The present address of Zafir Buraei is the Department of Biological Sciences, Columbia University, New York, NY 10027.
The separation of antagonist from agonist effects of trisubstituted purines on CaV2.2 (N-type) channels
Article first published online: 22 JAN 2008
© 2008 The Authors. Journal Compilation © 2008 International Society for Neurochemistry
Journal of Neurochemistry
Volume 105, Issue 4, pages 1450–1461, May 2008
How to Cite
Buraei, Z. and Elmslie, K. S. (2008), The separation of antagonist from agonist effects of trisubstituted purines on CaV2.2 (N-type) channels. Journal of Neurochemistry, 105: 1450–1461. doi: 10.1111/j.1471-4159.2008.05248.x
- Issue published online: 22 JAN 2008
- Article first published online: 22 JAN 2008
- Received October 10, 2007; revised manuscript received January 16, 2008; accepted January 16, 2008.
- slowed deactivation;
Dihydropyridines can affect L-type calcium channels (CaV1) as either agonists or antagonists. Seliciclib or R-roscovitine, a 2,6,9-trisubstituted purine, is a potent cyclin-dependent kinase inhibitor that induces both agonist and antagonist effects on CaV2 channels (N-, P/Q- and R-type). We studied the effects induced by various trisubstituted purines on CaV2.2 (N-type) channels to learn about chemical structure–function relationships. We found that S-roscovitine and R-roscovitine showed similar potency to inhibit, but agonist activity of S-roscovitine required at least a 20-fold higher concentration, suggesting stereospecificity of the agonist-binding site. The testing of other trisubstituted purines showed a correlation between CaV2.2 inhibition and cyclin-dependent kinase affinity that broke down after determining that a chemically unrelated inhibitor, kenpaullone, was a poor CaV2.2 inhibitor, and a kinase inactive analog (dimethylamino-olomoucine; DMAO) was a strong inhibitor, which together support a kinase independent effect. In fact, like dihydropyridine-induced L-channel inhibition, R-roscovitine left-shifted the closed-state inactivation versus voltage relationship, which suggests that inhibition results from CaV2 channels moving into the inactivated state. Trisubstituted purine antagonists could become clinically important drugs to treat diseases, such as heart failure and neuropathic pain that result from elevated CaV2 channel activity.
not statistically different
Cytokinins are plant hormones that also regulate plant cell cycle (reviewed in Gray 2004). They were found to be non-specific kinase inhibitors and their modification lead to the discovery of olomoucine (6-(benzylamino)-2-[(2-hydroxyethyl)amino]-9-methylpurine) as a specific inhibitor of cyclin-dependent kinases (CDKs) that are common to all eukaryotic cells (Havlicek et al. 1997). CDKs regulate cell cycle and their deregulation is a major cause of tumorigenesis (reviewed in Blagden and de Bono 2005). This motivated further efforts leading to the synthesis of R-roscovitine (6-(benzylamino)-2(R)-[1-(hydroxy-methyl)propyl]amino]-9-isopropylpurine) (see Fig. 1 for structures), which was a more potent and selective CDK inhibitor (Meijer 1996, Havlicek et al. 1997). R-roscovitine, a trisubstituted purine (TSP) is now a commonly used inhibitor of CDK5, the principal CDK in brain (reviewed in Smith et al. 2001). Furthermore, some TSPs have been shown to inhibit growth of over 20 human tumor cell lines (Mihara et al. 2002, Mgbonyebi et al. 1998, McClue et al. 2002, Lee et al. 1999, Iseki et al. 1998, Iseki et al. 1997) and roscovitine (Seliciclib, CYC202) is currently undergoing phase II clinical trials as an anticancer drug (Wesierska-Gadek et al. 2005, Hahntow et al. 2004). de Azevedo et al. (1997) showed that R-roscovitine and olomoucine, both adenine analogues, are competitive antagonists of CDKs that bind to the ATP binding pocket.
Recently, it was shown that R-roscovitine also has a direct agonist action on CaV2 channels (N-, P/Q- and R-type) (Buraei et al. 2005a, Buraei et al. 2007). The agonist action results from R-roscovitine-induced slowed closing (deactivation) of CaV2 channels, which dramatically enhances/prolongs action potential-induced calcium influx (Buraei et al. 2005a). This agonist effect of R-roscovitine is a CDK-independent effect, since (a) intracellularly applied R-roscovitine (expected to access CDKs) failed to affect CaV2 channels (i.e., to slow deactivation) and failed to occlude the effect of extracellularly applied R-roscovitine; (b) the onset of R-roscovitine-induced slow deactivation was fast (∼2 s); (c) olomoucine, a related CDK inhibitor, failed to slow deactivation; and (d) roscovitine slowed calcium channel deactivation in neurons lacking activated CDK5 (Yan et al. 2002). Based on the combination of experimental data and model simulations, we determined that R-roscovitine binds exclusively to open CaV2 channels to induce slowed deactivation (Buraei et al. 2005a). Furthermore, we determined that slowed closing resulted largely from a decreased transition rate constant between roscovitine-bound open states as opposed to roscovitine unbinding per se. As expected for enhancement of CaV2 channel activity, R-roscovitine enhances neurotransmitter release (Tomizawa et al. 2002, Monaco and Vallano 2005, Cho and Meriney 2006), and this enhancement is independent of CDK inhibition (Yan et al. 2002, Cho and Meriney 2006).
R-roscovitine exhibits multiple effects as it also has an antagonist effect on CaV2 channels (Buraei et al. 2005a). This is separate from the agonist effect since it requires higher concentrations and develops much slower (minutes) than the agonist effect (seconds) (Buraei et al. 2005a). The mechanism of this action remains unknown, but the slow time course supports kinase involvement in the R-roscovitine-induced CaV2 current inhibition (Buraei et al. 2005a). Our previous experiments were inconclusive in helping to determine kinase involvement (Buraei et al. 2005a), which was part of the motivation for further investigation.
