Thirteen is enough: the myosins of Dictyostelium discoideum and their light chains

Identification of Dictyostelium discoideum myosins

The Dictyostelium cDNA database in Japan [21] was searched for the new myosin genes to confirm their expression (exclusion of pseudogenes) and gene structure, as the newly identified myoF, myoH, and myoG genes contain several introns (Table 1). However, the cDNA clones only cover the region around the last of the introns of Myo5A (MyoH). The extremely high AT content of Dictyostelium introns and the help of a multiple sequence alignment of over 1700 myosin motor domains (M. Kollmar, unpublished data) nevertheless allowed the unambiguous identification of the introns and, subsequently, the protein coding regions.

Expression pattern of the new myosin genes

The cDNA database is based on libraries of cells obtained from several different stages of the developmental cycle of Dictyostelium [23]. One library contains data from clones obtained at the so-called first finger stage (14 h – 16 h) of development. A second library was constructed from vegetatively growing cells. A third library contains clones derived from "sexually competent cells" (cells are cultured in liquid medium in the dark, and are competent for fusion with opposite mating type cells), cells that are roughly equivalent to growth phase cells. The data supposed to contain full-length genes is based on libraries from cells at the following stages of the developmental cycle: axenically growing cells, cells developed on nitrocellulose filter to aggregation stage (8 hours), slug stage (16 hours) and early culmination stage (20 hours). For Myo5A (MyoH) and MyoG cDNA fragments have been obtained in all libraries containing the potentially full-length genes indicating that these two myosins are expressed in all stages of the developmental cycle. However, only one gene fragment has been obtained for Myo1F from the library of full-length genes of axenically growing cells. The cDNA data is not supposed to reveal the complete expression pattern of all proteins at the different developmental stages. But the number of obtained clones and the occurrence in a specific library indicates that Myo1F might not be strongly expressed, and primarily expressed in vegetative cells. These conclusions have of course to be confirmed by further experimental data.

Classification, nomenclature, and phylogenetic analysis

The classification and suggested revised nomenclature of the Dictyostelium myosins is summarized in Table 1 and shown in Figure 1. The general nomenclature for myosin proteins uses the term Myo followed by the class number (Arabic numeral) and the variant (Arabic letter). The class-II myosins are exceptions as Mhc or Myh is used as abbreviation (leaving out the class number) followed by the variant designations as either Arabic letters or Arabic numerals. To not severely increase the number of classes, myosins that do not have a homolog in at least one other organism should be referred to as orphans. So far, the Dictyostelium myosins have not consistently been named according to that nomenclature. The phylogenetic analysis of over 1700 myosin motor domains (M. Kollmar, unpublished data) together with the completed sequence of the Dictyostelium genome now allows a revision. The new nomenclature does not severely change the old names as the class-II myosin stays untouched, the class-I myosins only get the class designation added with the exception of MyoK that will now be referred to as Myo1G, and MyoM is still an orphan myosin and will not be renamed as long as it cannot be grouped to a certain class. The newly identified MyoG (preliminarily named according to its locus) is also an orphan myosin and might be renamed as soon as further homologous myosins are derived. However, the classification of the other myosins, MyoH, MyoJ, and MyoI, is not that unambiguous.

A phylogenetic tree of 180 myosins of the classes V, VIII, and XI (including the former class-XIII myosins) does not group MyoH and MyoJ to any of the already assigned classes (Figure 2, additional file 1). Instead of assigning these myosins a new class, as it happened in the past e.g. for the classification of the Acetabularia cliftonii class-XIII myosins, I suggest to name MyoH and MyoJ Myo5A and Myo5B, respectively. Both myosins have a similar domain organisation as the class-V and the class-XI myosins. But because Dictyostelium separated from the Fungi/Metazoa lineage after the separation of the plants, MyoH and MyoJ should rather be referred to as class-V than class-XI myosins. This classification is supported by the analysis of the over 1700 myosins that revealed MyoH and MyoJ to be closer related to the class-V myosins than to the plant myosins (M. Kollmar, unpublished data).

