Thiophosphorylation-induced Ca²⁺ sensitization of guinea-pig ileum contractility is not mediated by Rho-associated kinase

Abstract

1. Incubation of β-escin-permeabilized guinea-pig longitudinal ileal smooth muscle with ATPγS under conditions that do not lead to thiophosphorylation of regulatory light chains of myosin (r-MLC) increased subsequent Ca²⁺ sensitivity of force and r-MLC phosphorylation. In this study we tested whether this is due to activation of the Rho and/or Rho-associated kinase (ROK) as it is the case in agonist-induced Ca²⁺ sensitization.

2. The increase in Ca²⁺ sensitivity induced by pretreatment with ATPγS at pCa > 8 with the myosin light chain kinase (MLCK) inhibitor ML-9 in rigor solution was associated with ³⁵S incorporation into the regulatory subunit of myosin light chain phosphatase (MLCP), MYPT1, and several other high molecular mass proteins. No thiophosphorylation of r-MLC, MLCK, caldesmon, calponin and CPI-17 was detected.

3. While the relatively specific inhibitor of ROK, Y 27632, inhibited the carbachol-induced increase in Ca²⁺ sensitivity with an IC₅₀ of 1.4 μm, the ATPγS-induced increase in Ca²⁺ sensitivity and thiophosphorylation of MYPT1 was not inhibited. Inhibiton of Rho by exoenzyme C3 also had no effect.

4. Only staurosporine (2 μm), but not the PKC inhibitor peptide 19–31, nor genistein nor PD 98059, inhibited the ATPγS-induced Ca²⁺ sensitization of force, r-MLC phosphorylation, and the ³⁵S incorporation into MYPT1.

5. The staurosporine-sensitive kinase(s) appeared to be tightly associated with the contractile apparatus because treatment of Triton-skinned preparations with ATPγS also induced a staurosporine-sensitive increase in Ca²⁺ sensitivity of contraction. Since there was very little immunoreactivity with antibodies to p²¹-associated kinase (PAK) in Triton-skinned preparations, the staurosporine-sensitive kinase most probably is not PAK.

6. GTPγS had an additive effect on ATPγS-induced sensitization at saturating concentrations of ATPγS. The additional effect of GTPγS was inhibited by Y 27632.

7. We conclude that treatment with ATPγS under ATP-free conditions, unmasks a staurosporine-sensitive kinase which induces a large increase in Ca²⁺ sensitivity that is most likely to be due to thiophosphorylation of MYPT1. The kinase is distinct from ROK. The physiological significance of this kinase, which is tightly associated with the contractile apparatus, is not known at present.

According to current thinking, contractile activity of smooth muscle is mainly regulated through the reversible phosphorylation and dephosphorylation of the regulatory light chains of myosin (r-MLC) at Ser-19, which are respectively catalysed by the Ca²⁺-calmodulin-dependent myosin light chain kinase (MLCK) and a type 1 phosphatase (MLCP; for review Arner & Pfitzer, 1999). The latter enzyme is targeted to myosin by a regulatory subunit, MYPT1 (Hartshorne, 1998). The extent of r-MLC phosphorylation and, hence, the amplitude of force production depends on the relative activities of these two enzymes. Many studies with intact or permeabilized smooth muscle have shown that the dependence of r-MLC phosphorylation and force on intracellular [Ca²⁺] is not unique (for review cf. Somylo & Somlyo, 1994). This is because MLCK and MLCP are both substrates for other signalling pathways which modulate the respective activities at a given Ca²⁺ concentration (for reviews cf. Horowitz et al. 1996; Arner & Pfitzer, 1999). Stimulatory agonists typically shift the relation between force, r-MLC phosphorylation and Ca²⁺ towards lower Ca²⁺ concentrations, i.e. they increase Ca²⁺ sensitivity when compared to activation by depolarization only (Morgan et al. 1984; Himpens et al. 1990).

The intracellular signalling pathways mediating agonist-induced Ca²⁺ sensitization are incompletely understood. Studies in α-toxin- or β-escin-permeabilized smooth muscle, in which the coupling between membrane-bound receptors and intracellular effectors is functional while the Ca²⁺ concentration surrounding the myofilaments can be tightly controlled, have shown that a key event in Ca²⁺ sensitization is the G protein-dependent inhibition of MLCP (Kitazawa et al. 1991; Kubota et al. 1992; Trinkle-Mulcahy et al. 1995), which may be mediated by protein kinase C (Li et al. 1998), arachidonic acid (Gong et al. 1992) and Rho-associated kinase (ROK; Kimura et al. 1996), one of the effectors of the monomeric GTPase, RhoA (Bishop & Hall, 2000). In vitro, MLPC is inhibited by a not yet identified endogenous protein kinase which is present in crude preparations of MLCP (Ichikawa et al. 1996). For both protein kinase C and ROK an important role in Ca²⁺ sensitization of contraction has been demonstrated (for reviews Horowitz et al. 1996; Somlyo & Somlyo, 2000). However, the mechanisms of inhibition of MLCP appear to be different. Inhibition of MLCP by protein kinase C appears to involve the phosphorylation of an endogenous inhibitory peptide of MLCP, CPI-17 (Li et al. 1998). In contrast, inhibition of MLCP by ROK and the endogenous kinase is due to phosphorylation of MYPT1 (Ichikawa et al. 1996; Kimura et al. 1996; Feng et al. 1999a, Feng et al. 1999b). ROK in turn is activated by the monomeric GTPase, RhoA (Kimura et al. 1996), which has been shown to induce Ca²⁺ sensitization of force and enhancement of r-MLC phosphorylation (Hirata et al. 1992; Noda et al. 1995; Gong et al. 1996). It is also activated by arachidonic acid (Feng et al. 1999b).

The association between phosphorylation of MYPT1, inhibition of MCLP activity and increase in Ca²⁺ sensitivity was in fact first demonstrated by Trinkle-Mulcahy et al. (1995). These authors showed that treatment of α-toxin-permeabilized portal vein with ATPγS under conditions that did not lead to a significant thiophosphorylation of r-MLC increased Ca²⁺ sensitivity of subsequent force and r-MLC phosphorylation. The increase in Ca²⁺ sensitivity was associated with a decrease in the rate of dephosphorylation of phosphorylated r-MLC indicating that treatment with ATPγS inhibited MLCP activity. In line with this, they observed an increase in the thiophosphorylation of MYPT1 by a then unidentified protein kinase.

