“Mitotic cell rounding” describes the rounding of mammalian cells before dividing into two daughter cells. This shape change requires coordinated cytoskeletal contraction and changes in osmotic pressure. While considerable research has been devoted to understanding mechanisms underlying cytoskeletal contraction, little is known about how osmotic gradients are involved in cell division. Here we describe cytoplasmic condensation preceding cell division, termed “premitotic condensation” (PMC), which involves cells extruding osmotically active Cl− via ClC-3, a voltage-gated channel/transporter. This leads to a decrease in cytoplasmic volume during mitotic cell rounding and cell division. Using a combination of time-lapse microscopy and biophysical measurements, we demonstrate that PMC involves the activation of ClC-3 by Ca2+/calmodulin-dependent protein kinase II (CaMKII) in human glioma cells. Knockdown of endogenous ClC-3 protein expression eliminated CaMKII-dependent Cl− currents in dividing cells and impeded PMC. Thus, kinase-dependent changes in Cl− conductance contribute to an outward osmotic pressure in dividing cells, which facilitates cytoplasmic condensation preceding cell division.
- chloride channel
before dividing into two daughter cells, mammalian cells transform shape from flat during interphase to round during M-phase, typifying a process termed “mitotic cell rounding.” Mitotic cell rounding is an intrinsic and stereotypic phenomenon associated with cell division (35). The importance of mitotic cell rounding is manifold: it stabilizes and positions the mitotic spindle, a curved plasma membrane exposes lipids important in cell cycle progression, and it ensures that both daughter cells receive the correct chromosomal set with high fidelity (30). Therefore mitotic cell rounding seems to be important in the mechanics as well as signaling of cell division. A recent study demonstrated that mitotic cell rounding is primarily driven by two opposing forces: an inward cytoskeletal contraction versus an outward osmotic force (35). Actomyosin cytoskeletal dynamics promoting mitotic cell rounding have been well studied. Moesin is activated during mitosis and attaches actin filaments to the plasma membrane, increasing cortical rigidity and cell rounding (14). Additionally, RhoA, a small GTPase, activates Rho kinase, leading to increased actomyosin contractility (19). However, relatively little is known about the outward osmotic force contributing to mitotic cell rounding.
Several studies performed in human glioma cells demonstrate that Cl− channels substantially contribute to this outward osmotic pressure. M-phase cells have elevated Cl− currents as compared with interphase cells (9). These elevated Cl− currents in dividing cells are associated with a decrease in cytoplasmic volume, termed “premitotic condensation” (PMC) (9). As osmotically active Cl− exits dividing cells, water is osmotically driven out of the cytosol, leading to an outward osmotic pressure and decrease in cellular volume (9). Incubation of dividing cells in hypertonic media significantly decreases this outward osmotic pressure and hampers PMC (9). The molecular identity of Cl− channel responsible for this Cl− efflux in glioma cells was determined to be ClC-3, a voltage-gated Cl− channel/transporter (8). ClC-3 has previously been implicated in shape and volume changes in a variety of cells, including glioma cells (26). Additionally, ClC-3 activity facilitates proliferation in a number of cell types including nasopharyngeal carcinoma cells (39) and smooth muscle cells (25). However, how ClC-3 becomes activated in dividing cells leading to premitotic condensation is unknown. Previous reports have demonstrated that ClC-3 can be regulated by kinases, including Ca2+/calmodulin-dependent protein kinase II (CaMKII) (11).
Using a combination of imaging and biophysical techniques, we demonstrate that CaMKII activates ClC-3 in dividing cells, promoting premitotic condensation. ClC-3 and CaMKII strongly colocalize on the plasma membrane of dividing cells during mitotic cell rounding. Time-lapse microscopy revealed that inhibition of CaMKII or ClC-3 impedes premitotic condensation to the same extent. We directly measured ClC-3 activity in dividing cells using whole cell patch-clamp electrophysiology and found that dividing cells had an increased Cl− conductance that was CaMKII dependent. Inhibition of CaMKII impaired PMC selectively in ClC-3-expressing cells, suggesting an important role of CaMKII modulation of ClC-3 in PMC.
METHODS AND MATERIALS
D54 human glioma cells, derived from a glioblastoma, are a World Health Organization grade IV cell line. These cells were gifted by Dr. D. Bigner (Duke University, Durham, NC). Cells were passaged in Dulbecco's modified Eagle's medium (DMEM)-F-12 (Invitrogen), supplemented with 2 mM glutamine and 7% fetal bovine serum (Hyclone, Logan, UT), and incubated in 10% CO2 at 37°C. All reagents were purchased from Sigma-Aldrich (St. Louis, MO) unless otherwise specified.