MyoI does not group to any of the designated classes containing myosins with MyTH4/FERM domains as is shown by the phylogenetic analysis of 126 myosin motor domains of classes VII, X, XV, and XXII (Figure 3, additional file 2). It does not group to the class-XXII myosins although it branches from the class-XXII myosins in the tree shown. The reason is that the branching occurs very close to the separation point of the other classes, and in phylogenetic trees including all classes MyoI also often branches very early from the class-X myosins. A close view at the protein sequence in the multiple sequence alignment shows that MyoI shares several class specific features off all four classes that prevent a better classification. In addition to the motor domain sequence, the domain organisation of MyoI is unique compared to members of the other classes (Figure 4). In contrast to class-VII myosins, MyoI does not have an N-terminal SH3-like domain, it has four instead of five IQ-motifs and it misses the first FERM domain. The tails of class-X myosins are different to that of MyoI as they are characterised by two consecutive PH (p leckstrin h omology) domains followed by the MyTH/FERM tandem domain. The domain organisation of the class-XV myosin tails is similar to that of MyoI except that the mammalian myosins have a very long N-terminal domain, while the insect myosins miss the SH3 domain. Like MyoI, the class-XXII myosins do not have an N-terminal domain, but their tail domain is different containing two complete MyTH/FERM tandem domains but no SH3 domain. Thus, MyoI cannot be grouped to any of the already designated classes and should be considered as an orphan myosin. This implicates that MyoI cannot be considered as a specific model for class-VII myosins as it has been suggested earlier [11, 24]. The Dictyosteliida diverged before the evolution of the Metazoa. MyoI therefore rather resembles a common ancestor of the four classes instead of grouping to one of them.

Domain structure of the known myosins

Most of the tail domains of the known myosins have already been described and functionally analysed in some detail [25, 26]. While there is agreement on the determination of the larger tail domains like the class-I myosin membrane-binding and SH3 domains, or the Myo5B (MyoJ) DIL (dil ute) domain, there are contradicting predictions of the IQ motifs. The IQ motif is a short sequence motif of alpha-helical structure that is able to bind calmodulin or calmodulin-like proteins. IQ motifs have been predicted in the past by using the general pattern IQxxxRGxxxR [3]. The multiple sequence alignment of the whole myosin family now allows a revision of the motif (M. Kollmar, unpublished data). According to the revised motif several IQ motifs are found in the Dictyostelium myosins that have not been recognised before (Figure 1 and Figure 5). The starting isoleucine in the motif is often substituted by other large hydrophobic amino acids. The glutamine at the second position is mainly conserved, except for some cases where it can be substituted by lysine, glycine, or glutamate. A very important position is the residue before the first arginine, that is almost always a large hydrophobic amino acid, in most cases an aromatic one. The first arginine is also highly conserved, except for a few cases where it is substituted by lysine, leucine, or isoleucine. The following glycine is not very conserved in IQ motifs of myosin tails. The position after that glycine is mainly occupied by large hydrophobic amino acids, in most cases aromatic residues, but histidines, asparagines, and glutamines are also found. The second arginine of the initial motif is also not very conserved. Using the revised motif, no IQ motif is predicted for Myo1G, one is predicted for Myo1B and Myo1D, and two are predicted for the other class-I myosins. Myo1C might contain a third IQ motif but then the packing of the light chains would be relatively dense. However, Myo1B binds a light chain that consists of only two EF-hand motifs [half the size of a normal calmodulin-related protein, [19]]. Two similarly small myosin light chains could easily bind to the two closely located IQ motifs of Myo1C. According to the new motif description MyoI is now predicted to contain four IQ motifs in contrast to earlier predictions of three IQ motifs [11], and Myo5B (MyoJ) contains six. The domain compositions of the newly identified myosins are described in more detail below.

Domain structure of the newly identified and analysed myosins

Structural features of the myosin motor domains

The myosins of Dictyostelium contain several protein specific extensions to surface loops of the motor domain of class-II myosin (Fig. 6). The most prominent loop-extension is the insertion of ~130 amino acids into loop-1 of Myo1G. Except for members of an arthropoda specific myosin class, that contain loop-1 extensions of up to 300 amino acids [30], this is by far the longest loop-1 of all myosins. Myo5B (MyoJ) also contains a considerably longer loop-1 (20 residues in addition to loop-1 of MhcA). Loop-1 has been implicated in influencing access to the nucleotide-binding site. It has been shown in an analysis of chimeric loop-1 mutants of smooth muscle myosin that the mobility of this loop correlates to the rate of ADP release [31]. Larger and more flexible loops resulted in faster rates of ADP release. For class-V myosins, the rate-limiting step in the catalytic cycle is ADP release [32], a necessity for these myosins to walk over long distances along the actin filaments without detaching. Based on these results, Myo5B (MyoJ) might be a very unconventional class-V myosin with a fast rate of ADP release. It will be very interesting to see whether Myo5B (MyoJ) is still a long-distance cargo transporting myosin as the other homologs of the class.

