC, San Diego, CA). Unless otherwise stated, all measurements are reported with standard error (se) and n refers to the number of cells analyzed.

RESULTS

Cytoplasmic Condensation is a Normal Part of mitosis in immature and transformed glial cells

We previously described a novel phenomenon in glioma cells, namely a cytoplasmic volume condensation just prior to cell division, which we termed pre-mitotic cytoplasmic condensation (PMC). Notably, this process is functionally linked to the condensation of chromatin as cells of the D54-MG glioma line approach cell division (Habela and Sontheimer, 2007). To examine the relevance of PMC, to glioma and glial cell biology in general and investigate the mechanism behind this process, we utilized fluorescent time-lapse microscopy to measure volumetric changes in live, spontaneously dividing, and adherent cells throughout the cell cycle. These experiments are based upon the concept that changes in cytoplasmic volume can be measured by changes in the fluorescence intensity of freely diffusible dyes or fluorescent proteins as fluorescence varies inversely in proportion to the concentration of the fluorescent molecule. This relationship between volume and fluorescence intensity is illustrated in Figure 1 in which D54-GFP cells were imaged in 0.5 μm optical sections throughout their depth to generate 3 dimensional objects in a time-lapse series (Fig. 1A, left). Relative volumetric information extracted from a single cell in these 3D images is plotted against time normalized to the last time point prior to division in Figure 1B (open circles). Shown on the same graph (closed squares) is the relative intensity of the GFP fluorescence within a single place of the same stack (Fig. 1A, right panels). As the volume decreases over a relatively short period of time, the intensity of the GFP is increased due to the concentration of the molecule. Changes in fluorescence intensity, therefore, can be used as a surrogate for volumetric measurements when 3D measurements are impractical due to required time resolution and number of fields to be analyzed.

Using such 2D time-lapse techniques, we confirmed that, as D54-MG glioma cells stably expressing GFP freely proliferate under normal culture conditions, the entrance into mitosis and approach to metaphase is accompanied by an increase in the fluorescence intensity ratio (F/F_m) such that the peak F/F_m value is reached at the last time point prior to anaphase (t = 0) (Fig. 1C closed squares). Although GFP intensity can be affected by changes in pH, we determined using the ratiometric pH indicator BCECF that no significant intracellular pH changes occurred during the analysis period (data not shown) and fluorescence intensity changes were truly representative of a significant change in cell volume. These data, therefore, indicate that a large and rapid change in cytoplasmic volume occurs as D54-MG cells progress to division. This process precedes and is closely associated with the process of DNA condensation that occurs between prophase and metaphase of mitosis and can be measured by the relative increase in the intensity of the nucleic acid binding dye, Hoechst 33342, which is concentrated as the DNA condenses (Fig. 1A, right; Fig. 1C open circles). This phenomenon of PMC coupled to DNA condensation is a novel observation which opposes the commonly held belief that cells achieve there largest volume just prior to division (Jorgensen and Tyers, 2004). This concept is important as it accentuates structural and physical changes as factors which may be equally as important as growth factor signaling and phosphorylation cascades to regulating the proliferative capacity of these cells.

Importantly, this phenomenon is not specific to cells of the D54 cell line, as it was also observed when we analyzed mitotic events in a different glioma cell line (U251-MG, Fig. 1D) or in immature cultured spinal cord astrocytes (Fig. 1E). The example experiment in Figure 1D demonstrates that U251-MG cells loaded with the fluorescent cytoplasmic dye CMFDA (closed symbols) undergo a similar volume decrease within the hour preceding division as occurs in GFP expressing D54-MG cells. Likewise, when spinal cord astrocytes, loaded with CMFDA were examined by 2D time-lapse microscopy at 3–5 days in culture, PMC was also observed (Fig. 1E, closed symbols). Primary rodent astrocytes were used as a model for non-malignant glial cells because they have been studied extensively in the glial literature and show a high rate of proliferation immediately after plating. By comparison, human astrocytes cultured from adult biopsy tissues rarely divide making such studies technically challenging. However, the observation that PMC occurs similarly in gliomas and non-malignant rodent astrocytes suggests that this process is likely to resemble an integral part of cell proliferation in neural cells although this phenomenon warrants further examination in human astrocytes in the future

The PMC seen in D54-MG and U251-MG cells, as well as in cultured astrocytes, was temporally linked to DNA condensation (Figure 1C, D and E, open symbols). These data imply that PMC is a normal part of cell division in normal and malignant cells in the brain. Since we previously established in D54 cells (Habela and Sontheimer, 2007) that the rate of this process plays a role in determining the rate of DNA condensation and therefore progression through mitosis, the observation that this process occurs in multiple cell types is important. Cellular characteristics which determine proliferative capacity are important to normal brain development and maintenance in addition to disease processes. We, therefore, sought to investigate the mechanism by which such a significant change in volume is coupled to the nuclear changes that occur during mitosis. The condensation of the cytoplasmic volume requires the coordinated efflux of cytoplasmic water driven by K⁺ and Cl⁻. In most cell types, volume reduction is intrinsically associated with the activation of chloride channels.