Cells were monitored for spontaneous divisions as previously described (8). First, D54 cells were stably transfected with green fluorescent protein (GFP) with the peGFP-N1 (Clontech) plasmid and maintained with 0.25 mg/ml G418 (Invitrogen). Cells were then grown on four-chamber no. 1 German borosilicate glass slides (Nalge Nunc International) and cultured for 1–3 days before imaging. Images were acquired with an Axiocam MRm camera mounted on an Axiovert 200M microscope (Carl Zeiss) in 5- to 10-min intervals for 48 h with a ×20 Fluar air objective. The stage was housed in an incubator maintained at 37°C with 5% CO2. We used a metal halide X-cite 120 light source (EXFO Photonics Solutions) with a FITC filter cube (excitation, 480 ± 15 nm; emission, 535 ± 20 nm; Chroma Technologies). Neutral density filters were used to reduce phototoxicity. The acquired 16-bit TIF images were then screened for dividing cells. Regions of interest (ROIs) in the cytoplasm of dividing cells were selected in National Institutes of Health WCIF ImageJ to measure changes in cytoplasmic mean pixel intensity, which is a reflection of cell volume as previously demonstrated (9). Mean pixel intensities were compiled in Microsoft Excel 2007, then the time points and fluorescence intensities were normalized. For each cell, we designated the time point just before cell division as time = 0 and normalized all time points to time = 0. Therefore, if a cell began changing volume at − x minutes, it indicates that x minutes before division, the cell began to change its volume. Each cell's fluorescence intensity (F) was normalized to its own fluorescence intensity at time = 0 (F0). The normalized averages were then compiled into Origin 6.0 (MicroCal, Northampton, MA). To calculate Fmin, Fmax, and t50, the data were fit to a sigmoidal curve with a linear X scale (Boltzmann fit; R2 > 0.99 for all conditions). The Fmin and Fmax were the lowest and highest F/F0 intensities for each condition, respectively.
Cells were immunolabeled as previously described (4). D54 cells were grown for 2–4 days on glass coverslips (Macalaster Bicknell). After culturing, cells were washed with PBS and fixed for 10 min with 4% paraformaldehyde. After blocking for 30 min at room temperature with PBS containing 10% normal goat serum and 0.3% Triton X-100, cells were incubated in primary antibodies overnight at 4°C. The next day, cells were washed and incubated in secondary antibodies for 90 min at room temperature. Cells were washed again and incubated in Hoechst 33342 at 100 ng/ml for 20 min at room temperature. Cells were then washed and mounted onto glass slides (Fisher Scientific) with Fluoromont (Sigma).
ClC-3 was labeled with polyclonal rabbit anti-ClC-3 (Alomone Labs, Jerusalem, Israel; lot no. ACL001AN0702) used at 1:250. CaMKII was labeled with monoclonal mouse anti-CaMKII (Abcam, Cambridge, MA) used at 1:250. Autophosphorylated (p)CaMKII was labeled with polyclonal rabbit anti-pCaMKII (T286; Abcam) used at 1:250. Secondary antibodies used were goat anti-rabbit Alexa 488 and goat anti-mouse Alexa 546 (Invitrogen) at 1:500. Images were acquired with Slidebook software (version 5.0, Intelligent Imaging Innovations) using a Hamamatsu IEEE1394 digital charge-coupled device camera mounted on an Olympus IX81 microscope with an Olympus scanning disc unit to remove out-of-focus light. A ×60 oil immersion lens (numerical aperture, 1.42) was used to acquire 0.5-μm optical sections to generate 10-μm image stacks. Alexa 488 fluorescence was imaged with a FITC filter set (excitation, 482 ± 17 nm; emission, 536 ± 20 nm), Alexa 546 fluorescence was imaged with a tetramethylrhodamine isothiocyanate filter set (excitation, 543 ± 22 nm; emission, 593 ± 20 nm), and Hoechst fluorescence was imaged with a DAPI filter set (excitation, 387 ± 5.5 nm; emission, 447 ± 30 nm) (Semrock). Images were interpolated, displayed in “three-view,” and linearly normalized with Slidebook software.
Before recordings were performed, cells were loaded with Hoechst 33342 at 100 ng/ml in cell culture media for 25 min to aid in identification of M-phase cells with condensed DNA. For whole cell patch-clamp recordings, D54 cells were recorded from after 1–3 days in culture, with pipette tips ranging from 3 MΩ to 5 MΩ from thin-borosilicate glass (TW150F-4, World Precision Instruments, Sarasota, FL). Using an Axopatch 200B amplifier (Axon Instruments), series resistance was compensated to 80%, and cells with a series resistance of >12 MΩ were omitted because of poor electrical access. pClamp 9.0 software (Axon Instruments) was used to acquire and store data on a desktop computer. Standard KCl pipette solution consisted of 140 mM KCl, 1 mM MgCl2, 0.2 mM CaCl2, 10 mM EGTA, and 10 mM HEPES sodium salt. The pipette solution was adjusted to an osmolarity of 302–304 mosM and pH 7.2 with Tris base. The bath solution consisted of 130 mM NaCl, 5 mM KCl, 1 mM CaCl2, 10.55 mM glucose, and 32.5 mM HEPES acid, and was adjusted to 308–312 mosM and pH 7.4 with NaOH. All recordings were performed in the presence of 2 μM paxilline (Santa Cruz Biotechnology) to block BK channels (38). 5-Nitro-2-(3-phenylpropylamino)benzoic acid (NPPB) was bath applied at 200 μM to block Cl− channels. Cells were incubated in myristoylated autocamtide-2-related inhibitory peptide (AIP; Enzo Life Sciences) for 10 min before recordings were performed.