Loop-2 has been shown to be involved in both weak and strong binding interactions with actin [33]. According to this study, especially positively charged residues strengthen the binding to actin. MyoG and MyoM contain long extensions of loop-2 compared to MhcA, but the sequences contain mainly glycines, prolines, and polar amino acids. Thus, both myosins are not expected to have considerably different actin-binding properties. Loop-4 is the loop that is furthest removed from the actin surface as has been suggested from actomyosin models derived from electron microscopy. Except for some class-I, the insect class-V, some class-XVII, and some apicomplexa myosins, almost all myosins have a loop-4 of similar length. The loop has been suggested to be involved in either interactions with actin or regulatory proteins that are bound to actin [34]. For the class-I myosin myr1 from rat, it has been shown that a head fragment localizes to the same highly dynamic actin structures at the cell cortex as the full-length construct and not to the actin filaments that are regulated by tropomyosin [35]. An extended loop-4 might be responsible for this localisation as it might hinder binding to the tropomyosin stabilized less dynamic actin filaments as they occur in stress fibers. Myo5A (MyoH) contains one of the longest loop-4 of all myosins and might therefore only bind to the dynamic actin structures at the cell cortex and not to actin structures that are stabilized by other proteins.

MyoM has an extended loop at the same position where the class-VI myosins have one of their prominent insertions. This loop has been suggested to affect nucleotide binding by changing the conformation of a following loop [36]. This loop therefore protrudes within the nucleotide-binding pocket, resulting in a decrease in nucleotide accessibility. The functions of the other two surface loops, for which Myo5A (MyoH) has long extensions, have not been analysed so far.

Phylogenetic analysis of calmodulin-like proteins in Dictyostelium

An iterative TBLASTN search of the Dictyostelium genome data, starting with the sequence of Dictyostelium calmodulin A (Dd CalA), revealed 35 CBPs (c alcium-b inding p rotein) that exclusively contain EF-hands motifs (Table 3). While 32 CBPs contain four EF-hands, three contain only two EF-hand motifs (MlcB, MLC-1, and CBP10). 22 CBPs have already been described in the literature. To classify the remaining 13 CBPs and to identify those that could potentially function as myosin light chains, the Dictyostelium EF-hand proteins were compared with EF-hand containing proteins from Saccharomyces cerevisiae, Schizosaccharomyces pombe, Caenorhabditis elegans, Drosophila melanogaster, Homo sapiens, and some specific myosin light chains from other organisms. The obtained 176 calmodulin-related proteins were manually aligned and a phylogenetic tree created (Figure 7, additional file 3).