PMC is associated with an outwardly rectifying chloride current

The process of PMC was shown to be associated with an increase in the magnitude of an outwardly rectifying, time-dependent-inactivating, NPPB sensitive current which reversed at 0mV (Habela and Sontheimer, 2007). This suggested the involvement of a chloride current and prompted us to further study the electrophysiological profile of mitotic cells in the process of condensation. M phase and growth phase cells were distinguished based on cell body morphology and Hoechst labeling to determine chromatin condensation for whole-cell patch-clamp recordings (Fig. 2A). Entrance potential measurements indicated a mean resting membrane potential (RMP) of −26.1 ± 1.1 (n = 58) for M phase cells and −32.5 mV ± 1.8 (n= 40) for the bipolar growth phase cells. M phase cells were significantly more depolarized than the bipolar cells (P = 0.001, Two-Tailed T Test). Additionally, the measured whole cell capacitance for M phase cells was 40.1 pF ± 0.93 (n=58) and for bipolar cells was 55.6 pF± 2.4 (n= 40) (P< 0.0001, Two-Tailed T Test). This decrease in capacitance in the mitotic cells is indicative of a decreased membrane surface area and is consistent with the observed decrease in cell volume in these cells.

Figure 2.

Mitotic cells have increased chloride current densities. A, Example cells for bipolar (top) and M phase (bottom). Mitotic cells have distinct cell body morphology (phase) and DNA condensation is evident (Hoechst). Magnification = 40X. B, Example current traces from recordings of M phase cells before (top) and after (middle) chloride channel inhibitors. Bottom traces are drug sensitive currents. All traces are to scale at top left.

In order to examine membrane chloride current profiles, voltage steps were applied for 100 ms from −140 to + 140 mV at 20 mV intervals to cells with discernable chromatin condensation in pro-metaphase or metaphase (Fig. 2A bottom). The resulting currents were compared to those elicited from large bipolar cells with round uncondensed nuclei (Fig. 2A top). In the presence of 2 uM paxilline (Pax) to inhibit the major K⁺ conductance in these cells (Ransom and Sontheimer, 2001), we were able to isolate the majority of the remaining current with either of the nonspecific chloride channel inhibitors, 4,4′-Diisothiocyanatostillbene-2,2′-disulfonic acid disodium salt hydrate (DIDS) and 5-nitro-2-(3-phynylpropylamino)benzoic acid (NPPB). As shown by the example currents from M phase cells in Figure 2B, 200 μM DIDS inhibited a qualitatively similar current to that inhibited by 200 μM NPPB, and both the NPPB and the DIDS sensitive currents were significantly larger at in M phase cells compared to bipolar cells. There was an approximate twofold increase in the NPPB sensitive current for the M phase cells compared to the bipolar cells (M = 14.6 pA/pF ± 2.289, n = 35 vs Bipolar = 9.195 pA/pF ± 1.706, n = 27; One-Tailed T-Test P < 0.05) and a threefold increase in the DIDS sensitive current density in M phase cells (M phase = 6.5 pA/pF ± 1.4, n = 23 vs Bipolar = 2.2 pA/pF ± 0.9, n= 13; One-Tailed T-Test P < 0.05). Consistent with the current being carried by chloride, it was unaltered if extracellular Cl⁻ was replaced by Br⁻, which is known to permeate most chloride channels (data not shown). This substitution brought about a significant −3.05 mV ± 0.87 shift in the reversal potential of the whole cell Pax insensitive currents (n = 10, One PopulationT Test, p < 0.01) indicating that permeability to Br⁻ is slightly greater than to Cl⁻ with a P _Br-/P_Cl- ratio of 1.1 calculated from the constant field potential equation as described by Olsen et al. (2003). The biophysical characteristics of an outwardly rectifying, time- and positive voltage-dependent inactivating Cl⁻ current that is sensitive to both NPPB and DIDS is consistent with either I_ClC3 or I_{Cl swell} (Duan et al., 1997; von Weikersthal et al., 1999; Jin et al., 2003; Olsen et al., 2003). These currents and the means to definitively distinguish between the two, however, have been a source of controversy.

