A FERM domain autoregulates Drosophila myosin 7a activity

Abstract

Full-length Drosophila myosin 7a (myosin 7a-FL) has a complex tail containing a short predicted coiled coil followed by a MyTH4-FERM domain, an SH3 domain, and a C-terminal MyTH4-FERM domain. Myosin 7a-FL expressed in Sf9 cells is monomeric despite the predicted coiled coil. We showed previously that Subfragment-1 (S1) from this myosin has MgATPase of V_max ≈ 1s⁻¹ and K_ATPase ≈ 1 μM actin. We find that myosin 7a-FL has V_max similar to S1 but K_ATPase ≈ 30 μM. Thus, at low actin concentrations (5 μM), the MgATPase of S1 is fully activated, whereas that of myosin 7a-FL is low, suggesting that the tail regulates activity. Electron microscopy of myosin 7a-FL with ATP shows the tail is tightly bent back against the motor domain. Myosin 7a-FL extends at either high ionic strength or without ATP, revealing the motor domain, lever, and tail. A series of C-terminal truncations show that deletion of 99 aa (the MyTH7 subdomain of the C-terminal FERM domain) is sufficient to abolish bending, and the K_ATPase is then similar to S1. This region is highly conserved in myosin 7a. We found that a double mutation in it, R2140A-K2143A, abolishes bending and reduces K_ATPase to S1 levels. In addition, the expressed C-terminal FERM domain binds actin with K_d ≈ 30 μM regardless of ATP, similar to the K_ATPase value for myosin 7a-FL. We propose that at low cellular actin concentrations, myosin 7a-FL is bent and inactive, but at high actin concentrations, it is unfolded and active because the C-terminal FERM domain binds to actin.

Myosins comprise a superfamily of actin-based motor proteins with >20 subclasses (1, 2). They participate in various cellular functions including cytokinesis, endocytosis, adhesion, and transport of intracellular cargo such as pigment granules, secretory vesicles, and mRNA (3). All myosins contain a conserved motor domain, usually located near the N terminus, that binds actin and hydrolyzes ATP, a lever region that binds calmodulin or calmodulin-like light chains via IQ motifs and a tail region that is largely responsible for the anchoring or targeting of the motor to its intracellular localization and may be involved in regulation of enzymatic activity. The tails of some myosins (e.g., myosin 2 and myosin 5) contain long coiled-coil motifs, which self-associate to produce 2-headed molecules. The predicted coiled-coil motifs of some other myosins such as myosins 10 and 6 do not induce dimerization but instead form stable single α-helices, termed SAH domains (4, 5). Other myosins lack obvious dimerization domains in their tails, and some, like myosin 1 isoforms, are monomeric, as demonstrated by physicochemical techniques and electron microscopy (6). Often, the tail domains of myosins are rich in functional domains, such as PH domains, MyTH4 domains, FERM domains, or SH3 domains, which enable the molecules to interact with specific proteins or phospholipid-containing membranes (3).

Both human and Drosophila have 2 myosin 7 genes, termed 7a and 7b (2, 7). In mammals, myosin 7a mutations cause Usher syndrome 1B, the most common deaf/blindness disease (for review, see ref. 8). Myosin 7a is localized along the length of the stereocilia in hair cells and in the apical region of retinal epithelia cells (9, 10). Drosophila myosin 7a mutants, termed crinkled (ck), are lethal at larval stage (11). There is a small fraction of “escapers” that are sterile and deaf with abnormal bristle morphology (12).

All myosin 7 isoforms have a conserved motor domain, 5 IQ motifs, and a tail that contains a short region predicted to form a coiled coil, followed by 2 MyTH4-FERM domains separated by an SH3 domain (Fig. 1A). FERM domains normally contain 3 subdomains (13). In myosin 7a, the third subdomain, which is ≈100 aa in length, shows little homology with the third subdomain of FERM domains from other proteins such as ezrin or even myosins 10 or 15. However, this subdomain is highly conserved among myosin 7 isoforms [supporting information (SI) Fig. S1]. It shows 89% identity between Drosophila and human isoforms compared with only 65% identity between the motor domains of the 2 proteins (Fig. 1B) and has been termed MyTH7 (11). We will refer to this domain as the MyTH7 subdomain (of the FERM domain).

Fig. 1.

