The myosin interacting-heads motif present in live tarantula muscle explains tetanic and posttetanic phosphorylation mechanisms

Significance

Muscle contracts when myosin heads on the thick filaments bind to actin in the thin filaments, producing force, thin filament sliding, and sarcomere shortening. Two important questions on how muscle works are: 1) How myosin heads of a thick filament are “turned on” to enable their binding to actin; and 2) how, after a tetanus, a muscle produces a stronger twitch force than the twitch force produced before the tetanus (posttetanic potentiation). Muscles from tarantula provide insights into both questions: 1) Tarantula thick filaments are not primarily activated by mechanical stress, as in vertebrates, but by phosphorylation; 2) posttetanic potentiation in tarantula muscles occurs by phosphorylated heads that remain released and disordered.

Abstract

Striated muscle contraction involves sliding of actin thin filaments along myosin thick filaments, controlled by calcium through thin filament activation. In relaxed muscle, the two heads of myosin interact with each other on the filament surface to form the interacting-heads motif (IHM). A key question is how both heads are released from the surface to approach actin and produce force. We used time-resolved synchrotron X-ray diffraction to study tarantula muscle before and after tetani. The patterns showed that the IHM is present in live relaxed muscle. Tetanic contraction produced only a very small backbone elongation, implying that mechanosensing—proposed in vertebrate muscle—is not of primary importance in tarantula. Rather, thick filament activation results from increases in myosin phosphorylation that release a fraction of heads to produce force, with the remainder staying in the ordered IHM configuration. After the tetanus, the released heads slowly recover toward the resting, helically ordered state. During this time the released heads remain close to actin and can quickly rebind, enhancing the force produced by posttetanic twitches, structurally explaining posttetanic potentiation. Taken together, these results suggest that, in addition to stretch activation in insects, two other mechanisms for thick filament activation have evolved to disrupt the interactions that establish the relaxed helices of IHMs: one in invertebrates, by either regulatory light-chain phosphorylation (as in arthropods) or Ca²⁺-binding (in mollusks, lacking phosphorylation), and another in vertebrates, by mechanosensing.

Muscle contraction results from the force exerted by the myosin heads on Ca²⁺-activated thin filaments, following release of the heads from the thick-filament backbone. We want to understand how the helically ordered heads on the relaxed thick filament sense the Ca²⁺-activation of the thin filaments so that their release is coordinated with thin filament activation. Initial evidence for such a “thick-filament activation” mechanism was experimentally found in 2015 when mechanosensing was proposed for vertebrate muscle (1). Our first goal here is to inquire if a vertebrate-like mechanosensing mechanism is also present in muscles from invertebrates.

The structure of thick filaments from vertebrate and invertebrate muscle has been studied extensively by electron microscopy (EM) negative staining (2, 3) and low-angle X-ray diffraction (XRD) (4–6). Invertebrate filaments (Limulus, scallop, tarantula) were found to be particularly amenable to structural studies, as their heads were well ordered in true helices, in contrast to the poorer ordering in pseudohelices found in vertebrates (7). Three-dimensional (3D) recontructions from frozen-hydrated tarantula thick filaments showed that the helices of heads on the filaments were formed by a two-headed assemblage that we called the myosin interacting-heads motif (IHM) (8, 9), formed by the interaction of a blocked head (BH) and a free head (FH) (10) with the subfragment-2 (8) (Fig. 1A). The head–head interaction was essentially identical to that found in single molecules of myosin II in the relaxed state (10, 11). The switch-2 “closed” state of the myosin head was found to be required for establishing the helical arrangement of IHMs around the thick filament backbone, as shown by XRD (12) and EM (13). The IHM has been found to be present in all animals studied so far by EM of isolated myosin molecules (14, 15) and isolated thick filaments of skeletal, cardiac, and invertebrate smooth muscle (16). In all thick filaments the IHMs are located approximately parallel to the filament axis, the only exception being insect flight muscle (Lethocerus), with a perpendicular IHM (17). The IHM has been key to the concept of superrelaxation (18) and understanding the structural pathogenesis of hypertrophic and dilated cardiomyopathy (19). We show that an IHM-based structure (Fig. 1C) [PDB ID code 3JBH (20)] explains the XRD pattern from relaxed skinned tarantula muscle (21). Our second goal here is to assess if the IHM is present in living relaxed tarantula muscle as well as skinned muscle by recording XRD patterns from resting leg muscle in the whole animal (in vivo) or from excised legs (ex vivo).

Fig. 1.

Tarantula thick-filament 3D reconstruction and model for interpreting XRD patterns. (A) Longitudinal and transverse views of 2.0-nm resolution frozen-hydrated 3D map [EMD-1950 (20)] showing densities of four helices of myosin heads forming IHMs (tan). Two IHM densities are highlighted with BH (green) and FH (blue), and subfragment-2 (pink). IHMs are arranged every 14.5 nm in crowns, with densities of the backbone including myosin tails (gray) and a central PM core (orange). (Scale bar, 14.5 nm.) (B) Densities of the 3D map composed of (a) IHM crowns (tan), and (b) a backbone, comprising (c) an annulus of tails (gray), and (d) a central PM core (orange). (Scale bar, 14.5 nm.) (C) Pseudothick filament model with heads represented by PDB 3JBH (shown in A), and tails modeled using the tarantula MHCII sequence [GenBank KT619079 (20)].