ClC3 localization is altered in a cell cycle dependent manner

There are no specific inhibitors for either of these currents and the molecular identity of the channel which conducts I_{Cl Swell} is to date unknown. However, a long and a short form of the voltage dependent Cl⁻ channel family member, ClC3, have been cloned (Kawasaki et al., 1994; Borsani et al., 1995; Shimada et al., 2000; Huang et al., 2001) and characterized in knockout animals (Yamamoto-Mizuma et al., 2004; Dickerson et al., 2002; Stobrawa et al., 2001; Yoshikawa et al., 2002) and by transient knockdown with antisense oligonucleotides and shRNA (Olsen et al., 2003; Wang et al., 2002a; Mao et al., 2008; Wang et al., 2000). Further, this channel has been previously shown to be present in multiple glioma cell lines by RT-PCR and western blot analysis (Olsen et al., 2003; Ernest et al., 2005). Its presence and activity on the cell surface of D54-MG cells has been determined using electrophysiology in combination with oligonucleotide antisense, and by electron microscopy (Olsen et al., 2003) and its involvement in processes of physiological volume regulation and cell cycle progression in other cell types (Wang et al., 2002a; Wang et al., 2000; Jin et al., 2003; Duan et al., 2001; von Weikersthal et al., 1999) makes it a good candidate for the molecular entity in this process.

If ClC3 is the channel responsible for the currents observed in mitotic cells, then one possible explanation for the increase in magnitude would be that there is differential membrane localization of the channel in mitotic cells compared to normal growth phase cells. We examined this using confocal immunofluorescence microscopy. D54-MG cells were paraformaldehyde fixed 72 hours after plating, and fluorescently stained with either a polyclonal rabbit antibody to the C-terminus of ClC3 alone or in combination with FITC-labeled Phalloidin staining. Figure 3A shows a confocal section taken of an example field which contains large growth phase cells and a mitotic cell at metaphase, demonstrated by Hoechst labeling (blue). In growth phase cells, consistent with work by others (Zhao et al., 2007), ClC3 (red) appeared to be colocalized with the cortical actin skeleton at the cell membrane (Fig. 3A Box 1 and Fig. 2B) in addition to localizing to punctuate endosomal-like structures in the cytoplasm (Fig. 3B and 3C). Interestingly, there was also strong ClC3 labeling in the nucleus associated with the Hoechst labeled DNA (Fig. 3A Box 2 and Fig. 3C).

Figure 3.

ClC3 is present on the plasma membrane and associates with DNA. A, Immunofluorescence images of fixed D54- MG cells stained with an antibody to ClC3 (15K), FITC- conjugated Phalloidin and Hoechst 33342. This field contains both growth phase cells and a cell undergoing mitosis. Boxes 1, 2 and 3 correspond to images in B, C and D. B, ClC3 colocalizes with the subcortical actin. C, ClC3 localizes to the nucleus in growth phase cells. D, E, examples of ClC3 localization in two different M phase cells. Arrows indicate the plasma membrane. Arrowheads indicate spindle structures delineated by ClC3. Scale bars = 10 μm. Images are representative of 5 independent experiments using 2 different antibodies to ClC3.

In the mitotic cells, as demonstrated by Figure 3D and 3E, ClC3 immunofluorescence throughout the cell was significantly more intense, as would be expected for a cell in which the cytoplasmic volume has been condensed. However, the brightest staining was in the region of the cortical actin (arrows) as well as associated with the condensed chromosomes. This suggests that in the condensing (Fig. 3E) and fully condensed (Fig. 3D) mitotic cells ClC3 is preferentially localized to the membrane, which may account for an increase in the amplitude of membrane currents. Additionally, the ClC3 localization was reminiscent of the spindle formation (Fig. 3D and 3E, arrowheads). In an attempt to put this staining into context, we counterstained D54-MG cells with a FITC labeled monoclonal antibody to α-tubulin. Figure 4A demonstrates single plane confocal images of ClC3 localization in 4 separate mitotic cells which were stained with two antibodies recognizing two separate epitopes in the C- terminus of the ClC3 protein (Alomone and 15K). Interestingly, the spindle morphology illustrated by the α-tubulin labeling was mimicked by the discrete DNA associated staining from both antibodies in the ClC3 panels. This colocalization is most evident in the merged panels in which ClC3 is depicted in red, α-tubulin is depicted in green, and pixels in which the fluorescence overlaps appear yellow.