Myosin 7a constructs and the conserved MyTH7 subdomain. (A) Domain analysis of myosin 7a-FL and the truncation constructs. The numbers at the right indicate the number for the last amino acid residue of each construct. (B) Alignments of the conserved second MyTH7 subdomain in different species. Dm, Drosophila melanogaster; Aa, Anopheles arabiensis; Dr, Danio rerio; Mm, Mus musculus; Ss, Sus scrofa; Hs, Homo sapiens. Blue shading indicates very highly conserved amino acids, and paler blue shading indicates those that are fairly well conserved. PxxP at the top indicates the conserved PxxP motif in which the 2 flanking proline residues were mutated to alanines in the PxxP mutant. The “RK” at the bottom indicates the position of the conserved arginine and lysine residues that were mutated to alanines in the RK/AA mutant.

In this study of a purified full-length myosin 7a, we show that full-length Drosophila myosin 7a is monomeric, even though it contains a short region of predicted coiled coil and that intramolecular bending of the tail regulates its enzymatic activity. In the presence of ATP, the tail region of full-length Drosophila myosin 7a bends back on the motor domain, as shown by negative-stain electron microscopy and single-particle image processing. Under these conditions, the enzymatic activity and the in vitro motility of this bent myosin is inhibited. We demonstrate that the molecule is reversibly extended by raising the ionic strength, which results in uninhibited actin translocation. Finally, small deletions in the C-terminal tail region and selected point mutations reveal that the second MyTH7 subdomain is required for autoinhibition of the myosin's enzymatic activity.

Results

Expression and Purification of Full-Length Myosin 7a with Calmodulin.

Full-length Drosophila myosin 7a heavy chain with a C-terminal FLAG purification tag (myosin 7a-FL) was cloned into pFast-Bac1 vector and was coexpressed with Drosophila calmodulin in the baculovirus/Sf9 system. Approximately 1 mg of protein is obtained from 10⁹ cells. Purification results indicated calmodulin (3–4 mol/mol heavy chain) copurifies by anti-FLAG affinity resin as was previously found for Drosophila myosin 7a S1 and S1-SAH (previously called HMM) constructs (14) (Fig. 2A).

Fig. 2.

Purification and ATPase activity of myosin 7a-FL. (A) SDS-polyacrylamide electrophoretogram of myosin 7a. Samples were run on a 4–20% gradient gel. Lane 1, FLAG-affinity purified myosin 7a-FL consisting of a 230-kDa heavy chain and 17-kDa calmodulin light chains; lane 2, molecular mass markers (Mark12, Invitrogen). (B) Representative actin titration of the steady-state actin-activated ATPase activity of S1 (squares) and myosin 7a-FL (circles). Lines are fits to the Michaelis–Menten equation. See Table 1 for the averaged kinetic parameters derived from 3 independent preparations and assays.

Myosin 7a-FL Exhibits Enzymatic Properties Distinct from S1.

The actin-dependent MgATPase activity of myosin 7a-FL shows important differences compared with that of the S1 fragment reported previously (14, 15). The V_max of both constructs is ≈1 s⁻¹, but myosin 7a-FL has a high K_ATPase of 30 ± 12 μM, unlike S1 where the K_ATPase is 1.1 ± 0.1 μM (Fig. 2B and Table 1). The implication of this is clear if one examines the MgATPase activity at 5 μM actin where the MgATPase values are almost maximal for the S1 but are very low for myosin 7a-FL. Given that the S1 fragment lacks the entire tail region we suspected that this region is able to autoinhibit the enzymatic activity of the full length myosin.

Table 1.

Summary of steady-state actin-activated MgATPase activities and in vitro actin motility rates of myosin 7a-FL and C-terminal truncation mutants

Myosin 7a constructs	Vmax, s−1	KATPase, μM	Actin sliding, nm/s
Myosin7a-FL(WT)	1.2 ± 0.4	40.0 ± 14	17.0 ± 0.5*
Myosin7a-S1	1.0 ± 0.1	1.1 ± 0.1	N/A
Myosin7a-S1-SAH	1.4 ± 0.1	1.0 ± 0.1	9.0 ± 2.0
Myosin7a-TD1	1.1 ± 0.1	1.2 ± 0.05	8.7 ± 1.0
Myosin7a-TD2	1.5 ± 0.2	1.9 ± 0.1	8.3 ± 1.1
Myosin7a-TD3	1.2 ± 0.1	2.0 ± 0.3	8.6 ± 0.8
Myosin7a-TD4	1.3 ± 0.2	2.5 ± 0.3	8.1 ± 1.2
Myosin7a-FL(RK/AA)	0.6 ± 0.1	2.3 ± 0.2	10.2 ± 0.2
Myosin7a-FL(PP/AA)	1.8 ± 0.5	60.0 ± 20	N/A

N/A, not assayed.