In the investigation presented in this paper, we used the same approach to test whether ROK is responsible for the ATPγS-induced Ca²⁺ sensitization in a preparation in which the Rho-ROK pathway appears to be the predominant mechanism to induce Ca²⁺ sensitization, namely longitudinal ileal smooth muscle from the guinea-pig (Lucius et al. 1998; Sward et al. 2000). As has been reported for the rabbit portal vein (Trinkle-Mulcahy et al. 1995), we found a large increase in force at submaximal Ca²⁺ concentrations after treatment with ATPγS. This was associated with an increase in thiophosphorylation of MYPT1 but not of r-MLC. To our surprise neither the Ca²⁺ sensitization nor thiophosphorylation of MYPT1 was inhibited by the relatively specific ROK inhibitor Y 27632 (Uehata et al. 1997) or exoenzyme C3, which inactivates Rho. Of the protein kinase inhibitors tested, only staurosporine inhibited these events. The staurosporine-sensitive protein kinase was still present in preparations heavily skinned with Triton X-100 indicating a tight association with the myofilaments or cytoskeletal structures.

METHODS

Tissue preparation

Guinea-pigs of either sex (Dunkin Harley, 250–350 g) were anaesthetized with halothane and killed by exsanguination with procedures approved by the Institutional Animal Care and Use Committee. The ileum was rapidly removed and carefully flushed with physiological salt solution (PSS) equilibrated with 100 % O₂. Strips from the longitudinal muscle layer were dissected, mounted in a myograph, and permeabilized with β-escin after an equilibration period of 30–60 min as described (Otto et al. 1996). In brief, strips were incubated in Ca²⁺-free PSS containing 2 mm EGTA for 20 min followed by incubation in relaxation solution for 5 min. Permeabilization with β-escin was performed in relaxation solution containing 50 μmβ-escin for 35 min. Heavily permeabilized strips were obtained by chemical skinning for 4 h at 4 °C with 1 % (v/v) Triton X-100 solution with the following additions (mm): imidazole 20, EGTA 5, dithioerythritol (DTE) 1, KCl 50, and sucrose 150, pH 7.4 with KOH. After this, the preparations were washed thoroughly with relaxing solution containing 50 % (v/v) glycerol and stored at −20 °C (Pfitzer et al. 1993). All experiments were carried out at room temperature (21–24 °C) unless stated otherwise.

Solutions

PSS contained (mm): NaCl 118, KCl 5, Na₂HPO₄ 1.2, MgCl₂ 1.2, CaCl₂ 1.6, Hepes 24, glucose 10, pH 7.4 at room temperature (21–24 °C). Relaxing solution for β-escin-permeabilized smooth muscle consisted of (mm): imidazole 20, EGTA 10, magnesium acetate 10, ATP 7.5, creatine phosphate 10, NaN₃ 5, DTE 2, leupeptin 0.001, and 0.5 μm calmodulin, pH 7.0 at room temperature. Ionic strength was adjusted to 150 mm with potassium methanesulfonate. The contracting solution contained in addition 10 mm CaCl₂. The desired [Ca²⁺] was obtained by mixing contracting and relaxing solution in the appropriate ratio, and [Ca²⁺] was calculated as in Andrews et al. (1991). Relaxing solution for Triton-skinned preparations had the following composition (mm): imidazole 20, EGTA 4, MgCl₂ 10, ATP 7.5, NaN₃ 1, DTE 2, and 0.5 μm calmodulin, pH 6.7 at room temperature. The contracting solution contained in addition 4 mm CaCl₂. Intermediate Ca²⁺ concentrations were obtained as above.

Thiophosphorylation experiments

After permeabilization, a submaximal (pCa 6.16 unless otherwise stated) and nominally maximal (pCa 4.35) control contraction was elicited. The preparations were then relaxed followed by incubation with the MLCK inhibitor ML-9 (300 μm) for 5–10 min in relaxing or ATP-free rigor solution. Rigor solution had the same composition as the respective relaxing solution for β-escin or Triton-permeabilized preparations except it contained no ATP and 3 mm magnesium acetate (β-escin) or 3 mm MgCl₂ (Triton). The strips were then incubated in rigor solution containing ATPγS (1 mm unless otherwise specified) and ML-9 for 10 min. ATPγS and ML-9 were washed out in relaxing solution for 10 min and thereafter contraction was again elicited at pCa 6.16 and 4.35. Inhibitors of protein kinases were added 10 min before addition of the ML-9-containing relaxing solution and were continuously present during the ATPγS treatment protocol. They were washed out together with ATPγS and ML-9.

Identification of thiophosphorylated proteins

Muscle strips mounted isometrically were subjected to the same protocol as above except that the ATPγS solution contained only 100 μm ATPγS with 1–1.5 mCi ml⁻¹[³⁵S]ATPγS for 10 min. Control experiments showed that Ca²⁺ sensitization of force induced by 100 μm ATPγS was 49.6 ± 9 % of F_max (n = 6) and was not significantly different from experiments carried out a 1 mm ATPγS. Reactions were stopped at the end of the ATPγS incubation period by immersion of the strips in ice cold 15 % trichloroacetic acid (TCA) for 10 min. The strips were washed several times in imidazole (20 mm, pH 7.0) to remove the TCA and homogenized in Laemmli buffer containing 50 mm Tris-HCl, pH 6.8, 4.0 m urea, 1 % (w/v) sodium dodecyl sulfate (SDS), 20 mm DTE. The strips were then subjected to SDS-PAGE on 8 % or 7.5–20 % gradient gels. Equal amounts of protein (25 μg) were loaded on each lane. The protein concentration was determined using the method of Bradford (1976) with bovine serum albumin (BSA) as standard. For 2D-PAGE the strips were homogenized in a buffer containing 9.2 m urea, 10 mm Tris-HCl, pH 7.5, 10 mm DTE, and a mixture of ampholines (2 % pH 5–7 and 1 % pH 3–10). After isoelectric focusing (600 V, 4 h), the proteins were separated on 10 % SDS-PAGE. The gels were stained with Coomassie Brilliant Blue and dried on filter paper with a gel dryer (model 583, BioRad). The gels were exposed in a phosphorimager (MolekularDynamics, Amersham Pharmacia) for 7 days or to an X-ray film (Biomax, Kodak). Radioactivity of the 130 kDa band was corrected for slight variations in protein loading levels by densitometrical scanning of the stained gels (caldesmon band) by using a desktop scanner (GT-9600, Epson) and Phoretix software (Biostep, Jahnsdorf, Germany).