The modulation of skeletal muscle contraction by myosin light-chain phosphorylation has been extensively studied (22, 23). Studies on purified myosin from Limulus skeletal muscle (24) [an arthropod, like tarantula, exhibiting a similar thick-filament structure (25)] showed it was regulated by a Ca²⁺-calmodulin myosin light-chain kinase (MLCK)-dependent phosphorylation of its myosin regulatory light chain (RLC), and being dephosphorylated by a myosin light-chain phosphatase (MLCP). While myosin from vertebrate skeletal and cardiac muscle can also be phosphorylated, however, there was no direct effect on the in vitro actin-activated Mg-ATPase activity (26). In purified myosin from tarantula muscle, unlike Limulus, phosphorylation does not directly regulate myosin activity but is associated with enhanced actin-activated myosin Mg-ATPase activity (27). EM of negatively stained thick filaments showed that RLC phosphorylation induced the release and disordering of the helically ordered heads (27), consistent with changes in the equatorial X-ray intensities from skinned tarantula muscle (28) and with similar studies of Limulus (29) and rabbit skeletal muscle (30, 31). Thus, in vitro, release of heads upon phosphorylation is associated with either switching on of thick filament activity (Limulus) or enhancement of thick filament activity (tarantula and vertebrate). Further structural studies showed that tarantula thin filaments are regulated by actin/tropomyosin, as in vertebrate skeletal muscle (32).

Tarantula muscle has a phosphorylatable RLC (27) that can be mono- or biphosphorylated (33) on serines Ser35 and Ser45, as detected by mass spectrometry (MS) (34). On this basis, an IHM-based phosphorylation mechanism was proposed in which phosphorylation of the FH makes possible phosphorylation of the BH from the next myosin molecule along the helix, the so called cooperative phosphorylation activation (CPA) mechanism (34, 35), which can account for the thick-filament structural changes we observed in tarantula filament homogenates upon Ca²⁺-activation (27). In this mechanism (Fig. 2B), the selective action of MLCK on the RLC of the BH and FH (Fig. 2A) allow their sequential cooperative release from the thick-filament backbone by Ser35/Ser45 mono- or biphosphorylation (Fig. 2B).

Fig. 2.

CPA mechanism proposed for tarantula skeletal muscle. (A) Biochemical pathway of myosin RLC phosphorylation/dephosphorylation by MLCK/MLCP. (B, Left) Precursor IHM with FH (blue) and BH (green), both with non-P RLCs. Initially, PKC constitutively mono-Ps FHs at Ser35 creating the relaxed state (a), in which FHs can sway away and back by Brownian motion (swaying heads, double curved arrowhead). BHs are not phosphorylated because of steric hindrance from the FH of the next crown. (b) In the active state, when [Ca²⁺] is high for a brief time, swaying FHs can bind actin and produce force. (c) If [Ca²⁺] remains high for a longer time, swaying FHs are biphosphorylated by MLCK on Ser45. These bi-P FHs cannot dock back onto their partner BHs, exposing the Ser45 of the neighbor BHs to allow MLCK to mono-P them at Ser45, so the BHs become swaying heads as well (double curved arrowhead in d). (e) After a longer [Ca²⁺] exposure, most FHs are bi-P and BHs Ser45 mono-P.

The CPA mechanism (Fig. 2B) proposes that the FHs of relaxed muscle are constitutively monophosphorylated (mono-P) by PKC on Ser35 (34) (Fig. 2 B, a), supported by our ¹⁸O MS studies that show that on Ca²⁺-activation, de novo ¹⁸O-ATP incorporation occurs only on Ser45 but not on Ser35 (36). In vitro motility assays show that unregulated rabbit F-actin filaments slide along relaxed isolated thick filaments of tarantula muscle in the absence of Ca²⁺ (34). We therefore proposed that the Ser35 mono-P allows the FHs to move away from and toward the filament backbone by Brownian motion, becoming so-called swaying heads (Fig. 2 B, a, double curved arrows) (34). Constitutively ON heads (1) [or sentinel heads (37)], functionally similar to the swaying heads of tarantula, have also been hypothesized for vertebrate thick filaments, supported by in vitro motility evidence (38). In the CPA model, the BHs are nonphosphorylated because of steric blocking of their Ser35 by the FHs (Fig. 2 B, a) (34, 35). In the presence of Ca²⁺ (Fig. 2 B, b), thin filaments become activated, and the swaying (Ser35 mono-P) FHs produce initial force. Simultaneously, Ca²⁺ activates MLCK and the FHs slowly become phosphorylated on Ser45 (adding to their Ser35 phosphorylation, so that they are fully released from the filament backbone) (Fig. 2 B, c). These biphosphorylated heads cannot dock back to the filament as their RLC N-terminal extensions (NTEs) containing the phosphorylatable Ser45s are elongated (39) and rigid (40) (Fig. 2 B, c), which can only be reversed by dephosphorylation by MLCP. This semipermanent release of the FHs removes the steric hindrance of the BH Ser45, so that some BHs become mono-P by MLCK on Ser45 and are released and sway away (due to the mono-P) from the filament backbone (Fig. 2 B, d) and are now able to produce additional force (Fig. 2 B, e). Brito et al. (34) suggested that this mechanism may be limited to arthropods, which—as with tarantula—also exhibit RLCs with long NTEs having consensus sequences for two constitutive/potentiating (PKC/MLCK) phosphorylatable serine pairs, in contrast with vertebrate muscles, with short NTEs and a single MLCK-phosphorylatable serine. As the CPA mechanism was devised in the context of filament homogenates or isolated filaments, our third goal here was to test it in live muscle by assessing its structural predictions, using time-resolved synchrotron XRD patterns during a tetanus to evaluate the location and disorder of the heads, and measuring the non-, mono-, and bi-P RLC phosphorylation using urea-glycerol (U-G) gels.