*The actin sliding rate in this experiment was measured under 200 mM KCl, whereas all others were done under 50 mM KCl.

Second MyTH7 Subdomain of Myosin 7a Is the Key to Regulate Motor Activity.

To explore which regions of the tail domain of myosin 7a are required for regulating the motor activity, sequential C-terminal deletions were expressed and analyzed (Fig. 1A). All of these truncated constructs have V_max values of ≈1 s⁻¹ and K_ATPase values of ≈1 μM, which are similar to the kinetic constants obtained for myosin 7a S1, and very different from those obtained for myosin 7a-FL (Table 1). Moreover, all of the truncated constructs move actin filaments with similar speeds in motility assays at low ionic strength (50 mM KCl) (Table 1). These data suggest that the second MyTH7 subdomain is required for regulating the activity of myosin 7a-FL.

Tail Domain of Myosin 7a Folds Back and Interacts with Motor Domain.

To examine whether myosin 7a-FL is in a bent conformation we observed the molecules in the electron microscope by negative stain under conditions where actin-activated MgATPase activity is low (100 μM ATP and at 100 mM KCl) (Fig. S2). Image processing was used to produce class averages (Fig. 3), and examples of raw images are shown in Fig. S3. The majority of the myosin 7a-FL molecules were compact, making it difficult to identify the motor domain (Fig. 3A). However, when myosin 7a-FL was stained in the absence of ATP, the molecules were extended, and the motor domain at the top of the molecules was in line with the lever (Fig. 3B). Evidence that the mass at the top of the bent structure is a motor domain is 3-fold. First, its appearance is consistent with that of the molecular model of myosin 2 motor domain in the presence of ADP plus vanadate fused with a model of a 5 IQ lever domain (Fig. 3I). Second, when myosin 7a-FL is imaged in the absence of ATP the molecule extends and the presumptive motor domain is in line with the lever as opposed to being angled (Fig. 3B). Data from a number of myosins show that, in general, motor domains are at an angle to the lever in the presence of ATP and are in line with the lever in its absence (see Fig. 3I and J for molecular models) (16, 17). Finally, images of the S1-SAH construct in ATP show a single-headed molecule with a motor domain and lever that matches the top and left side appearance of myosin 7a-FL in ATP (Fig. 3I). Comparison of the images shown in Fig. 3I allow us to be confident that the motor domain is at the top of the folded molecules, and is at an angle to the lever. These results suggest that ATP is required for the compact, bent conformation. The size and appearance of these molecules indicates that myosin 7a is a single-headed motor under these conditions. The raw images of single molecules (Fig. S3) show that there is no additional density, flexibly attached to the compact structure, that has been lost by image averaging. However, comparing the high mass of myosin 7a-FL (≈350 kDa, including the 5 calmodulins that this myosin is predicted to bind) with the small size of the molecules in the averaged or raw images, it appears that the entire molecule is not picked out by negative staining. This could be, in part, that a portion of the molecule projects out of the plane of the image. However, the lever, which is predicted to have 5 calmodulins bound to it, is shorter than expected. A comparison with myosin 7a S1-SAH in which the entire lever with its 5 calmodulins should be visible, shows a shorter lever than expected (Fig. 3H) especially when compared with myosin 5 S1, which has a lever known to bind 6 calmodulins (Fig. 3I). Densitometry from gels using myosin 5 HMM, which is known to bind 6 calmodulins per heavy chain as a standard, we calculate that myosin 7a S1-SAH binds only 3 calmodulins, helping to account for the shorter lever observed by EM. Thus, 1 or more of the C-terminal IQ motifs seem not to bind calmodulin.

Fig. 3.

