Ca²⁺-dependent and -independent mechanisms of calmodulin nuclear translocation

Calmodulin (CaM) is a ubiquitous Ca²⁺-binding protein that exists as a monomer of 16.8 kDa and belongs to the EF-hand homolog family, which includes 160 different Ca²⁺ binding proteins (Persechini et al., 1989; Pruschy et al., 1994). CaM can transduce a Ca²⁺ signal into a cellular response to regulate processes in the cytoplasm and in the nucleus such as cytoskeletal structure changes, metabolic activity, gene expression, synaptic plasticity and nitric oxide synthesis (Mayer et al., 1992; Teruel et al., 2000; Deisseroth et al., 2003). During cell division, following nuclear entry, CaM remains associated with the mitotic apparatus (Török et al., 1998a; Li et al., 1999; Yu et al., 2004). A high concentration of CaM has been found in the nucleus of all cell types and it is believed that CaM functions through regulating the activities of CaM-binding proteins located there at high concentration (Bachs et al., 1994).

The mechanism of molecular transport into the nucleus is not fully understood, but is thought to occur via the nuclear pore complex (NPC), which regulates nucleocytoplasmic trafficking of ions, proteins and RNA in eukaryotic cells (Liao et al., 1999; Wang and Clapham, 1999). Molecular transport can occur via active or passive mechanisms depending on the size of the molecule, requirements for energy and the presence of a nuclear localization signal (NLS) (Perez-Terzic et al., 1999). It has been reported that molecules no larger than 45-60 kDa (depending on the cell type) can transverse the NPC by passive diffusion, whereas larger molecules that possess an NLS sequence require energy-dependent multi-step processes to enter and exit the nucleus (Allen et al., 2000; Perez-Terzic et al., 1999; Weiss, 1998). For a detailed review of NLS-dependent transport see Allen et al. (Allen et al. 2000). Although the translocation of NLS-containing proteins has been extensively studied, little is known about the mechanisms that regulate passive diffusion. However, pore permeability can be regulated by emptying the Ca²⁺ stores of the endoplasmic reticulum (ER), which then prevents the passive diffusion of molecules larger than 10 kDa (Greber and Gerace, 1995; Lee et al., 1998; Perez-Terzic et al., 1999; Strübing and Clapham, 1999). The blockade of medium-sized molecules can be reversed by adding Ca²⁺ and ATP to a bath, containing isolated nuclei (Wang and Clapham, 1999). Depletion of Ca²⁺ is thought to be sensed by the nucleoporin gp210 that, through its large lumenal domain, containing several putative Ca²⁺-binding sites, could induce conformational changes on the NPC (Perez-Terzic et al., 1997). According to Greber and Gerace, NLS-signal-mediated transport cannot occur when Ca²⁺ is released from intracellular stores (Greber and Gerace, 1995). However, Strübing and Clapham found that in HM1 cells (a human embryonic kidney cell line) ER-store depletion by carbachol, thapsigargin, ionomycin or 1,2-bis-(o-aminophenoxy) ethane-N,N,N′,N′-tetraacetic acid-acetoxymethyl ester (BAPTA-AM) had no effect on the nuclear import of the NLS-containing mitogen-activated protein (MAP) kinase-activated protein kinase 2 (MK2) (Strübing and Clapham, 1999).

Previous studies, investigating how CaM enters the nucleus have so far produced conflicting data. CaM lacks an obvious NLS motif and, because of its small size, import could occur by passive diffusion, depending on the diameter of the nuclear pores. In permeabilized or intact A7r5 cells, CaM translocation into the nucleus is not blocked by wheat germ agglutinin (WGA) or chilling, and does not require cytosolic factors or an ATP regenerating system (Liao et al., 1999). Also, nuclear accumulation of CaM is blocked by a CaM antagonist peptide, M13, and the amount of CaM in the nucleus increases with increasing concentrations of free Ca²⁺. CaM accumulation in the nucleus was also shown to be similar to the rate of uptake of a 20-kDa fluorescein-dextran (FL-dextran), consistent with a first order process. It was therefore concluded that nuclear entry is owing to passive diffusion and the steady state concentration of CaM in the nucleus depends on both the intracellular concentration of free Ca²⁺ and the relative concentration and affinity of CaM-binding proteins (Liao et al., 1999). However, Pruschy et al. found that CaM import is inhibited by WGA and chilling, but that ATP depletion did not affect nuclear uptake in PtK1 cells and this indicated a facilitated but not active transport mechanism (Pruschy et al., 1994). It was therefore suggested that CaM could be in a complex with a CaM-binding protein that was too large to enter the nucleus by diffusion. This idea was supported by evidence gained from experiments in smooth muscle cells that only 5% of total endogenous CaM was free, even at resting intracellular Ca²⁺, and the observed hindered diffusion of CaM was owing to a pool of CaM that was tightly bound to a target. Upon elevation of intracellular Ca²⁺ levels, a large portion of CaM was released and was thought to bind again, this time to Ca²⁺-dependent CaM binding sites in the nucleus (Luby-Phelps et al., 1995). Diesseroth et al. demonstrated that, in hippocampal neurons, brief bursts of activity caused fast translocation (within approximately 1 minute) of CaM from the cytoplasm to the nucleus (Diesseroth et al., 1998). Rapid chilling abolished this rapid translocation of CaM, although a persistent and slow leakage of CaM into the nucleus was observed for a long time after the stimulus. It was concluded that CaM translocation into the nucleus was not merely by diffusion but facilitated. Mermelstein et al. suggested that CaM interacts with Ca²⁺/CaM-dependent protein kinase kinase (CaMKK) for translocation (Mermelstein et al., 2001). In hippocampal neurons, the CaM kinase inhibitor KN-93 blocked CaM translocation into the nucleus, suggesting that CaM kinases or an isoform of CaMKK is required for CaM translocation. Furthermore, translocation of a fluorescently tagged CaM was prevented by fusing it with M13 peptide. This further suggested that CaM requires a Ca²⁺-dependent association with an endogenous CaM-binding protein to allow CaM translocation into the nucleus.