It has been known for over 150 y (41) that force produced by skeletal muscle can be potentiated by its previous activity. There have been several explanations proposed for this so-called posttetanic potentiation (PTP), with the early work reviewed in ref. 42. The potentiation of force in vertebrate skeletal muscle stimulated either at lower frequency (treppe or staircase) or higher frequencies resulting in fused tetani (reviewed in refs. 22, 23, and 43), as well as postactivation potentiation and postactivation performance enhancement in human muscle (reviewed in ref. 44), remain an important target of physiological and biophysical research. Our last goal here was to explore the structural basis of PTP in tarantula muscles using XRD.

Here we report: 1) That in thick filaments of live tarantula muscle, the backbone elongation induced by muscle force is much smaller than in frog, implying that mechanosensing is either absent or only of minor importance in tarantula thick filament activation; 2) the IHM structure is present in live, relaxed tarantula muscle and fully explains the observed XRD pattern; 3) there is structural evidence in live tarantula muscle for an IHM-based phosphorylation mechanism of thick filament activation; and 4) this mechanism also provides a structural basis for PTP.

Results and Discussion

The IHM Structure Is Present in Relaxed Live Tarantula Muscle.

We recorded XRD patterns from muscles of the femur either in the intact, living animal (in vivo), in whole legs removed from the animal (ex vivo), or with the muscle removed from the femur and skinned (Fig. 3 A–C). The in vivo preparation (Fig. 3E) had the advantage that the femur muscles are intact and, in contrast to the ex vivo preparation (Fig. 3D), irrigated by hemolymph carrying O₂ and metabolites. The in vivo and ex vivo patterns are essentially identical, displaying (see labels in Fig. 3A): nine 43.35-nm–based myosin layer lines (MLL1 to MLL9); three meridional reflections due to the 14.45-nm–based axial separation of myosin heads (M3, M6, M9); four troponin meridional reflections (Tn1 to Tn4) at 37.21, 19.12, 12.76, and 9.56 nm; two actin ∼35.4-nm–based layer lines at 5.9 and 5.1 nm (ALL6 to ALL7); and four equatorial reflections 10, 11, 20, and 30. The myosin layer line intensities (I_MLL1-I_MLL6) of the ex vivo and skinned XRD patterns (Fig. 4A) after background subtraction (45) show peaks at the same reciprocal radii, indicating that skinning has no effect on the average radial position of the myosin heads above the backbone. The axial spacings of the M3 and M6 meridional reflections (S_M3 and S_M6) are 0.11- and 0.05-nm longer, respectively, than in vertebrate (frog) muscle (1), consistent with other invertebrate muscles (46). The M3 and M6 splitting observed in frog, due to the interference between myosin head arrays on opposite ends of the thick filaments (1), was not detected, possibly because the thick filaments in tarantula muscle are much longer, about 4 to 5 µm (47), so that the interference fringes become too close to be resolved (48, 49). Tarantula skeletal muscle lacks the accessory protein MyBP-C (50) so there was no meridional reflection at 44.2 nm.

Fig. 3.

XRD patterns of muscles of tarantula femurs. (A) Ex vivo pattern of resting relaxed femur muscles. Equatorial, meridional and off-meridional intensity profiles in blue (Top), orange (Left), and green (Right); 23 °C, exposure 200 ms. (B) Pattern of in vivo femur muscles recorded directly from the femur of an intact live tarantula; 23 °C, exposure 1,000 ms. (C) Pattern of a skinned femur muscle in relaxing solution; 23 °C, exposure 1,000 ms. (D) Set-up for recording patterns from ex vivo femur (seen through yellow window in chamber), allowing electrical stimulation and tension recording through a metallic linker (hook near the knee) connected to a transducer. (E) Set-up for recording patterns from in vivo femur muscle.

Fig. 4.

Comparison of I_MLL1-I_MLL6 profiles with the IHM-based thick-filament model, and frozen-hydrated 3D-map densities. (A) I_MLL1-I_MLL6 profiles of ex vivo (red) and skinned (purple) XRD patterns with the IHM-based model (green), and the densities of the 2.0-nm resolution 3D map EMD-1950 (Fig. 1B; blue). Note excellent agreement with the strong layer lines (M1 and M4). Imperfect agreement for layer lines 2 and 5 is of minor significance, given the low intensity on these layer lines. Lattice sampling at 10, 11, and 20 is shown by black arrows. (B) Comparison of the I_MLL1-I_MLL6 profiles calculated from the 2.0-nm resolution 3D-map densities (blue), with the partial contributions from the backbone alone (orange) and only its PM core (brown).