Estimation of Q₁₀ and K_m values

These experiments were carried out in Triton-skinned strips. In the case of determination of Q₁₀, the strips were incubated after a maximal contraction and relaxation at 23 °C in ML-9-containing rigor solution followed by incubation with ML-9 and 3 mm ATPγS for 2.5, 5 and 10 min at either 13 or 23 °C followed by a submaximal and maximal contraction at 23 °C. Cooling to 13 °C was performed before thiophosphorylation protocol and temperature was increased to 23 °C in relaxing solution after treatment, both within 5 min. The pH of the rigor solution was adjusted at the lower temperature. For determination of the K_m values of thiophosphorylation of MYPT1 and subsequent Ca²⁺ sensitization, the strips were treated with hexokinase (400 U ml⁻¹) and glucose (10 mm) for 10 min to deplete endogenous ATP before the thiophosphorylation protocol. The strips were incubated for 2.5 min with 3, 10 or 100 μm ATPγS and a specific activity of 15 mCi μmol⁻¹[³⁵S]ATPγS (determination of thiosphosphorylation). For determination of K_m of Ca²⁺ sensitization, the strips were in addition pretreated with 1 mm ATPγS. Pretreatment with ATPγS was followed by either a submaximal (pCa 6.39) and maximal contraction (force determination) or termination of the reaction by incubation in TCA, separation of the proteins by 8 % SDS-PAGE and evaluation of MYPT1 thiophosphorylation as above.

Western blotting

For Western blot analysis the proteins separated by SDS-PAGE were transferred to a nitrocellulose membrane (0.2 μm, Schleicher and Schuell Dassel, Germany). Proteins (25 or 60 μg per lane) were visualized with Ponceau Red (Sigma). The membranes were blocked for 5 h with Tris-buffered saline (TBS)-Tween (10 mm Tris-HCl, pH 8, 150 mm NaCl, 0.05 % (v/v) Tween) containing 5 % dry milk or 1 % BSA (blots for caldesmon) or 2 % BSA (blots for PAK). After washing 3 times for 10 min with TBS-Tween, the membranes were treated with the primary antibodies overnight at 4 °C as follows: anti-MYPT1 (1:20000 dilution in TBS-Tween), anti-MLCK (1:50000 in TBS-Tween with 5 % dry milk), anti-caldesmon (1:200 in TBS-Tween with 1 % BSA), anti-PAK (1:500 in TBS-Tween with 2 % BSA), anti-RhoA (1:500 in TBS-Tween containing 5 % dry milk) and anti-ROK2 (1:100 in TBS-Tween with 5 % dry milk). After washing (3 times for 1 h), the membranes were treated with horseradish peroxidase conjugated secondary antibody for 1 h at room temperature as follows: anti-rabbit IgG for MYPT1, caldesmon, PAK and RhoA (1:15000 in TBS-Tween containing 2 % dry milk), and anti-goat IgG for ROK2 (1:5000 in TBS-Tween). In some cases, bound antibodies were removed by stripping the blots with glycine (0.2 m, pH 2.2), SDS (0.1 %) and Tween (1 %) for 1 h at room temperature. After washing the blots 3 times for 5 min with TBS-Tween they were then reprobed with a second primary and secondary antibody. Immunoreactive protein bands were detected with enhanced chemiluminescence (ECL, Amersham) and quantified by densitometry of the autoradiograms. To correct for slight variations in protein loading levels, the Ponceau Red stained membranes were also scanned (calponin band).

Myosin light chain phosphorylation

For determination of r-MLC phosphorylation, strips mounted in the myograph were quick frozen in a 15 % trichloroacetic-acetone slurry pre-cooled to −80 °C at the desired time points. The muscle strips were processed for two-dimensional gel electrophoresis as described previously (Lucius et al. 1998) using a mini-gel system (Biometra, Göttingen, Germany). The relative amounts of phosphorylated and non-phosphorylated r-MLC were determined by densitometry of the silver stained gels (Bio-Rad) using Phoretix software.

Materials and chemicals

[³⁵S]ATPγS (> 1000 Ci mmol⁻¹) was purchased from Hartmann Analytic (Braunschweig, Germany). ML-9 (1-(5-chloronaphthalene-1-sulphonyl)-1H-hexahydro-1,4-diazepine), carbachol, leupeptin, β-escin, and phorbol-12,13-dibutyrate (PDBu) were purchased from Sigma. Hexokinase, NAD⁺ (nicotinamide), Li₄-ATPγS and Li₄-GTPγS were obtained from Boehringer-Mannheim. Staurosporine and PD98059 were purchased from Calbiochem. Exoenzyme C3 was obtained from Cytoskeleton (TEBU, Frankfurt, Germany), PKC-inhibitor peptide (19–31) from Bachem, Genistein from ICN Biomedicals, and Ampholine from Amersham Pharmacia. All secondary antibodies were purchased from Jackson ImmunoResearch (Dianova, Hamburg). Primary antibodies anti-PAK (C-19, rabbit polyclonal), anti-RhoA (119, rabbit polyclonal) and anti-Rok2 (K-18, goat polyclonal) were purchased from Santa Cruz Biotechnology, Inc. The anti-MLCK (clone K36, mouse monoclonal) antibody was obtained from Sigma. Anti-MYPT1 was kindly provided by Dr D. Hartshorne (University of Arizona) and anti-caldesmon IgG by Dr W. Lehman (Boston University). R-(+)-trans-N-(4-pyridyl)-4-(1-aminoethyl)-cyclohexanecarboxamide dihydrochloride monohydrate (Y 27632) was a generous gift from Welfide Corporation (Osaka, Japan). Calmodulin was purified from bovine testicle by using a modification of the procedure of Gopalakrishna & Anderson (1982). All other chemicals were of the highest grade commercially available.

Statistics

Values are shown as means ±s.e.m. (n is the number of observations). Difference of responsiveness among groups was analysed by ANOVA followed by the Newman-Keuls test. Student's t test was used when appropriate. P < 0.05 were considered to indicate significant differences.