First, we used electroporated HeLa, cortical glial and neuronal cells to study the permeability of the nuclear pores using differently sized FL-dextrans. Second, we investigated CaM translocation into the nucleus and binding to nuclear targets in HeLa cells using a novel rhodamine-labeled CaM probe. We demonstrate that CaM entry into the nucleus occurs via two separate mechanisms, of which one is Ca²⁺-dependent and one is Ca²⁺-independent.

The expression vector pAED₄, containing CaM cDNA of human liver subcloned in the restriction sites NdeI for 5′ and PstI for 3′, was a kind gift from S. Marston (Imperial College, London, UK). CaM was expressed using standard procedures and purified as described by Török and Trentham (Török and Trentham, 1994). The purified CaM was desalted on a Sephadex PD-10 column (Amersham Biosciences) and then lyophilized. The typical yield was 20 mg/l culture. A mutant form of CaM (CaM1234) with mutations in the four Ca²⁺-binding lobes that prevent Ca²⁺-binding (Kasri et al., 2003), was prepared like CaM and labeled with 6ROX similarly to wild-type CaM (data not shown).

Human liver CaM expressed in Escherichia coli was labeled at a concentration of 1 mg/ml (60 μM) in 200 mM Tris-HCl pH 8.5, containing 20 mM CaCl₂ with 120 μM 6-carboxy-X-rhodamine, succinimidyl ester (6ROX, SE) (Molecular Probes). Labeling was carried out in the dark for 1 hour at 22°C for the formation of Lys75-6-carboxy-X-rhodamine-labeled CaM (6ROX-CaM) and the reaction was terminated by gel filtration on a PD10 column equilibrated with H₂O. Purification by HPLC was carried out as described in Török et al. with absorption measured at 215 nm and fluorescence monitored at 568 nm excitation and 602 nm emission (Török et al., 2002). In HPLC analysis the second of two fluorescent peaks corresponded to Lys75-labeled 6ROX-CaM. Absorption was measured at 575 nm and the extinction coefficient ϵ_o for 6ROX-CaM was taken as 82,000 M^–1cm^–1 (Molecular Probes).

Digestion mixtures consisted of 100 mM NH₄HCO₃ pH 9.0, 2 mM EGTA, 1 mg/ml 6ROX-CaM and 10 μg/ml trypsin. All digestions were performed at 37°C for 20 hours.

Sinapinic acid was used as a matrix for the ionisation of 6ROX-CaM. α-cyano-4-hydroxycinnamic acid was used as a matrix for the ionisation of polypeptides from the tryptic digest of 6ROX-CaM. A saturated solution of 10 mg α-cyano-4-hydroxycinnamic acid in 700 μL acetonitrile and 300 μL 0.1% TFA in an amber eppendorf and left to settle under gravity. Samples were desalted using C₁₈ ZipTips pipette tips (Millipore) and 0.5 μl of the appropriate matrix was loaded onto the MALDI-TOF analysis plate with 10 pmol 6ROX-CaM or 5 pmol of peptide sample. Samples were analysed on an Axima-CFR MALDI-TOF mass spectrometer (Kratos Analytical); the system was calibrated with angiotensin II (1046.2 Da) for tryptic digest analysis and human liver CaM (16706.4 Da) for undigested CaM analysis.

The fluorescence of 6ROX-CaM was measured spectrofluorometrically (SLM Aminco®) upon titration of free Ca²⁺. Excitation was set to 568 nm and emission was measured at 605 nm. Measurements were performed in a solution of 50 mM 2-(N-morpholino) ethanesulfonic acid K⁺ salt (K-MES) pH 7.0, 100 mM KCl, 2 mM MgCl₂ and 5 mM EGTA at 21°C, containing 100 nM 6ROX-CaM. Tyr fluorescence of 100 nM unlabeled CaM and CaM1234 was measured in the same buffer. Excitation was set to 278 nm and emission was measured at 305 nm. [Ca²⁺] concentrations were calculated as described before (Tzortzopoulos et al., 2004).

A cell permeant CaM-binding peptide based on the acetylated Trp peptide (Ac-RRKWQKTGHAVRAIGRL-CONH₂) (Török and Trentham, 1994; Török et al., 1998a; Török et al., 1998b) was generated by N-terminal myristoylation instead of acetylation. This myristoylated peptide (mTrp) was HPLC-purified to homogeneity and its K_d for 2-chloro-(ϵ-amino-Lys75)[6-4-(N,N-diethhylamino)phenyl]-1,3,5-triazin-4-yl-CaM (TA-CaM) was determined by stopped-flow kinetic experiments in a Hi-Tech PQ/SF-51MX multimixing stopped-flow spectrofluorimeter (Hi-Tech Scientific, UK) in the solution conditions above with the exception of EGTA being replaced by 100 μM CaCl₂.

A cell permeant analogue of the Ca²⁺/CaM-dependent protein kinase II (CaMKII) inhibitor AIP, myristoyl-AIP (mAIP, myristoyl-NKKALRRQGAVDAL-OH) (Ishida et al., 1995) from Calbiochem was also tested as an inhibitor of CaM nuclear translocation to see if CaMKII was involved in the translocation mechanism.

Unlabeled, human liver CaM that had been expressed in E. coli, 6ROX-CaM and 6ROX-CaM1234 were measured and their activating effects on CaM-free cAMP PDE of the heart were compared. One activator unit of CaM is defined as the amount of CaM that is necessary to raise the activity of 3 mU of Ca²⁺-free PDE to 50% of its maximum at CaM saturation. Measurement value was the decrease in adenosine extinction at 265 nm. For detailed methodology see Török et al. (Török et al., 1998b).

HeLa cells (ATCC lot number 2056308) were cultured in minimum essential medium (MEM) (Sigma) supplemented with 2 mM L-glutamine, 1.5 g/l sodium bicarbonate, 0.1 mM non-essential amino acids, 1 mM sodium pyruvate, 100 U/ml penicillin and 100 μg/ml streptomycin and 10% heat inactivated fetal bovine serum. Cells were maintained in a humidified atmosphere of 5% CO₂ at 37°C and were passaged every 3 days. For imaging, cells were grown on 35×10 mm culture dishes and synchronized with 10 ng/ml aphidicolin.