Figure 4.

ClC3 colocalizes with α-tubulin at the mitotic spindle. A, Immunofluorescent images of four different cells in various stages of mitosis. Cells are labeled with either 15K anti-ClC3 or Alomone anti-ClC3 antibodies (red) and FITC-labeled anti- α-tubulin (green). DNA is stained with Hoechst 33342 (blue). Colocalization is indicated by yellow in merged panels. Scale bar = 10 μm. Images representative of 5 independent experiments. B, Representative western blot of a mouse anti- α-tubulin immunoprecipitation probed with Abcam’s anti ClC3 antibody (top) and rabbit anti- α-tubulin (bottom). WC = whole cell lysates, SN = supernatant, IP Tub = α-tubulin immunoprecipitation, IP Neg = no antibody control. Similar results observed in 3 independent experiments.

To determine whether this spindle associated staining was in fact due to an interaction between ClC3 and tubulin, we immunoprecipitaed α-tubulin with a monoclonal antibody and protein G agarose beads. Whole cell lysates as well as precipitates where subject to SDS-PAGE and Western blot analysis with polyclonal rabbit antibodies against ClC3. As shown in an example blot in Figure 4B, ClC3 antibodies recognize a ~120 kDa band in the immunoprecipitate but not in the negative control. This is similar to the molecular weight which has been published for glycosylated ClC3 and indicates that it physically interacts with tubulin. This implies that in addition to being a candidate for the channel responsible for mitotic chloride currents, ClC3 may play a more direct role in the structural changes which occur in the DNA as mitosis progresses.

M phase chloride currents are diminished by ClC3 knockdown

Decreased proliferation rates have been reported for multiple normal and cancerous cell types in the presence of NPPB and/or DIDS (Chen et al., 2002; Shen et al., 2000; Wondergem et al., 2001; Zheng et al., 2003) supporting a possible role for membrane expressed CLC3 in the progression of the cell cycle. Yet the promiscuity of chloride channel inhibitors makes linking pharmacology to molecular entities very difficult. We therefore used shRNA constructs to determine the role of ClC3 in PMC. D54-MG cells were transfected with three pGIPZ-shRNAmir vectors containing short hairpin sequences which target 3 unique sites of the human ClC3 gene and which are included in both the long and the short splice variants. Selecting for the vector encoded puromycin resistance gene, we generated 4 stable lines corresponding to each of the three shRNA constructs (H8, F3, and A4) in addition to a nonsense control cell line (NS). To determine if protein levels were effectively knocked down by the shRNA constructs, whole cell lysates from the H8, F3, and A4 lines were compared to lysates from the NS cell line by SDS-PAGE and western blot analysis with antibodies to various regions of the ClC3 protein. The example blots shown in Figure 5A and 5B were probed with two separate antibodies and demonstrate a significant knockdown of two bands at 116 and 80 kDa compared to control cells in which a nonsense sequence was transfected. These bands are within the published range of molecular weights for various states of post translational modification of the ClC3 isoforms (Stobrawa et al., 2001; Huang et al., 2001). Similar results were consistently observed and quantification of 3 independent experiments (Fig. 5B) indicates that ClC3 protein levels were substantially reduced, although not fully eliminated, in the knockdown cell lines.

Figure 5.

CLC3 expression is reduced by shRNA contructs. A and B, western blots probed with either Abcam or 14K anti-ClC3 and actin as a loading control. A. Lysates from control (NS) and ClC3 shRNA expressing (A4, H8, and F3) cells were probed with Abcam anti-ClC3 antibody. B, The blot from A was stripped and then re-probed with the 14K antibody. C, Densitometry quantification of Abcam anti-ClC3 bands between 116 and 80 kDa normalized to loading controls. n = 3 independent experiments, One-way ANOVA P < 0.0001, Tukey-Kramer post- test (***) P < 0.001.

