Thiophosphorylation of myosin light chain increases rigor stiffness of rabbit smooth muscle

Abstract

1. The effect of thiophosphorylation of the regulatory myosin light chain (MLC₂₀) on rigor stiffness was determined in permeabilized rabbit bladder smooth muscle.

2. Rigor stiffness of α-toxin-permeabilized smooth muscle was significantly increased by thiophosphorylation of MLC₂₀. This increase may have been due to partial shortening (melting) in the proximal rod region and/or stiffening of the regulatory domain of the myosin head.

3. We suggest that phosphorylation of MLC₂₀, by increasing the stiffness of the S1 lever arm and/or S2 hinge regions of the myosin molecule, favours separation of the two phosphorylated heads and consequent deinhibition of motor domain activity.

Contraction of smooth muscle is initiated by phosphorylation of the regulatory myosin light chains (MLC₂₀) by a site-specific myosin light chain kinase (MLCK) activated by the Ca₄-calmodulin complex (Hartshorne, 1987; Somlyo & Somlyo, 1994). However, the structural changes induced in smooth muscle myosin by phosphorylation, and which are responsible for increased ATPase activity, force development and muscle shortening, are not known. Thiophosphorylation of MLC₂₀ of smooth muscle myosin induces a small, but significant, shortening in the tail region of the myosin molecule (Zhang et al. 1997), which could be associated with a change in the mechanical properties of muscle, possibly through melting of the coiled-coil. It has also been suggested (Uyeda et al. 1996) that changes in the elastic properties of the lever arm, a region of the myosin heavy chain surrounded by the essential and regulatory chains (Rayment et al. 1993; Holmes, 1996), could modify the elementary force generated by the myosin motor. The aim of the present study was to determine whether MLC₂₀ phosphorylation can alter the mechanical properties of cross-bridges in smooth muscle, as estimated with rigor stiffness measurements, used as an averaged index of the mechanical state of cross-bridges attached to actin. A preliminary report has been presented to an American Biophysical Society meeting (Khromov et al. 1998).

METHODS

Bundles of smooth muscle, having homogenous cross-sectional areas along their length (∼150–200 μm in diameter and 2.4–3.6 mm in length), were dissected from bladders of New Zealand White rabbits anaesthetized with halothane and killed by exsanguination through the carotid artery, as approved by the Animal Care and Use Committee. The ends of the strips were clamped in aluminium T-clips and glued (Super Bonder 416, Loctite Corp., Newington, CT, USA) to the tungsten hooks of the transducer (AE 801; Senso Nor A.S., Horten, Norway) and the length-adjusting servo-controlled motor (6800, Cambridge Technology, Watertown, MA, USA). The length of both hooks was reduced as much as possible, in order to eliminate the compliance in the apparatus. The resultant resonant frequency of the transducer in air was 1.0 kHz. The contribution to compliance of the instrumentation (transducer and hook), determined by its response to displacement with a known tungsten bar, was 1.7 × 10⁻² m N⁻¹; this was ∼9 % of the measured compliance of the strip in rigor, and is the lower limit of our underestimate of muscle stiffness.

Experiments were conducted on preparations permeabilized with Staphylococcus aureusα-toxin (130 rat units ml⁻¹ for 45 min at 20°C) in Ca²⁺-activating solution at pCa 6.5 with continuous stirring. After permeabilization the strips were washed in Ca²⁺-free (0 Ca²⁺) relaxing solution containing 2 mm MgATP for 10–15 min.

Solutions

The composition of the relaxing (pCa ≥ 8.0) and Ca²⁺-activating (pCa 5.0) solutions (ionic strength 0.2 m, pH 7.1) is shown in Table 1. The composition of rigor solution was similar, but MgATP was substituted by potassium methanesulphonate such that the ionic strength was 0.2 m. All solutions contained 2 mm free Mg²⁺.

Table 1.

