Binding of Ca²⁺/CaM to the two Myo1d IQ-motifs

The light chain binding domain of rat Myo1d contains two IQ-motifs that are quite distinct in sequence (Fig. 1B). To investigate their interaction with the Ca²⁺-sensor molecule CaM, we expressed different recombinant Myo1d proteins in HeLa cells (Fig. 1A). The recombinant Myo1d proteins included the head domain and either no, one or both IQ-motifs. In addition, we expressed a construct that lacked the first IQ-motif and contained only the second IQ-motif (Myo1d-IQ2). This construct was further modified in that the two C-terminal amino acids (LV) of the converter domain not present in the Myo1d-head construct were deleted (Myo1dΔLV-IQ2, Fig. 1). Because these two amino acids are part of a α-helix continued by the IQ-motifs, the IQ-motif is rotated counterclockwise by 200° and moved ≈ 3 Å closer towards the head domain. Following expression of these Myo1d constructs and affinity purification in the absence of free Ca²⁺, we assessed the stoichiometry of CaM bound to the two IQ-motifs. The stoichiometries of CaM associated with Myo1d-head (0 : 1; Fig. 2B), Myo1d-IQ1 (1 : 1; 2, 3) and Myo1d-IQ1.2 (1.74 : 2), respectively, have been described previously [22]. Adjusting the free Ca²⁺ concentration to 0.1 mm did not cause a release of CaM bound to IQ 1 in the Myo1d-IQ1 heavy chain construct (Fig. 3). This demonstrates that IQ 1 has a high affinity for CaM both in the absence and presence of free Ca²⁺. In contrast, the binding of CaM to IQ 2 was clearly distinct. The deletion constructs Myo1d-IQ2 and Myo1dΔLV-IQ2 could be purified under identical conditions using antibodies directed against the C-terminal FLAG epitope albeit with a lower protein yield (2, 4). Densitometric analysis of the CaM content in affinity purified Myo1d-IQ2 and Myo1dΔLV-IQ2 preparations revealed a ratio of only 0.24 ± 0.03 CaM per Myo1d-IQ2 heavy chain and 0.39 ± 0.14 CaM per Myo1dΔLV-IQ2 heavy chain, respectively (2, 4). However, a higher protein yield and a stoichiometric association of CaM with Myo1d-IQ2 were achieved when this construct was purified in the presence of free Ca²⁺ (data not shown), indicating that IQ2 binds CaM more tightly in the presence of Ca²⁺.

To test if the IQ2 in Myo1d-IQ2 and Myo1dΔLV-IQ2 could be saturated with CaM in the absence or presence of free Ca²⁺, we mixed purified Myo1d-IQ2 or Myo1dΔLV-IQ2 with 10 µm exogenous CaM (4, 5). Free CaM was separated from CaM bound to Myo1d-IQ2 or Myo1dΔLV-IQ2 by cosedimentation of the myosin with F-actin. In the absence of free Ca²⁺ Myo1d-IQ2 and Myo1dΔLV-IQ2 had stoichiometric amounts of CaM bound (molar ratio of 0.92 ± 0.12 CaM/Myo1d-IQ2 heavy chain and of 1.1 ± 0.1 CaM/Myo1dΔLV-IQ2 heavy chain) (4, 5). The 1 : 1 ratio of CaM binding to Myo1d-IQ2 and Myo1dΔLV-IQ2 did not change upon the addition of 100 µm free Ca²⁺ to the assay mixture (4, 5). Therefore, we supplemented purified Myo1d-IQ2 and Myo1dΔLV-IQ2 preparations for all further experiments with 10 µm CaM to guarantee the stoichiometric binding of CaM.

Regulation of the actin-activated ATPase of recombinant Myo1d proteins by Ca²⁺/CaM

In the absence of free Ca²⁺, the basal ATPase activity of the Myo1d-head without an IQ-motif was 0.01 s⁻¹. The V_max of the actin-activated ATPase was 2.6 s⁻¹ and the K_actin 38 µm[22] with an apparent second-order rate constant K_app of 0.68 × 10⁵ s⁻¹·m⁻¹(Fig. 6). These values were virtually identical for the Myo1d-IQ1 and Myo1d-IQ1.2 constructs that contained in addition either IQ1 or IQ1 and IQ2. The two IQ-motifs were even exchangeable, as the Myo1d-IQ2 fusion protein also exhibited identical basal and actin-activated ATPase activities (Fig. 6A,B). However, deletion of the two C-terminal amino acids of the converter domain in the construct Myo1dΔLV-IQ2 lead to a pronounced inhibition of the actin-activated ATPase activity while the basal ATPase activity was unaltered (Fig. 6B). The actin-activated ATPase activity increased almost linearly in the range of the actin concentrations tested. It exhibited a K_app of 0.12 × 10⁵ s⁻¹·m⁻¹ that was about six-fold reduced in comparison with Myo1d-IQ2, indicating a change in coupling between the actin and nucleotide binding sites.