Protein name	Gene name	Size in aa	No. of introns	Chr.	Gene accession (GenBank)	Protein accession (GeneBank)	dictyBaseID (DDB...)	References, comments
ELC	mlcE	150	1	3	AAFI01000052	EAL68102	0214813	[56, 57]
RLC	mlcR	161	2	2	AAFI01000032	EAL69338	0185146	[58]
MlcB	BC5V2_0_01417	73	2	5	AAFI01000224	EAL62399	0188713	[19]
MlcD	mlcD	147	2	3	AAFI01000052	EAL68131	0214812	[18]
MLC1	BC5V2_0_01163	80	2	5	AAFI01000214	EAL62662	0219456
CalA	calA	152	3	3	AAFI01000060	EAL67642	0214955	[59]
CalB	calB	149	2	1	AAFI01000011	EAL71901	0191193	[60]
Calcineurin1	BC4V2_0_01538	183	3	4	AAFI01000129	EAL64441	0218775
Calcineurin2	cnbA	180	3	1	AAFI01000009	EAL73175	0191204
CBP1	cbpA	156	1	2	AAFI01000042	EAL68881	0185026	[61]
CBP2	CBP2	168	2	1	AAFI01000009	EAL73181	0191196	[62]
CBP3	cbpC	166	3	4	AAFI01000093	EAL65573	0191405	[63], cluster with cbpC, cbpG, cbpF, CBP14
CBP4a	cbpD1	162	3	4	AAFI01000087	EAL65815	0219925	[64]
CBP4b	cbpD2	162	3	4	AAFI01000086	EAL65870	0191416	[64]
CBP4c	BC4V2_0_01277	145	2	4	AAFI01000124	EAL64575	0186519	pseudogene?
CBP5	cbpE	180	2	2	AAFI01000027	EAL69942	0185183	[37]
CBP6	cbpF	174	3	4	AAFI01000093	EAL65572	0191383	[37]
CBP7	cbpG	169	3	4	AAFI01000093	EAL65571	0191382	[37]
CBP8	cbpH	165	2	5	AAFI01000181	EAL63115	0191153	[37]
CBP9	cbpI	163	3	2	AAFI01000022	EAL71051	0185027	[37]
CBP10	JC2V2_0_03060	107	0	2	AAFI01000044	EAL68782	0169106
CBP11	BC4V2_0_02040	192	1	4	AAFI01000145	EAL63932	0187233	No EST
CBP12	BC4V2_0_00355	171	3	4	AAFI01000093	EAL65589	0185595
CBP13	BC4V2_0_00374	102	1	4	AAFI01000093	EAL65599	0185615	pseudogene?
CBP14	BC4V2_0_01173	139	1	4	AAFI01000121	EAL64796	0186418
Frequenin1	cbpJ	194	1	1	AAFI01000011	EAL72455	0231009	[65]
Frequenin2	cbpK	192	1	2	AAFI01000027 AY655132	EAL70072 AAT72748	0167922	[65]
Frequenin3	cbpL	191	2	5	AAFI01000187	EAL63014	0231012	[65]
Frequenin4	cbpM	183	2	4	AAFI01000135	EAL64210	0231014	[65]
Frequenin5	ncsA	186	2	2	AAFI01000030	EAL69716	0231007	[65]
Frequenin6	JC2V2_0_01524	185	1	2	AAFI01000027	EAL70071	0167921	cluster with cbpK
Frequenin7	JC2V2_0_01529	225	3	2	AAFI01000027	EAL70283	0217470
Calfumirin	cafA	169	3	3	AAFI01000052	EAL68086	0214954	[66]
CentrinA	cenA	151	3	5	AAFI01000171	EAL63239	0219932
CentrinB	cenB	150	1	5	AAFI01000056	EAL67875	0220501

According to the phylogenetic tree, 16 of the Dictyostelium CBPs (CBP1-13, calfumirin) do not belong to an already named class but are closely related to the frequenins. CBP4c is very similar to CBP4a and CBP4b, but the N-terminus could not be identified. CBP13 is a protein fragment, missing the N-terminus as well as the C-terminus, and is very similar to CBP12. CBP4c and CBP13 are therefore most probably pseudogenes. Those members of this group, that have already been identified, belong to developmentally regulated genes. Their distinct spatial expression patterns suggested that they might be involved in morphogenesis [37]. Seven of the Dictyostelium CBPs belong to the frequenin class. The members of the frequenin class are also highly developmentally regulated. An interaction with a myosin family protein has not been described in the literature for any member of this group of any organism. Dictyostelium also contains two centrins and two proteins grouping to the calcineurin family.

So far, only the ELC and RLC myosin light chains as well as calmodulin subfamily proteins have been shown to bind to the IQ motifs of myosins. The exceptions are the myosin light chain of Toxoplasma gondii Myo14A [TgMLC-1, [16]], the light chain of Accanthamoeba castellanii Myo1C [AcMIMLC, [17]], and, identified in Dictyostelium, a light chain of Myo1D that is phylogenetically related the Ac Myo1C light chain [MlcD, [18]], and a light chain that binds to Myo1B [MlcB, [19]]. Next to MlcD and MlcB, Dictyostelium contains one ELC and one RLC myosin light chain that bind to MhcA. For Saccharomyces cerevisiae it has been shown that the ELC (also termed Mlc1p) not only binds to the class-II myosin, but also to the class-V myosin Myo2p. Chicken myosin-5A even binds two different ELCs next to calmodulin. Therefore, the Dictyostelium ELC is also expected to participate in binding to the class-V myosins. Dictyostelium also contains two members of the calmodulin subfamily that have not specifically been shown to bind to myosins, but are highly expected to in accordance with results obtained for other organisms. The analysis also revealed another CBP containing two EF-hand motifs that is most closely related to MlcB. It is therefore also highly expected to bind to a myosin, and termed MLC-1. As Myo1B and Myo1D bind a specific light chain each, and Myo1C is the closest homolog of the remaining class-I myosins, MLC-1 might be the specific light chain for Myo1C, but this has of course to be proven by biochemical experiments.