Having established that these cells expressed reduced levels of protein, we next sought to examine how this change would affect the current profiles of mitotic cells. In the presence or absence of 200 μM NPPB we examined chloride current from cells of either the ClC3 knockdown or the control lines under whole-cell patch-clamp conditions. Mitotic cells were selected based on morphology and nuclear architecture and matched as closely as possible between cell lines (Fig. 6A). Currents were elicited in the presence of 2 μM Pax to inhibit K⁺ currents using the voltage step protocol described for Figure 2. As demonstrated by the representative example traces in Figure 6B, in nonsense transfected cells, a large proportion of the remaining current could be inhibited by 200 μM NPPB. The subtracted drug sensitive current in these cells (Fig. 6B, bottom left) displayed outward rectification and time dependent inactivation and was both qualitatively and quantitatively similar to the currents present in the untransfected cells (Fig. 2A). In cells that were stably transfected with the shRNA constructs, this current was greatly reduced and often absent (Figure 6B, bottom right). The mean current-voltage (IV) relationship of currents from nonsense transfected cells (n = 7) compared to the H8 shRNA cell line (n = 7) demonstrated in Figure 6C illustrates that NPPB inhibits qualitatively similar, but quantitatively smaller currents in the mitotic cells of the knockout cell line. Significantly, the mean current density at 100 mV was decreased ≥ 50% for all three knockdown lines compared to control cells (H8 control = 7.231 pA/pF ± 2.846 n = 8 vs. H8 = 0.5917 pA/pF ± 0.8665 n = 9, P=0.0165; F3 Control = 15.066 pA/pF ± 3.856 n = 9 vs. F3 = 7.430 pA/pF ±1.726 n = 9, P= 0.0392; A4 Control = 15.699 pA/pF ± 4.812 n = 10 vs. A4 = 4.564 pA/pF 2.906 n = 8, P = 0.0344; One-tailed T-Tests). These diminished currents as a result of decreased channel expression implicate ClC3 as the channel mediating chloride currents in mitotic cells.

Figure 6.

Mitotic chloride currents are reduced by ClC3 shRNA. A, Example M phase cells from control and shRNA cell lines. B, Example current traces from a control cell (left) and a shRNA expressing cell (right). Voltage steps were made in the presence (middle traces) or absence (top traces) of 200 μM NPPB and the drug sensitive current was determined by subtraction (bottom traces). Scale applies to all traces. C, Current voltage relationship of the NPPB sensitive current from Control (closed squares) and H8 shRNA (open circles) cells. n = 9 of each cell type.

ClC3 associated current is required for efficient PMC

To determine the effect of knocking down the ClC3 channel on the ability of cells to undergo PMC, we conducted 2D time-lapse experiments in which the control cell line was imaged simultaneously with the shRNA transfected cell lines. Under these conditions, the rate of PMC could be compared in relation to increased or decreased levels of the ClC3 protein. The time-lapse images in Figure 7A show an example control cell (left) compared to a ClC3 shRNA transfected cell (right) undergoing the process of mitotic cytoplasmic condensation demonstrated by increases in the intensity of the GFP fluorescence. The cytoplasmic condensation occurs rapidly in the control cell with the process of condensation occurring within the course of forty minutes. In the shRNA expressing cell this process is prolonged and the time required for condensation exceeds two hours. This was quantified for n = 45 cell divisions per condition captured from three independent experiments for the H8 shRNA line (circles) compared to control cells (squares) and plotted in Figure 7B. It is apparent from these curves that the cytoplasmic GFP in the ClC3 knockdown cells (closed circles) began to condense earlier, at approximately 120 minutes prior to division, and took longer to reach maximal condensation (F/F_m = 1) than in the control cells (closed squares) which condensed within the last 60 minutes of the cell cycle. The increase in the time that the cells took to condense is also reflected in a significant decrease (One Way ANOVA P = 0.0002) in the slope of the change in GFP intensity over time for all three knockdown lines compared to the control line (Fig. 7C). The control cells undergo cytoplasmic condensation at almost twice the rate of the ClC3 shRNA expressing cells do, indicating that ClC3 plays a key role in this process. Interestingly, although the loss of ClC-3 causes a significant decrease in the rate of condensation, the volume condensation ultimately does occur, albeit much delayed. This may be explained by an incomplete knockdown of the channel protein, which, on average, leaves 40% of the original protein remaining. More likely, however, this suggests that a secondary, redundant mechanism for cell condensation exists. Consistent with this notion, a previous paper demonstrated a significant contribution of KCC1 to Cl⁻ efflux in the context of regulatory volume decrease in osmotically challenged glioma cells (Ernest et al., 2005). Indeed, we were able to reduce the rate of PMC by up to 50% in the presence of the KCC inhibitor, R(+)-Butylindazone R-(+)-[(2-n-Butyl-6,7-dichloro-2-cyclopentyl-2,3-dihydro-1-oxo-1H-inden-5-yl)oxy]acetic acid (DIOA) (Supplemental Figure 1). As the process of PMC appears to be a critical regulator of the cell cycle, and hence population growth, the fact that multiple mechanisms are in place to ensure a successful volume change and hence cell division is to be expected.