Electron microscopy of myosin 7a constructs. (A–H) Averaged images of full-length or modified myosin 7a molecules. Averages used 80–250 molecules. Representative raw image are shown in Fig. S3. (A) Myosin 7a-FL in the presence of ATP. (B) Myosin 7a-FL without ATP. (C) Myosin 7a-TD3 with ATP. (D) Myosin 7a-TD3 without ATP. (E) Myosin 7a-RK/AA mutant with ATP. (F) Myosin 7a-RK/AA mutant without ATP. (G) Myosin 7a-FL with ATP and 400 mM KCl. (H) Myosin 7a S1-SAH with ATP. (I–J) Comparison of atomic models of myosin 7a head with image averages of myosin 5 S1, myosin 7a S1-SAH, and myosin 7a-FL with (I) and without (J) ATP. The models use scallop myosin 2 ADP.Vi or apo motor domains, respectively, together with a lever that has 5 calmodulins spaced at intervals of 23-aa residues along the heavy chain, as in the myosin 7a sequence. (Scale bars: 20 nm in H applies to A–H; 10 nm in J applies also to I.)

Evidence for an Ionic Strength-Dependent Conformational Change in Myosin 7a-FL.

The in vitro movement of actin filaments by Drosophila myosin 7a S1-SAH does not depend on ionic strength, and filaments are moved at a range of KCl concentrations from 10 mM to 400 mM (Fig. S4 and Movie S1). In contrast, the ability of myosin 7a-FL to move actin filaments depends strongly on the ionic strength. At 50 mM KCl, myosin 7a-FL binds actin filaments to the surface but does not move them. However, when the KCl concentration is increased to 200 mM, myosin 7a-FL moves actin filaments at the rate obtained with S1-SAH under these conditions (20 ± 5 nm/s). This ionic strength dependence is reversible because lowering the KCl concentration back to 50 mM stops actin movement. Because increasing ionic strength is known to extend myosins 2 and 5 (18, 19), it seems likely that at low ionic strength myosin 7a is bent and inhibited from moving actin filaments, but becomes extended at higher ionic strengths and is then able to move actin filaments. The off state of other myosins such as unphosphorylated (inactive) smooth muscle also tether immobile actin filaments to the surface (20).

The basal MgATPase rate measured in the absence of actin also shows a strong dependence on ionic strength with a steep transition occurring in the range of 200 mM (Fig. S5). Previous studies on smooth muscle myosin 2 showed a similar ionic strength dependence of the basal MgATPase activity that was correlated with a switch from a bent to an extended structure (21).

Microscopy of myosin 7a-FL at 400 mM KCl in the presence of ATP, showed the molecules to be largely extended with a globular feature at one end of a narrow rod that is likely to be the motor domain attached to the lever at a prepowerstroke angle (Fig. 3G). At the other end of the lever is a variable mass, presumably comprising the MyTH4 and FERM domains. These molecules look similar to those at low ionic strength in the absence of ATP, except that the motor domain is angled with respect to the lever, as expected for molecules in the presence of ATP (see above). This is consistent with the above data showing changes in the enzymatic and motile properties of the myosin at higher ionic strength.

Myosin 7a-FL forms aggregates at 0.1M KCl in a time- and concentration-dependent manner, and this precluded attempts to study putative conformational changes by using analytical ultracentrifugation. We used controlled proteolysis to show that there is an ATP-dependent difference in the rate of proteolysis of the heavy chain at 0.1 M KCl, but not at 0.5 M KCl (Fig. S6).

MyTH7 Subdomain Is Required for the Folded Conformation.

The structure of the truncated and unregulated myosin 7a-TD3, in which the C-terminal FERM/MyTH7 domain, is missing was also examined at 100 mM KCl and 100 μM ATP. In contrast to myosin 7a-FL, the TD3 molecules were mostly extended with a single, clearly discernable motor domain that is strongly angled to the rest of the molecule (Fig. 3C). The tail domains appear to be in multiple conformations.

Conserved SH3-Binding Motif in the MyTH7 Subdomain Is Not Involved in the Regulation.

There is a conserved PxxP motif in the second MyTH7 subdomain (Fig. 1B). In other proteins, PxxP motifs bind to SH3 domains (reviewed in ref. 22). We therefore tested the possibility that the PxxP motif stabilizes the bent myosin 7a by interacting with the SH3-like domain located at the N terminus of the molecule or the SH3 domain of the tail (Fig. 1A). Both of the prolines were mutated to alanines. The mutant myosin behaved similarly to myosin 7a-FL in MgATPase assays, indicating that this motif is not involved in the regulation (Table 1).

Point Mutations in the Second MyTH7 Subdomain Deregulate Myosin 7a-FL.