RESULTS

Inhibition of the Rho/ROK pathway inhibits carbachol- but not thiophosphorylation-induced Ca²⁺ sensitization

Activation of β-escin-permeabilized longitudinal ileal smooth muscle with threshold [Ca²⁺] (pCa 6.16) elicited a small contraction that peaked and then declined to a steady state level (Fig. 1). The peak of the contraction amounted to 8.1 ± 1.2 % (n = 18) of the nominally maximal contraction elicited subsequently at pCa 4.35 (F_max). After incubation with ATPγS in rigor solution, subsequent force at pCa 6.16 was increased to 36.8 ± 2.7 % of F_max (n = 18). Treatment with ATPγS also increased half-time of relaxation from F_max from 19.6 ± 1.8 to 42 ± 2.6 s (n = 9), indicative that the treatment inhibited myosin phosphatase as has been suggested previously (Trinkle-Mulcahy et al. 1995). However, inhibition was not complete since in the presence of microcystin-LR, the preparations relaxed much slower (data not shown).

Figure 1. Effect of Y 27632 and staurosporine on Ca²⁺ sensitization induced by treatment with ATPγS.

A submaximal (pCa 6.16) and nominally maximal contraction (pCa 4.35, F_max) was elicited. The strips were then incubated for 10 min with 300 μm ML-9 and 1 mm ATPγS in ATP-free rigor solution. ATPγS was washed out in relaxing solution followed by incubation in pCa 6.16 and pCa 4.35 solution. Right panels show summary of results. Force is normalized to F_max before treatment with ATPγS. Values are given as means ±s.e.m. for control strips (n = 18), Y 27632- (100 μm, n = 5) and staurosporine- (2 μm, n = 7) treated strips. Staurosporine but not Y 27632 inhibited submaximal force after treatment with ATPγS. ***P < 0.001 compared to force at pCa 6.16 in control strips after treatment with ATPγS. Note slower rise of force in the staurosporine-treated strips compared to the control strips.

In most experiments, we observed a decrease in force at pCa 4.35 after treatment with ATPγS. Therefore, we normalized all contractions to F_max before treatment with ATPγS unless otherwise stated. This perhaps leads to an underestimation of the degree of sensitization, i.e. force at pCa 6.16 expressed relative to maximal force after treatment with ATPγS amounted to 78.7 ± 1.7 % (n = 18). In control experiments, ATPγS was omitted from the rigor solution. In this case, no increase in submaximal force was observed while the decrease in F_max was of similar magnitude. The cause of the decrease in F_max is not clear at present but it does not appear to be due to treatment with ATPγS. We suspect that it is due to loss of signalling proteins from the β-escin-permeabilized preparations rather than a non-specific deterioration (cf. Otto et al. 1996). To be able to compare our results with those of Trinkle-Mulcahy and coworkers (1995) we used high concentrations of ML-9. Since this may have other effects than inhibiting MLCK (Kureishi et al. 1999) we also performed some control experiments with the more specific MLCK inhibitor wortmannin, which yielded comparable results (Table 1).

Table 1.

Effect of protein kinase inhibitors on ATPγS- and carbachol-induced increase in force at constant pCa 6.16

Conditions	ATPγS-induced sensitization (%)	Carbachol-induced sensitization (%)
Control	36.8 ± 2.7(18)	100
Staurosporine (2 μm)	11.6 ± 2.0 (7)***	4.5 ± 3(4)***
Y 27632 (10 μm)	n.d.	4 ± 2.6 (4)***
Y 27632 (100 μm)	37.4 ± 2.3 (5)	n.d.
PKC-I (10 μm) with Y 27632 (100 μm)	32.2 ± 1.8 (6)*	n.d.
PKC-I (30 μm) with Y 27632 (100 μm)	36.5 ± 2.9 (4)	n.d.
Genistein (100 μm)	30.7 ± 3.2 (7)	33.3 ± 9.3 (7)***
Wortmannin (2 μm)	33.7 ± 8 (3)	n.d.
PD 98059 (10 μm)	32.2 ± 3.6 (6)	n.d.
PD 98059 (100 μm) with 1% DMSO	50.8 ± 6.6 (8)	n.d.
Control with 1% DMSO	40.2 ± 2.9 (6)	n.d.

ATPγS-induced sensitization was elicited in the presence of respective protein kinase inhibitors as detailed in Methods. Force at pCa 6.16 is expressed as a percentage of F_max before treatment with ATPγS and inhibitors. In the case of wortmannin, ML-9 was omitted. The increase in Ca2+-activated force after thiophosphorylation in the presence of 100 μM PD 98059 was not significant with respect to the appropriate control. Carbachol-induced sensitization was induced at constant pCa 6.16 as in Fig. 2 and after a plateau was reached the respective kinase inhibitors were added. Force is expressed as a percentage of carbachol-induced force. Values are given as means ± s.e.m.; number of experiments indicated in parentheses.

^*P < 0.05,

^***P < 0.001. n.d., not determined.

Incubation with ATPγS was performed under conditions (see Methods) that should not support thiophosphorylation of r-MLC. Indeed, very little incorporation of ³⁵S into r-MLC was observed even in the absence of ML-9 (Fig. 6) and no force was produced when the strips were transferred to a relaxing solution (pCa > 8) containing 7.5 mm MgATP (Fig. 1). In some experiments the observation period was extended to 30 min. In contrast, the phosphatase inhibitor microcystin-LR induced a slow contraction in relaxing solution which started after a lag period of 6 ± 0.4 (1 μm, n = 3) and 5 ± 0.7 min (10 μm, n = 4) and reached near maximal force with a half-time of 18.6 ± 1.2 and 12.4 ± 1.5 min (1 and 10 μm, respectively). Addition of pCa 4.35 at the plateau of a microcystin-induced contraction caused only a slightly further increase in force (cf. also Sward et al. 2000).

Figure 6. Separation and identification of thiophosphorylated proteins following treatment with [³⁵S]ATPγS.

β-Escin-permeabilized ileal strips were incubated with [³⁵S]ATPγS as described in Methods. A, Coomassie-stained SDS-PAGE (7.5–20 %) of total muscle homogenate; left lane, molecular weight markers. B, corresponding autoradiogram; arrow marks band which is immunoreactive with an antibody against MYPT1. Control incubations in [³⁵S]ATPγS were performed with and without ML-9. Note, ³⁵S incorporation into r-MLC is very low. Concentration of staurosporine and Y 27632 was 2 and 100 μm, respectively. C, results (means ±s.e.m., n = 7) of densitometry scans of autoradiograms of the MYPT1 immunoreactive band. ***P < 0.001. D–F, Western blots of 2D-PAGE using the antibody to MYPT1; G–I, corresponding autoradiograms; D–I, representative of 2–4 experiments.