Analysis of the XRD patterns in relaxed and contracting tarantula muscle is simplified by the almost total absence of lattice sampling [due to random rotational orientations of the thick filaments (5)], unlike vertebrate skeletal muscle where the filament organization gives rise to significant lattice sampling. We used a thick-filament model (Fig. 1C), based on a cryo-EM 3D reconstruction that clearly resolves the FHs and BHs of the IHM (Fig. 1 A and B, a), to assess the separate contribution of BHs and FHs to the XRD pattern. To evaluate the correctness of this IHM-based model as a description of the thick-filament structure in the live muscle lattice, we compared the calculated I_MLL1-I_MLL6 (with all FHs docked) of this IHM-based model with those of the ex vivo and skinned relaxed muscle (Fig. 4A). We also did the comparison with the densities of the 2.0-nm 3D map, used to fit the 3JBH IHM-based model (20) (Fig. 1 B, a) and the higher resolution (1.3 nm) (51) thick-filament frozen-hydrated 3D maps (Fig. 4A and SI Appendix, Fig. S1). The striking similarity of the layer line profiles shows that the average position of the heads above the backbone and their conformation in ex vivo and skinned muscle is very similar to the ones in both cryo-EM 3D density maps and in the 3JBH-derived IHM-based model. The correctness of the fitting of the model to the observed XRD intensities was assessed using the crystallographic R-factor (Materials and Methods). The R-factor for the IHM-based model and the ex vivo data was low (4.9%), while those for the model and the 2.0- and 1.3-nm 3D-map densities were 4.0% and 8.4%, respectively, indicative of generally good fits. There are two significant differences between the observed and calculated layer line profiles: 1) The strong, narrow M3 meridional reflection in both experimental XRD patterns, and 2) the sampling at the positions of the equatorial 10, 11, and 20 reflections in the ex vivo MLL3 (Fig. 4A, arrows), caused by the filament lattice in sarcomeres, while all calculations were performed for a single thick filament in the absence of a lattice. The first difference is because the interference of the X-rays scattered by neighboring thick filaments in the ordered lattice results in a higher meridional peak on MLL3, as explained in Koubassova and Tsaturyan (45). The second difference is because the filament lattice in ex vivo is much better ordered than in skinned muscle.

We conclude that in tarantula live muscle, the 3D-maps obtained by cryo-EM and the IHM-based model derived from these maps fully explain the recorded XRD pattern. Thus, the thick filaments in live tarantula muscle have a structure similar or identical to that observed by cryo-EM, in which myosin heads are in the IHM conformation. This contrasts with the conclusion that IHM-based thick filaments are not consistent with XRD patterns of bony fish skeletal muscle and Lethocerus flight muscle (52). The reason for this discrepancy is not yet clear, as cryo-EM studies unambiguously show the presence of the IHM structure in isolated, frozen-hydrated Lethocerus thick filaments, while no cryo-EM studies have yet been reported on fish thick filaments. Because studies of isolated thick filaments in relaxing conditions from a broad evolutionary range of species show the IHM as a prominent feature (16), our results suggest that it is likely that this will also be the case with all these species in vivo.

Time-Resolved XRD Suggests No Significant Change in Backbone Length in Contracting Tarantula Muscle.

In SI Appendix, Supplementary Results, we show that we can use our 3D tarantula thick-filament reconstructions to demonstrate that the M6 meridional reflection contains information from both the thick-filament backbone and myosin head periodicities (SI Appendix, Fig. S1), so changes in S_M3 can be used as a measure of thick-filament length changes. The time course of S_M3 (Fig. 5B) shows a small, statistically not significant, decrease of 0.02 nm (0.13%) during the tetanus, while S_M6 (Fig. 5C) increases by 0.02 nm (0.27%) during force development (Fig. 5A). This increase of S_M6 in tarantula is much smaller than the 1.8% increase for frog muscle, associated with mechanosensing (49), whereas the 0.13% decrease in S_M3 in tarantula contrasts with a 1.61% increase in frog. In frog it has been proposed that a small population of constitutively ON motors generates a mechanical stress in the thick filament that lengthens the filament and releases the remaining motors from the OFF state, allowing the development of the full isometric force (1). For tarantula this would imply that the Ser35 mono-P swaying FHs should generate a similar stress releasing the remaining FHs and BHs. However, the lengthening of the tarantula thick-filament backbone as estimated from S_M6 is only 15% of the lengthening reported for frog. The very small S_M6 change that we observe in tarantula suggests that the thick-filament backbone in these muscles is very rigid, consistent with the presence of a paramyosin (PM) core that is absent in frog. We conclude that the small change in the S_M6 and S_M3 in tarantula compared with frog suggests that a mechanosensing mechanism based on changes in backbone length would, if present, be of minor importance or at a minimum, be very different from the vertebrate mechanism.

Fig. 5.

Time-resolved XRD patterns of an ex vivo femur during a twitch, and before, during, and after a tetanus. (A) Force during a supramaximal stimulus train eliciting a pretetanic twitch (not visible due to small size), a 60-Hz 2-s tetanus, and three posttetanic twitches at 4, 7, and 13 s (visible due to potentiation). Time course showing S_M3 (B) and S_M6 (C) changes, with right Insets showing statistics (M3 n = 5, M6 n = 6) in the resting (black circles, green horizontal bar on A), plateau (black squares, red bar), and posttetanic (post-T; black triangles, blue bar) 12 to 13 s. (D) XRD patterns of resting, plateau, 12 to 13 s and 5 min after plateau (exposure 200 ms each). (E–H) Time courses of d₁₀ and d₁₁ (E), I₁₀ (F), I₁₁₊₂₀ (G), and I₁₁₊₂₀/I₁₀ ratio (H) during 13.5 s. (I–L) I_M3 (I), I_MLL4 (J), I₁₁₊₂₀ (K), and I₁₀ (L) during the resting, tetanic plateau, 12 to 13 s, and 2 to 3, 4 to 5, and 6 min after the end of the tetanus

The Myosin RLC Is Highly Mono-P and Bi-P during the Tetanic Plateau in Tarantula Muscle.