Knowing that the bent inactive conformation had a strong electrostatic basis, we searched the second FERM/MyTH7 domain for conserved charges. R2140 and K2143 are highly conserved in the myosin 7a subfamily, and thus we mutated both these residues to alanines (Fig. 1B). (Note that whereas glutamine replaces the arginine in Danio rerio myosin 7a, there is 1 D. rerio myosin 7a sequence (XP 001921522) in which there is an arginine at this position.) The V_max of this mutant was 0.6 s⁻¹ with a K_ATPase value of 2.3 μM, demonstrating that regulation has been lost (Table 1). Correspondingly, electron microscopy revealed that, even in the presence of ATP at low ionic strength, the mutant remains extended and resembles the myosin 7a-TD3 construct (Fig. 3 E and F). No extra density is observed below and to the right of the motor domain.

Second FERM Domain in the Myosin 7a Tail Binds with Actin Filament.

The FERM domain in the tumor suppressor protein, merlin, binds weakly but directly to F-actin (23). To test whether the entire second FERM domain including the MyTH7 subdomain in the myosin 7a tail binds to actin filaments, we used actin cosedimentation experiments. These revealed that the second FERM domain bound to actin in a saturable manner with a K_d of 28 ± 6 μM (Fig. 4). Interestingly, this affinity is close to the value of the K_ATPase determined for the activation of the MgATPase activity of myosin 7a-FL by actin.

Fig. 4.

Binding of the C-terminal FERM construct to actin. (A) SDS PAGE bands showing the distribution of C-terminal FERM between pellet (P) and supernatant (S) on Western blots after ultracentrifugation. (B) Percentage of the construct present in the pellet increases with actin concentration, and the dataset could be fit with a hyperbolic curve with a dissociation constant (K_d) of 28 ± 6 μM. Data were collected from 2 independent experiments.

Discussion

Electron microscopy shows that full-length recombinant Drosophila myosin 7a, purified from Sf9 cells, is monomeric despite the presence of a short sequence of predicted coiled coil. The predicted coiled-coil motif in myosin 10 has been shown to form a stable single α-helix (SAH domain) rather than a dimeric coiled coil (4). A fragment of myosin 10 that contained the motor domain and all of the predicted coiled-coil domain was shown to be largely monomeric by electron microscopy and had a neck region substantially longer than that predicted for a myosin with 3 bound calmodulins. A similar domain has recently been shown to be present in myosin 6, which is also monomeric when expressed as the full-length molecule in the baculovirus/Sf9 system (5, 24). The sequence of 70 aa of myosin 7a-FL downstream of the IQ motifs, including the predicted coiled coil, would be able to form 26 intrahelical ionic bonds, which are the major stabilizers of SAH domains. By contrast, there is poor evidence of a hydrophobic seam in the predicted coiled-coil region, so it is unlikely to dimerize. Thus, the construct containing this sequence, previously called myosin 7a HMM (14), is better called S1-SAH. It is possible there are binding partners or other in vivo mechanisms that cause Drosophila myosin 7a-FL to dimerize in vivo, but none has been reported as yet.

The regulatory mechanisms have been studied extensively for smooth muscle myosin 2 and myosin 5. Both of these myosins are dimers with respect to heavy chains and their regulated off states involve intramolecular bending where elements of the tail interact with the motor domains, although the mechanisms differ substantially between the 2 myosin types (see ref. 25 for review). Here, we show that monomeric myosin 7a exhibits a unique regulatory mechanism, which also involves an intramolecular bending. Negative-staining electron microscopy shows that myosin 7a has a compact structure. The orientation of the motors, levers, and the tails can be discerned in the negatively stained images of the regulated conformation of myosin 2 and myosin 5 (26, 27). For bent myosin 7a-FL we can identify the motor domain, but further work is required to unambiguously identify the tail domains and how they interact with the motor in the bent structure. Unlike myosin 5, in which the heads are not in a prepowerstroke state, the bent structure of myosin 7a requires ATP, and in the absence of ATP it extends. Like myosin 5 and myosin 2, bending is inhibited at higher ionic strengths (>200 mM KCl). The overall length of the bent molecule is less than expected even for the motor plus 5 IQ lever, suggesting that the lever itself may be bent or that not all IQ motifs are occupied by calmodulins (Fig. 3I). We note in this context that the third IQ motif is noncanonical in ending RSxxxS rather than RGxxxR, which may weaken binding to the N-lobe of calmodulin. This could explain why we find only 3–4 calmodulins bound per molecule for myosin 7a-FL and S1-SAH. Immunoprecipitated mouse myosin 7a also appears to bind <5 calmodulins (28).