Composition (mm) of solutions

Solution	KMs	MgMs2	Na2ATP	K2H2EGTA	K2CaEGTA	CP	Pipes
Relaxing	61.7	5.42	2.3	10.0	0	10	30
Ca2+-activating	62.3	4.72	2.3	0.25	9.74	10	30
Ca2+-rigor	103.5	2.0	0	0.25	9.74	—	30
0 Ca2+-rigor	102.9	2.0	0	10.0	0	—	30

CP, creatine phosphate; Ms, methanesulphonate.

Protocol

The length of the strips in relaxing solution was adjusted before the experiment, by giving a series of small increases in stretch/release pulses until a change in force was just detectable (< 5 % of maximal force). The mechanical measurements were performed in Ca²⁺-free (0 Ca²⁺) solution on strips in high-tension rigor in which MLC₂₀ was either thiophosphorylated or unphosphorylated (Somlyo et al. 1988). The majority of measurements were performed as paired experiments on the same strip: first in the dephospho- and subsequently in the thiophospho-rigor state. Nucleotides were removed by washing the strips with Ca²⁺-containing rigor solution (pCa 5.0, two to three changes in 10 min), followed by washing with 0 Ca²⁺ rigor solution (three to five changes in 30 min) and the ‘dephospho-rigor’ state was developed. Rigor force was ∼6.3 × 10³ N m⁻². All experiments were conducted at 20°C.

After the strips had been in 0 Ca²⁺ rigor for 30 min, when the decline in rigor force was slow and stable (∼10–15 % of initial rigor force in 30 min), a sequence of release-stretch steps (0.1–2.5 % of the initial length, L₀), generated with the aid of the length-adjusting motor triggered by command signal from a function generator (HP 33120A, Hewlett-Packard, San Diego, CA, USA), was imposed. Both releases and restretches of up to 3 % of L₀ were completed in 0.4 ms. The duration of the release was 10–20 ms and after each release the strip was stretched back to the initial length, allowing rigor force to recover to the original level. After completion of the stiffness measurements in dephospho-rigor, a thiophospho-rigor state was induced by thiophosphorylation of MLC₂₀ (10 min in 2 mm ATPγS, pCa 5.0), followed by removal of Ca²⁺ with 0 Ca²⁺ rigor solution (three to five changes in 30 min), and the stiffness measurements were repeated at a level of force ∼5.8 × 10³ N m⁻². In order to prevent the rise of tension during thiophosphorylation (probably due to residual ATP contamination), hexokinase (10 IU ml⁻¹, Sigma), glucose (2 mm) and a myokinase inhibitor (Ap₅A, 100 μm, Sigma) were included in the rigor solution. The level of the solution in the experimental trough was maintained constant to avoid the influence of surface tension on the amplitude of rigor force. In a few strips, to eliminate the possibility of time-dependent effects, the thiophospho-rigor protocol was carried out immediately after permeabilization, and in other experiments the dephospho-rigor protocol was carried out twice, with the second protocol time-matched to the measurement shown in Fig. 2 and Table 2.

Figure 2. Accumulated (n = 12) length-force plots constructed from the releases imposed on dephospho- (•) and thiophospho-rigor (○) bladder strips.

The initial 40 % of decrease of force in both plots was approximated by linear regression and data are shown together with their confidence limits (P < 0.05). The inset shows the results of fitting of the experimental data to simulated length-force relationships, describing total elasticity of the rigor strip as that consisting of linear in-series with exponential components. Confidence limits for both types of rigor were calculated by varying total stiffness(es) within the limit of ± s.e.m.

Table 2.