To investigate whether the Myo1d ATPase is regulated by Ca²⁺/CaM, we determined the actin-activated ATPase of Myo1d constructs in the presence of 22 µm-free Ca²⁺. Interestingly, the V_max of the actin-activated ATPase activity of the Myo1d-head was reduced by roughly 20% whereas the K_actin was unaltered (Fig. 6A). This Ca²⁺-dependent inhibition was not reversible when free Ca²⁺ was chelated with EGTA (data not shown). In the Myo1d-IQ1 construct, free Ca²⁺ inhibited the actin-activated ATPase activity by ≈ 75% (Fig. 6C). The calculated K_app was 0.12 × 10⁵ s⁻¹·m⁻¹. As the two Myo1d IQ-motifs have different CaM-binding properties, we analyzed if IQ1 and IQ2 regulate the Myo1d ATPase activity differently. However, in the construct Myo1d-IQ2 that has IQ1 exchanged for IQ2, free Ca²⁺ caused a comparable inhibition of the ATPase with a V_max of 0.7 s⁻¹(6, 7). The actin affinity was not affected significantly with a determined K_actin = 44 µm. The K_app derived from the initial slope of the hyperbola was 0.15 × 10⁵ s⁻¹·m⁻¹.

In the construct Myo1dΔLV-IQ2 that exhibited already a reduced ATPase activity in the absence of free Ca²⁺, the addition of free Ca²⁺ reduced its ATPase activity further by about 40% and a K_app of 0.07 × 10⁵ s⁻¹·m⁻¹ was determined.

Inhibition of the Myo1d actin-activated ATPase activity as a function of the concentration of free Ca²⁺

Next, we determined the ATPase activities for the different Myo1d constructs as a function of the free Ca²⁺ concentration (0.001–158 µm) (Fig. 7). The slight reduction of the actin-activated ATPase activity of the Myo1d-head exhibited an IC₅₀ of pCa ≈ 6 (0.3 µm free Ca²⁺). All of the Myo1d constructs that contained either one or two IQ motifs, specifically Myo1d-IQ1, Myo1d-IQ1.2 and Myo1d-IQ2, were inhibited with an IC₅₀ of pCa ∼7 (0.05–0.08 µm Ca²⁺) (Fig. 7). This IC₅₀ value corresponds to the affinity of the pair of Ca²⁺-binding sites in the C-terminal lobe of CaM. The Myo1dΔLV-IQ2 protein that exhibited already a reduced ATPase activity in the absence of free Ca²⁺ did not show any significant changes in ATPase activity with increasing free Ca²⁺ concentrations.

Ca²⁺-independent and Ca²⁺-dependent binding of CaM to IQ-motifs

The two IQ motifs in Myo1d are supposed to belong to different classes of IQ motifs [8,23]. Whereas IQ1 in Myo1d conforms well to the consensus sequence of IQ-motifs, IQ2 is less well conserved and possesses hydrophobic residues at positions 1, 5, 8 and 14 as is typical for Ca²⁺-dependent CaM-binding motifs [1]. Indeed, we found that IQ1 has a higher affinity for ApoCaM than IQ2. Based on structural studies of ELC binding to IQ1 of a myosin II [24,25] and Mlcp1 binding to the IQ-motifs in Myo2p [23], we presume that the C-terminal lobe of ApoCaM binds to the N-terminal parts of IQ1 (LQKVWR) and IQ2 (IIRYYR), respectively, in a semi-open conformation. We expect that the N-terminal lobe of ApoCaM binds in a closed conformation to the C-terminal part of IQ1 (GTLAR). The reduced affinity of IQ2 for ApoCaM as compared to IQ1 is probably due to a lack of interaction between the N-terminal lobe of CaM and the C-terminal part of IQ2 (RYKVK). The exchange of Gly at position 7 for an Arg with a bulky side chain in IQ2 is likely to interfere sterically with the binding of the N-lobe of ApoCaM as has been demonstrated for Mlc1p binding to IQ4 of Myo2p [23]. ApoCaM has also been reported to bind weakly to the single IQ motif present in Myo VI [26]. This IQ motif deviates from the consensus sequence (IQXXXRGXXXR/K) in that the glycine consensus residue at position 7 is changed to a methionine. Charge repulsion due to a stretch of positively charged amino acids in the C-terminal part of IQ2 (Arg733, Lys735, Lys737) may additionally repel the N-lobe farther away from IQ2. The N-lobe of CaM might thus be free to interact with sequences in the Myo1d tail or with other proteins.