CBP14 does not group to any specific class but is more similar to the members of the calmodulin/ELC/RLC part of the phylogenetic tree then to the developmentally related CBPs. If it were able to function as a myosin light chain, then it would be the founding member for another specific myosin light chain class.

Identification of Dictyostelium myosins and calmodulin related proteins

The full-length sequences of ten of the thirteen Dictyostelium myosins have been reported in the literature (Table 1). Partial sequences of Myo1F (MyoF) and Myo5A (MyoH) have already been reported [27] and were used as basis for the manual assembly of the genes from clones published by the Dictyostelium Genome Sequencing Project. These genes are in consistence with the assembly of the recently published genome [28]. The two sequences, as well as MyoG that has also been derived from genomic data [38], are therefore predicted sequences that have not been verified by complete cDNAs. The Japanese cDNA project [21, 23] includes only small parts of these new myosins, which are, however, consistent with the predicted sequences. The Japanese cDNA project also includes at least fragments of all other myosins. No additional myosin genes have been found in the genome of Dictyostelium, and thus the reported "myoL" gene locus [27] might have originated from experimental artefacts.

The Dictyostelium calmodulin-related genes have been identified in an iterated TBLASTN search of the completed Dictyostelium genome starting with the protein sequence of CalA. Thus, all solely EF-hand motif-containing proteins have been collected (Table 3). The predicted sequences have been verified by searches against the Japanese cDNA database [23]. The EF-hand containing proteins from Saccharomyces cerevisiae, Schizosaccharomyces pombe, Caenorhabditis elegans, Drosophila melanogaster, and Homo sapiens have been obtained in iterative TBLASTN searches of the corresponding genomes. Specific myosin light chains from other organisms have been obtained from the protein database at NCBI.

Building trees

The complete analysis of the myosin motor domains derived from the NCBI non-redundant database and the EST and genomic sequences of more than 270 eukaryotes will be published elsewhere (Kollmar, unpublished data). All these myosin sequences, including the Dictyostelium myosins, together with their accession numbers and additional references will be accessible through the newly designed CyMoBase [[39], F. Odronitz and M. Kollmar, submitted]. The database comprises over 1700 myosin sequences (Feb. 2006) that have been used for the phylogenetic classification of the Dictyostelium myosins. The underlying phylogenetic tree has been built of a structure-guided multiple sequence alignment. The phylogenetic trees of the class-V/-VIII/-XI and the class-VII/-X/-XV/-XXII myosins have been constructed from corresponding sequences of this alignment. All phylogenetic trees are unrooted and were generated using neighbour joining and the Bootstrap (1,000 replicates) method as implemented in ClustalW [standard settings, [40]] and drawn by using TreeView [41].

The phylogenetic tree of the calmodulin-related proteins has been calculated based on a manual sequence alignment. The manual sequence alignment has been improved by iteratively creating phylogenetic trees and adjusting the alignment. The resulting phylogenetic tree is unrooted and was generated using neighbour joining and the Bootstrap (1,000 replicates) method as implemented in ClustalW and drawn by using TreeView.

Domain and motif prediction

Protein domains were predicted using the SMART [42, 43] and Pfam [44, 45] web server. The prediction of protein motifs (coiled coils, leucine zipper, prenyl-group binding motifs) is mainly based on the results of the predict-protein server [46, 47]. The IQ-motifs and N-terminal domains were predicted manually based on the homology to similar domains of other myosins included in the multiple sequence alignment of the myosins. The recognition motifs included in the SMART and Pfam databases are too restrictive, as the motifs have been created based on the small datasets available some years ago.