We stably knocked down endogenous ClC-3 expression in D54 human glioma cells as previously described (8). A pGIPZ-lentiviral small hairpin mir vector containing a hairpin sequence targeting CLCN3 (Open Biosystems) was transfected with the Amaxa Biosystems nucleofection technique. After generating clonal populations, we screened multiple targeting constructs for ClC-3 knockdown by quantitative real-time (qRT-PCR). Multiple clones with ClC-3 knockdown were selected and probed for protein knockdown with Western blotting. We then selected a subset of clones to functionally test for cell migration, which requires ClC-3 activity (4). Experiments in this paper used a representative clone with the hairpin sequence 5′-TGCTGTTGACAGTGAG CGCCCTACCTCTTTCCAA AGTATAT AGTGAAGCCACAGATGTATATA CTTTGGAAAGAGGTAGGATGCC TACTGCCTCGGA-3′. We used a nontargeting sequence as a negative control: NT, 5′-TGCTGTTGACAGTGAGCGATCTCGCTTGG GCGA GAG TAAGTAGTGAAGCC ACAGATGTACTTACTCTCGCCCAAGCGA GAGTG CCTA CTGCCTCGGA-3′. Transfected cells were maintained on 10 μg/ml puromycin.
Western blot analysis.
Western blot analysis was performed as previously described (4). Protein concentrations were determined using a NanoDrop 1000 spectrophotometer (Thermo Scientific), and 15 μg of protein were loaded per well. Sample buffer (6×; 60% glycerol, 300 mM Tris pH 6.8, 12 mM EDTA, 12% SDS, 864 mM 2-mercaptoethanol, 0.05% bromophenol blue) was added to samples, which were then loaded on 10-well, 10% gradient precast SDS-polyacrylamide gels (Bio-Rad). After gels were run at 120 mV for 75 min, they were transferred to polyvinylidene difluoride paper (Millipore, Bedford, MA) at 200 mA for 120 min. Blots were then blocked with blocking buffer containing 10% nonfat dried milk and Tris-buffered saline with 0.1% Tween 20. Primary antibodies [rabbit anti-ClC-3 (Alpha Diagnostic International; 1:1,500), rabbit anti-ClC-3 (Alomone Labs; 1:1,500), and mouse anti-glyceraldehyde-3-phosphate dehydrogenase (Abcam; 1:5,000)] were incubated for 1 h at room temperature. After washes, blots were incubated in respective horseradish peroxidase-conjugated secondary antibodies at 1:1,500 (Santa Cruz Biotechnology). After further washes, blots were then developed with Supersignal West Femto (Thermo Scientific) or Luminol Reagent (Santa Cruz Biotechnology) and imaged on an Eastman Kodak Image Station 4000MM. ClC-3 immunoreactivity was normalized to glyceraldehyde-3-phosphate dehydrogenase.
Cells were placed on six-well plates at a density of 1 × 104. On day 0, a well was collected to normalize all future cell collections. Cell numbers were measured using a Coulter Counter Multisizer 3 (Beckman Coulter, Miami, FL). Drugs, including 40 μM NPPB and 2.5 μM KN-93 or vehicle, and media were changed every other day. Wells were measured in triplicate.
Cells were cultured on 12-mm glass coverslips (Macalaster Bicknell, New Haven, CT) in a 24-well plate for 1–3 days. Cells were then pulsed with 25 μM bromodeoxyuridine (BrdU) for 30 h. After 2 washes with PBS, the cells were fixed at room temperature with 4% paraformaldehyde for 15 min. Cells were again washed with PBS. DNA was then denatured with 2 M HCl in PBS with 0.3% Triton X to permeabilize the plasma membrane for 30 min at room temperature. Cells were then blocked with 10% normal goat serum and 0.3% Triton X-100 for 30 min at room temperature. Mouse anti-BrdU antibody (1:100, Molecular Probes, Eugene, OR) was then incubated overnight at 4°C. The next day, goat anti-mouse Alexa 350 (1:500, Invitrogen) was incubated for 2 h at room temperature. Coverslips were then washed in PBS and mounted on glass slides.
Electrophysiological data were quantified with Clampfit (Axon Instruments). Origin 6.0 (MicroCal, Northampton, MA) or Microsoft Excel was used to compile and graph data. GraphPad Instat (GraphPad Software) was used to perform statistical analyses. All data are reported as means ± SE.
Mitotic cell rounding in glioma cells is associated with CaMKII-dependent cytoplasmic condensation.