Cortical neurons and glial cells were prepared from neonatal rat pups (Wistar) using a papain based dissociation method. Cells were grown on 25 mm glass coverslips coated with 10 μg/ml poly-L-lysine and 1 μg/ml laminin and incubated in a humidified atmosphere (95% air and 5% CO₂). Cells were grown in basal medium Eagle (BME) with Earle's salts, without glutamine (Gibco BRL, UK) without L-glutamine and supplemented with Earle's salts, N₂ supplement, 10% horse serum, 20 mM glucose, 100 U/ml penicillin and 100 μg/ml streptomycin. Cells were used for imaging after 6-10 days of culture.

Cells were loaded by electroporation as described before (Török et al., 2002). For a detailed review on electroporation see Teruel and Meyer (Teruel and Meyer, 1997). HeLa, glial and neuronal cells were washed and kept in electroporation buffer, containing 135 mM NaCl, 5 mM KCl, 1.3 mM MgSO₄, 10 mM glucose, 2 mM EGTA and 20 mM HEPES (pH 7.4) for 15 minutes. Just before electroporation, cells were put into 1 ml electroporation buffer to which 0.5 μM tetrodotoxin (TTX) was added and cells were electroporated with 6ROX-CaM, 6ROX-CaM1234 and/or 4.3-77 kDa FL-dextrans (Molecular Probes) (estimated final concentration 4 μM) or fluo-4 (Molecular Probes) (estimated final concentration 3 μM) by applying two 25 ms 105 V cm^–1 pulses of opposite polarity. Cells were then washed and put back into electroporation buffer for 15 minutes before imaging.

To inhibit facilitated transport, cells were electroporated with WGA to an estimated final concentration of 20 μg/ml. In CaM inhibitory experiments, 10 μM mTrp peptide was added to electroporation buffer and the cells were incubated for 15 minutes before stimulation

To observe Ca²⁺-dependent CaM translocation, cells that had been kept in 2 mM EGTA were stimulated by the addition of 2.1 mM Ca²⁺ to the external medium. Intracellular Ca²⁺ transients were induced in HeLa cells by the addition of 50 μM histamine to the external medium (Bootman et al., 1997).

Where specified, as a means of stimulation, 4 mM Ca²⁺ was added to the cells in electroporation buffer to give a final concentration of 2 mM Ca²⁺ and observed for the specified amount of time up to 1 hour.

Cells were imaged on a Leica SP system with a 63×0.9 water immersion lens, using a 568 nm krypton laser to excite 6ROX-CaM and a 488 nm argon laser to excite fluo-4 and FL-dextrans. Images were taken at 7-second intervals for 1400 seconds at 22°C in electroporation buffer (SB) and had a box size of 512×512 pixels.

Results are expressed as the mean±s.e.m. and n=number of experiments. All statistical analysis was performed using the parametric test ANOVA-one way. Significant difference was accepted at P<0.01.

Human liver CaM expressed in E. coli was labeled with 6-ROX, in conditions that had previously been used to generate Lys75-labeled TA-CaM and FL-CaM (Török and Trentham, 1994; Török et al., 1998a; Török et al., 2002). The reagent 6ROX was chosen to simultaneously image changes in intracellular [Ca²⁺] by fluo-4 and CaM. Additionally, 6ROX was expected to be more photostable than fluorescein, which had previously been used for FL-CaM and is excited by visible, rather than UV light. MALDI-TOF analysis of HPLC-purified 6ROX-CaM (Fig. 1A) showed that labeling occurred at a single site (Fig. 1B). Analysis of the tryptic digest of 6ROX-CaM revealed that the Lys75-Lys77 and the Lys75-Arg86 peptides were bound to 6ROX (Fig. 1C,D). Post-source decay sequence analysis identified Lys75 as the labeled residue (data not shown) as observed previously for TA-CaM and FL-CaM. No other masses were found in the 6ROX-CaM tryptic digests to suggest that 6ROX was bound to any other Lys residues. Similar results were obtained for the CaM mutant 6ROX-CaM1234, in which labeling also occurred at Lys75 (data not shown).

PDE was activated by 6ROX-CaM with similar K_m and V_max values as unlabeled CaM expressed in E. coli. The K_m for CaM was 6.5±0.7 nM and 8.3±0.9 nM for 6ROX-CaM (Fig. 1E,F). 6ROX-CaM1234, however, did not activate PDE. 6ROX-CaM showed a small (6 nm) hypsochromic shift in emission (604 to 598 nm) upon the increase of free Ca²⁺ to 1.74 μM, fluorescence was increased 1.46-fold (Fig. 1G). The addition of more Ca²⁺ had no further effect on the fluorescence, which reduced to base level again upon the addition of 4 mM EGTA (data not shown). Unlabeled CaM showed a 3-fold increase in the Tyr fluorescence upon increasing [Ca²⁺] – with a midpoint of fluorescence enhancement at 535 nM and a Hill coefficient of n=2 – whereas unlabeled CaM1234 showed no significant increase, consistent with its inability to bind Ca²⁺ (Fig. 1H).

The previously described Trp peptide (Török and Trentham, 1994; Török et al., 1998a; Török et al., 1998b) was myristoylated at the N-terminus to make it membrane permeant. The Ca²⁺/CaM-binding properties of mTrp peptide were determined by stopped-flow fluorimetry using TA-CaM. Similarly to Trp peptide, mTrp had a high affinity, with a K_d of 15 pM for Ca²⁺/TA-CaM (Fig. 1I).

The nuclear pore size has been reported to be affected by the level of Ca²⁺ within the intracellular Ca²⁺ stores (Perez-Terzic et al., 1997). Therefore, we examined the state of the Ca²⁺ stores under our conditions. HeLa cells were loaded with 3 μM of the cell permeant fluo-4AM ester (F14217 Molecular Probes) for 15 minutes and stimulated with 50 μM histamine under a range of extracellular and intracellular Ca²⁺ conditions. In the presence of 2 mM extracellular Ca²⁺, histamine induced Ca²⁺ oscillations, initially occurred at 40-second intervals whose frequency and intensity decreased over time (Fig. 2A). Cells incubated with 2 mM extracellular EGTA and stimulated after 15 minutes with 50 μM histamine, also showed oscillations that initially started at 40-second intervals but decreased in frequency and intensity. However, under the above conditions, oscillations ceased more rapidly than in the presence of 2 mM extracellular Ca²⁺ (Fig. 2B) because refilling of intracellular stores required Ca²⁺ influx through plasma membrane ion channels. Cells that had been electroporated with fluo-4 while 2 mM EGTA was present in the medium showed no response to 50 μM histamine. This lack of response might indicate that, electroporation of cells while EGTA is present in the medium can result in either the depletion of Ca²⁺stores or in the buffering of released Ca²⁺ by EGTA entering the cell during electroporation (Fig. 2C). Cells stimulated with 2.1 mM Ca²⁺ after they had been loaded with fluo-4, showed a large single transient. Addition of 50 μM histamine after this transient had diminished caused another single, small Ca²⁺ transient, typically after about 20-25 minutes (n=3), indicating that the intracellular stores had partially refilled with Ca²⁺ (Fig. 2D).