Under conditions where myosin 7a adopts the bent conformation, its enzymatic activity is strongly inhibited, suggesting that the interaction of some portion of the tail domain inhibits the motor function. The results from a series of tail truncations showed that the C-terminal MyTH7 subdomain is required for regulation. Its removal both abolishes the ability to bend and dramatically reduces the concentration of actin required for activation of MgATPase activity. This subdomain is highly conserved among myosin 7a family members (11). Larger truncations, even ones that remove all of the tail up to the IQ motifs, have the same enzymatic signature as this minimal truncation. A nonsense mutation has been described for human myosin 7a associated with Usher 1B syndrome (29) that would truncate the molecule after residue 1896, resulting in a molecule similar in length to the TD3 construct, and we would predict this mutant to be unregulated. Raising the ionic strength or mutating the highly conserved, positively charged amino acids R2140 and K2143 to alanines in myosin 7a-FL both extended myosin 7a and allowed activation of the MgATPase activity by low concentrations of actin, suggesting that, as with myosin 5, specific electrostatic interactions between the tail and the motor domain are essential for regulation.

The enzymatic mechanism of the regulation of myosin 7a differs from that of myosin 5 in an important respect. The V_max of the actin-activated MgATPase activity of switched off preparations of myosin 5, even at high levels of actin, is much lower than that of myosin 5 that has been activated by addition of calcium ions or of the cargo receptor melanophilin (19, 30–32). In contrast, at high concentrations of actin, the MgATPase activity of full-length myosin 7a is the same as that of unregulated, truncated fragments. The difference between myosin 7a-FL and its truncated fragments is in the weaker apparent affinity for actin (K_ATPase of 30 μM compared with 1 μM, respectively). We have shown that the tail fragment comprising the second FERM domain including its MyTH7 subdomain binds directly to actin with a K_d of ≈30 μM. This similarity in the K_ATPase of myosin 7a-FL to the affinity of the second FERM domain for actin suggests that as actin concentration increases, the tail binding site binds to actin and causes the molecule to extend. This would result in an increase in the ATPase as the newly extended molecule would now be in the presence of essentially saturating actin allowing it to hydrolyze ATP at a rate of 1 s⁻¹.

Based on these results, we propose the following model for one possible way in which Drosophila myosin 7a might be regulated (Fig. 5). In the absence of binding partners that might activate the MgATPase of myosin 7a similar to that described for the interaction of myosin 5 and melanophilin (32), Drosophila myosin 7a is likely to be bent and inactive inside most regions of the cell. However, in regions in which actin density is high, myosin 7a could bind to actin via its second FERM domain, unfold, and become activated, allowing the motor domain to interact in an ATP-sensitive manner with an adjacent actin filament. Interestingly, myosin 7a is found in regions of cells that contain actin-rich structures such as Johnston's organ in Drosophila, which facilitates hearing, and the stereocilia of the inner ear in vertebrates (8, 9, 12, 33).

Fig. 5.

A model for how myosin 7a is active in actin-rich areas. In regions of low actin concentration, myosin 7a-FL tail bends back and inhibits motor activity in the presence of ATP. The ATP-insensitive, low-affinity actin-binding site in the tail binds to actin at the high actin concentrations in an actin bundle, as in this filopodium, allowing the motor domain to actively interact with actin. Alternatively, the FERM domains may bind to membrane components triggering movement of these components along actin. Finally, a protein binding partner (shown as a mauve hexagon) may bind to the tail of the myosin and cause it to extend and become active.

An alternative regulatory mechanism is that binding partners in the cell may bind to the tail region of myosin 7a, extending the molecule and activating its enzymatic and motile properties (Fig. 5). We have shown that myosin 7a will move processively along actin filaments if artificially dimerized by the addition of a leucine zipper (14). Perhaps binding partners exist that functionally dimerize 2 motor domains of myosin 7a. Such proteins have been speculated to exist as partners for myosin 6 (see ref. 34 for review). In this regard, it is interesting that numerous binding partners have been found for mammalian myosin 7a, although the effects of these proteins on the structure or enzymatic properties of this myosin have not yet been determined (for review, see ref. 35).

Materials and Methods

Protein Expression, Purification, and Reagents.