This study was designed to address two unresolved questions concerning the effect of R-roscovitine on CaV2.2 (N-type) channels: (i) What is the mechanism of CaV2 channel inhibition, and to what extent are kinases involved? (ii) What are the functional TSP moieties that contribute to the agonist versus antagonist effects, and to what degree can slowed CaV2 channel deactivation (agonist effect) be separated from CaV2 inhibition (antagonist effect)? We show that alterations of the C2 position in TSPs (roscovitine and its analogues) can profoundly affect the agonist effect, while having a smaller impact on the antagonist effect. In addition, our evidence supports a kinase independent mechanism for TSP-induced CaV2.2 channel inhibition. The inhibitory mechanism appears to involve direct TSP-channel binding that drives CaV2.2 channels into the inactivated state. This mechanism is similar to that of dihydropyridine (DHP) antagonists on L-type channels (CaV1.2). Because of this, and other similarities, we argue that TSPs could become as important for the study of CaV2 channels as DHPs have become for CaV1 (L-type) channels.
Materials and methods
We performed studies on both native and heterologously expressed N-channels. Bullfrog sympathetic neurons, whose whole-cell calcium current is comprised of 90% N-type (CaV2.2) current, are well suited for studying native N-channels since additional calcium channel isolation procedures are unnecessary (Liang and Elmslie 2001, Elmslie et al. 1992). Neurons from paravertebral sympathetic ganglia of adult bullfrogs (R. catesbeiana) were dissociated using collagenase/dispase digestion and trituration (Elmslie et al. 1994). The method of killing was approved by the Institutional Animal Care and Use Committee. Cells were maintained in L-15 medium supplemented with 10% fetal bovine serum and penicillin/streptomycin at 4°C until use (usually 2–14 days).
For heterologously expressed N-channels, we used tsA201 cells stably expressing CaV2.2 channels (α1B, β3 and α2δ) obtained from Dr. Diane Lipscombe (Brown University, Providence, RI). The N-channel variant expressed in these cells is CaV2.2e[Δ24a, 31a] (previously referred to as rnα1B-a) (Lin et al. 2004). These cells were maintained in Dulbecco’s modified Eagle’s medium with Glutamax® (Invitrogen) with 10% fetal bovine serum, 100 × antibiotic/antimycotic and the selection agents Zeocin, Blasticidin, and Hygromycin B at 37°C and 5% CO2.
Neurons were voltage-clamped using the whole-cell configuration of the patch-clamp technique. Pipettes were pulled from Schott 8250 glass (Garner Glass, Claremont, CA, USA). The series resistance ranged from 1.3 to 2.5 MΩ and was compensated at 90–95%. Currents were recorded using an Axopatch 200A amplifier (Molecular Devices, Sunnyvale, CA, USA) and digitized with a MacAdios II analog digital converter (GW Instruments, Somerville, PA, USA). Experiments were controlled by a Macintosh Quadra 800 computer (Apple Computer, Cupertino, CA, USA) running S3 data acquisition software written by Dr. Stephen Ikeda (National Institutes of Health, National Institute on Alcohol Abuse and Alcoholism, Bethesda, MD, USA). Leak current was subtracted online using a P/4 protocol. All recording was at 25°C.
For bullfrog sympathetic neurons, the internal solution contained (in mM) 61.6 NMG-Cl, 6.0 MgCl2, 14 Creatine-PO4, 2.5 NMG-HEPES, 5 Tris2-ATP, 10 NMG2-EGTA, and 0.3 mM Li2-GTP. The extracellular solution contained (in mM) 117.5 NMG-Cl, 10 NMG-HEPES, and 3 BaCl2. In some external solutions 3 mM BaCl2 was replaced by either 10 mM CaCl2 or 30 mM BaCl2. The NMG-Cl concentration of these solutions was reduced to preserve osmolarity, which was 240 mOsm for the external and 200 mOsm for internal solutions. All solutions were titrated to pH 7.2 with NMG base.
For the tsA201 cells, the internal solution contained (in mM) 104 NMG-Cl, 6.0 MgCl2, 14 Creatine-PO4, 10 NMG-HEPES, 5 Tris2-ATP, 10 NMG2-EGTA, and 0.3 mM Li2-GTP, and the external solution contained 145 NMG-Cl, 10 NMG-HEPES, and 5 BaCl2. The osmolarity was 300 mOsm for external and 280 mOsm for internal solutions, the pH was titrated to 7.4 using NMG base.
Data were analyzed using Igor Pro (WaveMetrics, Lake Oswego, OR, USA) running on a Macintosh computer. Step currents were measured as the average of 10 points at the end of voltage steps. Slowed deactivation was quantified using one of two methods that when compared gave essentially identical results. For most treatments, tail currents were fit with double exponential equations with τFast and τSlow fixed to ∼0.6 and 6 ms, respectively (Buraei et al. 2007). The rationale for using fixed deactivation τ’s was that our previous data demonstrated that CaV2.2 channels exist in two states in the presence of R-roscovitine (bound and unbound). The unbound channels deactivate with normal kinetics, while the bound channels deactivate with kinetics dictated by the drug (Buraei et al. 2005a). Thus, the fraction of R-roscovitine affected channels can be determined by the fraction of slowly deactivating channels (Buraei et al. 2005a). The value chosen for τFast represents the deactivation τ of control channels as determined from single exponential fits to tail current in the absence of drugs. The value of τSlow was obtained by fitting slowed deactivation induced by 300 μM R-roscovitine, the maximum effective dose, with a single exponential. The deactivation τ’s were determined for each cell separately. If data from 300 μM was not available for a cell, then τSlow was obtained from double exponential fitting of the tail current in 100 μM R-roscovitine with τFast fixed to that for control deactivation. The two methods yielded similar results for τSlow when compared in the same cells (n = 3). For Figs 4 and 5, slowed deactivation was measured from the late tail current (2.5 ms following repolarization) by averaging current over 0.5 ms (Buraei et al. 2005a). Dose–response relationships, normalized as indicated on each ordinate, were fitted using a single site binding isotherm to determine the EC50. For the fits, only the base value was fixed (to zero) while the maximum value was near 1 (range 1–1.1). The fit Group data were calculated as mean ± SD throughout the article. Paired t-test was used for in-cell comparison. anova with Tukey honestly significant difference posthoc analysis was used when comparing more than two data sets.