Recently, it was shown that mitotic cell rounding is associated with a decrease in cytoplasmic volume (8), a process termed premitotic condensation (PMC). In this paper, we hypothesize that CaMKII regulates Cl− efflux via ClC-3 channels, promoting mitotic cell rounding and cell division. To test this hypothesis, we transfected D54 human glioma cells, a patient-derived glioma cell line that is representative of glioblastoma cells (8), with GFP. We were able to quantitatively assess changes in cell volume by taking advantage of the fact that fluorescence of soluble GFP is inversely proportional to cell volume, as a constant number of GFP molecules are dissolved in the cytoplasm (9). Therefore a decrease in cell volume manifests as an increase in GFP intensity. Transfected glioma cells were imaged and monitored for spontaneous cell divisions, and a representative cell dividing under control conditions is depicted in Fig. 1A. For each cell, we designated the time point just before cell division as time = 0 and normalized all time points to time = 0 for each cell. Therefore, if a cell began changing volume at −x minutes, it indicates that x minutes before division, the cell began to change its volume. We found that as cells approached division, the relative intensity of GFP fluorescence increased, indicating a decrease in cytoplasmic volume (Fig. 1, A and B). Compiled averages are graphed in Fig. 1B, where the data were normalized to F0, which is the GFP intensity at time = 0. Each cell's fluorescence intensity (F) was normalized to its own fluorescence intensity at time = 0 (F0). As the cells approached division, the GFP fluorescence intensity (F/F0) rapidly increased, indicating cytoplasmic condensation as expected during PMC. Changes in GFP fluorescence were not attributed to pH, because as previously reported, no changes in pH were detected using the ratiometric pH indicator 2′,7′-bis-(2-carboxyethyl)-5(6)-carboxyfluorescein (BCECF) in dividing D54 glioma cells (8). The time taken for glioma cells to decrease volume by 50% (t50) was 28.4 ± 1.6 min (Fig. 1D). Interestingly, inhibition of CaMKII with 10 μM AIP significantly prolonged the time needed for PMC. AIP (10 μM) is specific for CaMKII and does not block PKC, PKA, or CaMKIV (12). When dividing glioma cells were incubated in AIP, the rate of condensation was slower (Fig. 1, A and B). While the minimum and maximum GFP intensities were not significantly different (Fmin= 0.4 ± 0.01 for control and 0.44 ± 0.02 for AIP; Fmax= 1.02 ± 0.02 for control and 1.01 ± 0.02 for AIP; Fig. 1C), the time taken for 50% volume condensation was significantly greater than control (39.6 ± 2.6 min; P < 0.001; Fig. 1D). Taken together, these data indicate that cytoplasmic condensation during mitotic cell rounding is facilitated by CaMKII activity.
ClC-3 and CaMKII strongly colocalize in dividing glioma cells.
We hypothesized that CaMKII facilitated PMC by increasing ClC-3 activity. ClC-3 localizes to the plasma membrane and cytoplasm in human glioma cells (4). Given the hypothesized activation of ClC-3 by CaMKII in dividing glioma cells, we labeled glioma cells with antibodies targeted against residues 592–661 of ClC-3 and residues 305–410 of CaMKII to determine whether these proteins were located in the same subcellular regions in dividing cells. We also labeled cells with Hoechst to visualize DNA, which condenses during M-phase. Images acquired as 0.5-μm optical sections with a spinning disc confocal microscope demonstrated ClC-3 (green) and CaMKII (red) colocalization as seen in a representative field-of-view (Fig. 2A). Digital zooms of individual cells (boxed cells in Fig. 2A) revealed robust colocalization of ClC-3 (green) and CaMKII (red) on the plasma membrane, especially in dividing M-phase glioma cells with condensed Hoechst-labeled DNA (blue; Fig. 2, B and C). Strong colocalization of ClC-3 and CaMKII was also seen throughout the 10-μm imaging stack as seen in three-view in the x,z (top) and z,y (left) planes (Fig. 2B). Thus ClC-3 and CaMKII are in a prime location to interact in mitotic cells. ClC-3 is expressed on the plasma membrane of dividing glioma cells, as previously demonstrated by colocalization with phalloidin-labeled cortical actin (8). Cortical actin provides a cytoskeletal framework for the plasma membrane and colocalizes with several membrane-bound proteins. We labeled cortical actin with phalloidin (Fig. 2; red) and found that it colocalizes with a fraction of CaMKII (Fig. 2; green), suggesting that CaMKII is also present at the plasma membrane, especially in dividing cells (Fig. 2, D and F). We then asked whether the CaMKII on the plasma membrane of dividing glioma cells was the activated form of the kinase. For CaMKII to be active, its autoinhibitory domain must be separated from the kinase domain. This is achieved through binding of Ca2+/calmodulin to the calmodulin-binding domain of CaMKII, allowing kinase activity (28). CaMKII can then autophosphorylate at the Thr286 position (pCaMKII) making the kinase active, independent of Ca2+/calmodulin binding. We used an antibody specific for phosphorylated Thr286 and found robust binding on the plasma membrane of dividing glioma cells as compared with nondividing interphase cells (green; Fig. 3). Dividing and nondividing cells expressed relatively little CaMKII in the nucleus (Fig. 2), but this CaMKII appears activated as it bound to pCaMKII antibody (Fig. 3). Dividing glioma cells with condensed chromatin were round and had enhanced pCaMKII labeling (green) throughout the 10-μm optical section in the x,z (top) and z,y (left) planes (Fig. 3B) as compared with nondividing cells (Fig. 3C). These data cumulatively indicate that activated CaMKII and ClC-3 colocalize in the proper cellular domains where this could facilitate premitotic condensation.
Dividing glioma cells have large CaMKII-dependent Cl− currents.