An intracellular Ca²⁺ transient was generated by the addition of free Ca²⁺ to the extracellular solution as demonstrated above (Fig. 2). Upon addition of 2.1 mM Ca²⁺, fluo-4 fluorescence was increased by 10.2±1.2 (n=16) in the nucleoplasm of the cells. Fluo-4 fluorescence was also higher in the nucleoli and other nuclear structures. 1300 seconds after Ca²⁺ stimulation, the fluorescence-level dropped to 2.7±0.3 above the original level.

To measure diffusion from the cytoplasm into the nucleus, we used 19.5-kDa FL-dextran, which is similar in size to CaM, and measured the change in the ratio of nucleoplasmic to cytoplasmic fluorescence (F_np:F_cytoplasm). Measurements were carried out over 1 hour in glial cells that had been kept in 2 mM EGTA and also in glial cells to which 2.1 mM Ca²⁺ was added. In 2 mM EGTA only, the change of the F_np:F_cytoplasm ratio was 0.003±0.0002/minute (n=4) (Fig. 3A). However, after the addition of 2.1 mM Ca²⁺, the change of ratio increased to 0.008±0.001/minute (n=4) (Fig. 3A). This indicated that the higher intracellular Ca²⁺ concentration was causing the nuclear pores to be in a more open conformation.

We examined the permeability of the nuclear pores with 4.3-, 9.5-, 19.5-, 40- and 77-kDa FL-dextrans, by observing their distribution at low intracellular [Ca²⁺] and after stimulation by Ca²⁺. HeLa cells in EGTA that had been electroporated with fluorescent dextrans were imaged after 15 minutes, cells in 2 mM free Ca²⁺ were imaged after 1 hour. Also, cells were electroporated with WGA to determine whether it has any effect on the passive transport of FL-dextrans in the presence and absence of intracellular Ca²⁺.

In the case of 4.3-kDa FL-dextran, the F_np:F_cytoplasm ratio of cells in 2 mM EGTA (4.3 kDa: 1.23±0.1, n=5) (Fig. 3B) was not significantly different compared with cells in 2 mM Ca²⁺ (4.3 kDa: 1.3±0.03, n=6) (Fig. 3G). Thus, this small dextran was able to freely enter the nucleus, even at the more-closed state of the nuclear pores. However, significantly different F_np:F_cytoplasm ratios were seen between cells in 2 mM EGTA and cells in 2 mM Ca²⁺, when 9.5-, 19.5- and 40-kDa FL-dextran was used. The use of 9.5-, 19.5- and 40-kDa FL-dextran, respectively, gave ratios of 1.0±0.03 (n=6), 0.56±0.03 (n=8) and 0.11±0.01 (n=7) for cells in 2 mM EGTA (see Fig. 3C-E), and increased ratios of 1.28±0.05 (n=6), 1.03±0.02 (n=7) and 0.39±0.02 (n=6) for cells kept in 2 mM Ca²⁺ for 1 hour (see Fig. 3B, Table 1).

The addition of 10 μM BAPTA-AM before electroporating cells with 19.5 kDa dextran did not alter the F_np:F_cytoplasm ratios further; they were 0.54±0.03 (n=5) in BAPTA-AM, compared with 0.56±0.03 (n=8) of cells in 2 mM EGTA only (Fig. 3B). The 77-kDa FL-dextran was excluded from the nucleus regardless of the [Ca²⁺]. Cells in 2 mM Ca²⁺ had a F_np:F_cytoplasm ratio of 0.1±0.02 (n=6), whereas cells kept in EGTA had a ratio of 0.14±0.05 (n=6) (Fig. 3G, Table 1). WGA, which is known to block NLS-facilitated nuclear import but not passive diffusion (Dabauvalle et al., 1988) did not significantly inhibit the diffusion of any FL-dextran in high or low [Ca²⁺].

Glial cells were electroporated with 4.3-, 9.5-, 19.5- or 77-kDa FL-dextrans and 2 mM EGTA. When electroporated with the 4.3-kDa FL-dextran there was no significant difference in the F_np:F_cytoplasm ratio of cells in 2 mM EGTA (1.34±0.03, n=6) compared with cells in 2 mM Ca²⁺ for 1 hour (1.42±0.04, n=4) (Fig. 3H). However, 9.5- and 19.5-kDa FL-dextran showed significant increases. Ratios, using 9.5- and 19.5-kDa FL-dextran, respectively, were 0.96±0.05 (n=7) and 0.6±0.02 (n=5) in cells kept in 2 mM EGTA, and 1.34±0.04 (n=5), 1±0.04 (n=5) in cells kept in 2 mM Ca²⁺ for 1 hour (see Fig. 3H). 77-kDa FL-dextran was excluded from the nucleus regardless of the [Ca²⁺]. For cells kept in 2 mM Ca²⁺, the F_np:F_cytoplasm ratio was 0.18±0.01 (n=7), whereas for cells in EGTA it was 0.14±0.01 (n=8) (Fig. 3H, Table 1). WGA did not significantly inhibit the diffusion of any FL-dextran in high or low [Ca²⁺].