The drugs utilized in this study are shown in Fig. 1. Iso-olomoucine, S-roscovitine, bohemine, N9-isopropyl-olomoucine, and R-olomoucine II were obtained from Calbiochem (La Jolla, CA, USA). R-roscovitine and olomoucine were from either Calbiochem or LC Labs (Woburn, MA, USA). Dimethylamino-olomoucine, indirubin-3′-monoxime and kenpaullone were from Axxora Biochemicals (San Diego, CA, USA). All drugs were dissolved in 100% dimethylsulfoxide (DMSO) at a concentration of 50 mM and maintained at −30°C. Thawed aliquots were freshly dissolved in external solution and the same volume of DMSO was added to the control solution. For dose–response experiments, the DMSO concentration was equalized in all solutions. The DMSO in the control solutions had no effect on the whole-cell calcium currents. All other chemicals were obtained from Sigma (St. Louis, MO, USA). Test solutions were applied from a gravity-fed perfusion system with seven inputs and a single output. The minimum exchange time for this system approximately 2 s.
Figure 2(a) demonstrates the two effects of R-roscovitine on N-type calcium current (CaV2.2) in bullfrog sympathetic neurons. One effect is to slowly (minutes) inhibit current (antagonist effect), which is evident from the decreased step current amplitude (at 0 mV). The second effect is to rapidly (∼2 s) slow channel deactivation (agonist effect) as evidenced from the slow decay of the tail current at –40 mV (Buraei et al. 2005a). Note that the tail currents do not completely deactivate since −40 mV is the threshold for N-channel activation under our conditions (3 mM Ba2+), which results in a non-zero steady-state current at this voltage (Liang and Elmslie 2001, Elmslie et al. 1992, Elmslie et al. 1990, Buraei et al. 2005a). To measure the agonist effect in isolation from the separate antagonist effect, we fit tail current deactivation using double exponential equations with τFast and τSlow fixed as described in Materials and methods, while the amplitudes of the two exponentials varied (Buraei et al. 2007). We previously demonstrated that the roscovitine-bound channels contribute the slowly deactivating component (and drug-free channels contribute the fast-deactivating component) of tail current so that the ratio of the amplitudes slow/(slow+fast) yielded the fraction of drug-bound channels (Buraei et al. 2005a, Buraei et al. 2007). Here we utilize this technique to determine if other TSPs have an agonist effect on CaV2.2 channels with the implicit assumption that differing off-rates among the various TSPs have a negligible effect on the slowed deactivation τ. This assumption was supported by our modeling, which showed that R-roscovitine-induced slowing of deactivation resulted from altered channel kinetics and did not, largely, depend on the unbinding kinetics of the drug. According to our modeling (Buraei et al. 2005a), if slowed deactivation resulted from roscovitine unbinding, deactivation time constants would reach a plateau at voltages below −100 mV, instead of monotonically decreasing with hyperpolarization as observed in our recordings of tail currents down to −180 mV (Buraei et al. 2005a). In addition, tail current deactivation was well fit by double exponential equations with the same fixed τs for all TSPs that at least partially slowed deactivation (Fig. 2a).
Differential effects of R- versus S-roscovitine
Racemic DHP variants have differential effects on CaV1 (L-type) channels (Franckowiak et al. 1985). For example, the agonist (-)BayK 8644 prolongs L-channel openings, while the (+) enantiomer inhibits currents. We wondered if roscovitine, which has a chiral carbon atom in the C2 side chain (Fig. 1), also exhibits stereospecificity. The main effect of S-roscovitine was to inhibit N-type calcium current (Fig. 2a), while there was no obvious effect on deactivation. However, it is possible that a small effect on deactivation could have been obscured by the inhibition. Using the double exponential fitting method (Materials and Methods) we found that 100 μM S-roscovitine slowed deactivation of only 13 ± 4% of CaV2.2 channels, compared to 89 ± 5% for 100 μM R-roscovitine (p < 0.05, n = 6, Fig. 2a). However, the magnitude of slowed deactivation induced by S-roscovitine was significantly different from zero (p < 0.05). As previously demonstrated (Buraei et al. 2005a), activation was also slowed by R-roscovitine for voltage steps ≤ 0 mV. Indeed, our modeling showed that slowed activation resulted from R-roscovitine bound channels converting from a low Po open state to a high Po open state (Buraei et al. 2005a). Interestingly, activation was not affected by S-roscovitine (95 ± 4% in S-roscovitine vs. 123 ± 12% slower activation τ at 0 mV in R-roscovitine, p < 0.05). This supports our previous conclusion that slowed activation induced by R-roscovitine is correlated with the agonist effect and not inhibition (Buraei et al. 2005a). We next generated dose–response curves for both S- and R-roscovitine to compare the potency to induce slowed deactivation and inhibition. Figure 3 shows that the S-roscovitine dose–response curve for slowed deactivation is right-shifted to yield an EC50 > 500 μM, which is at least 20-fold larger than that for R-roscovitine (28 μM). As for the potency to induce inhibition, S-roscovitine was either equipotent or slightly less potent than R-roscovitine depending on the batch of sympathetic neurons examined. The dose–response relationship showed a slightly larger IC50 for S-roscovitine (219 μM) compared to that measured for R-roscovitine (130 μM) (Fig. 3). Thus, the agonist-binding site on CaV2.2 channels is stereosensitive for roscovitine. S-roscovitine is an effective antagonist, but an ineffective agonist, which further supports the idea of separate mechanisms for the agonist versus antagonist effects of R-roscovitine (Buraei et al. 2005a). Since R- and S-roscovitine are equally effective CDK antagonists, S-roscovitine can be used to differentiate between kinase inhibition and CaV2 channel-enhancement effects of R-roscovitine.
