Calmodulin regulates dimerization, motility, and lipid binding of Leishmania myosin XXI

Abstract

Myosin XXI is the only myosin expressed in Leishmania parasites. Although it is assumed that it performs a variety of motile functions, the motor's oligomerization states, cargo-binding, and motility are unknown. Here we show that binding of a single calmodulin causes the motor to adopt a monomeric state and to move actin filaments. In the absence of calmodulin, nonmotile dimers that cross-linked actin filaments were formed. Unexpectedly, structural analysis revealed that the dimerization domains include the calmodulin-binding neck region, essential for the generation of force and movement in myosins. Furthermore, monomeric myosin XXI bound to mixed liposomes, whereas the dimers did not. Lipid-binding sections overlapped with the dimerization domains, but also included a phox-homology domain in the converter region. We propose a mechanism of myosin regulation where dimerization, motility, and lipid binding are regulated by calmodulin. Although myosin-XXI dimers might act as nonmotile actin cross-linkers, the calmodulin-binding monomers might transport lipid cargo in the parasite.

Keywords: motor properties; unconventional myosin.

Fig. 1.

Calmodulin binding prevents the dimerization of myosin XXI. (A) Domain structure of full-length myosin XXI. By comparing the sequence with well-studied myosins, we identified the highly conserved regions of the motor domain (gray). In addition to the calmodulin-binding IQ motifs identified in a previous study (10), we found another IQ motif (yellow) immediately following the converter domain. Sequence analysis revealed four potential coiled-coil regions (in blue). The predicted propensity to oligomerize was scored between 0 and 10 using Scorer 2.0 (11); positive numbers predict dimer and negative numbers predict trimer formation. (B) Size exclusion chromatography using a Superose-6 column and silver staining of SDS/PAGE gels (C) of the elution fractions indicate that the motor is monomeric when expressed in high levels of calmodulin and dimeric or in a higher oligomeric state when expressed at low levels of calmodulin. Aggregated myosin XXI would appear in the void volume. (D) The Western blot does not resolve endogenous calmodulin in the purified myosin-XXI preparations that were expressed in the absence of added calmodulin virus (i.e., at low levels of calmodulin). By varying the amount of calmodulin virus added during expression, the amount of calmodulin bound to the expressed myosin XXI could be controlled. The apparent difference in molecular weight of calmodulin in C and D is due to the differences in buffers.

Fig. 2.

Analysis of calmodulin binding to predicted calmodulin-binding motifs. Tryptophan fluorescence (excitation at 290 nm, emission at 323 nm) was used to study calmodulin binding to target peptide sequences. (A) The synthetic peptide of the first predicted calmodulin-binding motif C-terminal of the converter (aa 754–769) bound to human calmodulin (CamH) and to Leishmania CamL1 with a 1:1 stoichiometry and a K_d of 20 nM in the presence of calcium (pCa 4). (B) For this target peptide, calmodulin binding was reversible and calcium-dependent. (C) Leishmania CamL2 also bound to the peptide in the presence of calcium, but remained bound when free calcium was subsequently lowered to less than 100 nM.

Fig. 3.

Analysis of myosin-XXI tail constructs to investigate domains involved in dimerization. (A) SEC data of myosin-XXI tail constructs. The UV signals (in arbitrary units) were normalized to the highest peak for each experiment. Each construct was loaded at ∼30 μM. (B) SEC showed that, in the presence of the N-terminal RFP tag, both monomer and dimer conformations were observed for the tail constructs. The ratio of monomer to dimer was dependent on protein concentration. The calculated molecular weight for RFP and CFP is 27 kDa. (C) For the RFP-830–930 construct, the presence of the N-terminal RFP tag was insufficient to induce dimerization. (D) FRET study of a control CFP-RFP fusion protein containing a cleavable TEV site. CFP-donor excitation was performed at 445 nm. The emission spectrum was recorded from 450 to 650 nm (in gray for the uncleaved fusion protein and in black for the fusion protein following enzymatic cleavage). Protein concentration before cleavage was 7 μM; measurements were done at 22 °C. (E) CFP-donor emission at 475 nm (blue) and RFP-acceptor emission at 605 nm (red) for a mixture of 2-μM CFP-730 tails and 7-μM RFP-730 tail constructs in the presence of 100 μM calmodulin and 1 mM CaCl₂. (F) In the absence of calcium calmodulin, the mixture of 2-μM CFP-730-tails and 7-μM RFP-730 tail constructs resulted in a FRET signal due to the formation of heterodimers. (G) Summary of the FRET experiments with 1:1 mixtures of RFP-730 and CFP-730 tails (2 μM each) in the presence and absence of 100 μM calmodulin and of the control CFP-RFP fusion protein (2 μM) before and after enzymatic cleavage. The FRET efficiency was calculated as described in Materials and Methods. Each experiment was performed at least three times with three different preparations.