Given that cytoplasmic condensation is facilitated by CaMKII, which strongly colocalizes with ClC-3 on the plasma membrane of dividing cells, we asked whether CaMKII modulated ClC-3 activity in mitotic cells. Mitotic cells have previously been demonstrated to have larger Cl− currents, allowing for enhanced Cl− efflux and obligated osmotic water release to potentiate cytoplasmic volume condensation (9). However, how Cl− currents become activated in dividing cells has been unknown. We performed whole cell patch-clamp electrophysiology on human glioma cells to assess Cl− channel activity. We patched onto nondividing and dividing glioma cells, holding the cells at −40 mV and stepping from −100 mV to +120 mV in 20-mV increments (Fig. 4 column a). Dividing M-phase cells were identified by Hoechst-labeled chromatin condensation and cell morphology, as previously described (8). We then washed on 200 μM NPPB, a nonspecific anion channel blocker (Fig. 4 column b), to assess NPPB-sensitive Cl− channel activity (Fig. 4 column a–b). Interestingly, dividing glioma cells had larger NPPB-sensitive Cl− currents (Fig. 4B) as compared with nondividing control cells (Fig. 4A). At 100 mV, the condensed, dividing cells have a current amplitude of 19.4 ± 4.4 pA/pF, which is significantly greater than nondividing cells (2.6 ± 1.7 pA/pF; P < 0.01; Fig. 4F). These NPPB-sensitive Cl− currents were slightly outwardly rectifying and inactivated at depolarized potentials, similar to Cl− currents attributed to ClC-3 (32), and reversed at ∼0 mV as predicted by symmetric Cl− concentrations in the pipette and bath solutions. However, when glioma cells were preincubated in 10 μM AIP, Cl− current amplitudes were back to baseline levels in dividing (7.2 ± 1.7 pA/pF; P < 0.05; Fig. 4, C and F) and nondividing cells (4.8 ± 2.2 pA/pF; P < 0.01; Fig. 4, D and F). This indicates that the larger Cl− currents seen in dividing cells are CaMKII dependent. As seen in Fig. 4E, dividing cells have approximately threefold elevated NPPB-sensitive Cl− currents, as compared with nondividing cells, or cells with inhibited CaMKII activity (corresponding with Fig. 4, column a–b). Thus dividing cells have enhanced CaMKII-dependent Cl− currents.
ClC-3 knockdown eliminates CaMKII-dependent Cl− currents in dividing glioma cells.
To determine whether CaMKII specifically enhanced PMC (Fig. 1) and activated Cl− currents (Fig. 4) by enhancing ClC-3 activity, we stably knocked down endogenous ClC-3 expression in glioma cells using stably expressed short hairpin RNA (shRNA). Glioma cells were transfected with ClC-3 shRNA in a pGIPZ lentiviral vector. Clonal populations were then screened for ClC-3 knockdown with qRT-PCR. Multiple clones with multiple targeting sequences were then isolated, and ClC-3 protein expression and activity were assessed. As seen in a representative Western blot, ClC-3 protein expression (∼120 and 100 kDa bands) was robustly knocked down in cells transfected with ClC-3 shRNA as compared with cells transfected with nontargeting shRNA (Fig. 5A). The smear and multiple adjacent bands in the NT shRNA lane are due to posttranslational modifications of ClC-3. As previously reported, the multiple bands of ClC-3, including the band at 120 kDa, correspond to multiple glycosylation states of ClC-3 (29). Therefore it appears that knockdown of ClC-3 decreased protein expression and the associated glycosylation. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control (∼40 kDa band). We quantified ClC-3 protein knockdown by normalizing ClC-3 band intensity to GAPDH band intensity (relative band intensity; Fig. 5, B and C). Probing with the ClC-3 antibody from Alomone Labs targeted to an intracellular domain demonstrated decreased band intensity to 0.2 ± 0.05 as normalized to NT shRNA (Fig. 5B). Probing with a ClC-3 antibody from Alpha Diagnostic International targeting an extracellular domain demonstrated decreased band intensity to 0.44 ± 0.09 as compared with control (Fig. 5C).
After knocking down ClC-3 protein expression, we again performed whole cell patch-clamp electrophysiology 1) to determine whether the larger Cl− currents in dividing cells were mediated by ClC-3 and 2) to determine whether the CaMKII-dependent Cl− currents in dividing cells were mediated by ClC-3. We patched onto ClC-3 knockdown cells and stepped from −100 mV to +120 mV in 20-mV increments (Fig. 6 column a). NPPB (200 μM) was then washed on to inhibit Cl− currents (Fig. 6 column b), and the NPPB-sensitive Cl− currents were subtracted out (Fig. 6 column a-b). After knocking down ClC-3, there was no elevation of Cl− currents in dividing M-phase cells as compared with nondividing interphase cells (Fig. 6, A, B, and E). Importantly, this indicates that the increase in Cl− current activity in dividing cells (Fig. 4) is specifically mediated by ClC-3. Additionally, CaMKII inhibition did not eliminate Cl− currents after ClC-3 knockdown (Fig. 6, C–E), indicating that CaMKII specifically acts at ClC-3 to increase Cl− currents in dividing glioma cells. After ClC-3 knockdown there was no significant difference in Cl− current density at 100 mV in dividing versus nondividing cells (3.9 ± 1.5 pA/pF vs. 0.3 ± 1.3 pA/pF, respectively) or in the presence of 10 μM AIP in dividing versus nondividing cells (−3.0 ± 1.4 pA/pF vs. 0.1 ± 0.9 pA/pF, respectively; Fig. 6F). These data indicate that ClC-3 is necessary for CaMKII-activated Cl− currents in dividing glioma cells.