Other tri-substituted purines
We further investigated the role of the C2 position in agonist activity of TSPs by testing two additional compounds, bohemine and N9-isopropyl-olomoucine. Bohemine has a structure similar to that of roscovitine, but with a more distal OH group (3-aminopropanol group at C2 replacing the 2-aminobutanol of roscovitine) (Fig. 1). 300 μM bohemine showed both agonist and antagonist effects that were similar to S-roscovitine. This concentration slowed deactivation of 22 ± 3% of CaV2.2 channels (n = 5), which was much weaker than R-roscovitine (Figs 2b and 3a). The slowly developing inhibition was also weaker (51 ± 5%) than that induced by 300 μM R-roscovitine (75 ± 7%), but again was similar to S-roscovitine (Figs 2b, 3a and 4c). Shortening the C2 side chain by one carbon from that of bohemine yielded N9-isopropyl-olomucine (Fig. 1). At 300 μM, N9-isopropyl-olomucine failed to induce an agonist effect (6 ± 3%), and generated a smaller antagonist effect than bohemine (23 ± 6% inhibition, Figs 2c, 3a and 4c). Thus, the C2 side chain not only determines agonist affinity, but plays a role in modulating antagonist affinity for CaV2.2 channels. The two effects co-vary with C2 substitutions with the agonist effect being more sensitive than the antagonist effect. We also investigated TSPs with changes at the N9 position. We found that replacing the isopropyl group with a methyl group did not significantly alter TSP-induced inhibition. The step inhibition induced by 300 μM N9-isopropyl-olomoucine (23 ± 6%) was not statistically different from that induced by 300 μM olomoucine (N9 methyl group, 19 ± 4%, Fig. 4), which suggests the N9 group is not critical for inhibition. Nevertheless, relocation of the methyl group from N9 (olomoucine) to the N7 position (iso-olomoucine, Fig. 1) completely abrogated inhibition of CaV2.2 channels (4 ± 3%, n = 3, Fig. 4). Incidentally, this relocation renders N7 unavailable to form a hydrogen bond thought to be crucial for TSP binding to the CDK ATP-binding site (de Azevedo et al. 1997), which could explain the loss of antagonist activity on CaV2.2 channels (see next section).
The R-isomer of olomoucine II is identical to R-roscovitine except for a hydroxyl group on the C6 benzyl group (Fig. 1). R-olomoucine II slowed deactivation but its effect was weaker than that of R-roscovitine (Fig. 3). The dose–response for slowed deactivation was right-shifted compared to R-roscovitine to yield an EC50 of 80 μM, which is roughly threefold larger than that for R-roscovitine. Inhibition was also weaker (IC50 = 180 μM) than that induced by R-roscovitine (IC50 = 130 μM, p < 0.01, Fig. 4), but was similar to that observed for S-roscovitine (IC50 = 219 μM). Thus, the hydrophilic addition of a hydroxyl group to the C6 side chain similarly impairs slowed deactivation (agonist effect) and CaV2.2 channel inhibition (antagonist effect). Figures 3 and 4c summarize the potencies of various TSPs for CaV2.2 inhibition and enhancement.
Involvement of CDKs in CaV2.2 channel inhibition
As a result of the slow time course of the development and recovery from roscovitine-induced CaV2.2 channel inhibition, we previously speculated that CDK block was a possible mechanism (Buraei et al. 2005a). The results presented so far, in particular the failure of the inactive CDK blocker iso-olomoucine to inhibit CaV2.2 channels, appear to support this hypothesis. We next examined whether there is a correlation between the potency of TSP-induced CaV2.2 inhibition with the IC50 for CDK block, as expected if CDK block mediates CaV2.2 current inhibition. We found a significant correlation (r = −0.845, p < 0.037, n = 3–9) in a plot of inhibition induced by 300 μM R-roscovitine, S-roscovitine, R-olomoucine II, bohemine, N9-isopropyl-olomoucine, olomoucine and iso-olomoucine versus CDK1 affinity (Fig. 3c). Affinity for CDK1 was used for this comparison because it is a prototypical CDK that is easily purified (Meijer et al. 1997, de Azevedo et al. 1997), and CDK1, 2 and 5 have been shown to have similar affinities for olomoucine and R-roscovitine (Meijer et al. 1997, Vesely et al. 1994).