Neurons were electroporated with 4.3-, 9.5-, 19.5- or 77-kDa FL-dextrans and 2 mM EGTA. When electroporated with the 4.3-kDa FL-dextran, there was no significant difference in the F_np:F_cytoplasm ratio of cells in 2 mM EGTA (1.37±0.04, n=6) compared with cells in 2 mM Ca²⁺ for 1 hour (1.4±0.05, n=4) (Fig. 3I). However, 9.5- and 19.5-kDa FL-dextran showed significant differences in the F_np:F_cytoplasm ratios of cells in 2 mM EGTA compared with cells in 2 mM Ca²⁺ for 1 hour. Ratios, using 9.5- and 19.5-kDa FL-dextran, respectively, were 0.92±0.02 (n=6) and 0.61±0.02 (n=5) for cells in 2 mM EGTA and 1.22±0.04 (n=5) and 0.95±0.02 (n=5) for cells in 2 mM Ca²⁺ (Fig. 3I). 77-kDa FL-dextran was excluded from the nucleus regardless of the Ca²⁺conditions. In 2 mM Ca²⁺ the F_np:F_cytoplasm ratio was 0.12±0.01 (n=6), whereas the ratio in EGTA was 0.17±0.01 (n=6) (Fig. 3I, Table 1). WGA did not significantly inhibit the diffusion of any FL-dextran in high or low [Ca²⁺].

The above results show that similar changes of F_np:F_cytoplasm ratios are obtained in HeLa cells, glial cells and neurons. When cells had been electroporated with 9.5- and 19.5-kDa FL-dextrans and were left in high [Ca²⁺] for 1 hour, F_np:F_cytoplasm ratios were significantly different owing to different diffusion rates of the differently sized dextrans into the nucleus. In all cell three cell types, the 9.5-, 19.5- and 40-kDa FL-dextrans showed significant differences in F_np:F_cytoplasm ratios in cells kept in EGTA and cells kept in Ca²⁺, providing evidence that Ca²⁺ increased the nuclear pore size.

Upon loading the cells with 6ROX-CaM in 2 mM EGTA, the F_np:F_cytoplasm ratios for glial, neuronal and HeLa cells were 1.26±0.03 (n=10), 0.97±0.02 (n=4) and 1.18±0.03 (n=19), respectively (Fig. 4A). These ratios were significantly higher than those for the similarly sized 19.5-kDa FL-dextran (0.56±0.03, n=8), suggesting that passive diffusion might not account for the observed nuclear presence of CaM. Electroporation of 6ROX-CaM into HeLa cells, in the presence of an increased EGTA concentration of 10 mM, gave a F_np:F_cytoplasm ratio of 1.21±0.02 (n=5) (Fig. 4B), not significantly different from the ratio obtained in the presence of 2 mM EGTA. Glial, neuronal and HeLa cells, loaded with 6ROX-CaM and 20 μg/ml WGA in the presence of 2 mM EGTA displayed F_np:F_cytoplasm ratios of 0.39±0.02 (n=16), 0.33±0.06 (n=6) and 0.23±0.02 (n=17), respectively (Fig. 4A). This indicates a significant decrease in the concentration of 6ROX-CaM in the nucleus and also a facilitated translocation mechanism that is Ca²⁺-independent.

The addition of 10 μM mTrp peptide – which interacts with CaM in a Ca²⁺-dependent manner – before electroporation, gave an F_np:F_cytoplasm ratio of 1.21±0.01 (n=5) (Fig. 4B), confirming the Ca²⁺-independence of the nuclear entry of 6ROX-CaM.

WGA was used to block facilitated entry of 6ROX-CaM into the nucleus in cells that had been electroporated in the presence of 2 mM EGTA. Under these conditions, the rate of change of the F_np/F_cytoplasm ratio was 0.004±0.0003/min (n=4) (Fig. 4C), similar to that of 19.5-FL-dextran (Fig. 3A).

Upon loading the cells with 6ROX-CaM1234 in the presence of 2 mM EGTA, the F_np:F_cytoplasm ratio was 1.31±0.04 (n=15). Cells loaded with 6ROX-CaM1234 and 20 μg/ml WGA in the presence 2 mM EGTA, and imaged 15 minutes after electroporation, displayed a F_np:F_cytoplasm ratio of 0.34±0.06 (n=9). This was a significant decrease in the concentration of 6ROX-CaM1234 in the nucleus (Fig. 4B), which indicates a facilitated transport mechanism for CaM that did not require Ca²⁺.

Upon stimulation with 2.1 mM Ca²⁺ there was a 1.2- to 1.3-fold increase in 6ROX-CaM fluorescence throughout the cell. This can be explained by the sensitivity of 6ROX-CaM fluorescence to Ca²⁺ as shown in Fig. 1G. There was also strong association of the 6ROX-CaM with the nucleoli from 6ROX-CaM already in the nucleoplasm, which was particularly noticeable in HeLa and glial cells. So, Ca²⁺ stimulation probably induced immediate binding of 6ROX-CaM that was present in the nucleoplasm before stimulation.

Typically there was a delay of up to 40 seconds in the translocation of 6ROX-CaM to the nucleus from the cytoplasm compared to that of Ca²⁺, which was also noted by Craske et al. (Craske et al., 1999). The time to reach half-maximum fluorescence (t_1/2) of 6ROX-CaM translocation was 80±12 seconds (n=12). The F_np:F_cytoplasm ratio showed a significant increase from 1.11±0.04 (t₀) to 1.63±0.05 (t₃₁₇) and the nucleolic-cytoplasmic fluorescence ratio had markedly increased from 0.92±0.04 to 2.8±0.2. The fluorescence of 6ROX-CaM in the cytoplasm after 317 seconds was also significant and the measured relative fluorescence dropped from 1 to 0.68±0.04. An example of Ca²⁺-dependent nuclear translocation of 6ROX-CaM in a HeLa cell is shown in Fig. 5 (a representative Ca²⁺ signal is shown below in Fig. 8D).

Similarly to HeLa cells, stimulation of electroporated glial cells with 2.1 mM Ca²⁺ caused 6ROX-CaM-binding to nucleoli and translocation completed with a t_1/2 of 82±18 seconds (n=5). The F_np:F_cytoplasm ratio showed a significant increase from 1.27±0.03 (t₀), to 1.8±0.12 (t₂₈₂). The nucleolic-cytoplasmic fluorescence had also significantly increased within 282 seconds from 1.23±0.06 to 2.02±0.04. The decrease in 6ROX-CaM fluorescence in the cytoplasm was also significant, with its relative fluorescence dropping from 1 to 0.75±0.06 within 282 seconds. An example of Ca²⁺-dependent nuclear translocation of 6ROX-CaM and the Ca²⁺ signal monitored by fluo-4 fluorescence in a glial cell is shown in Fig. 6.