Phosphorylation and dephosphorylation of the myosin regulatory light chains via MLCK, MLCP, and calmodulin has been a common feature of vertebrate and invertebrate skeletal muscle (53), including tarantula (50), since the earliest times of evolution. In tarantula, RLC phosphorylation has been demonstrated using multiple complementary techniques: SDS/PAGE gels (27, 33–35), U-G gels (27, 33–35), autoradiography (27, 33), Western blots using anti-RLC antibodies (35), iso-electric focusing (54), and MS detection of phosphorylatable serines (34–36). RLC phosphorylation was investigated in tarantula filament homogenates using U-G gel electrophoresis, which allows discriminating RLC molecules by their charge as non-P, mono-P, and bi-P, detected as three gel bands (27, 33–35), with the relative quantities of phosphorylated RLC molecules quantified by the relative optical densities of the bands. The BH and FH RLC molecules are distributed between the three bands according to the activation state of the preparation, and their numbers can be accounted for by the CPA mechanism (20). We used a similar approach to determine the phosphorylation levels in rapidly frozen tarantula muscles at the end of a tetanic plateau (SI Appendix, Fig. S2 A, b), showing a substantial increase in mono-P and bi-P and almost complete disappearance of non-P as compared with relaxed live muscles (SI Appendix, Fig. S2 A, a and B, a). These changes contrast with the ones we have seen on Ca²⁺-activation in filament homogenates (35) (SI Appendix, Fig. S2 B, b vs. SI Appendix, Fig. S2 B, a), suggesting that the MLCK is more effective in the ex vivo muscle because of its higher activity and concentration in the sarcoplasm and possible structural advantages of the sarcomere lattice in vivo (SI Appendix, Fig. S3A).

Time-Resolved XRD and U-G Gels Add Key Evidence for an IHM-Based Phosphorylation Mechanism.

To test the IHM-based CPA mechanism structurally (Fig. 2B), we determined the radial distance of the heads from the backbone and their degree of disorder by recording seven time-resolved XRD patterns (Fig. 5) from ex vivo leg muscles. Patterns were recorded before and during a stimulus train that elicited an initial twitch (pretetanic), followed by a tetanus, and then several posttetanic twitches (up to 13.5 s after the end of the tetanus) (Fig. 5 E–H), and finally 2 to 3, 4 to 5, and >6 min later, during recovery after the stimuli (Fig. 5 I–L). I_MLL4, the second strongest layer line coming from the helical organization of myosin heads was used as a measure of helical order; and a decrease in I₁₀ and increase in I₁₁₊₂₀ were used as indicators of movement of heads away from the thick-filament surface.

During the tetanus plateau, I_MLL4 decreased to a low (but nonzero) residual value compared to the maximum value before the tetanus (Fig. 5J), while I₁₀ decreased to a low value (Fig. 5 F and L) and I₁₁ increased to a high value (Fig. 5 G and K). We conclude that the heads progressively move away from the thick-filament backbone toward the thin filaments during the tetanus and, in the process, become disordered. The presence of residual I_MLL1 intensity (∼20 to 25%) during contraction has been reported in electrically stimulated skeletal muscle from frog (55, 56) and mouse (57) during tetanic contraction and has been interpreted as indicating that during the tetanus plateau ∼50% of heads are disordered while ∼50% remain ordered in their helical positions as in the resting muscle. Here we interpret the observed residual I_MLL4 and equatorial changes in tarantula similarly to refs. 55 and 56, but in the context of the CPA mechanism (Fig. 2B) and the IHM-based thick-filament structural model (Fig. 1C). First, we interpreted the observed XRD pattern changes (Fig. 5) by associating the disordered heads as FHs and the ordered ones as BHs (Fig. 2 B, c–e). We then calculated the effect on I_MLL1 and I_MLL4 of progressively removing FHs (SI Appendix, Fig. S3A; compare with SI Appendix, Fig. S3B) for interpreting the observed I_MLL1 and I_MLL4 changes (Fig. 5). The purpose was to simulate the swaying of the FHs between a docked position in the ordered IHM for a given percentage of the time (t_IHM) and a released (disordered) position for the balance of the time (t_rel); the duty cycle [t_IHM/(t_IHM + t_rel)] for the ensemble of heads would correspond to the fraction of FHs remaining docked in the simulation (SI Appendix, Fig. S3C), being greatest if the FH spends all of its time in the IHM and least if it spends all of its time released (SI Appendix, Fig. S3C) (19). The simulation results for I_MLL1 and I_MLL4 (SI Appendix, Fig. S3C) showed residual values of ∼34% and 14%, respectively, when all FHs were removed (while all BHs are still present) (SI Appendix, Fig. S3B), consistent with our proposal that the residual I_MLL1 and I_MLL4 come from the (more stable) BH helices (SI Appendix, Fig. S3B), while all bi-P FHs are disordered during a tetanus. Therefore, we interpret the observed decrease of I_MLL4 intensity to ∼18% (Fig. 5J), as reflecting the release and disordering of most FHs while most BHs remain helically ordered. This association is supported also by the observed concomitant changes in I₁₀ and I₁₁₊₂₀ (Fig. 5 F–H).

We previously found by EM of isolated filaments that RLC phosphorylation induced the release and disordering of the helically ordered heads (27), as shown also by changes in the equatorial reflection intensities from skinned tarantula muscle (28). Similarly, XRD studies on skinned rabbit psoas muscle (31) showed decreases in I_MLL1 upon RLC phosphorylation, consistent with partial loss of the helical order of myosin heads, and increases in I₁₁/I₁₀, supporting the idea that phosphorylation of myosin RLC causes the myosin heads to move away from the thick filaments toward the thin filaments, thereby enhancing the probability of interaction with actin. Our diffraction (Fig. 5) and U-G gel electrophoresis (SI Appendix, Fig. S2 A, b and B, a) results in live, intact tarantula muscle are consistent with the proposal that at the end of the plateau (Fig. 2 B, e), most BHs are Ser45 mono-P (SI Appendix, Fig. S2 A, b and B, a); that is, the filament-docked non-P BHs become swaying Ser45 mono-P BHs contributing to the X-ray pattern (Fig. 2 B, d, double curved arrows) with their own docked/released duty cycle (SI Appendix, Fig. S3D), similar to the swaying Ser35 mono-P FHs in the relaxed state (Fig. 2 B, b, double curved arrows), while a significant proportion of FHs are bi-P, released and fully disordered.