Figure 7.

PMC and DNA condensation is impaired by reduced ClC3 expression. A, Images of GFP and Hoechst 33342 fluorescence over time for an example control and an example H8 shRNA cell at 20X magnification. B, Quantification of PMC based on relative GFP fluorescence (closed symbols) and DNA condensation based on relative Hoechst fluorescence (open symbols) for Control (squares) and H8 shRNA (circles) cells from three independent experiments. n = 45 cells each condition. C, Quantification of the rate of PMC in Control cells compared to shRNA expressing cells. The PMC curve was integrated and the slope at −30 min was determined for n = 69 cells (control), n = 45 cells (H8), n= 39 cells (F3), and n = 45 cells (A4). The experiment was repeated a minimum of three independent times for each condition. One-way ANOVA P < 0.001, Dunnet post- test. D, Quantification of the number of cell divisions occurring over a 48 hour period in shRNA expressing cells compared to controls from 3 independent experiments. One way ANOVA P < 0.01, Dunnets Post- test. * indicates P < 0.05, ** indicates P < 0.01 for all graphs.

As we have shown in previous work, decreasing the rate of PMC also decreases the rate of the condensation of the DNA into metaphase chromosomes (Habela and Sontheimer, 2007). We therefore analyzed the mitotic DNA condensation by changes in Hoechst intensity (Fig. 7A and 7B) and found that in the shRNA lines the F/F_m curve for Hoechst intensity (open circles), was slowed to an equal extent as that for the cytoplasmic GFP (closed circles) compared to controls. This is indicative of chromatin condensation not beginning until the onset of cytoplasmic condensation. These results were observed in all three cell lines and, as the result of prolonging any stage of the cell cycle is to increase cell cycle length, we measured the number of divisions occurring during each imaging period. We observed significantly fewer (One Way ANOVA, P = 0.0065) divisions occurring among the knockdown cells (Fig. 7D) indicating a longer time period between divisions and suggesting that the slowed PMC brought about an increase in cell cycle length. This confirms our hypothesis that PMC is linked to DNA condensation and therefore cell cycle progression and supports a role for ClC3 in this process.

The above data implicate the ClC3 channel in PMC, yet it does not definitively prove that the size of the membrane currents is deciding the rate of PMC. The immunoprecipitation data and immunostaining in Figure 4 localize intracellular ClC3 to the region of the mitotic spindle and loss of activity there in the knockdowns could possibly explain a decreased PMC rate and number of cell divisions. To clarify this, we acutely treated both the control cells and the ClC3 knockdown cells with 200 μM NPPB and imaged dividing cells by time-lapse microscopy within the first four hours of treatment (Figure 8). NPPB at this concentration should block the mitosis associated chloride current, but have no effect on the overall ClC3 composition of the cells. Analysis of the similarities and differences between NPPB treatment and ClC3 knockdown should therefore differentiate between effects mediated by the channel associated plasma membrane currents and those mediated by other actions of the ClC3 molecule. Maximal inhibition of the NPPB sensitive chloride current significantly altered the kinetics of the observed PMC curve in nonsense transfected cells (Fig. 8A open squares) compared to vehicle treated controls (Fig. 8A, closed squares). The PMC curve for the H8 ClC3 shRNA cells (Fig. 8A, closed circles) was also shifted from the nonsense control cells such that the PMC slopes of the NPPB treated nonsense cells and the control ClC3 shRNA cells were not significantly different (Fig. 8B). Although there was a trend towards an increased effect at early time points, addition of NPPB to the H8 cells did not significantly alter this effect. Comparing the NPPB treated nonsense cells to the NPPB treated H8 shRNA cells or the H8 cells alone there was not a significant difference in the mean slope values (Fig. 8B). However all were significantly decreased from the DMSO treated nonsense transfected controls (Fig. 8B; One Way ANOVA, P = 0.0001, Tukey-Kramer Multiple Comparisons Test). ClC3 may have additional roles in the cell cycle and even in DNA condensation, but this data indicates that the plasma membrane chloride mediated by ClC3 plays a major role in determining the rate at which these cells undergo PMC and therefore progress through mitosis.