CaMKII activation of ClC-3 facilitates premitotic condensation.
After finding that CaMKII activates ClC-3 to enhance Cl− currents in dividing glioma cells, we determined whether CaMKII activation of ClC-3 also facilitates premitotic condensation. Using GFP cells transfected with NT shRNA and ClC-3 shRNA (Fig. 5), we monitored premitotic condensation associated with mitotic cell rounding. As compared with NT shRNA-transfected cells, ClC-3 knockdown cells took significantly more time to condense cytoplasmic volume preceding cell division (Fig. 7, A and B). Additionally, incubation of control cells in AIP, a CaMKII inhibitor, significantly prolonged PMC to the same level as ClC-3 knockdown (Fig. 7, A and B). Importantly, inhibition of CaMKII in ClC-3 knockdown cells did not lead to a further prolonging of PMC (Fig. 7, A and B), indicating that CaMKII inhibition prolongs PMC via its interaction with CaMKII. While ClC-3 inhibition and/or CaMKII inhibition did not lead to a significant change in Fmax, there was a slight change in Fmin indicating that ClC-3 and CaMKII may function in maintenance of the basal size of glioma cells (Fig. 7C; for Fmin NT shRNA = 0.41 ± 0.01, NT shRNA + AIP = 0.36 ± 0.01, ClC-3 shRNA = 0.40 ± 0.01, and ClC-3 shRNA + AIP = 0.45 ± 0.01; P < 0.01). Significantly, there was an increase in the time taken for PMC upon ClC-3 or CaMKII inhibition. The t50 increased from −29.3 ± 0.7 min in NT shRNA cells to −38.2 ± 0.7 min in ClC-3 knockdown cells, −43.5 ± 1.5 min in control cells incubated in 10 μM AIP, and −37.9 ± 1.2 min in ClC-3 knockdown cells incubated in AIP (Fig. 7D; P < 0.001). These data strongly indicate that CaMKII activates ClC-3 to increase Cl− currents in dividing cells and facilitate premitotic condensation.
ClC-3 and CaMKII inhibition reduces glioma cell proliferation.
Experimentation at the single-cell level demonstrates that CaMKII activation of ClC-3 facilitates premitotic condensation. Given that CaMKII and ClC-3 inhibition impairs premitotic condensation, we hypothesized that this would also impede glioma cell proliferation. We determined population growth by counting cells each day over a 6-day period. Cells were cultured with vehicle, 40 μM NPPB to block Cl− channels, 2.5 μM KN-93 to block CaMKII, or 40 μM NPPB and 2.5 μM KN-93. KN-93, a small-molecule inhibitor, was used instead of AIP, which is a peptide sensitive to degradation during long incubations at 37°C. Cl− channel inhibition decreased glioma cell number at day 6 to 38.9 ± 5.34 cells as compared with 62.75 ± 4.67 cells in the control condition. CaMKII inhibition reduced the number of cells at day 6 to 36.24 ± 5.68, and simultaneous Cl− channel and CaMKII inhibition reduced the number of cells to 12.8 ± 1.12 (Fig. 8A; P < 0.05). Thus, Cl− channels and CaMKII appear to facilitate glioma cell proliferation. Simultaneous Cl− channel and CaMKII inhibition decreased proliferation by day 6 to numbers below that of Cl− channel blockade or CaMKII blockade alone (Fig. 8A; P < 0.05). This may be due to CaMKII regulation of cell division independent of ClC-3. CaMKII inhibits p21, leading to p53 degradation and increased proliferation (17). Increases in intracellular calcium concentration ([Ca2+]i) that activate CaMKII can also lead to phosphorylation of CDC25C, which in turn dephosphorylates CDC2 to trigger entry into mitosis (23, 33). Thus CaMKII may have pleiotropic effects on cell cycle progression independent of Cl− channels.
Additionally, NPPB is not a specific inhibitor of ClC-3; NPPB blocks intermediate-conductance K+ channels, which can also regulate proliferation (4a, 37). Therefore, we also studied the role of ClC-3 in glioma cell proliferation using a genetic approach (Fig. 5). At day 6, ClC-3 knockdown decreased proliferation from 53.59 ± 3.96 cells in control conditions to 34.46 ± 4.85 cells (Fig. 8B). Additionally, CaMKII inhibition with 2.5 μM KN-93 significantly decreased glioma cell proliferation to the same levels as ClC-3 inhibition at day 6 (Fig. 8B; P < 0.01). This provides further indication that CaMKII facilitates proliferation by activating ClC-3.