The affinity of IQ2 for ApoCaM was not affected when the two C-terminal amino acids (LV) of the converter domain α-helix were deleted. This deletion is predicted to introduce a counterclockwise rotation by 200° and a shift by ≈ 3 Å towards the head domain of the CaM bound to the IQ-motif. We conclude that no steric hindrance for CaM binding was introduced by this deletion.

We have shown previously that 1.74 CaM molecules were associated with affinity purified Myo1d-IQ1.2. This stoichiometry of bound CaM is higher than the sum of CaM molecules bound to Myo1d-IQ1 (1.1) and Myo1d-IQ2 (0.24). This result indicates that CaM binds in a cooperative manner to the light chain binding domain of Myo1d. The binding of CaM to IQ1 may induce a stabilization of the IQ2 α-helical structure and thereby facilitate binding of the C-terminal lobe of ApoCaM to the N-terminal half of IQ2.

We found that purified Myo1d-IQ2 and Myo1dΔLV-IQ2 could be fully saturated by the addition of exogenous ApoCaM or Ca²⁺/CaM. This finding allowed us to investigate the effects of Ca²⁺-binding to ApoCaM associated with the IQ2. Myo1d-IQ2 affinity purified under Ca²⁺ conditions contained stoichiometric amounts of copurified CaM demonstrating that IQ2 actually has a higher affinity for Ca²⁺/CaM than for ApoCaM. This result is in accordance with previous observations in a gel overlay assay [8] and the above mentioned sequence similarity of IQ2 with Ca²⁺-dependent CaM-binding motifs. The two lobes of Ca²⁺/CaM may both bind in an open conformation to IQ2, explaining the increased affinity.

Mechanism of inhibition of the Myo1d ATPase activity by Ca²⁺/CaM

As reported previously, in the absence of free Ca²⁺ Myo1d-head, Myo1d-IQ1 and Myo1d-IQ1.2 exhibited very similar basal and actin-activated ATPase activities. In the presence of actin, the ATPase activities reached V_max values of 2.6–3.1 s⁻¹[22]. The addition of one or two IQ motifs to the head domain did not affect ATPase rates and actin affinities. In the present study, we show that IQ1 can even be replaced by IQ2 without that the ATPase rates and actin affinities get significantly altered. However, binding of Ca²⁺ to CaM associated with the IQ motif directly following the head (converter) domain induced a significant inhibition of the Myo1d actin-activated ATPase activity. This inhibition was independent of whether one or two IQ motifs were present or IQ1 or IQ2 were directly fused to the head region. The free Ca²⁺ concentrations that were necessary to induce the inhibition of the actin-activated ATPase activity correlated well with the reported affinity of the C-terminal lobe of CaM for Ca²⁺. The N-terminal lobe of CaM exhibits a 10-fold lower affinity for Ca²⁺[19]. These results support the notion that the C-terminal lobe of CaM is bound to both IQ1 and IQ2 in a semi-open conformation and switches upon Ca²⁺-binding to an open conformation. The open conformation inhibits the actin-activated ATPase activity of Myo1d. The C-terminal lobe of CaM bound in an open conformation to the IQ motif directly following the head region might inhibit the actin-activated ATPase activity by specific interactions with the head region. The ELC bound to IQ1 of smooth muscle myosin II has been observed to contact in the prepower stroke state a loop in the head domain that modulates nucleotide affinity [27]. Therefore, a change in nucleotide affinity might be the reason for the reduced ATPase activity. Changes in nucleotide affinity have also been reported for Dictyostelium discoideum myosin II head constructs with varying lengths of the C-terminal α-helix of the converter domain that is continuous with the IQ-motifs [28]. Therefore, Ca²⁺/CaM may modulate the actin-activated ATPase activity by affecting the stability or flexibility of the α-helix N-terminal to the IQ-motif(s).