Figure 8.

Chloride channel inhibition has similar effects on PMC as ClC3 knockdown. A, Mean PMC curves for control cells (squares) and H8 shRNA cells (circles) treated with 200 μM NPPB (open symbols) or vehicle (closed symbols). B, Quantification of the rate of PMC at −30 min from experiments in A. Experiments repeated 3 independent times. n = 25 cells for control, n = 23 cells for control treated with NPPB, n= 26 cells for H8 and n= 23 cells for H8 treated with NPPB. One-way ANOVA P < 0.01, Dunnet post-test * indicates P < 0.05, ** indicates P < 0.01.

DISCUSSION

In this study, we established that a cytoplasmic volume condensation accompanying progression through mitosis is a normal part of cell division in both malignant glioma cells and immature cultured astrocytes. This pre-mitotic condensation (PMC) is associated with a chloride current that has the biophysical properties of ClC3. Involvement of this channel is supported by immunofluorescence localization and knockdown of ClC3 in glioma cells not only decreased the chloride currents associated with mitotic cells, but slowed PMC, prolonging M phase and hindering DNA condensation. This mechanistically links ClC3 to cell division with a specific role in the cytoplasmic and DNA condensation phase of mitosis.

ClC3 mediated chloride currents are involved in mitotic progression of glioma cells

This is the first study that describes a specific mechanism for chloride channel modulated progression through M phase and furthers our knowledge of the functions of ClC3 in the brain. Although previous studies have analyzed the relative chloride conductance in pharmacologically arrested cells as well as the effect of chloride channel inhibitors on cell cycle arrest (Chen et al., 2007; Doroshenko et al., 2001; Zheng et al., 2003; Wondergem et al., 2001; Klausen et al., 2007; Shen et al., 2000; Chen et al., 2002; Wang et al., 2002b) the lack of specific inhibitors and the use of population data has caused confusion as to which channels are involved at particular transition points and by what mechanism. As a result, currents with the same profile as I_Cl-Swell have been implicated in the G0/G1 transition, S phase and mitosis, but it is difficult to distinguish between I_Cl-Swell and I_Cl-ClC3 based on pharmacology and electrophysiology alone. Complicating the matter further, some controversy surrounds the postulation that ClC3 is the molecular entity behind I_Cl-Swell (Duan et al., 1997; Jin et al., 2003; Wang et al., 2000; Duan et al., 2001; Weylandt et al., 2001; Weylandt et al., 2001; Li et al., 2000) and there is no consensus as to how to definitively identify either without genetic manipulation.

We demonstrated that an NPPB and DIDS sensitive current which is carried as well by bromide as by chloride is increased in mitotic cells compared to bipolar cells. These characteristics established that the current was a chloride current. Further, depolarized RMPs in M phase cells compared to bipolar cells suggests that these channel have enhanced conductance in M-phase cells, i.e. are more likely to be open at rest. We relied on shRNA mediated knockdown to definitively identify ClC3 as the channel responsible for conducting this current.

ClC3 was originally cloned as a 760AA protein in the rat and shown by northern blot to be present throughout the body with particularly high levels in the brain (Kawasaki et al., 1994). In situ hybridization demonstrated greatest expression in the hippocampus, olfactory bulb and cerebellum. Subsequently, a long form containing a 58AA n-terminus insertion was cloned and shown to be the predominant form in the brain (Shimada et al., 2000). Currents elicited from cells expressing the cDNA for these two isoforms demonstrated a permeability sequence of I⁻ > Br⁻ > Cl⁻ based on reversal potentials and outward rectification (Shimada et al., 2000; Matsuda et al., 2008; Kawasaki et al., 1995). Additionally, the channel displayed time and voltage dependent inactivation (Shimada et al., 2000; Huang et al., 2001). The currents which we observed in M phase cells were consistent with this description and were decreased in the shRNA lines. This effect was specific as it was observed with three different sequences. These sequences targeted both long and the short form and two bands at 80 and 116 kDa that were consistently present on western blots of control cells and were knocked down with ClC3 shRNA. Our results establish ClC3 as the channel behind the chloride conductance in mitotic cells.

PMC and DNA condensation are dependent on ClC3 expression

The involvement of ClC3 in the progression of mitosis was established using a single cell, quantitative imaging technique which allowed us to capture the dynamic changes occurring during mitosis. The rapid volume decrease which we previously found to be functionally linked to DNA condensation was significantly slowed by the shRNA mediated knockdown of ClC3. The importance of ClC3 in mitotic progression was further supported by the fact that in ClC3 shRNA expressing cells, the condensation of the DNA was hindered to the same extent as the volume decrease.