This positive correlation suggested CDK involvement, and motivated us to investigate this possibility more deeply. One result supporting kinase involvement was the lack of CaV2.2 inhibition induced by iso-olomoucine. We tested a second kinase inactive analog, DMAO (Fig. 1) (Walker et al. 1999), but surprisingly, unlike iso-olomoucine, DMAO strongly inhibited CaV2.2 channel activity with a potency and time course similar to that of R-roscovitine. 100 μM DMAO inhibited N-current by 26 ± 12% (n = 4), while 300 μM increased the inhibition to 67 ± 7% in 3 additional cells (Fig. 4). Consistent with our early conclusion that the agonist effect is more sensitive to changes at the C2 side chain, DMAO failed to induced slowed deactivation (Fig. 4a and b). We next investigated two structurally unrelated kinase blockers, indirubin-3′-monoxime and kenpaullone, which have CDK affinities similar to that of R-roscovitine (Hoessel et al. 1999, de Azevedo et al. 1997, Leost et al. 2000). For indirubin-3′-monoxime, kenpaullone and R-roscovitine the IC50 for block of CDK1 was 0.18, 0.4 and 0.45 μM, respectively, while the IC50 for block of CDK5 was 0.1, 0.85 and 0.16 μM, respectively. If CDK1 is involved, indirubin-3′-monoxime should be a more potent blocker than R-roscovitine, while kenpaullone should be equipotent. If CDK5 is involved indirubin should be equipotent and kenpaullone less potent than R-roscovitine. Using tsA201 cells stably expressing CaV2.2 channels, we found that, suggestive of neither CDK1 nor CDK5 involvement, indirubin inhibited CaV2.2 channels more potently, while kenpaullone inhibited less potently than R-roscovitine (inhibition was 31 ± 9%, 17 ± 5% and 35 ± 9% for 30 μM indirubin, 100 μM kenpaullone and 100 μM R-roscovitine respectively, n ≥ 6, Fig. 5). Unlike R-roscovitine, neither indirubin nor kenpaullone induced slowed deactivation (agonist effect, Fig. 5a). Note that the inhibition of expressed CaV2.2 channels induced by 100 μM R-roscovitine was similar to that for native CaV2.2 current recorded from bullfrog sympathetic neurons (Fig. 3), which shows that data from these two preparations are comparable. However, one problem in the interpretation of our results is that each of these inhibitors can affect several different kinases, which weakens conclusions based on the above analysis. An alternative approach that can determine if any kinases are involved is to intracellularly apply the drug to directly affect the intracellularly-localized kinases (Belevych et al. 2002, Buraei et al. 2005a). Using bullfrog sympathetic neurons, we examined whether internally applied 100 μM R-roscovitine could abrogate the inhibition induced by externally applied R-roscovitine (100 μM), as expected if kinase block was involved. Externally applied R-roscovitine inhibited CaV2.2 current by 42 ± 7% in cells dialyzed for at least 30 min with R-roscovitine versus 41 ± 0.5% in cells dialyzed with 0.2% DMSO (n = 6 and 3 respectively, ns; compare also with R-roscovitine in Fig. 5c). If R-roscovitine equilibrated in the cytoplasm at the internally applied concentration, the intracellularly located kinases should have been inhibited and, at least partially, unavailable for block by external R-roscovitine. When taken together, the evidence supports a kinase independent mechanism where TSPs bind to an extracellularly exposed site to inhibit CaV2.2 channels. We next examined the mechanism by which R-roscovitine binding could inhibit channel activity.
R-roscovitine-induced inhibition is holding potential sensitive
Guided by the somewhat similar action of TSPs on N-channels to that of DHP on L-channels (i.e., the stereospecificity and the slowing of deactivation), we wondered if, like DHPs (Sanguinetti and Kass 1984, Bean 1984), TSP-induced inhibition of CaV2.2 channels involved inactivation. We initially investigated this by measuring CaV2.2 current inhibition induced by a range of R-roscovitine doses at HP = –80 versus –120 mV in a within cell comparison (Fig. 6a). The reduced inhibition at HP –120 mV resulted from a right-shift in the dose–response relationship (Fig. 6b and c) yielding an IC50 of 300 μM compared to 130 μM for inhibition at HP –80 mV. In contrast, slowed deactivation was not HP sensitive (Fig. 6d), which was expected since this effect requires roscovitine binding to the open state (Buraei et al. 2005a) and the occupancy of that state in actively gating channels will not be affected by HP. Interestingly, a sustained enhancement of step current was noted in 30 μM R-roscovitine at HP –120 mV (arrowhead, Fig. 6b). Such an enhancement was observed only transiently in our previous work because of the stronger inhibition at HP –80 mV (Buraei et al. 2005a).
The right-shift in IC50 for R-roscovitine-induced inhibition supports the involvement of closed-state inactivation. This was directly tested by examining the effect of 100 μM R-roscovitine on the closed-state inactivation versus voltage relationship, which was generated using a triple pulse protocol with test pulses to 0 mV bracketing a 5-s inactivating step ranging from either −160 or −140 mV to −40 mV (Fig. 7) (Klemic et al. 2001, Cox and Dunlap 1994, Degtiar et al. 2000). Inactivation was measured as the ratio of the two 0 mV steps (IPost/IPre). This work was done using CaV2.2 channels stably expressed in tsA201 cells along with CaVβ3 and α2δ. We have found that inactivation for CaV2.2 channels expressed in these cells is faster than in native cells, which facilitated data collection. The closed-state inactivation versus voltage relationship was well fit by a single Boltzmann relationship with V½ = −95 mV for control versus −133 mV in 100 μM R-roscovitine (Fig. 7c and d), which was significantly different (p < 0.01, paired t-test) and further supports the involvement of closed-state inactivation in TSP-induced inhibition of CaV2.2 channels. One additional finding was that 5 s at −160 mV failed to completely recover control current amplitude (Fig. 7b). While it is possible that an additional mechanism exists, it is likely that the incomplete recovery resulted from a failure to reach steady state during the 5 s step. We conclude that roscovitine inhibits by driving N-channels into the closed-inactivated state, which is supported by both the left-shift in the inactivation versus voltage relationship induced by R-roscovitine and right-shift in the R-roscovitine dose–response relationship with HP hyperpolarization.
We studied structure-function of TSPs to determine structures responsible for slowed deactivation and inhibition of CaV2.2 channels, and to better understand the mechanism for inhibition. We found that TSPs form the first family of small molecules that act as CaV2.2 (N-type) channel agonists and antagonists. Moreover, we found that the C2 side-chain seems key for slowed deactivation (agonist effect) and identified several TSPs that isolate antagonist from agonist activity. The study also finds that the kinase-inactive TSP DMAO inhibits CaV2.2 channels, which maybe useful in designing specific CaV2 antagonists (that do not inhibit CDKs). We demonstrate that inhibition favors inactivated CaV2.2 channels, as the apparent TSP potency increases with depolarized HP and the V½ for closed-state inactivation is significantly left-shifted by R-roscovitine.