We also measured proliferation by pulsing GFP-expressing glioma cells with BrdU, a synthetic analog of thymidine, for 30 h. We then fixed the cells (Fig. 8C; green) and probed for BrdU incorporation (Fig. 8C; blue). In control glioma cells, nearly all glioma cells labeled for BrdU, indicating that almost all cells had replicated DNA at least once during the 30-h BrdU pulse period (Fig. 8C; NT shRNA arrows). However, upon ClC-3 knockdown, there was a significant increase in the number of cells not incorporating BrdU during the 30-h pulse period (Fig. 8C; ClC-3 shRNA arrows). The number of cells not incorporating BrdU increased from 4.0 ± 1.9% in control conditions to 20.1 ± 6.2% in ClC-3 knockdown cells (Fig. 8D; P < 0.05). However, CaMKII inhibition with 2.5 μM KN-93 did not significantly increases the percentage of nonproliferating BrdU-negative cells, even after ClC-3 knockdown (Fig. 8D; NT shRNA + KN-93 = 5.8 ± 2.2% and ClC-3 shRNA + KN-93 = 17.5 ± 3.8%; P > 0.05), indicating that the decreases in cell number due to CaMKII inhibition (Fig. 8, A and B) may partially be attributed to ClC-3-independent regulation. Interestingly, most of the nonproliferating cells induced by ClC-3 knockdown appeared larger and flatter than dividing cells (Fig. 8C). This provides further evidence that ClC-3, which is activated by CaMKII, facilitates premitotic condensation and glioma cell division.
Here we demonstrate that CaMKII activation of ClC-3 facilitates premitotic condensation in human glioma cells. ClC-3 and CaMKII robustly colocalize on the plasma membrane of dividing cells, and specifically, auto-activated CaMKII appears more abundant in M-phase cells versus interphase cells. Inhibition of CaMKII prolonged cellular condensation preceding cell division, but only when ClC-3 was expressed. Importantly, the condensation of glioma cells was associated with enhanced Cl− currents mediated by CaMKII activation of ClC-3. These Cl− currents are associated with shape and volume changes in glioma cells (26) and facilitate cytoplasmic volume decrease during PMC (8).
A link between mitotic cell rounding and premitotic condensation.
An elegant study recently demonstrated that mitotic cell rounding is driven by two opposing forces: an inward contraction of the actomyosin cytoskeleton versus an outwardly directed osmotic pressure (35). Disruption of either process perturbed the rounding ability of mitotic cells (35). Data presented here indicate that CaMKII activation of ClC-3 may be a source for this outward osmotic pressure, at least in glioma cells. Activation of ClC-3 by CaMKII facilitates cytoplasmic condensation during mitotic cell rounding (Figs. 1 and 7) and is associated with larger Cl− currents (Figs. 4 and 6). As osmotically active Cl− exits the cell via ClC-3, water is drawn out of the cell, leading to cytoplasmic volume condensation (7). Glioma cells accumulate [Cl−]i to 100 mM through Na+-K+-Cl− cotransporter 1 (NKCC1) activity, creating an outwardly directed gradient for Cl− movement (6). This gradient can be used to drive water out of the cell, creating an outward osmotic pressure and facilitating PMC (7). While we did not assess the contribution of cytoskeleton contraction to PMC in glioma cells, our data indicate that PMC is facilitated by efflux of osmotically active ions in glioma cells, and through a similar mechanism, different channels/transporters that flux osmotically active ions may be of importance in other proliferative cells.
Actomyosin contraction and hydrodynamic volume changes may work in concert to enhance macromolecular crowding.
Considerable attention has been given to the role of cytoskeletal dynamics during mitotic cell rounding. Actomyosin contraction is essential to generate an inward pressure to balance the outward osmotic pressure generated in rounding mitotic cells (35). Additionally, RhoA, a small GTPase, activates Rho kinase, which phosphorylates myosin II regulatory light chain leading to cortical rigidity and retraction in mitotic cells (19). Independent of myosin II activity, moesin proteins link the actin cortex to the plasma membrane, inducing cell rounding and membrane rigidity in mitotic cells (14). Several other actin-binding proteins such as l-caldesmon, actin interacting protein 1 (AIP1), and spectrin also play an integral role in actin dynamics in dividing cells (5). The importance of mitotic cell rounding may lie in its ability to form a symmetric mitotic spindle, assuring that both daughter cells receive the correct set of chromosomes with high fidelity (30). While these studies elucidate a mechanism through which cells can attain a spherical morphology via actomyosin contraction, it does not account for intracellular contents: do cells maintain the same cytosolic composition, or do cells modify this intracellular milieu? Here we find that glioma cells condense cytoplasmic volume by extruding Cl− ions, allowing for the osmotic release of water and cellular volume decrease. Thus, given that ∼70% of a cell is water (18), channels extruding osmotically active ions contribute to volume condensation and may work in concert with actomyosin contraction to control PMC during mitotic cell rounding. While the interaction between ClC-3 and CaMKII appears to accelerate PMC in glioma cells, it is likely that alternate channels fluxing osmotically active ions may facilitate PMC and cell division in other cell types. Additionally, while we found that inhibition of CaMKII activation of ClC-3 impedes PMC, cells (though fewer in numbers, see Fig. 8) nevertheless condensed and progressed through M-phase. This indicates that PMC is integral to cell division, and dividing cells have redundant mechanisms to ensure PMC occurs.