Fusion of the IQ2 directly to the Myo1d head construct led to the deletion of the two amino acids immediately N-terminal to the IQ-motif. This deletion caused a similar but Ca²⁺-independent inhibition of the Myo1d actin-activated ATPase activity. This deletion is predicted to rotate the CaM associated with the IQ-motif by 200° on the α-helix emanating from the converter domain and to shorten this α-helix by roughly 3 Å. The fact that this construct demonstrates a similar reduction in ATPase activity might either be explained by coincidence or by the assumption that it mimics the Ca²⁺-dependent changes induced in the other IQ-motif constructs. However, the latter possibility seems only feasible when the inhibitory mechanism does not involve a stereospecific interaction of CaM with the head region. A common mode of regulation may include effects on conformation and/or flexibility of the light chain binding domain and N-terminal α-helix that are transduced to the head region. Such effects may become more pronounced when the head is bearing strain.

As reported here for Myo1d, the first of several IQ-motifs was demonstrated to control the Ca²⁺-sensitivity of the kinetics of Myo1b (MI¹³⁰; myr 1) [29]. Myo1e (myr 3; Myosin IC) has only a single IQ-motif and Ca²⁺-binding to the CaM associated with this IQ-motif negatively regulates the basal ATPase activity [21]. Therefore, the first IQ-motif in class I myosins might serve generally as a Ca²⁺-regulatory element. On the other hand, conversion of the CaM C-lobe upon Ca²⁺-binding from a semi-open to open IQ-motif binding configuration is unlikely to provide a general inhibitory mechanism for class I myosin ATPase activities. Although binding of Ca²⁺ to the C-lobe of CaM has been reported to inhibit actin gliding powered by Myo1c (myosin Iβ, myr 2), Myo1c ATPase activity was not affected at this Ca²⁺ concentration and was enhanced at 10-fold higher Ca²⁺ concentrations simultaneously with the dissociation of one CaM [19]. The actin-activated ATPase activities of Myo1a (brush border myosin I) and Myo1b (MI¹³⁰; myr 1) were actually increased in the presence of Ca²⁺/CaM [20,30]. In Myo1e the basal ATPase activity was reduced with a Ca²⁺-sensitivity suggestive of a contribution by both the C- and N-terminal lobes of CaM [21]. In conclusion, the regulatory functions of Ca²⁺/CaM in different class I myosin molecules appears to be quite diverse and no common mechanisms have emerged yet. The molecular basis for these differences remains to be elucidated. The detailed characterization of Myo1d regulation by Ca²⁺/CaM provided here represents a necessary step towards this goal.

Plasmid construction

Construction of the expression plasmids Myo1d-head-FLAG, Myo1d-IQ1-FLAG and Myo1d-IQ1.2-FLAG which encode amino acids 1–697, 1–721 and 1–743 of rat Myo1d, respectively, has been described previously [22].

Plasmids Myo1d-IQ2-FLAG and Myo1dΔLV-IQ2-FLAG have the first of the two IQ-motifs deleted, so that the second IQ-motif is fused to the head domain directly. Myo1dΔLV-IQ2-FLAG is further missing the last two codons for amino acids 698–699 (LV) of the head. For the generation of these two deletion constructs, a two step PCR strategy was employed. At first, two overlapping fragments were amplified by PCR. The sequences flanking the deleted region were fused in the reverse primer used for amplification of the 5′-fragment. To construct Myo1d-IQ2-FLAG, the two overlapping fragments were amplified with the two primer pairs 5′-GGCAAACTTGATGATGAGCGCTGC-3′ (forward 1)/ 5′-CAGAGCTGCCTTGACGA-GCATCTG-3′ (reverse 1) and 5′-CAGATGCTCGTCAAGGCAGCTCTG-3′ (forward 2)/ 5′-ATTCCAGCACACTGGTCACTT-3′ (reverse 2). To obtain Myo1dΔLV-IQ2-FLAG, the two primer pairs forward 1/5′-CAGAGCTGCCTTCATCTGGGCGCG-3′ (reverse 1′) and 5′-ATTCCAGCACACTGGTCACTT-3′ (forward 2′)/reverse 2 were used, respectively. After annealing and extension of the two overlapping fragments, the resultant fragment served as the template in a final PCR using the primer pair forward 1/reverse 2. The resulting products were cloned into pIRES Myo1d-head-FLAG using the unique BstXI and Eco47III restriction sites. PCR derived fragments were verified by sequencing.