The ATPγS-induced increase in Ca²⁺ sensitivity was not significantly inhibited by the relatively specific ROK inhibitor, Y 27632 (100 μm, Fig. 1). The inhibitor was preincubated 20 min before addition of ATPγS and was continuously present during incubation with ATPγS. This incubation period is sufficient to completely inhibit the carbachol-induced increase in Ca²⁺ sensitivity (Fig. 2). The calculated IC₅₀ value for inhibition of carbachol-induced sensitization was 1.4 μm (95 % confidence intervals were 0.3–5 μm). Y 27632 also relaxed strips when added at the plateau of the carbachol-induced force at constant pCa of 6.16 with a similar potency (data not shown). Since it appeared that high concentrations of Y 27632 might inhibit Ca²⁺-induced contraction (Fig. 2), we tested whether Y 27632 inhibits Ca²⁺-activated force in Triton-permeabilized preparations. Cumulative addition of 10 and 100 μm Y 27632 at the plateau of a submaximal contraction (85 ± 4 % of F_max) induced a small, but significant, inhibition of force by 7.6 ± 0.7 % and 11 ± 0.4 %, respectively (n = 5, P < 0.05).

Figure 2. Inhibition of carbachol-induced Ca²⁺ sensitization by Y 27632.

Ileal smooth muscle strips were incubated in pCa 6.16 () followed by cumulative addition of GTP (10 μm, ) and carbachol (10 μm, □). Y 27632 was preincubated in relaxing solution for 5 min and was present during activation. Each strip was challenged with one concentration of Y 27632 only. Results (means ±s.e.m., n = 4–8) are expressed as a percentage of maximal contraction induced by pCa 4.35. *P < 0.05, **P < 0.01, ***P < 0.001.

The ATPγS-induced sensitization was, however, inhibited by 2 μm staurosporine (Fig. 1). Staurosporine also induced a small, but not significant inhibition of F_max. It also slowed the rate of rise in force (Fig. 1). In the control strips, force at pCa 6.16 after incubation with ATPγS rose to a peak with a half-time of 54 ± 6 s and then declined to a steady state value. In the staurosporine-treated strips, force rose to the steady state force with a half-time of 160 ± 38 s (n = 6, P < 0.05).

Since RhoA activates several protein kinases (Bishop & Hall, 2000), we tested, whether inhibiting Rho directly with exoenzyme C3 would inhibit the ATPγS-induced Ca²⁺ sensitization. As previously shown (Otto et al. 1996), the carbachol-induced increase in Ca²⁺ sensitivity was completely inhibited by exoenzyme C3 (Fig. 3). In contrast, the ATPγS-induced Ca²⁺ sensitization was not inhibited by C3 (Fig. 3).

Figure 3. The effect of exoenzyme C3 on carbachol- and ATPγS-induced Ca²⁺ sensitization.

Control ileal smooth muscle strips permeabilized with β-escin were challenged twice with carbachol (CCh, 10 μm with 10 μm GTP) at constant pCa 6.16. They were then incubated with ATPγS (1 mm) and ML-9 (300 μm) in rigor solution as in Fig. 1 followed by wash out of ATPγS and ML-9 in relaxing solution and activation at pCa 6.16 and 4.35. A, examples of force traces: top trace, control; bottom trace, treatment with exoenzyme C3 (5 μg ml⁻¹ with 10 μm NAD⁺) for 30 min before the 2nd stimulation with carbachol. Note complete inhibition of carbachol-induced (***P < 0.001) but not of ATPγS-induced increase in Ca²⁺ sensitivity. B, results (means ±s.e.m., n = 3–4) are expressed relative to the maximal force (pCa 4.35) after treatment with ATPγS. □, force elicited by pCa 6.16; , force at constant pCa 6.16 with carbachol.

ATPγS-induced increase in r-MLC phosphorylation is inhibited by staurosporine but not by Y 27632

The increase in force at pCa 6.16 after treatment with ATPγS was associated with an increase in phosphorylation of r-MLC which was comparable to the value at pCa 4.35 before treatment with ATPγS (Fig. 4). This increase in phosphorylation was significantly inhibited by 2 μm staurosporine (P < 0.01) but not by 100 μm Y 27632 (P > 0.05). The increase in the phosphorylation levels in relaxing solution after ATPγS treatment was not significantly different from the resting values before ATPγS (P > 0.05).

Figure 4. Myosin light chain (r-MLC) phosphorylation under various conditions.

Ileal strips were mounted in the myograph and subjected to the same experimental conditions as in Fig. 1. They were quickly frozen 5 min after incubation at the indicated pCa before (□, n = 6–8) and after treatment with ATPγS: control (▪, n = 7–9), incubation with ATPγS in the presence of 100 μm Y 27632 (, n = 7–10) or 2 μm staurosporine (, n = 7–9). Results shown are means ±s.e.m., **P < 0.01.

Effect of other protein kinase inhibitors on ATPγS-induced Ca²⁺ sensitization

Carbachol-induced sensitization was inhibited by genistein (Table 1; Steusloff et al. 1995). Genistein, however, did not inhibit the ATPγS-induced Ca²⁺ sensitization. It was also not inhibited by PD 98059, an inhibitor of MAP kinase kinase (Alessi et al. 1995), and by the more specific inhibitor of MLCK, wortmannin (Table 1).

Low concentrations (0.2 μm) of staurosporine, which are more specific for PKC, had no effect on the ATPγS-induced Ca²⁺ sensitization. To further probe whether protein kinase C might be involved we used the relatively specific PKC inhibitor peptide 19–31 (PKC-I). We hypothesized that perhaps ROK and PKC have to be simultaneously blocked in order to inhibit the ATPγS-induced increase in force. The strips were incubated with PKC-I (10 μm) in the presence of the ROK inhibitor (100 μm), which produced a small but significant (P < 0.05) inhibitory effect of about 10 % which was not further enhanced by increasing the concentration of PKC-I to 30 μm (Fig. 5). To verify the activity of the peptide inhibitor, it was tested on the phorbol ester-induced increase in Ca²⁺ sensitivity. As has been reported recently (Sward et al. 2000) the phorbol ester PDBu (3 μm) does not increase force at constant submaximal pCa of 6.16 in the ileum (n = 4, data not shown). In contrast, in small mesenteric arteries, force in the presence of 3 μm PDBu was increased at constant pCa 6.75 from 0.43 ± 0.17 to 26.9 % of F_max. This increase in force was completely inhibited by 30 μm PKC-I (n = 4, data not shown).