The addition of 2.1 mM Ca²⁺ to neurons electroporated with 6ROX-CaM, caused nuclear translocation of 6ROX-CaM and was completed with a t_1/2 of 90±20 seconds (n=4). The F_np:F_cytoplasm ratio showed a significant increase from 0.97±0.02 (t₀) to 1.7±0.2 (t₂₉₀). The relative fluorescence of 6ROX-CaM in the cytoplasm, 290 seconds after stimulation, decreased from 1 to 0.82±0.13. Fig. 7 shows 6ROX-CaM nuclear translocation and a representative Ca²⁺ signal monitored by fluo-4 in a cortical neuron.

The cell-permeant peptide mTrp was added to the cells to inhibit Ca²⁺-dependent nuclear translocation. HeLa cells were electroporated with 6ROX-CaM and left for 15 minutes in electroporation buffer supplemented with 10 μM mTrp peptide (n=7). In these cells, the addition of 2.1 mM Ca²⁺ into the extracellular medium caused no Ca²⁺-dependent nuclear translocation or binding of 6ROX-CaM to nucleoli, even though a normal Ca²⁺ transient was observed (Fig. 8A-D).

AIP is a potent inhibitor of CaMKII and has been reported to be highly specific (Ishida et al., 1995). AIP also prevented Ca²⁺-dependent nuclear translocation of 6ROX-CaM in HeLa cells, even though a normal Ca²⁺ transient was observed (n=15). This suggested that CaMKII is responsible for the translocation of CaM into the nucleus, either by acting as a carrier or by phosphorylating CaM. However, myristoylated (mAIP) prevented 6ROX-CaM from binding to nuclear targets in about 80% of the cells tested. This was surprising and suggested that mAIP directly interacts with CaM, in a way similar to mTrp. TA-CaM in 100 μM Ca²⁺, 100 mM KCl, 2 mM MgCl₂ and 50 mM KMES pH 7.0 at 21°C was thus titrated with mAIP in a fluorimetric assay. Additions of mAIP up to 140 nM caused a 1.67-fold increase in TA-CaM fluorescence. A K_d value of 68 nM was obtained by numerical fitting to the data. However, in the presence of 2 mM EGTA there was no increase in TA-CaM fluorescence upon addition of up to 140 nM mAIP. This indicated that mAIP can interact with TACaM in a Ca²⁺-dependent manner (Fig. 8E).

Passive diffusion of 6ROX-CaM was measured in HeLa cells that had been stimulated with 2.1 mM Ca²⁺ in the presence of 10 μM mTrp peptide. Under these conditions, the F_np:F_cytoplasm ratio change was 0.01±0.002/min (n=5) (Fig. 8F). Thus, the rate of diffusion of Ca²⁺-6ROX-CaM-mTrp was similar to that of FL-dextran (Fig. 3A).

WGA was electroporated into HeLa, glial and neuronal cells at an estimated final concentration of 20 μg/ml together with 6ROX-CaM or 6ROX-CaM1234, to determine whether CaM entry upon stimulation with 2.1 mM Ca²⁺ was by passive diffusion or by a facilitated mechanism. Upon stimulation there were no significant increases in the F_np:F_cytoplasm ratios in any of the cells. The rate of change in F_np:F_cytoplasm of 6ROX-CaM was 0.012/min. This rate was similar to 6ROX-CaM in 10 μM mTrp stimulated with 2.1 mM Ca²⁺ (0.01/min), indicating that even when Ca²⁺/CaM was free to bind targets in the nucleus, the rate of diffusion into the nucleus was the same as when binding was inhibited by mTrp. Similarly, HeLa cells loaded with 6ROX-CaM1234 showed no immediate nuclear translocation upon Ca²⁺ addition. These results demonstrated that Ca²⁺-dependent nuclear translocation (like Ca²⁺-independent nuclear translocation) occurred via a facilitated mechanism and passive diffusion did not play a significant role in either the localization of CaM in cells under resting conditions or in response to stimuli (Fig. 9).

We have used novel CaM probes and a membrane-permeant peptide antagonist to determine the Ca²⁺-dependent and -independent mechanisms of the nuclear transport of CaM. 6ROX-CaM activated PDE similarly to unlabeled, bacterially expressed human liver CaM and, therefore, the 6ROX label had little effect on the ability of 6ROX-CaM to bind to targets.

Previous work on CaM nuclear translocation has produced conflicting results and, as we present here, this is partly because CaM appears to have a multitude of mechanisms regulating its entry to the nucleus. Not fully taken into account by previous reports is the state of the NPCs during CaM translocation. It is well documented that the nuclear pores are in a more-closed state when intracellular Ca²⁺ stores are depleted, preventing a passive diffusion of molecules with a molecular mass >10 kDa (Stehno-Bittel et al., 1995; Perez-Terzic et al., 1999). By using our methods, we found that, in HeLa cells, FL-dextrans with the molecular mass of 9.5-40 kDa were significantly inhibited from passively diffusing into the nucleus. It is, however, not certain whether a closing-up of the nuclear pores was owing to store depletion or a result of considerable Ca²⁺-buffering in the cell. Neither the application of BAPTA-AM before electroporation nor a higher concentration of EGTA, reduced the F_np:F_cytoplasm ratio further than the one found with 2 mM EGTA.

Slow store-refilling has been described before (Hofer et al., 1998). In pancreatic acinar cells, acetylcholine (ACh) stimulation kept Ca²⁺ stores depleted as long as the agonist was present. Upon removal of ACh, the concentration of Ca²⁺ in the ER lumen began to rise but did not reach the original levels for minutes. From the results shown here it was probable that significant store-refilling occurred after the addition of 2.1 mM Ca²⁺ to the extracellular medium. It was, however, not possible to conclude whether store-refilling affected the nuclear pore diameter or whether the direct binding of Ca²⁺ to NPC was sufficient.

Fluo-4 fluorescence was typically higher in the areas around the nucleoli. This phenomenon might originate in the different binding affinities of fluo-4 for subcellular compartments (Thomas et al., 2000). However, it could be related to recently described nuclear tubules or nucleoplasmic reticulum, which either transect the nucleus entirely or terminate near to the nucleoli and are responsible for the release of free Ca²⁺ into the nucleus (Echevarria et al., 2003).