The effect of R-roscovitine on other ion channels
Previously we examined the effect of R-roscovitine on other voltage-dependent ion channels. We showed that the agonist effect of roscovitine was an exclusive property of CaV2 channels (N-, P/Q- and R-type), while inhibition was found for L-type (CaV1.2) calcium channels and various potassium (Kv1.5, 2.1 and 4.3) channels (Buraei et al. 2007). Voltage-dependent sodium channels were relatively weakly inhibited. Interestingly, the mechanism of inhibition is unique for each channel class (CaV1.2, CaV2.2 and Kv). We demonstrated that potassium channels were inhibited by an open channel blocking mechanism (Buraei et al. 2007), whereas L-type calcium channels were inhibited by roscovitine-induced slowed activation and enhanced voltage-dependent inactivation (VDI) (Yarotskyy and Elmslie 2007). Unlike that for N-type channels, we found that R-roscovitine-induced L-channel inhibition was not affected by holding potential (Yarotskyy and Elmslie 2007). In addition, calcium-dependent inactivation was not affected by R-roscovitine (Yarotskyy and Elmslie 2007). Together these results led us to conclude that roscovitine enhanced open-state inactivation (VDI) of L-type channels without affecting either closed-state inactivation or calcium-dependent inactivation. In the present work, we conclude that inhibition of N-type channels primarily results from roscovitine-induced enhancement of closed-state inactivation.
Structural requirements for enhancement
The available evidence demonstrates that the C2 side chain is crucial to the agonist effect of TSPs on CaV2.2 channels. For non-branching moieties, the extension of the C2 side chain by one carbon over N9-isopropyl-olomoucine (C2 = 2-hydroxyethylamino) yields bohemine (C2 = 3-hydroxypropylamino, Binarova et al. 1998), which can induce slowed deactivation at high concentrations (300 μM) (Fig. 2). This suggests that the agonist-binding site requires either an extended OH or a more hydrophobic C2 side chain. The comparison of bohemine with R-roscovitine supports the latter since R-roscovitine does not have an extended OH group, but strongly induces slowed deactivation. The stereospecificity of the enhancement site further indicates that the orientation of groups along C2 is important for slowed deactivation. Modeling of stereo-specific ligands with their respective binding sites shows that three out of the four groups around a chiral carbon are generally involved in ligand binding (Mesecar and Koshland 2000). Conveniently, the stereospecificity of the CaV2 channel agonist-binding site for roscovitine (i.e., the > 20-fold higher agonist potency of R- versus S-roscovitine) provides a method for determining whether a given R-roscovitine effect results from CaV2 current enhancement or CDK inhibition (the latter being of similar affinity for the two isoforms). Thus, 30–100 μM R-roscovitine can significantly enhance CaV2 channel activity, while the same concentration of S-roscovitine will have little or no enhancement effect. Indeed, based on a preliminary report of this work (Buraei et al. 2005b), Cho and Meriney (2006) utilized the differential effect of R- versus S-roscovitine to show that the R-roscovitine-induced enhancement of neurotransmitter release at the neuromuscular junction results from enhanced CaV2 channel activity as opposed to kinase block.
Hydrophobicity of the C6 side chain may be important for agonist activity. The addition of an OH group on the C6 benzyl ring yields R-olomoucine II, which has a nearly threefold reduced EC50 for slowed deactivation of CaV2.2 channels compared to R-roscovitine (EC50 = 80 vs. 28 μM, respectively). Interestingly, this addition increases affinity for CDK1 20-fold over R-roscovitine (Krystof et al. 2002) while significantly decreasing the inhibition of CaV2.2 current, which provides additional support that CDKs are not involved in the roscovitine-induced inhibition, as further discussed below.
CaV2.2 current inhibition
We previously noted that the R-roscovitine-induced CaV2.2 current inhibition was sufficiently slow (Buraei et al. 2005a) to be consistent with phosphorylation/dephosphorylation reactions (Hartzell et al. 1991, Frace et al. 1993, Fitzgerald 2000). Indeed, there is precedent for kinase effects on CaV2 channels. CDKs have been shown to phosphorylate P/Q-type calcium channels (CaV2.1) in hippocampal pyramidal neurons (Tomizawa et al. 2002). However, it was concluded that the effect of phosphorylation was to reduce the interaction between SNARE proteins and the channel to decrease neurotransmitter release. The inhibition of map-ERK kinase, from the Ras-MAPK pathway, can cause an inhibition of calcium current in rat dorsal root ganglion neurons on a slow time scale (10–15 min) (Fitzgerald 2000). We initially examined kinase involvement by testing a number of drugs with a wide range of CDK affinities, and these results showed a significant correlation between CDK1 affinity and CaV2.2 channel inhibition. In addition, the structurally unrelated CDK blocker indirubin-3′-monoxime strongly inhibited CaV2.2 current, which was inline with relative CDK1 affinity (Hoessel et al. 1999). However, that correlation broke down when we investigated kenpaullone and DMAO. Kenpaullone blocks CDK1 with an affinity similar to R-roscovitine (Leost et al. 2000), but could muster an inhibition only 50% as large as that induced by R-roscovitine. More striking was the kinase inactivate analog, DMAO (Walker et al. 1999), which inhibited CaV2.2 current as well as R-roscovitine. It is possible that a non-CDK is involved in TSP-induced CaV2.2 channel inhibition. For example, Tyrosine kinases are more potently inhibited by R-roscovitine compared with olomoucine (Meijer et al. 1997). In addition, the involvement of kinases not previously tested with roscovitine, such as map-ERK kinase, cannot be excluded. Nevertheless, our results with internally applied R-roscovitine support the conclusion that kinases are not involved in TSP-induced inhibition of CaV2.2 channels. This technique was successfully used to show that the tyrosine kinase inhibitor genistein induced a kinase independent block of CaV1.2 (L-type) current in cardiac myocytes (Belevych et al. 2002). Taken together with the inactivation-dependent mechanism of inhibition discussed below, our results support a kinase independent mechanism for TSP-induced CaV2.2 channel inhibition.