Total (s) and exponential (q) components of stiffness of permeabilized bladder smooth muscle estimated by the linear and model-fit methods

Condition	Stiffness (Lo−1)	n
Dephospho-rigor
Model fit	s = 115 ± 5, q= 420± 20	12
Linear fit	s = 127 ± 4	12
Thiophospho-rigor
Model fit	s = 145 ± 9, q= 360± 25	10
Linear fit	s = 166 ± 4	10

MLC₂₀ phosphorylation measurements

After completion of the stiffness measurements, the strips were frozen in liquid N₂ and the levels of MLC₂₀ thiophosphorylation or phosphorylation, determined by two-dimensional electrophoresis, were 86 ± 5 and 15 ± 5 % (means ±s.e.m.; n = 4) for thiophosphorylated and dephosphorylated rigor smooth muscles, respectively. Approximately 5.6 % of apparently doubly phosphorylated MLC₂₀ in femoral artery smooth muscle is non-phosphorylated non-muscle myosin that co-migrates with phosphorylated smooth muscle myosin.

Data processing

The length and force signals were collected with Labview 3.1.1 data acquisition software (National Instruments Corp., Austin, TX, USA) at a sampling frequency of 10 kHz. Before analysis the recorded force signal was digitally filtered (1.0 kHz) to remove oscillations superimposed upon the force trace by undamped resonance of the system (an example is shown in Fig. 1, inset). The amplitudes of the force (F) immediately after the release, normalized to the rigor force (F₀) present before the releases and the change in length (ΔL; normalized to the initial length of the preparation, L₀), were used to construct length-force (L–F) plots. Usually 10–15 points were collected from each preparation for both the thiophospho- or dephospho-rigor states.

Figure 1. Force responses to release-stretch cycles of 0.03 % and 0.3 % of initial length L₀ in permeabilized strips of rabbit bladder smooth muscle in rigor.

The inset shows original (dots) and filtered (lines) force responses to release of 0.3 % of L₀.

With certain assumptions (Warshaw et al. 1988; Arheden & Hellstrand, 1991), the elasticity of muscle can be modelled as consisting of two components in series: one linear, ascribed to cross-bridges, and a second, exponential, ascribed to passive elasticity. The length-force relationship of smooth muscle can then be defined as

where F₀ and F are the rigor force before and immediately after the release of the strip from initial length L₀ to L, s is a normalized total stiffness at F₀ and q is a constant, describing the exponential stiffness of the passive component. Consequently, the linear compliance attributed to cross-bridges is 1/s–1/q. The experimental data were fitted to this equation with Sigma Plot 2 (San Rafael, CA, USA) using an iterative non-linear least-squares algorithm (Marquart, 1963).

Two approaches were used to analyse rigor stiffness measurements. For the ‘linear’ method, stiffness was estimated as the slope of the initial part (where after release force dropped by less than 40 % of maximal force) of the normalized length-force plot. In the ‘model-dependent’ method, the experimental data from all imposed releases were fitted to the model of length-force relationship described above. The advantage of the second approach, albeit being model-dependent, is that it yields an estimate of the distribution of the total stiffness between linear (cross-bridge-related) and exponential (passive) parts. Application of the model to data derived only from muscles in rigor is consistent with the assumption, not necessarily applicable when comparisons are from rigor with data from actively contracting muscles (Arheden & Hellstrand, 1991), that detachment of cross-bridges does not contribute to the apparent series elastic compliance. Both analyses were performed on the accumulated normalized length-force plot created by superposing the individual plots constructed separately for each strip in dephospho- or thiophospho-rigor.

Statistics

The linear method data were compared by Student's t test. P < 0.05 was taken as significant. Data are presented as means ±s.e.m.; n is the number of measurements.

RESULTS

The force responses to release steps are shown in Fig. 1; the normalized length-force plots obtained in dephospho- and thiophospho-rigor (Fig. 2) are curvilinear in the range of releases 0–0.025 of L₀. Similar results have been obtained in studies of taenia coli smooth muscle and single smooth muscle cells, probably due to the non-linearity of a series elastic component and tension recovery processes (Warshaw et al. 1988; Arheden & Hellstrand, 1991). Compliance, defined as the relative amount of shortening necessary to reduce force to basal level, was 0.0085 L₀ in dephospho-rigor, close to that measured in Triton X-100-permeabilized guinea-pig taenia coli in rigor (0.0078 L₀ in Arheden & Hellstrand, 1991), and somewhat less than in other studies (ranging from 0.017 to 0.012 of L₀ in Mulvany & Warshaw (1981), Pfitzer et al. (1982) and Warshaw & Fay (1983)). Possible reasons for such a discrepancy may include differences in rates of release and/or mounting procedures, and tissue specificity.