Volume condensation in dividing cells may allow for the bridging of kinases to their substrate proteins. As seen in Fig. 2, while ClC-3 and CaMKII do colocalize in nondividing cells, the colocalization is enhanced in condensed dividing cells. Thus, condensation of the cytoplasm may increase macromolecular crowding, leading to enhanced interaction between kinases and their substrate proteins. Changes in cellular volume can lead to shifts in actin polymerization/depolymerization or protein regulation, including activation of membrane ion channels (15). Several kinases become activated upon cellular condensation, including protein kinase C (16), which can then phosphorylate and activate the Na+/H+ exchanger in tumor cells (24). Additionally, Ca2+/calmodulin-dependent protein kinases phosphorylate NKCC, a Na+-K+-Cl− cotransporter, enhancing activity after cell shrinkage (22). Thus PMC, by mechanically shrinking the cytoplasmic space, may increase macromolecular crowding and interaction of kinases with their substrate proteins. We found a significant increase in the number of cells that did not incorporate BrdU and undergo mitosis after ClC-3 knockdown (Fig. 8). These nondividing cells were typically larger and flatter than dividing cells and did not undergo PMC after a 30-h period. This is in agreement with data from rat C6 glioma cells demonstrating that there are low and high cell size checkpoints for proliferation (31). Therefore, cells that cannot undergo PMC, such as after ClC-3 knockdown, may be senescent and larger. In a number of other cell types, including fibroblasts, senescent cells are larger and flatter as compared with dividing cells (1). An understanding of whether these cells lack mechanisms for mitotic cell rounding and PMC could yield further insight into cytoskeletal and hydrodynamic changes associated with cell division.
Therapeutically targeting CaMKII activation of ClC-3 in human glioma.
Identifying the mechanisms by which human glioma cells divide may lead to the development of novel therapeutic targets. Interference with CaMKII activation of ClC-3 in human glioma cells decreases glioma cell proliferation (Fig. 8), so further investigation of these proteins may lead to the development of novel therapeutics. Directly inhibiting Cl− channel activity in glioma cells is a promising therapeutic strategy. This has been achieved with chlorotoxin, a toxin in the venom of Leiurus quinquestriatus (deathstalker scorpion), which recently completed Phase II trials for the treatment of glioblastoma multiforme. Preclinical studies demonstrate that chlorotoxin inhibits Cl− channel activity in glioma cells (21), leading to impaired glioma cell migration (34). In Phase I studies, chlorotoxin demonstrated exquisite specificity, binding glioma tissue after intracavitary injection into patients with recurrent glioma (20). By inhibiting Cl− channel activity, chlorotoxin decreases glioma cell migration, and our data suggest that it may also be effective at inhibiting PMC and glioma cell proliferation.
We demonstrate here that ClC-3 activity in glioma proliferation is regulated by CaMKII. CaMKII is a Ca2+-sensitive kinase; therefore, CaMKII activation of ClC-3 can be inhibited by identifying upstream Ca2+ sources for CaMKII activation. One candidate expressed by gliomas is mutant epidermal growth factor (EGF) receptor (EGFR). The EGFRvIII mutation, caused by a deletion of exons 2–7 of the EGFR gene, is the most common and leads to constitutive EGFR signaling, enhancing proliferation (10). Given that, in epithelial cells, EGF stimulation leads to increase in [Ca2+]i and subsequent CaMKII activation (36), EGFRvIII mutations in gliomas may drive constitutive CaMKII activation. Additionally, transient receptor potential (TRP) canonical 1 (TRPC1) channels play an important role in Ca2+ dynamics and glioma cell division, as inhibition of TRPC1 in glioma cells impedes store-operated Ca2+ entry and proliferation (2). Other TRP channels, including TRPC3 and TRP vanilloid type 1 (TRPV1), increase [Ca2+]i to enhance CaMKII activity (3, 13), so an understanding of whether TRPC1-mediated Ca2+ influx activates ClC-3 via CaMKII could lead to novel mechanistic insight. During mitosis, several cell types transiently increase [Ca2+]i, including Pt K2 epithelial cells, which spike [Ca2+]i during metaphase for about 20 s (27). Clearly, future studies should investigate the upstream Ca2+ signaling that regulates the CaMKII-ClC-3 interaction during mitosis. These studies may reveal novel therapeutic targets to improve clinical management of gliomas.
The authors are grateful for the following grants from the National Institutes of Health (National Institute of Neurological Disorders and Stroke): RO1 NS-31234, RO1 NS-52634, and RO1 NS-36692 (to H. Sontheimer) and F31 NS-73181 (to V. A. Cuddapah).
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: V.A.C., C.W.H., and H.S. conception and design of the research; V.A.C., C.W.H., S.W., L.S.M., and T.-T.C.B. performed the experiments; V.A.C., C.W.H., S.W., L.S.M., and T.-T.C.B. analyzed the data; V.A.C., C.W.H., S.W., L.S.M., T.-T.C.B., and H.S. interpreted the results of the experiments; V.A.C. and L.S.M. prepared the figures; V.A.C. drafted the manuscript; V.A.C. and H.S. edited and revised the manuscript; V.A.C. and H.S. approved the final version of the manuscript.
- Copyright © 2012 the American Physiological Society