Inhibition is holding potential-dependent
The impact of holding potential on R-roscovitine-induced inhibition is reminiscent of the effect of depolarization to strengthen DHP binding to L-type calcium channels, which has been interpreted as DHP binding favoring the inactivated state (Sanguinetti and Kass 1984, Bean 1984). Thus, the increased potency of R-roscovitine with depolarized HPs is likely related to the relative availability of the inactivated state. Fewer CaV2.2 channels are inactivated at −120 mV relative to −80 mV (Boland et al. 1994), which results from either shorter dwell times in the inactivated state, lower frequency of entering the inactivated state or both. The higher roscovitine concentrations required at −120 mV to achieve inhibition effectively increase on-rate for R-roscovitine binding so that significant block can occur during the brief or infrequent sojourns of the channel into the inactivated state. This reasoning is supported by the R-roscovitine-induced left-shift in the closed-state inactivation versus voltage relationship. Closed-state inactivation is favored by more depolarized holding potentials, which results in stronger TSP-induced CaV2 channel inhibition. Note that the −80 mV holding potential used for our neuronal studies is hyperpolarized relative to resting membrane potentials of −50 to −70 mV typically recorded from neurons (Jones 1989, Jahnsen and Llinas 1984). Thus, our reported IC50 of 130 μM for R-roscovitine (HP −80 mV) is likely an under estimate, since inhibition will be more potent at typical neuronal resting membrane potentials.
The failure of intracellularly applied R-roscovitine to affect CaV2.2 channel activity has led us to conclude that the relevant sites are extracellularly located for both the TSP agonist and antagonist effects. The speed of the agonist effect is consistent with this idea, but the slow development of inhibition is not. However, the requirement for inactivation could explain the slow time course of inhibition. At steady state, an equilibrium exists for channels in the closed (activatable) versus inactivated states. If TSP binding effectively locks channels into the inactivated state, more channels will need to inactivate to re-establish that equilibrium. Thus, the time course of inhibition is likely influenced by speed of inactivation at the holding potential. Indeed, recovery from R-roscovitine-induced inhibition was found to be faster at HP −120 mV (τ = 0.21 ± 0.10 min, n = 5) than HP −80 mV (τ = 0.36 ± 0.06 min, n = 6, p < 0.05), which is easily explained if the time course of channel reactivation is a dominant factor.
Further work is required to find specific CaV2 channels agonists (that do not also induce CaV2 and/or CDK inhibition). On the other hand, we have identified DMAO as a CaV2 channel antagonist that fails to block CDKs. DMAO could serve as a focal point for further investigation of more potent CaV2 channel antagonists. Such antagonists could be important for treating heart failure, where high catecholamine levels resulting from sympathetic nervous system hyperactivity contribute to pathology (Lymperopoulos et al. 2007, Kaye and Krum 2007). In addition, a CaV2.2 channel toxin (Prialt®) is already used to treat neuropathic pain, but poor pharmacodynamics have resulted in significant side effects (Elmslie 2004). Thus, new CaV2.2 channels blockers could have a great impact on the treatment of neuropathic pain. Our results demonstrate the potential for TSPs to become the long sought small molecules to modulate CaV2 channel activity (Triggle 2003).
- 1984) Nitrendipine block of cardiac calcium channels: high-affinity binding to the inactivated state. Proc. Natl Acad. Sci. USA 81, 6388–6392. (
- 2002) Genistein Inhibits Cardiac L-Type Ca2+ Channel Activity by a Tyrosine Kinase-Independent Mechanism. Mol. Pharmacol. 62, 554–565. , and (
- 1998) Treatment of Vicia faba root tip cells with specific inhibitors to cyclin-dependent kinases leads to abnormal spindle formation. Plant J. 16, 697–707. , , , , and (
- 2005) Drugging cell cycle kinases in cancer therapy. Curr. Drug Targets 6, 325–335. and (
- 1994) omega-Conotoxin block of N-type calcium channels in frog and rat sympathetic neurons. J. Neurosci. 14, 5011–5027. , and (
- 2005a) Slowed N-Type Calcium Channel (CaV2.2) Deactivation by the Cyclin-Dependent Kinase Inhibitor Roscovitine. Biophys. J. 89, 1681–1691. , and (
- 2005b) Roscovitine specifically slows deactivation of CaV2 (P/Q-, N-, and R.-Type) channels to enhance action potential-induced Ca2+ influx. Program No. 265.3. 2005 Abstract Viewer/Itinerary Planner. Soc. Neuroscience, Washington DC. , and (
- 2007) Roscovitine differentially affects CaV2 and Kv channels by binding to the open state. Neuropharmacology 52, 883–894. , and (
- 2006) The effects of presynaptic calcium channel modulation by roscovitine on transmitter release at the adult frog neuromuscular junction. Eur. J. Neurosci. 23, 3200–3208. and (
- 1994) Inactivation of N-type calcium current in chick sensory neurons: calcium and voltage dependence. J. Gen. Physiol. 104, 311–336. and (
- 1997) Inhibition of cyclin-dependent kinases by purine analogues: crystal structure of human cdk2 complexed with roscovitine. Eur. J. Biochem. 243, 518–526. , , , , and (
- 2000) Syntaxin modulation of slow inactivation of N-type calcium channels. J. Neurosci. 20, 4355–4367. , and (
- 2004) Calcium channel blockers in the treatment of disease. J. Neurosci. Res. 75, 733–741. (