Figure 5. The effect of Y 27632 and PKC inhibitor peptide 19-31 on ATPγS-induced increase in force at pCa 6.16.

ATPγS-induced sensitization was induced in β-escin-permeabilized ileal strips at pCa 6.16 as before (Fig. 1). Y 27632 together with PKC inhibitor peptide was added to the strips 20 min before treatment with ATPγS and was continuously present during treatment. Results (means ±s.e.m., n = 4–6) are expressed relative to the maximal force (pCa 4.35) before treatment with ATPγS. pCa 6.16 before ATPγS (), pCa 6.16 after ATPγS () and pCa 4.35 after ATPγS (□). *P < 0.05.

Staurosporine inhibits ³⁵S incorporation into several protein bands

Incubation of the ileal strips with [³⁵S]ATPγS resulted in ³⁵S incorporation into several protein bands. The four major thiophosphorylated bands had molecular masses of about 160, 130, 97 and 85 kDa (Fig. 6). There was no incorporation of ³⁵S into protein bands corresponding to calponin, r-MLC or CPI-17. ³⁵S incorporation into all bands was reduced by staurosporine (2 μm) but not by Y 27632 (100 μm). The 130 kDa band was immunoreactive with specific antibodies against the regulatory subunit of MLCP, MYPT1. Since several proteins migrate in this range, the proteins were also resolved by 2D-PAGE showing that only the MYPT1 (Fig. 6D–I and Fig. 7D and E) but not the MLCK (Fig. 7A and B) or caldesmon-immunoreactive (Fig. 7C) band incorporated ³⁵S. Staurosporine (2 μm) but not Y 27632 (100 μm) inhibited ³⁵S incorporation into the MYPT1-immunoreactive band (Fig. 6).

Figure 7. Separation of high molecular mass thiophosphorylated proteins by 2D-PAGE.

A, Western blot with an antibody against MLCK; B, corresponding autoradiogram. C and D, Western blot of 2D-PAGE using an antibody against caldesmon and MYPT1, respectively; E, corresponding autoradiogram. Note, the blot was first incubated with the MYPT1 antibody, and after stripping, with the caldesmon antibody. The results shown are representative of 3 experiments.

The kinase responsible for ATPγS-induced Ca²⁺ sensitization may be associated with the myofilaments

In order to determine whether the protein kinase(s) mediating the ATPγS-induced Ca²⁺ sensitization is associated with the myofilaments, longitudinal smooth muscle strips were heavily permeabilized with Triton X-100. This treatment reduced the amount of RhoA immunoreactivity to about 8 % of the content in β-escin-permeabilized strips (Fig. 8). Similarly the extent of ROK and PAK immunoreactivity but not that of MLCK or MYPT1 was reduced (Fig. 8). Force at pCa 6.74, which induced no contraction before treatment with 1 mm ATPγS, was 56.5 ± 1 % of F_max afterwards (Fig. 9). The Q₁₀ for Ca²⁺ sensitization of force was approximately 2 between 13 and 23 °C. These experiments also showed that the increase in Ca²⁺ sensitivity is maximal after pretreatment with 3 mm ATPγS for 5 min. The apparent K_m values for Ca²⁺ sensitization of force and thiophosphorylation of MYPT1 derived from double reciprocal plots were 8.5 and 8.7 μm, respectively (n = 3). Similar to the results obtained in the β-escin-permeabilized preparation, the Ca²⁺ sensitization was significantly inhibited by staurosporine but not by Y 27632 (Fig. 9). Ca²⁺ sensitization was also not inhibited by chelerythrine (Fig. 9). Since chelerythrine is a non-competitive inhibitor with respect to ATP (Herbert et al. 1990), ATPγS was reduced to 100 μm. Since we noted that ATP inhibits the ATPγS-induced sensitization in particular at low concentrations of ATPγS, the strips were pretreated with hexokinase and glucose to deplete endogenous ATP before treatment with ATPγS in experiments estimating K_m values and with chelerythrine.

Figure 8. Distribution of different signalling proteins in ileal smooth muscle permeabilized with β-escin or skinned with Triton X-100.

A, Ponceau stained SDS-PAGE (15 %) of total muscle homogenate (60 μg lane) taken from β-escin- (1) or Triton-permeabilized (2) ileal strips in relaxing solution; left lane, molecular weight markers. B–F, Western blots using antibodies to RhoA, MLCK, MYPT1, PAK and ROK2. Results are representative of 2–4 independent experiments. ROK2 and ROK1 (not shown) antibodies gave very weak signals for unknown reasons.

Figure 9. The effect of protein kinase inhibitors on ATPγS-induced Ca²⁺ sensitization in Triton-skinned ileal smooth muscle.

Results (means ±s.e.m., n = 3–6) shown are force elicited at pCa 6.74 after incubation with ATPγS at the indicated concentrations alone or with 2 μm staurosporine, 100 μm Y 27632 or 100 μm chelerythrine. Experimental protocol as in Fig. 1. ***P < 0.001; n.s., not significantly different. There was no contraction at pCa 6.74 before treatment with ATPγS.

Effect of GTPγS on ATPγS-induced Ca²⁺ sensitization of force

It has been suggested that phosphorylation of Thr⁶⁹⁵ of MYPT1 by the endogenous kinase caused inhibition of MLCP activity. Thr⁶⁹⁵ is also a major functional site of phosphorylation by ROK (Feng et al. 1999a). We therefore tested whether GTPγS- and ATPγS-induced sensitization were additive. We used GTPγS rather than an agonist to activate all GTP binding proteins. Pretreatment with 100 μm GTPγS alone for 10 min caused subsequent Ca²⁺ sensitization (P < 0.05) after several washes, which is not surprising because GTPγS is poorly hydrolysed and should lead to a prolonged activation of G proteins. The sensitizing effect of GTPγS was, however, much smaller than the sensitization induced by thiophosphorylation with 1 or 3 mm ATPγS (Fig. 10A). Ca²⁺ sensitization induced by pretreatment with 1 mm ATPγS was augmented by addition of 100 μm GTPγS to the ATPγS solution. The effect of GTPγS was inhibited by Y 27632 (Fig. 10A) indicating that it is mediated by ROK. Note, that pretreatment with 3 mm ATPγS did not lead to a larger sensitization than pretreatment with 1 mm ATPγS indicating that the effect of ATPγS on Ca²⁺ sensitivity is maximal at 1 mm ATPγS. We then tested whether addition of 100 μm GTPγS to the pCa 6.16 solution once tension had reached a plateau would further increase force (Fig. 10B). In strips pretreated with both GTPγS and 1 mm ATPγS, force elicited at pCa 6.16 is not further enhanced by 100 μm GTPγS. However, in strips pretreated with 1 or 3 mm ATPγS alone, there was an additive effect of GTPγS on force. There was also a small additional effect in strips pretreated with GTPγS only. These experiments indicate that GTPγS has an additive effect on Ca²⁺ sensitivity even at saturating concentrations of ATPγS of similar magnitude independent or whether it is added during the thiophosphorylation protocol or afterwards to the submaximal Ca²⁺ solution.