The initial part of the plot in Fig. 2 can be approximated as linear for both curves (r² > 0.95). In both thiophospho-and dephospho-rigor, force reached basal level at releases of about 0.01 of initial length; a certain amount of negative tension (up to 20 % of maximal rigor force) developed with larger releases. The two curves corresponding to dephospho-and phospho-rigor states converged at shortening steps larger than 1.5 % of L₀. The results of the model-dependent stiffness estimates are summarized in Table 2, and the linear approximation of the initial part of both plots, together with the confidence limits (P < 0.05), are shown in Fig. 2.

The analysis based on the model shows that only the linear, and not the exponential component, which is presumably related to passive elasticity, was significantly increased by thiophosphorylation of MLC₂₀. The inset in Fig. 2 displays the difference in the shape of the simulated linear-force relationships for the dephospho- and thiophospho-rigor state. The relative increase in total stiffness was 30 or 26 %, depending on whether the linear or model approximation was used for the calculations; the relative increase in the linear component of stiffness (reciprocal of 1/s - 1/q; see Methods) was 53 %.

Controls

As described in the Methods, the stiffness measurements were usually performed ∼30 min later on the thiophosphorylated than on the dephosphorylated preparations. At this time, force in the dephospho-rigor state was not significantly different, but did show a slight tendency to be higher than in the thiophospho-rigor state (ratio of forces was 1.2 ± 0.2; n = 12). In order to determine whether this time difference and/or the additional decrease in rigor force (10–15 % of initial rigor force in 30 min) during prolonged incubation in rigor solution affected our results and caused an increase in stiffness, we performed control stiffness measurements on dephosphorylated preparations incubated in rigor solution for an additional 30 min. Although rigor force declined by ∼15 % during incubation, no significant difference in normalized stiffness was found between two consecutive estimates: total stiffness was 126 ± 3 L₀⁻¹ and 134 ± 3 L₀⁻¹ (n = 3, linear method) for the first and the second measurements, respectively. Additional control experiments were performed to determine whether binding of ATPγS to rigor cross-bridges, rather than thiophosphorylation of MLC₂₀, could affect the results. For this purpose, after the dephospho-rigor stiffness was measured, the preparations were incubated with 2 mm ATPγS in the absence of Ca²⁺ for 10 min and after washing out ATPγS over ∼30 min in the 0 Ca²⁺ rigor solution, stiffness was measured again. No significant difference between the first and the second measurement was found: s = 124± 4 L₀⁻¹vs. 129 ± 5 L₀⁻¹ (n = 3, linear method).

DISCUSSION

The increase in rigor stiffness caused by thiophosphorylation of MLC₂₀, the major finding of this study, indicates that this phosphorylation can modify an elastic element residing in smooth muscle myosin. The increase in total stiffness, 26–30 %, would appear modest, but is a very significant decrement (∼53 %) in the linear component of compliance. The change localized within the part of the molecule (S1-S2) projecting from the thick filament may be even greater than the above value if, as in striated muscles (review in Goldman & Huxley, 1994), a large part of the linear compliance resides in myosin and actin filaments. We will attempt, in the following discussion, to relate this finding to the mechanism of force generation.