Previous studies (56, 58, 59) suggested that in frog skeletal muscle, the recovery after a short tetanus takes at least several seconds, and that there was a discrepancy between the fall of tension and the return of myosin heads to their helical paths (58). Here we found that in tarantula muscle after a tetanus I_MLL4 and I₁₀ stayed low, and I₁₁ high until 12 to 13 s, slowly returning in >6 min to their initial resting values (Fig. 5 J–L). This slow recovery of the original helical arrangement of the myosin heads (several minutes) correlates with the slow dephosphorylation by the endogenous MLCP of first, the Ser45 mono-P swaying BHs, which allows them to dock back helically again to the filament, together with non-P BHs already docked (SI Appendix, Fig. S3B), and second, the fully released bi-P FHs, so they can dock back as swaying heads on the previously docked BHs, both increasing their contribution to the I_MLL4 and I₁₀ (Fig. 5 J and L). As with the helical layer lines, the I₁₁₊₂₀ remains high after the tetanus (Fig. 5K), consistent with the bi-P FHs (and some Ser45 mono-P BHs) remaining in the vicinity of the thin filament, recovering slowly to their resting value as dephosphorylation progresses as mentioned before.

Heads that Remain Released and Disordered after a Tetanus Produce PTP.

When muscle is stimulated soon after a tetanus, the twitch elicited is substantially stronger than a pretetanic twitch (PTP). It has been proposed that such PTP results from RLC phosphorylation and its accompanying structural effects, in which the myosin heads project from the filament backbone, facilitating their interaction with actin. This structural model has received experimental support from EM studies of isolated thick filaments (27, 30) and XRD of skinned vertebrate and invertebrate muscles (28, 31), but has not been demonstrated in live muscle. Here we have used ex vivo tarantula muscle to correct this deficit.

Fig. 6 shows that early after a tetanus, myosin heads remain disordered (weak I_MLL4) and away from the filament backbone (weak I₁₀, strong I₁₁₊₂₀, and high I₁₁₊₂₀/I₁₀), when almost all heads are still phosphorylated (mono- or bi-) (SI Appendix, Fig. S2 A, b and B, a). When the muscle is stimulated during this period, the twitch elicited is much stronger than the pretetanus twitch (i.e., exhibits PTP) (Fig. 6 C and D, force traces, green arrows) and is accompanied in each case by a further increase in I₁₁₊₂₀ (Fig. 6A, red arrows) and I₁₁₊₂₀/I₁₀ ratio (Fig. 6C, red arrows), while disorder remains high (Fig. 6D). This can be interpreted straightforwardly in terms of the CPA mechanism, which predicts that following a tetanus, most BHs will be still mono-P and some FHs bi-P, with few non-P heads (Fig. 2 B, e, compare with SI Appendix, Fig. S2 A, b and B, a). The bi-P FHs are disordered and project from the filament backbone, remaining this way until being dephosphorylated to Ser45 mono-P. The mono-P BHs are swaying, spending part of their time, defined by their [t_IHM/(t_IHM + t_rel)] duty cycle (see previous section and SI Appendix, Fig. S3D), on the helical tracks and thus contributing to the layer lines, similar to the swaying FHs in the relaxed pattern. When muscle is stimulated to contract during this time, these heads, already projecting away from the thick filament (reflected in the high I₁₁₊₂₀ and I₁₁₊₂₀/I₁₀) (Fig. 6 A and C), attach to actin (producing enhanced force, or PTP), shown by a further increase in I₁₁₊₂₀ and I₁₁₊₂₀/I₁₀ (Fig. 6 A and C red arrows). Thus, our experiments suggest that in live muscle a tetanus leads to phosphorylation of many heads (including FH and BH), which continue to project toward the thin filament (phosphorylation “memory”) and are therefore readily available for interaction with actin, producing a strong posttetanic twitch. During recovery from the tetanus, RLC phosphorylation gradually returns to baseline through MLCP activity, and we observe a decline in PTP. This follows a return of heads to the thick-filament backbone (I₁₁₊₂₀ and I₁₁₊₂₀/I₁₀ drop and I₁₀ increases) and restoration of their helical arrangement (I_MLL4 regains its strength) (Fig. 5 D and J). We have observed twitch and tetanic tension in freshly dissected muscle where the helical ordering is weak or absent. Thus, helical ordering of the heads establishing the IHM is not a requirement for contraction. Rather, our results show that the highest tension twitches come from heads that are disordered (i.e., during PTP).

Fig. 6.

Time-resolved XRD patterns of a PTP experiment. Time courses of the I₁₁₊₂₀ (A), I₁₀ (B), I₁₁₊₂₀/I₁₀ ratio (C), and I_MLL4 (D) during a pretetanic twitch (not visible due to small size), 2-s 60-Hz tetanus (recorded force shown in blue in C and D), and three PTP twitches (shown in blue in C and D, green arrows), showing that early after a tetanus, myosin heads remain disordered (weak I_MLL4) and away from the filament backbone (weak I₁₀, strong I₁₁₊₂₀, and high I₁₁₊₂₀/I₁₀); during this time, almost all heads are phosphorylated (mono- or bi-). The pretetanic twitch (0.4 mN) was barely above the force transducer (resolution of ∼1 mN) baseline. The first posttetanic twitch (57 mN) was potentiated ∼50 times. When the muscle is stimulated during this period, the twitch elicited is much stronger than the pretetanus twitch (i.e., exhibits PTP) and is accompanied in each case by a further increase in I₁₁₊₂₀ (red arrows) and I₁₁₊₂₀/I₁₀ (red arrows), whereas disorder (weak I_MLL4) remains high.