Figure 10. GTPγS augments ATPγS-induced Ca²⁺ sensitization.

A, ileal strips were pre-treated in the presence of ML-9 after a submaximal (pCa 6.16) and maximal activation with the indicated conditions. GTPγS and Y 27632 were both 100 μm. Submaximal force before treatment was not significantly different between the different groups and was 16 ± 3 % (mean of all experiments shown in Fig. 10, n = 18). Experimental protocol as in Fig. 1. B, the effect of addition of 100 μm GTPγS at the plateau of submaximal force (pCa 6.16) of strips from A pretreated with the indicated conditions. Force is expressed in percent of submaximal Ca²⁺-activated force. Results shown are means ±s.e.m., n = 3–4; n.s., not significantly different from force elicited by pCa 6.16. **P < 0.01, ***P < 0.001.

DISCUSSION

It has previously been reported that treatment of α-toxin-permeabilized portal vein with ATPγS under conditions that do not support thiophosphorylation of r-MLC induced an increase in Ca²⁺ sensitivity of contraction which was associated with a decrease in the activity of myosin phosphatase most probably due to thiophosphorylation of MYPT1 (Trinkle-Mulcahy et al. 1995). Furthermore, the GTPγS-induced thiophosphorylation of MYPT1 in permeabilized smooth muscle was inhibited by the ROK inhibitor HA1077, which is less specific than Y 27632 (Sward et al. 2000). Therefore, it was tempting to speculate that the ATPγS-induced Ca²⁺ sensitization was mediated by this mechanism. Using similar experimental conditions to those used by Trinkle-Mulcahy and coworkers (1995) the major finding of this study is that Ca²⁺ sensitization induced by treatment with ATPγS is not inhibited by the relatively specific ROK inhibitor, Y 27632, nor by exoenzyme C3, which inhibits Rho. The latter finding also indicates that other effectors of Rho (Bishop & Hall, 2000) do not mediate the ATPγS-induced sensitization. However, the sensitization appears to be mediated by a staurosporine-sensitive kinase(s). In line with this, ³⁵S incorporation into several high molecular mass proteins including the regulatory subunit of MLCP, MYPT1, is inhibited by staurosporine but not by Y 27632. In contrast, Ca²⁺ sensitization induced by the muscarinic agonist carbachol is completely inhibited by the ROK inhibitor with an IC₅₀ value that correlated with the known K_i values for inhibition of ROK (Uehata et al. 1997). The staurosporine protein kinase(s) is tightly associated with the contractile apparatus because treatment of Triton-skinned smooth muscle with ATPγS also induced an increase in Ca²⁺ sensitivity.

Treatment with ATPγS resulted in a staurosporine-sensitive increase in r-MLC phosphorylation at submaximal Ca²⁺ and a 2-fold decrease in the rate of relaxation suggesting that the most likely mechanism of the ATPγS-induced increase in Ca²⁺ sensitivity of force is inhibition of myosin phosphatase as has been suggested (Trinkle-Mulcahy et al. 1995). This is also supported by the observation that there was no ³⁵S incorporation into protein bands corresponding to regulatory or putative regulatory proteins of the contractile apparatus, MLCK, caldesmon and calponin. But we cannot exclude the possibility that the other high molecular mass thiophosphorylated protein bands which we have not yet identified contribute to sensitization.

Still, there remains the theoretical possibility that some remaining activity of ROK mediates the effect of ATPγS. This is because thiophosphorylated proteins are dephosphorylated very slowly and, hence, even very low protein kinase activities would eventually lead to a fully thiophosphorylated protein. However, we believe that this is unlikely since we used concentrations of C3 and Y 27632 in excess of those required for complete inhibition of carbachol-induced Ca²⁺ sensitization. The concentration of Y 27632 was 10-fold higher than required for complete inhibition of the carbachol-induced sensitization. At this high concentration, Y 27632 induced a small inhibition of force in the Triton-permeabilized preparation, most probably due to inhibition of MLCK. ROK is also inhibited by staurosporine with a K_i of 0.022 μm, which is 4-fold lower than that of Y 27632 (Feng et al. 1999b). However, only 100-fold higher concentrations inhibited the ATPγS-induced increase in Ca²⁺ sensitivity indicating that staurosporine does not act through inhibition of ROK. Furthermore, ATPγS also induced a staurosporine-sensitive increase in Ca²⁺ sensitivity in the Triton-permeabilized preparations in which very little immunoreactivity with RhoA or ROK could be detected. Previously, we also probed the presence of low molecular mass GTPases with [α³²P]GTP overlay assays (Satoh et al. 1992). While the amount of radioactivity bound to protein bands with a relative molecular mass between 20 and 22 kDa was similar in intact and β-escin-permeabilized preparations, virtually no radioactivity was bound in the Triton-skinned preparations.

Protein kinases inhibited by staurosporine that have the potential to modulate the Ca²⁺ sensitivity of contraction include PKC (reviewed in Horowitz et al. 1996), PAK (Zeng et al. 2000) and MAPK (Gerthoffer et al. 1996). We have no evidence that the ATPγS-induced increase in Ca²⁺ sensitivity is mediated by PKC for the following reasons: (i) low concentrations of staurosporine which are more specific for PKC had no effect, (ii) the peptide inhibitor of PKC in conjunction with Y 27632 had only a minor inhibitory effect of about 10 %, (iii) we have no evidence for ³⁵S incorporation into the PKC substrate, CPI 17. Finally, unlike in the arterial smooth muscle, phorbol esters did not induce an increase in Ca²⁺ sensitivity in the ileum as has also been reported previously (Sward et al. 2000). Thus, unlike in the arterial smooth muscle (Gailly et al. 1997; Buus et al. 1998), activation of conventional and/or novel PKC isoforms does not seem to play a major role in regulation of contraction in the ileum.