Fig. 4.

Lipid binding of myosin XXI. (A) Pull-down experiments of Folch mixed liposomes and myosin-XXI constructs showed that monomeric constructs bound to liposomes. (B) PLO experiments indicated lipid binding along the entire tail of monomeric myosin XXI, whereas dimers (730-Tail) did not bind to lipids. The PLO data also showed that myosin XXI binds to Folch mixed lipid preparations but not to pure PC or PE lipids. (C) Binding to lipid membrane strips showed that all constructs bound to PIP monophosphates. Binding to di- and triphosphates was found N-terminally of amino acid 830. Binding studies of RFP-730–830 confirmed that the monomeric form (fraction 7) bound to phospholipids, whereas the dimeric form (fraction 2) did not. The fractions collected from a SEC experiment on the RFP-730–830 construct were analyzed using silver-stained SDS/PAGE (Inset) and confirm that both peaks originate from a single protein. (D) The cartoon indicates capacity for phospholipid binding along the entire myosin-XXI tail, including the converter region (645–747).

Fig. 5.

Sequence analysis of myosin XXI to localize lipid binding and dimerization. (A) We identified a number of potential lipid-binding sites, including a PX domain that overlapped with the converter region of the motor. (B) The potential lipid-binding sites (green) overlapped with the potential dimerization domains (blue). (C) Six lipid-binding domains (LBD1–6) with basic residues flanking a hydrophobic patch of amino acids (25, 26) were found. (D) Using the basic-hydrophic scale (25), we identified a potential lipid-binding site following the PX domain and confirmed LBD5 as a site with high potential for phospholipid binding.

Fig. 6.

Structural analysis of the PX domain in the myosin-XXI converter. (A) We used the crystal structure of scallop muscle myosin II (30) and replaced the sequence in the converter domain with the sequence of L. donovani. The amino acids specific to the PX domain (in green) form a patch on the surface of the converter (in red), consistent with these residues interacting with membranes without interfering with the general structure and function of the converter of this myosin motor. Comparison of 130 sequences from 21 myosin classes (Dataset S1) suggest that myosin XXI is exceptional within the myosin family in terms of featuring a PX domain in the converter region. (B) PIP strips showing that site-directed mutagenesis of arginine tyrosine (RY) in the expressed PX mutant converter construct completely abolished the characteristic Pi(3,5)P₂ binding of the PX domain.

Fig. 7.

Model of myosin-XXI regulation of dimerization, lipid binding, and motility. (A) The model summarizes our findings. At least two different calmodulin-like proteins (CamL1 and CamL2) in L. donovani can bind to the same calmodulin-binding motif (754–769) following the converter and prevent dimerization. Their binding differs in K_d and calcium sensitivity. Monomeric, calmodulin-binding myosin XXI is motile and binds to phospholipids, whereas the dimeric motor does not and is nonmotile. (B) The velocity of actin filaments driven by monomeric myosin XXI unspecifically adsorbed to nitrocellulose is four times lower compared with motors bound to Folch lipid bilayers. (C) Negatively stained electron micrographs of single, monomeric myosin XXI in rigor (ATP < 1 μM) adsorbed to carbon-coated EM grid on their own and when bound to actin. Representative class averages are shown. Crystal structures of truncated myosin V in rigor with a single light chain bound (green) have been overlaid. The numbers state the number of images contributing to the class average. (D) Representative negatively stained electron micrograph of two actin filaments cross-linked by dimeric myosin XXI in rigor (ATP < 1 μM).