A Drosophila myosin 7a cDNA clone was provided by Daniel Kiehart (Duke University, Durham, NC). The length and domain structure of each construct shown on Fig. 1A and all constructs were subcloned into baculovirus transfer vector pFastBac1 (Invitrogen). Overlapping PCR was applied to make point mutations. A FLAG tag (sequence: DYKDDDDK) was fused to the C terminus of all constructs to aid purification. Creation and amplification of recombinant baculoviruses was performed as described previously (36). Drosophila calmodulin was coexpressed with each construct (36). Expressed proteins were purified as described (37). GST fused second FERM domain was subcloned into pGEX6P-1 bacterial expression vector (Amersham Biosciences), and the protein expression, purification, and GST tag cleavage were performed according to the manufacturer's instructions. All of the recombinant plasmids were confirmed by DNA sequencing. Actin was prepared from rabbit skeletal muscle (38). Antibody against Drosophila myosin 7a C terminus peptide (TNMNKNRTIRAN) was prepared from immunized rabbit serum by Invitrogen and was affinity purified by chromatography on Affi-Gel 15 (Bio-Rad) coupled with peptide. Other reagents were from Sigma.

ATPase Activity and Actin Sliding Assays.

Steady-state MgATPase activities were measured at 25 °C (39) in a buffer containing 10 mM Mops (pH 7.0), 2 mM MgCl₂, 0.1 mM EGTA, 1 mM ATP, and 50 mM KCl. Motility assays were performed at 25 °C in motility buffer containing 20 mM Mops (pH 7.4), 4 mM MgCl₂, 0.1 mM EGTA, 50 mM DTT, 1 mM ATP, various concentration of KCl and an oxygen scavenging system consisting of 25 μg/mL glucose oxidase, 45 μg/mL catalase, and 2.5 mg/mL glucose (37). Data fitting and analysis was performed by using ORIGINLAB 7.0 (Microcal, Amherst, MA).

Actin Binding Assay.

Actin was cosedimented with the second FERM domain of myosin 7a (GST-tag removed) at 100,000 × g for 20 min in 10 mM Mops (pH 7.0), 2 mM MgCl₂, 0.15 mM EGTA, 1 mM DTT and in the absence of ATP or in the presence of 1 mM ATP. Quantification was performed by Western blot analysis of the supernatant and the resuspended pellet with anti-rabbit IRDye 680 fluorescent antibody. Fluorescence intensity was analyzed by ODYSSEY Infrared Imaging System (LI-COR).

Electron Microscopy and Image Processing.

All constructs were supplemented with a 2-fold excess of calmodulin and diluted to ≈100 nM myosin in 100 mM KCl, 5 mM Mops, 1 mM EGTA (pH 7.5) with the addition of 100 μM ATP and 3 mM MgCl₂ where required. In addition, myosin 7a-FL was also diluted into buffer containing 400 mM KCl. Protein was applied to UV-treated, continuous carbon-coated grids and negatively stained with 1% uranyl acetate. Images were recorded on film at a nominal magnification of 40,000× by using a JEOL1200EX at 80 kV and digitized at 0.52 nm/pixel, as calibrated by using the 14.4-nm repeat of paramyosin filaments. All image processing was performed by using SPIDER as described (40). Datasets ranged from 1,857 to 6,654 particles. K-means classification was performed by using masks based on the variance of the global average, with the exception of the RK/AA mutant in the absence of ATP and the S1-SAH construct, for which a mask of just the tail region was used.

Atomic Model Building.

A 5IQ model lever was constructed by using the 2.5-Å resolution structure of calcium-free calmodulin bound to IQ motifs 1 and 2 of murine myosin 5 (2IX7) (41). By using MacPyMol, the backbone atoms of IQ motif 2 were superposed on those of IQ motif 1 from a second 2IX7 structure, to give a 3IQ model (RMSD 0.21 Å). This process was repeated with a third and a fourth 2IX7 structure to give a 5IQ model with all IQ motifs having 23-residue spacing, as in myosin 7a. Overlapping CaM and IQ motif sequences were deleted. To create apo and prepowerstroke head models, backbone atoms of the first IQ motif of the 5IQ lever model were superposed on the first IQ motif of scallop myosin, PDB ID codes 1SR6 and 1QVI, respectively (RMSD of 0.252 and 0.416 Å, respectively) (41, 42), and the scallop lever was then deleted.