MAP kinase (Gerthoffer et al. 1996) and PAK (Van Eyck et al. 1998) both phosphorylate caldesmon, which could lead to an increase in force at a given Ca²⁺ concentration (for review Arner & Pfitzer 1999). While phosphorylation of MLCK by several protein kinases inhibits the enzyme (Gallagher et al. 1997 for review), phosphorylation by MAPK activates MLCK (Morrison et al. 1996; Nguyen et al. 1999), which could also lead to an increase in Ca²⁺ sensitivity of contraction. As we could not detect thiophosphorylation of MLKC or caldesmon, these kinases appear not to be involved in the ATPγS-induced Ca²⁺ sensitization. Furthermore, the MEK and MAPK inhibitor PD 98059 (Alessi et al. 1995) does not inhibit sensitization of force. There is no specific inhibitor of PAK. However, in the Triton-skinned preparations, which show the ATPγS-induced Ca²⁺ sensitization, there is no immunoreactivity with PAK. Finally, tyrosine phosphorylation does not appear to be involved in the ATPγS-induced Ca²⁺ sensitization although genistein inhibits carbachol-induced sensitization (Steusloff et al. 1995).

Incubation of the permeabilized ileal strips with ATPγS was performed in a nominally Ca²⁺-free solution (pCa > 8) with a high concentration of the MLCK inhibitor ML-9. Under this condition we observed no or very little ³⁵S incorporation into r-MLC. Functionally the ileal smooth muscle remained relaxed after switching from the ATPγS- to ATP-containing solution at pCa > 8 (relaxing solution) indicating that Ca²⁺ sensitization was not due to thiophosphorylation of r-MLC. In contrast, inhibition of phosphatase activity by microcystin-LR induced a contraction in the absence of Ca²⁺ in guinea-pig ileum (this study; Sward et al. 2000), Triton-permeabilized rat caudal artery (Weber et al. 1999) and portal vein (Kureishi et al. 1999). In the arterial preparations this was associated with an increase in phosphorylation of r-MLC and was only inhibited by staurosporine but not by ROK or MLCK inhibitors (Weber et al. 1999; Kureishi et al. 1999). No information is available as to the phosphorylation of MYPT1 in these experiments. We propose that the kinase that mediates the Ca²⁺-sensitizing effect of ATPγS is distinct from the kinase which is unmasked by inhibition of phosphatase activity with microcystin because (i) we did not observe ³⁵S incorporation into r-MLC, and (ii) we did not observe a Ca²⁺-independent contraction. Interestingly in small renal and mesenteric small arteries treatment with ATPγS also induced a small contraction after re-addition of ATP (data not shown). We are currently investigating this difference between the ileum and the small arteries.

As the staurosporine-sensitive kinase is tightly associated with the contractile apparatus it is tempting to speculate that this kinase is identical to the not yet identified protein kinase that is present in partially purified myosin-bound phosphatase and is inhibited by chelerythrine (Ichikawa et al. 1996). In our hands, the ATPγS-induced sensitization of force was not inhibited by chelerythrine. However, this does not exclude the endogenous kinase as candidate because, in vitro, inhibition by chelerythrine is frequently not observed (D. J. Hartshorne and A. Muranyi, personal communication). Both the endogenous kinase and ROK phosphorylate MYPT1 at Thr⁶⁹⁵, which was suggested to be the major functional site (Feng et al. 1999a). It was also shown that GTPγS increased thiophosphorylation of MYPT1 in a ROK-dependent manner under conditions which were non-saturating in respect to ATPγS (Sward et al. 2000). Therefore, we hypothesized that activation of the Rho/ROK pathway with GTPγS under conditions which appeared to be saturating for ATPγS-induced sensitization (100- to 300-fold above the apparent K_m value) should not have be additive. This was clearly not the case. Incubation with GTPγS together with ATPγS had an additive effect on Ca²⁺ sensitivity. The effect of GTPγS was inhibited by Y 27632 indicating that it was mediated through ROK (Fu et al. 1998). Also, addition of GTPγS at the plateau of submaxial Ca²⁺-activated force increased force in strips that were pretreated with saturating concentrations of ATPγS in the absence but not in the presence of GTPγS. We cannot exclude the possibility that thiosphosphorylation of MYPT1 by ATPγS is not maximal although this appears to be unlikely because the effect on force was maximal. Hence, our results suggest that phosphorylation of other sites of MYPT1, e.g. Thr⁸⁵⁰, by ROK (Feng et al. 1999a) may contribute to inhibition of MLCP. Alternatively or even in addition, phosphorylation of other proteins of the contractile apparatus such as calponin and CPI-17 by ROK (Koyama et al. 2000; Kaneko et al. 2000) may contribute to Ca²⁺ sensitization of force. Future work will have to resolve this point. Since the strips remained relaxed after treatment with GTPγS, direct phosphorylation of the r-MLC by ROK, is unlikely (Amano et al. 1996; Iizuka et al. 1999; Sward et al. 2000).

In conclusion we have shown here, that ATPγS induced an increase in Ca²⁺ sensitivity in guinea-pig ileum which was associated with thiophosphorylation of the regulatory subunit of MLCP but not of other regulatory proteins of the contractile apparatus. Unlike microcystin-LR, treatment with ATPγS did not induce a Ca²⁺-independent contraction. The kinase responsible for ATPγS-induced sensitization appears to be distinct from Rho-associated kinase, protein kinase C, genistein-sensitive tyrosine kinases, MAPK and probably p²¹-activated protein kinase (PAK) and is tightly associated with the myofilaments. As carbachol-induced Ca²⁺ sensitization is completely inhibited by the relatively specific ROK inhibitor Y 27632 and partially by genistein, the physiological role of this myofilament-associated kinase remains to be determined.

Note added in proof

Since this work was submitted for publication, the endogenous kinase has been identified as ZIP-like kinase, which could act downstream of ROK, by J. A. MacDonald, M. A. Borman, A. Muranyi, A. V. Somlyo, D. J. Hartshorne & T. A. Haystead (Proceedings of the National Academy of Sciences of the USA 98, 2419–2424 (2001)).