Smooth muscle myosin ATPase is activated by actin upon phosphorylation of MLC₂₀; this phosphorylation removes the repressor effect of the non-phosphorylated light chain (Hartshorne, 1987; Cremo et al. 1995; Sata et al. 1996). The structure of myosin (Rayment et al. 1993) indicates that this mechanism requires information to be transmitted from the C-terminus of the S1 region of the myosin heavy chain, where MLC₂₀ is bound, to the catalytic domain of the myosin head. Furthermore, regulation requires both heads of myosin to be present, as single-headed myosin has ATPase activity even in the absence of regulatory light chain phosphorylation (Cremo et al. 1995; Sata et al. 1996). The specific changes that signal to the catalytic sites are not known, although cryo-atomic force microscopic studies show that MLC₂₀ phosphorylation increases the separation of the two heads of the myosin dimer relative to each other (Zhang et al. 1997) and phosphorylation inhibits the co-operativity of MgADP binding to myosin (Wu & Cremo, 1998). No immediately obvious and verifiable biochemical mechanism other than, perhaps, the removal of negative co-operativity between the two myosin heads (Persechini & Hartshorne, 1981; Sellers et al. 1983), connects MLC₂₀ phosphorylation to activation of actomyosin ATPase.

Thiophosphorylation of MLC₂₀ reduced the compliance of the force-transmitting elements. Given the high selectivity of (Ca²⁺-dependent) myosin light chain kinase for MLC₂₀, we can ascribe this, with some confidence, to a change in the linear component of rigor stiffness residing within the myosin molecule. In view of the similar amplitudes of rigor force in thiophosphorylated and non-thiophosphorylated muscles, it is unlikely that the different stiffnesses found in the two states were caused by differences in the number of cross-bridges attached, thought to be near 100 %. Shortening in the hinge region, at the subfragment 2 and light meromyosin (LMM) juncture, thought to reflect a helix coil transition, is the first relevant structural change known to be induced by thiophosphorylation (Zhang et al. 1997), but we have no direct evidence about its effect on myosin compliance. The lever arm is the second site where MLC₂₀ phosphorylation could increase stiffness (Uyeda et al. 1996). Indeed, a contribution of the regulatory light chain to the stiffness of the C-terminus of the long α-helix of S1 is indicated by electrical birefringence measurements (Eden & Highsmith, 1997) and by the loss in motility caused by the removal of MLC₂₀ from myosin molecules (VanBuren et al. 1994; Lowey & Trybus, 1995). A phosphorylation-induced tightening of MLC₂₀ that is wrapped around the C-terminus of the α-helix (lever arm) could increase its stiffness.

Current theories of cross-bridge cycling, based on structural findings (Whittaker et al. 1995; Gollub et al. 1996; review in Holmes, 1996), suggest that during the force-generating transition, the regulatory domain (‘lever arm’) of the myosin head undergoes rotation, while the immobile motor domain is strongly bound to actin. Effective transmission of force requires stiffness of both the lever arm and of the contiguous S1-S2 region of the myosin rod, which is connected to the myosin filament backbone by a second hinge that is thought to allow the myosin head to swing out from the backbone and allow force generation over a greater range of filament separations (Huxley, 1969). In smooth muscle myosin the movement of the lever arm is particularly sensitive to MgADP (Whittaker et al. 1995), consistent with the high MgADP affinity of smooth muscle cross-bridges (Nishiye et al. 1993).

Increased separation of two heads would not, by itself, increase rigor stiffness, but a stiffer lever arm could allow structural methods (Zhang et al. 1997) to capture the two catalytic domains of the myosin dimer in positions less close to each other. (Consider the case of two balls connected to a fulcrum by, respectively, strings or stiff rods.) Separation of the heads could be favoured by electrostatic repulsion resulting from the two negative charges added by phosphorylation if, and only if, the phosphorylation sites (Ser 19) of the two MLC₂₀ are closely adjacent. It is interesting that the activating effect of phospho-ser 19 can be mimicked, at least partially, by negatively charged amino acid substitution in the same position (Sweeney et al. 1994; Kamisoyama et al. 1994). Such separation of the catalytic domains may have a regulatory effect by removing negative co-operativity, if present, between the two heads of unphosphorylated myosin (Persechini & Hartshorne, 1981).

We conclude that thiophosphorylation of MLC₂₀ increases a linear component of stiffness residing in the myosin molecule, and our results are consistent with the possibility (Linari et al. 1998) that, at least in smooth muscle, the compliance of a given cross-bridge may vary with its functional states.