Cell culture

HtTA-1 HeLa cells [31] were cultured at 37 °C and 5% (v/v) CO₂ in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum, 100 U·mL⁻¹ penicillin and 100 µg·mL⁻¹ streptomycin. Cells were transfected by the addition of plasmid DNA precipitated with calcium phosphate. Single colonies were isolated after selection in 200 µg·mL⁻¹ hygromycin for 2 weeks. Cells were expanded and analyzed for expression of Myo1d constructs by immunoblotting with the rat Myo1d antibody SA 522. Selection by hygromycin was maintained continuously.

Protein expression and purification

Recombinant Myo1d proteins were purified as described in detail previously [22]. Briefly, cells were grown in 20–30 culture dishes (diameter 150 mm) to 80% confluence, washed with NaCl/P_i, collected by scraping and permeabilized with lysis buffer [150 mm NaCl, 20 mm Hepes pH 7.4, 2 mm MgCl₂, 1 mm EGTA, 0.5% (v/v) Triton X-100, 2 mm ATP, 0.1 mg·mL⁻¹ Pefabloc, 0.01 mg·mL⁻¹ leupeptin, 0.02 U·mL⁻¹ aprotinin] for 1 h on ice. After clearing the lysate by two subsequent centrifugation steps, the supernatant was mixed with 1 mL FLAG-antibody agarose (Sigma-Aldrich) and incubated for 2 h in the cold. The beads were washed twice with buffer WP (50 mm KCl, 10 mm Hepes pH 7.4, 2 mm MgCl₂, 1 mm EGTA, 1 mm 2-mercaptoethanol and 2 mm NaN₃) and finally, bound protein was eluted by buffer WP supplemented with 125 µg·mL⁻¹ soluble FLAG peptide (Sigma-Aldrich). In some cases, Myo1d-IQ2 was purified in the presence of 50–100 µm free Ca²⁺ concentrations. Occasionally, 5 µm calmodulin was added during the elution of Myo1d-IQ2 to enhance protein solubility. Eluted proteins were dialyzed against buffer WP and concentrated using microcon filters (cut-off 10 kDa) if necessary. All recombinant Myo1d proteins were cleared by ultracentrifugation (150 000 g for 20 min) immediately before use. To saturate all light chain binding sites, cleared Myo1d-IQ2 and Myo1dΔLV-IQ2 preparations were preincubated with 10 µm calmodulin for 20 min on ice. Densitometric analysis of Coomassie-stained protein bands on SDS gels was performed with an ultrascan laser densitometer. Values represent the mean of at least three different preparations. Actin was purified from rabbit skeletal muscle as described by Pardee and Spudich [32]. Purified calmodulin was purchased from Sigma-Aldrich.

ATPase assays

Steady-state ATPase activities were determined at 37 °C as described in detail previously [21]. All assay mixtures contained 30 mm KCl, 10 mm Hepes pH 7.4, 2 mm MgCl₂, 2 mm ATP, 3 mm EGTA, 1 mm 2-mercaptoethanol, 2 mm NaN₃ and various actin concentrations. To measure Ca²⁺-dependent activities, a constant actin concentration of 24 µm was used. Free Ca²⁺ concentrations between 0.001 and 158 µm were adjusted by adding the appropriate amounts of CaCl₂ to obtain the desired value in the presence of 3 mm EGTA. Actin-dependent ATPase activities were measured in the absence or presence of 22 µm free Ca²⁺ in a concentration range between 0 and 75 µm actin. Purified Myo1d constructs were used in a range between 54 and 270 nm. V_max and K_m values were determined by fitting the measured ATPase rates (v) to the Michaelis–Menten equation

(1)

with the program kaleidograph.

Sedimentation assays

To separate myosin-associated CaM from soluble CaM, myosin was incubated with 2 µm F-actin in a buffer containing 30 mm KCl, 10 mm Hepes pH 7.4, 2 mm MgCl₂, 3 mm EGTA, 1 mm 2-mercaptoethanol and 2 mm NaN₃ for 15 min on ice. Where indicated, the free Ca²⁺ concentrations were adjusted accordingly. Assay mixtures with Myo1d-IQ2 purified under calcium conditions contained 50–100 µm free Ca²⁺. Free Ca²⁺ was chelated in these preparations by the addition of appropriate amounts of EGTA. After high speed centrifugation at 150 000 g for 20 min, supernatants with soluble CaM were separated from pellets containing acto-myosin complexes with tightly bound CaM and analyzed by SDS/PAGE and densitometry.