FL-dextran distributions were similar in the three cell types studied. The 4.3-kDa FL-dextran freely entered the nucleus in all conditions. A similar result has been reported with a 3-kDa FL-dextran, which was able to freely diffuse into the nucleus regardless of the intracellular Ca²⁺ levels within the cell (Perez-Terzic et al., 1999). It was concluded that although the pore size was smaller because of reduced intracellular free Ca²⁺ concentrations, smaller molecules could still pass through the nuclear pores unregulated. Previous observations that molecules >60 kDa cannot enter the nucleus via passive diffusion (Weiss, 1998; Allen et al., 2000; Wei et al., 2003) were also seen to be true here. 77-kDa FL-dextran was inhibited from entering the nucleus in the presence or absence of Ca²⁺, indicating that molecules of this size need an NLS sequence to pass through the NPC. WGA, which blocks NLS-facilitated transport by binding to carbohydrate moieties on nucleoporins did, as expected, not have any effect on the passive diffusion of any FL-dextrans. This has also been seen in rat hepatoma cells using a fluorescently-labeled nucleoplasmin and a 10-kDa dextran (Dabauvalle et al., 1988). In WGA-containing cells, only nucleoplasmin was blocked from entering the nucleus, whereas the dextran was still able to enter. It was concluded that WGA does not constrict the nuclear pore, but binds to O-glycosidically-bound carbohydrates of specific nucleoporins to inhibit facilitated transport.

To observe nuclear translocation of 6ROX-CaM in our experiments, HeLa, glial and neuronal cells were, after electroporation in the presence of EGTA, immersed in an extracellular buffer containing 2 mM EGTA; then 2.1 mM Ca²⁺ was added. Stimulation of nuclear translocation of CaM by Ca²⁺ probably explains the results of Milikan and Bolsover, namely that FL-CaM had a significantly higher F_np:F_cytoplasm ratio compared with a 10-kDa FL-dextran in micro-injected dorsal root ganglion cells (Milikan and Bolsover, 2000). This was attributed to be a property of the FLCaM, although it is more probable that there was a Ca²⁺ transient associated with the injection process, which drove the fluorescent CaM into the nucleus. Furthermore, FL-CaM and TA-CaM were found to be localized in the nucleus of resting cells, which were loaded by patch pipettes. However, the cells were washed in a solution of high of [Ca²⁺], which again – as shown by our data – would have caused an increase of intracellular Ca²⁺ causing nuclear CaM translocation. Hardingham et al., found endogenous CaM and also microinjected FITC-CaM to be localized in the nucleus of resting hippocampal neurons (Hardingham et al., 2001). Our data show that, whereas the Ca²⁺-independent mechanism causes an equilibrium of CaM between the cytoplasm and the nucleus, any further nuclear accumulation of CaM requires a Ca²⁺ stimulus. Therefore, the hippocampal neurons were probably stimulated, either by microinjection or a change of extracellular solution, before the experiments to test gene expression were carried out. Nuclear localization was inhibited by WGA, demonstrating that CaM was transported into the nucleus by a facilitated mechanism consistent with this interpretation.

Because passive diffusion of 19.5 kDa-FL-dextran is very slow, 6ROX-CaM, which has an average mass of 17223.1 Da, would be expected to be blocked from entry into the nucleus. However, upon imaging the cells in low Ca²⁺ conditions (<10 nM), the concentration of 6ROX-CaM in the nucleus was similar to that in the cytoplasm. This contrasted the behavior of 19.5-kDa FL-dextran, which was at a significantly lower concentration in the nucleus than in the cytoplasm when the cells were kept in the same conditions. This indicated that 6ROX-CaM was entering the nucleus by using a different mechanism than 19.5 kDa FL-dextran. 6ROX-CaM could enter the nucleus by a facilitated mechanism, possibly by being bound to an NLS-containing transporter molecule, independently of [Ca²⁺]. This hypothesis was supported by our observation that WGA significantly inhibited the entry of 6ROX-CaM in resting cells that were kept in EGTA. A similar observation was made by Pruschy et al., who found that in digitonin permeabilized PtK1 cells, WGA was blocking a fluorescein labeled-CaM in Ca²⁺-free form and ATP depletion had no effect on CaM import (Pruschy et al., 1994). It was therefore concluded that CaM could be imported into the nucleus by a facilitated mechanism that was independent of ATP. Further evidence that CaM could be bound in apo conditions (in which CaM is expected to be a Ca²⁺-free form) came from the observations by Luby-Phelps and collegues, who reported that only 5% of total endogenous CaM is unbound in smooth muscle cells at resting intracellular [Ca²⁺] (Luby-Phelps et al., 1995). In Swiss 3T3 cells, bleaching within the nucleus caused rhodamine-labeled CaM-fluorescence to disappear, which then only reappeared with a slow t_1/2 of equilibration of 2-3 minutes. However, 4-kDa FL-dextran could not be reduced, suggesting that the nuclear import of CaM was constrained. It was suggested that CaM in the cytoplasm was in a complex with another protein (Luby-Phelps et al., 1985). In HeLa cells before electroporation, the addition of mTrp peptide, which interacts with CaM in a Ca²⁺-dependent manner, did not alter the F_np:F_cytoplasm ratios. This also indicates that this process is Ca²-independent. The Ca²⁺-independence of this process was also shown in HeLa cells electroporated with the mutant CaM1234 labeled with 6ROX, which cannot bind Ca²⁺ (Kasri et al., 2003). Ca²⁺ binding deficiency of the mutant was also demonstrated here by the observation that there was no increase in tyrosine fluorescence upon [Ca²⁺] titrations, whereas CaM gave a 3-fold increase upon [Ca²⁺] additions to 1 mM. As with 6ROX-CaM, the nucleus contained a similar concentration of 6ROX-CaM1234 compared with cells that had been electroporated in the presence of 2 mM EGTA; WGA was also found to significantly inhibit its nuclear import.