- 1992) Calcium current modulation in frog sympathetic neurones: L-current is relatively insensitive to neurotransmitters. J. Physiol. 456, 107–123. , and (
- 1994) Reevaluation of Ca2+ channel types and their modulation in bullfrog sympathetic neurons. Neuron 13, 217–228. , and (
- 1990) LHRH and GTP-gamma-S modify calcium current activation in bullfrog sympathetic neurons. Neuron 5, 75–80. , and (
- 2000) Regulation of voltage-dependent calcium channels in rat sensory neurones involves a Ras-mitogen-activated protein kinase pathway. J. Physiol. 527 (Pt 3), 433–444. (
- 1993) Rate-limiting steps in the beta-adrenergic stimulation of cardiac calcium current. J Gen Physiol. 101, 337–353. , , and (
- 1985) The optical isomers of the 1,4-dihydropyridine BAY K 8644 show opposite effects on Ca channels. Eur. J. Pharmacol. 114, 223–226. , , and (
- 2004) Hormonal regulation of plant growth and development. PLoS Biol. 2, E311. (
- 2004) Cyclin-dependent kinase inhibitor Roscovitine induces apoptosis in chronic lymphocytic leukemia cells. Leukemia 18, 747–755. , , , , , , and (
- 1991) Sympathetic regulation of cardiac calcium current is due exclusively to cAMP-dependent phosphorylation. Nature 351, 573–576. , , and (
- 1997) Cytokinin-derived cyclin-dependent kinase inhibitors: synthesis and cdc2 inhibitory activity of olomoucine and related compounds. J. Med. Chem. 40, 408–412. , , , , , and (
- 1999) Indirubin, the active constituent of a Chinese antileukaemia medicine, inhibits cyclin-dependent kinases. Nat. Cell Biol. 1, 60–67. , , , et al. (
- 1997) Cyclin-dependent kinase inhibitors block proliferation of human gastric cancer cells. Surgery 122, 187–194; discussion 194–185. , , , , and (
- 1998) A novel strategy for inhibiting growth of human pancreatic cancer cells by blocking cyclin-dependent kinase activity. J. Gastrointest. Surg. 2, 36–43. , , , and (
- 1984) Electrophysiological properties of guinea-pig thalamic neurones: an in vitro study. J. Physiol. 349, 205–226. and (
- 1989) On the resting potential of isolated frog sympathetic neurons. Neuron 3, 153–161. (
- 2007) Drug discovery for heart failure: a new era or the end of the pipeline? Nat Rev Drug Discov. 6, 127–139. and (
- 2001) U-type inactivation of Kv3.1 and Shaker potassium channels. Biophys J. 81, 814–826. , and (
- 2002) Synthesis and biological activity of olomoucine II. Bioorg. Med. Chem. Lett. 12, 3283–3286. , , , and (
- 1999) Induction of differentiation accompanies inhibition of Cdk2 in a non-small cell lung cancer cell line. Int. J. Oncol. 15, 161–166. , , , and (
- 2000) Paullones are potent inhibitors of glycogen synthase kinase-3beta and cyclin-dependent kinase 5/p25. Eur. J. Biochem. 267, 5983–5994. , , , et al. (
- 2001) E(f)-current contributes to whole-cell calcium current in low calcium in frog sympathetic neurons. J. Neurophysiol. 86, 1156–1163. and (
- 2004) Alternative Splicing in the Voltage-Sensing Region of N-Type CaV2.2 Channels Modulates Channel Kinetics. J. Neurophysiol. 92, 2820–2830. , and (
- 2007) Adrenal GRK2 upregulation mediates sympathetic overdrive in heart failure. Nat. Med. 13, 315–323. , , , and (
- 2002) In vitro and in vivo antitumor properties of the cyclin dependent kinase inhibitor CYC202 (R-roscovitine). Int. J. Cancer 102, 463–468. , , , et al. (
- 1996) Chemical inhibitors of cyclin-dependent kinases. Trends Cell Biol., 6, 393–397. (
- 1997) Biochemical and cellular effects of roscovitine, a potent and selective inhibitor of the cyclin-dependent kinases cdc2, cdk2 and cdk5. Eur. J. Biochem. 243, 527–536. , , , , , , , and (
- 2000) A new model for protein stereospecificity. Nature 403, 614–615. and (
- 1998) Roscovitine inhibits the proliferative activity of immortal and neoplastic human breast epithelial cells. Anticancer Res. 18, 751–755. , and (
- 2002) Cyclin-dependent kinase inhibitor (roscovitine) suppresses growth and induces apoptosis by regulating Bcl-x in head and neck squamous cell carcinoma cells. Int. J. Oncol. 21, 95–101. , , , and (
- 2005) Roscovitine triggers excitotoxicity in cultured granule neurons by enhancing glutamate release. Mol Pharmacol. 68, 1331–1342. and (
- 1984) Voltage-dependent block of calcium channel current in the calf cardiac Purkinje fiber by dihydropyridine calcium channel antagonists. Circ. Res. 55, 336–348. and (
- 2001) Cdk5 on the brain. Cell Growth Differ. 12, 277–283. , and (
- 2002) Cdk5/p35 Regulates Neurotransmitter Release through Phosphorylation and Downregulation of P/Q-Type Voltage-Dependent Calcium Channel Activity. J. Neurosci. 22, 2590–2597. , , , , , and (
- 2003) Drug targets in the voltage-gated calcium channel family: why some are and some are not. Assay Drug Dev. Technol. 1, 719–733. (
- 1994) Inhibition of cyclin-dependent kinases by purine analogues. Eur J Biochem. 224, 771–786. , , , et al. (
- 1999) The clam 3′ UTR masking element-binding protein p82 is a member of the CPEB family. RNA 5, 14–26. , , , and (
- 2005) Roscovitine-induced up-regulation of p53AIP1 protein precedes the onset of apoptosis in human MCF-7 breast cancer cells. Mol. Cancer Ther. 4, 113–124. , and (
- 2002) Roscovitine: a novel regulator of P/Q-type calcium channels and transmitter release in central neurons. J. Physiol. (Lond) 540, 761–770. , , , and (
- 2007) Roscovitine, a cyclin-dependent kinase inhibitor, affects several gating mechanisms to inhibit cardiac L-type (Ca(V)1.2) calcium channels. Br. J. Pharmacol. 152, 386–395. and (