In D54-MG cells, ClC3 colocalizes with the subcortical actin on the plasma membrane. We also observed, strikingly, that ClC3 localized to the mitotic spindle (Figure 4). This association was supported by the co-immunoprecipitation of ClC3 with α-tubulin and suggested the possibility that it is not plasma membrane currents but a direct effect on spindle assembly that mediates the protein’s catalysis of M phase. In order to resolve this question, we compared the rate of PMC in control or shRNA cells in the presence or absence of a chloride channel inhibitor. The similarities between the volume condensation curves for shRNA expressing cells with and without NPPB and control cells treated with NPPB suggests that the primary effect is due to plasma membrane currents as chloride channel inhibitors would not be expected to alter intracellular interactions of the protein.

A direct interaction between an N terminal portion of the ClC3 protein and clathrin has been demonstrated and is required for internalization (Zhao et al., 2007). Additionally, binding of a c-terminal intracellular portion of sClC3 to subcortical F-actin is required for hypotonic activation of the channel (McCloskey et al., 2007). A potential model for this process, therefore, is that actin rearrangements, occurring at the onset of mitosis, cause ClC3 to open at the cell membrane. At a membrane potential of −30 mV, opening of the channel causes an efflux of chloride which draws water out of the cytoplasm. This facilitates membrane collapse as ClC3 undergoes clathrin mediated endocytosis. The result is a decrease in cytoplasmic water content and membrane surface area. ClC3 has been implicated in endocytosis (Hara-Chikuma et al., 2005) and inhibition of clathrin mediated endocytosis has been shown to prevent a mitotic reduction in membrane surface area resulting in abnormal spindle morphology and multinucleate cells (Boucrot and Kirchhausen, 2007). Clathrin has been shown to play a structural role in spindle formation (Okamoto et al., 2000; Royle et al., 2005; Royle and Lagnado, 2006) and ClC3 may traffic with clathrin to the spindle. Alternatively nuclear ClC3 may facilitate clathrin recruitment. Both scenarios propose a secondary effect of ClC3 on mitotic progression at the spindles. However, the most direct interpretation of this study is that ClC3 dependent chloride efflux, causes a volume decrease which in turn facilitates DNA condensation possibly by decreasing the activation energy for important molecular interactions via molecular crowding.

ClC3 is important in normal as well as pathological states in the brain

When the long isoform of ClC3 was cloned in humans, it was determined to be regulated by CAMKII (Reeves and Andreoli, 1992), and a chloride conductance with similar properties was determined to modulate excitability in immature hippocampal neurons from wt but not ClC3^−/− animals (Wang et al., 2006). The importance of ClC3 to the normal physiology of these neurons is further supported by the fact that in all knockout models to date; the primary pathological finding is hippocampal degeneration (Stobrawa et al., 2001; Yamamoto-Mizuma et al., 2004; Yoshikawa et al., 2002; Dickerson et al., 2002). In light of our finding that ClC3 is a modulator of cell cycle progression, it is interesting that this region which is most affected by its knockdown is one of the two brain regions where neural stem cell (NSC) proliferation continues throughout life.

Self renewing NSCs with the capacity to differentiate into neurons, astrocytes, and oligodendrocytes have been identified (Gage, 2000; Zhao et al., 2008) and neurogenesis has been shown to occur in both the subventricular zone (Lois and Alvarez-Buylla, 1993; Doetsch and Alvarez-Buylla, 1996) and the subgranular zone of the hippocampus (Bayer et al., 1982; Gage, 2000). A growing body of evidence indicates that these cells are the precursors to glioma cells (Rao et al., 2003; Singh et al., 2003; Ma et al., 2008; Ignatova et al., 2002; Hemmati et al., 2003). Although this study does not directly address the relationship between gliomas and NSC’s, the implication of the hypothesized NSC origin of glioma is that normal immature cells in the brain and glioma cells may share common mechanisms to enhance their proliferative capacity and further investigation should be made into the biophysical alterations that accompany these proliferating cells. We hypothesize that normal immature cells in the brain may use the outwardly directed chloride gradient to drive this rapid volume change as a means facilitate rapid progression through mitosis as we have shown in glioma cells. Resolving this question will shed light in the normal biology of the brain as well as the mechanisms of cancerous proliferation.