Proposed IHM-Based PTP Structural Mechanism for Tarantula.

Based on our results and on the dual phosphorylation IHM-based CPA mechanism (Fig. 2B), we propose the structural model shown in Fig. 7 to explain force enhancement and PTP in the muscles of tarantula. This mechanism supports the general concepts for phosphorylation-based potentiation advanced previously for fast-twitch vertebrate muscle (22, 23, 43, 60–62), in which it was hypothesized that in live muscle, RLC phosphorylation releases the heads, similarly to what was found in isolated thick filaments and skinned muscle in tarantula (27, 28), Limulus (29), and rabbit (30, 31), thus facilitating their attachment to actin, providing an explanation for potentiation both during and after a tetanus (as PTP). The model, which includes the constraints of the tarantula thick-filament structure (34), suggests a specific structural mechanism that explains: 1) Force enhancement during a tetanus by recruiting additional bi-P FHs and Ser45 mono-P BHs (Fig. 7C vs. Fig. 7B), grading (i.e., potentiating) the achieved force, and 2) the PTP phenomenon (compare Fig. 7E vs. Fig. 7B), consistent with our experimental XRD (Figs. 5 and 6) and U-G results (SI Appendix, Fig. S2 A, b and B, a).

Fig. 7.

Model explaining PTP in tarantula, based on the IHM-based CPA mechanism. In relaxation (A) [Ca²⁺] is low so thin filaments and MLCK are switched OFF while MLCP, always on, dephosphorylates any phosphorylated Ser45. On activation, [Ca²⁺] is high, switching on thin filaments and MLCK. With brief high [Ca²⁺] exposure (twitches) (B), Ser35 mono-P swaying FHs bind to thin filaments producing force, while few FHs can be bi-P. On longer high [Ca²⁺] exposure (tetani) (C), MLCK can bi-P FHs and Ser45 mono-P some BHs. After tetanus, [Ca²⁺] is low, switching off thin filaments and MLCK. In posttetanus relaxation (D) bi-P FHs remain released and Ser45 mono-P BHs sway away until dephosphorylated by MLCP (time τ), and resting relaxation is recovered (G). If, in this state (D), a brief high [Ca²⁺] exposure occurs, bi-P FHs and Ser45 mono-P BHs (not yet dephosphorylated by MLCP) remain released or swaying away, bind to switched ON thin filaments, producing greater force (E) than pretetanic (B), structurally explaining PTP “memory.”

Arthropod striated muscles share similar thick-filament head arrangements and long RLC NTEs with dual PKC/MLCK phosphorylation sites. Apart from tarantula, RLC phosphorylation has been reported in Limulus (24), and PTP in scorpion (63), whose thick filaments (25, 64) are similar to tarantula. Therefore, the proposed mechanism could, in principle, be extended to the other species exhibiting long RLC NTEs with PKC/MLCK consensus sequences. RLC phosphorylation remaining after a tetanus has been reported in frog (65), rabbit (66), and mouse (62). Residual I_MLL1 after a tetanus has been observed in frog (55, 56) and mouse (57). Here we have associated, in intact tarantula muscle, residual RLC phosphorylation (SI Appendix, Fig. S2 A, a and B, a), residual I_MLL1 and I_MLL4, and residual released and disordered heads after a tetanus, with PTP (Fig. 7). Can the mechanism shown in Fig. 7 be extended as well to explain PTP in vertebrate muscles? Brito et al. (34) reported that vertebrate muscle RLCs exhibit only short NTEs, with MLCK consensus sequences and only one MLCK-phosphorylatable serine. Therefore, the proposed mechanism could be extended to vertebrate muscle but modified for allowing only monophosphorylation. Mono-PKC–phosphorylated, swaying heads in tarantula would correspond to the constitutively ON (nonphosphorylated) heads in vertebrate (1). In tarantula the recovery phase after a tetanus implies a slow dephosphorylation time τ (minutes) of the mono-P and bi-P heads (Fig. 7, Inset), that by remaining released and disordered “memorize” the previous muscle history. In vertebrates, this could be accomplished more quickly as the PTP recovery would involve the dephosphorylation time of only mono-P heads. Brito et al. (34) also reported that some species that show direct Ca²⁺-activation via the RLC, like scallop (67), exhibit RLCs with only short NTEs and a single serine, with no MLCK consensus sequence, suggesting that a phosphorylation-based mechanism like the IHM-based CPA mechanism (Fig. 2) and the PTP mechanism (Fig. 7) cannot be functional for them.

Evolution Has Resulted in Different Thick-Filament Activation Mechanisms.