We have described here a possible Ca²⁺-independent mechanism by which CaM might enter the cell by binding to an NLS-containing transporter molecule. Stimulation of HeLa and glial cells by increasing extracellular Ca²⁺ levels, immediately caused the 6ROX-CaM already present in the nucleus to strongly bind to the nucleoli. Typically, a delay of up to 40 seconds was seen between the increase of intracellular [Ca²⁺] and the nuclear translocation of CaM. This is similar to the delay of about 30 seconds following the addition ionomycin to pancreatic acinar cells (Craske et al., 1999). This may be of significance as it might allow the nuclear translocation mechanism to discriminate between brief and prolonged Ca²⁺ signals. Ca²⁺-dependent nuclear translocation of CaM occurred with a t_1/2 of 80 seconds, reflecting the rate of transfer of a CaM-transporter complex through the nuclear pores. A similar nuclear translocation time for fluorescein-labeled CaM was seen in pancreatic acinar cells upon ionomycin stimulation, after a 30-second delay, translocation was completed in about 260 seconds (Craske et al., 1999). Teruel et al. also reported that in RBL cells loaded with a fluorescein-labeled CaM and stimulated with ionomycin, nuclear translocation was completed after about 300 seconds. However, they concluded that this was the diffusion of free Ca²⁺-CaM from the cytoplasm to the nucleus (Teruel et al., 2000).

In our system, the nuclear translocation of CaM upon Ca²⁺ addition could not have been owing to diffusion of free Ca²⁺-CaM because CaM transport was almost completely blocked by WGA. To test 6ROX-CaM diffusion into the nucleus, rates were either measured at low Ca²⁺ levels and in the presence of WGA or at high Ca²⁺ levels and in the presence of mTrp peptide, to prevent the possibility of free Ca²⁺/6ROX-CaM being drawn to nuclear targets. This was compared with measurements at high Ca²⁺ levels in the presence of WGA, in which case free Ca²⁺/6ROX-CaM was able to bind to nuclear targets. As with the 19.5-kDa Fl-dextran, the rate of diffusion into the nucleus was 2.5-fold higher at conditions of elevated [Ca²⁺] compared with low [Ca²⁺]. The rate of facilitated transport of Ca²⁺/CaM into the nucleus was, however, 25-fold more rapid than passive diffusion. That the entrance of 6ROX-CaM and 6ROX-CaM1234 into the nucleus can be inhibited by WGA, suggests CaM bound to a transporter molecule with an NLS as the principle mechanism of CaM-entry into the nucleus. Our results contrast with the findings by Liao et al., who suggested that Ca²⁺/CaM was drawn to nuclear targets and freely diffused into the nucleus from the cytoplasm (Liao et al., 1999). In our case, free diffusion of Ca²⁺/CaM did not account for the translocation of CaM.

If CaM primarily enters the nucleus by an NLS-facilitated mechanism when nuclear pores are in a more-closed state, the one argument against this is the uncertainty of whether the depletion of intracellular Ca²⁺-stores prevents NLS-mediated transport as well as passive diffusion. Greber and Gerace certainly showed that it prevented not only passive diffusion but also signal-mediated uptake (Greber and Gerace, 1995), but other reports disagree. Marshall et al. demonstrated that in reconstituted Xenopus laevis nuclear envelopes, active nuclear import is independent of Ca²⁺-pools that can be released by ionophores (Marshall et al., 1997). Furthermore, Strübing and Clapham found that store depletion by carbachol, thapsigargin or ionomycin had no effect on nuclear translocation of GFP-tagged glucocorticoid receptor (Strübing and Clapham, 1999).

Nuclear entry of CaM in response to a Ca²⁺ signal occurred with t_1/2 values of approximately 80 seconds and was followed by CaM binding to nuclear targets over minutes. Slow binding might reflect the displacement of endogenous CaM from the target and the nuclear entry might also be limited by dissociation of CaM from cytoplasmic targets (Luby-Phelps et al., 1995). It has been demonstrated that CaM inactivation, following a rapid Ca²⁺-transient, occurs over a period of a minute (Milikan et al., 2002). The nuclear translocation rate of CaM is thus similar to the lifetime of its activated form. The Ca²⁺-dependent mechanism of nuclear translocation of CaM demonstrated here, is therefore capable of responding to rapid Ca²⁺ signaling with CaM translocating to the nucleus.

We propose that, CaM entry relies on a facilitated mechanism. The identity of the transporter, however, is still unknown. It has previously been reported that CaMKIIα and CaMKIIβ failed to show any striking redistribution upon depolarization of hippocampal neurons (Deisseroth et al., 1998). However, it was found that KN-93, which inhibits CaM kinases and CaMKKs by preventing CaM binding, also prevented CaM translocation in neurons (Mermelstein et al., 2001). It was suggested that CaM can translocate into the nucleus in a complex with a CaMKK and then phosphorylate and activate CaMKIV to phosphorylate CREB in hippocampal neurons (Mermelstein et al., 2001). KN-93, however, is not entirely specific to CaM kinases (Smyth et al., 2002) The CaM binding peptide M13 also prevented CaM nuclear translocation (Mermelstein et al., 2001). Consistent with that report, we found here that the membrane-permeant mTrp peptide inhibited Ca²⁺-dependent nuclear translocation and binding of 6ROX-CaM to nucleoli. AIP, however, proved to be less specific than previously thought and thus its inhibitory effect on nuclear translocation and binding of 6ROX-CaM can be explained by its binding to Ca²⁺/CaM rather than its inhibition of CaMKII.

In addition to the classical NLS-dependent protein import, alternative facilitated-diffusion pathways have now been put forward (Matsubayashi et al., 2001, Whitehurst et al., 2001). Nuclear import of the mitogen-activated protein (MAP) kinase ERK2 occurs without cytosolic factors and ATP, but is inhibited by WGA and was thus thought to be based on direct interaction of ERK2 with nucleoporins, and so was not thought to require an individual import carrier (Dabauvalle et al., 1988; Matsubayashi et al., 2001; Whitehurst et al., 2001).

We demonstrate here that, when the nuclear pores are closed CaM enters the nucleus possibly via two distinct facilitated mechanisms, one Ca²⁺-independent and the other Ca²⁺-dependent. This is certainly possible because of the vast number of targets CaM binds to. However, the identities of the transporter/s responsible for CaM translocation independent and dependent of Ca²⁺ are still not known.

This work was supported by the MRC, UK. We are grateful to Nael Nadif Kasri (University of Leuven, Belgium) for the E. coli expression plasmid of CaM1234 and for the functional assay of our purified protein.