Different molecular mechanisms for activation have evolved to control the release and disordering of myosin heads in striated muscles in order to actively recruit them to produce force: 1) Direct Ca²⁺-binding, 2) RLC phosphorylation, 3) mechanosensing, and 4) delayed stretch activation. Mechanosensing (Fig. 8C) is a prominent feature of vertebrate skeletal muscle (1). Here we show evidence (Figs. 5 and 6 and SI Appendix, Fig. S2) that in tarantula there is a primary phosphorylation mechanism (Fig. 8B), with at most a subsidiary role for mechanosensing because the lengthening of the backbone was much smaller (and statistically not significant) than in frog. In tarantula muscle, during a twitch, the swaying heads (due to their mono-P at Ser35) are intermittently extended ready to bind to Ca²⁺-activated thin filaments to produce initial force, and phosphorylation releases more heads. In frog, lacking constitutive phosphorylation, constitutively ON heads develop initial force, and inactive heads are then released by mechanosensing as force increases. In both muscles, PTP is due to RLC phosphorylation, possibly differing in how long the PTP lasts. Another important difference between the thick filaments of invertebrates and vertebrates is the presence of PM in the invertebrate filament backbone, suggesting that the lack of backbone lengthening in tarantula may be associated with the rigidity of its thick filaments, due to a core of tightly packed PM molecules (Fig. 1 B, d), in contrast to frog that lacks this rigid core. Some of these mechanisms may be widespread, such as the dual phosphorylation CPA mechanism in Arthropoda, with single phosphorylation variations present in vertebrate muscle as a modulatory mechanism. Other activation mechanisms are very specific, such as the Ca²⁺-binding activation mechanism, observed only in mollusks (Fig. 8A), and the stretch activation observed in the indirect flight muscles of Lethocerus and other flying insects. So far, mechanosensing appears to be confined to vertebrates.

Fig. 8.

Dual-filament regulation in invertebrate and vertebrate muscles could be associated with the presence of PM in the thick-filament backbone core. Thick filaments containing PM (scallop, tarantula) or not (frog) are activated either (A) directly by Ca²⁺ (blue circles), (B) by PKC/MLCK Ser35/Ser45 dual phosphorylation (circles in yellow: Ser35 mono-P, red: Ser45 mono-P, yellow/red: bi-P), or (C) mechanosensing. Thick and thin filaments are shown either switched OFF (orange) or ON (green), in relaxed (a) or contracting without load (b) or with load (c). Based on Linari et al. (1).

We conclude that the molecular mechanisms of thick filament activation and PTP evolved differently in invertebrate and vertebrate striated muscle as adaptations to the lifestyle of the organisms. Our results support the notion that helically ordered IHMs are the key to energy conservation in the resting relaxed state, brought about by the interactions between the heads that inhibit their activity (20). In tarantula live muscle, while the ATP turnover rates for the swaying FHs is either slow (250 to 300 s) while being transiently attached forming helices or fast (<30 s) while being transiently detached and disordered (20, 54), an important fraction of the heads (BHs)—permanently docked forming helices on the filament—exhibit a very slow ATP turnover rate (1,800 s), an energy conserving state. Force potentiation both during and after a tetanus in tarantula muscle could be a way to increase force when needed, according to the previous muscle activation history (PTP), while saving ATP when not (54), which could be energetically favorable in an arthropod that spends large amounts of time immobile, with only occasional bursts of energy required for moving when capturing food or avoiding predators, among other vital activities.

A key point to note is the existence of two different relaxed states that are structurally distinct: 1) The “resting” relaxed state in which heads are helically ordered in the IHM, with minimal usage of ATP (possibly equivalent to the superrelaxed state); and 2) the “posttetanic” relaxed state, in which heads are in a disordered, preactivated, non-IHM state, near to actin. Both states are mechanically relaxed, and the thin filaments are switched off. These states, determined experimentally, may correlate with different phases of the daily life of the tarantula. The animal spends most of its time at rest, hidden and motionless, conserving energy (resting relaxed state). When the need to capture food or escape a predator arises, the muscles become active and RLC phosphorylation is initiated, leading to enhanced force production. Relaxation still occurs when the thin filaments are switched off between contractions (posttetanic relaxed state), but more force is produced upon activation, as many heads are already near actin. Energy conservation is lower during these relaxed intervals but that is less important than having high force when needed. When activity ceases, the muscles return to their resting relaxed state.

Tarantula provides an excellent model system for understanding thick-filament structure and function in terms of the BHs and FHs of the IHM. EM yields key insights into the organization of the two heads in the IHM in the static state of relaxation, including their interactions, relative stability, and accessibility of their RLC phosphorylatable serines (8, 34, 35). XRD complements these studies, demonstrating the presence of this same IHM arrangement in intact muscle, and additionally revealing dynamic changes in the heads that underlie activation and potentiation. These studies suggest very different roles and dynamics for BHs and FHs in tarantula. A similar distinction between heads might also occur in vertebrate muscle, where the IHM is also present (68). Previous studies indicate a residual population of helically ordered heads in activated muscles, but not assigned to any particular head (blocked or free) (55–57). Our tarantula observations would suggest that the majority of residual ordered heads would be BHs, and that any constitutively active heads would come from the FH population.

Materials and Methods

X-Ray Diffraction.

Legs were excised from anesthetized Texas brown tarantulas and placed in an X-ray chamber. Muscles were stimulated through platinum electrodes using an Aurora Scientific 701A stimulator. XRD patterns from muscles in resting or activated states were obtained at the Advanced Photon Source, Argonne National Laboratory, and recorded on a Pilatus 3 1M detector (Dectris Inc.) with 20-ms exposures and 50 ms between exposures. For in vivo experiments, the live tarantula was constrained in a plastic tube mounted in the X-ray beam.

XRD Pattern Modeling.

Fourier transforms were calculated for the thick filaments assuming a myosin head spacing of 14.5 nm, a helical repeat of 43.5 nm, and fourfold rotational symmetry.

U-G gel electrophoresis was carried out according to Sulbarán et al. (35) on muscles frozen at the end of the tetanic plateau.

Full details of methods are provided in SI Appendix.

Data Availability.

Data were deposited in the Open Science Framework, https://osf.io/anxuy/?view_only=5491c6ba8bca4add9d97930614734e0e.