Molecular cloning of protozoan myosin I heavy chain genes

The myosins I are low molecular weight, nonfilamentous, actin-activated MgATPases, originally isolated from the soil amoeba Acanthamoeba castellanii almost 20 years ago by Tom Pollard and Ed Korn (1973) (reviewed by Korn and Hammer, 1988, 1989; Pollard et al. 1990). Recently, interest in these proteins has increased since it has been shown (1) that they are capable of supporting motile and contractile-like activities in vitro, (2) that Dictyostelium cells which contain myosin I but lack conventional type myosin (myosin II) retain the ability to locomote, chemotax and phagocytose (reviewed by Spudich, 1989), (3) that myosins I and II are differentially localized in both migrating and dividing cells and the pattern of localization is consistent with myosin I playing a crucial role in supporting myosin Il-independent motile functions (Fukui et al. 1989), and (4) that myosins I exist in higher eukaryotes (Montell and Rubin, 1988; Hoshimaru and Nakanishi, 1987; Garcia et al. 1989). In this article we describe briefly how the myosin I heavy chain (MIHC) genes of Acanthamoeba and Dictyostelium have been identified, what is known currently concerning the complexity of the MIHC gene families in these two protozoans, and what might be the best approaches for cloning MIHC genes in other organisms.

Acanthamoeba myosin I for several reasons. While a great deal was already known about the biochemistry of the amoeba myosins I, and while there was immunological evidence that they were true gene products, there remained a suspicion in the field that these nonfilamentous myosins might in fact be simply proteolytic breakdown products of a conventional-type amoeba myosin (in point of fact, type II myosins can be proteolyzed to a soluble active fragment, subfragment 1). Cloning the MIHC gene would resolve this question unambiguously, would indicate to what extent the primary structure of the myosin I heavy chain resembled that of conventional myosins and to what extent it differed, and would hopefully allow the use of molecular genetic techniques to examine myosin I function in living amoebae. In 1987 we published the sequence of Acanthamoeba myosin IC^* (Jung et al. 1987), one of the three known isoforms of Acanthamoeba myosin I (IA, IB and IC), and the first MIHC sequence described. Since then we have completed the sequence of myosin IB (Jung et al. 1989) and the partial sequence of a third isoform (J. Gu, G. Jung and J. A. Hammer III, unpublished observations).

All three of these MIHC genes, as well as an Acanthamoeba myosin II heavy chain gene (Hammer et al. 1987) and the heavy chain for an apparent high molecular weight form of myosin I (HMWMI; see below), were isolated as genomic clones from an Acanthamoeba Sau-3A partial-digest genomic DNA library constructed in phage Å2001 (Hammer et al. 1986). In every case, the genes were identified using a 2.7 kb BarnH.1 fragment from the nematode une 54 muscle myosin heavy chain gene (Kam et al. 1983) as a heterologous probe. This probe spans residues 35–760 of the nematode myosin heavy chain, or approximately 90 % of the globular head domain coding sequence. This portion of the nematode heavy chain contains several regions of sequence that are highly conserved among all myosins, e.g. the ATP binding site sequence and the sequence surrounding the reactive thiols. Unlike Dictyostelium (see below), nematode and Acanthamoeba share a similar codon bias (strong preference for G and C in the third position), which was crucial to the success of this approach. In addition, before actually screening the library, we found that this nematode probe would detect at moderate stringency both discrete bands in Southern blots of restricted Acanthamoeba DNA and RNA transcripts of the size expected for the MIHC in Northern blots of amoeba poly(A)⁺ RNA (Hammer et al.

In 1984 we set out to clone the heavy chain gene for 1986, 1987). These preliminary studies gave us confidence that this approach would work.

All of the Acanthamoeba MIHC genes cloned to date are highly interrupted by introns (myosin IB, 17 introns; myosin IC, 23 introns; HMWMI, 17 introns; the partially-sequenced MIHC isoform referred to above is a cDNA clone). For the portion of these genes that encodes the globular head-like sequence, intron positions and reading frame were determined by homology with other sequenced myosins. Determining the intron positions and reading frame for the tail domains, which show no homology to any portion of conventional myosin sequences, was particularly difficult for the first myosin I sequenced (IC) and required very careful sequencing as well as mapping of all intron splice sites by Sl-nuclease protection analyses (Jung et al. 1987). The sequencing of Acanthamoeba myosin IB, as well as several of the Dictyostelium MIHC isoforms (see below), was simplified by the fact that they demonstrate striking homology throughout their tail domains to Acanthamoeba myosin IC (see below).

The heavy chain of Acanthamoeba myosin IC revealed a polypeptide of approximately 127×10³M_r composed of a 77×10³M_r Sl-like domain fused to a 51×10³M_r C-terminal domain whose sequence shows no significant similarity to any portion of conventional myosin sequences and lacks completely the characteristic sequence repeats found throughout the coiled-coil rod-like tail of type II myosins. These results, in addition to proving unequivocally that myosin I is a true gene product, provided a way to integrate further the biochemical properties of the purified protein. Specifically, the presence of a slightly truncated Sl-like domain was consistent with the expression of actin-activated Mg²⁺-ATPase by myosin I and its ability to support actin-based contractile activities in vitro (indeed, SI alone has been shown to be sufficient to generate both movement and force in vitro (Toyoshima et al. 1987; Kishino and Yanagida, 1988)). In addition, the lack of a coiled-coil tail sequence was consistent with the monomeric nature of the protein (one heavy chain and one head per molecule) and its inability to self-associate into filaments at low ionic strength like myosin II. This result further supported the conclusion, drawn from studies of the purified protein (reviewed by Korn and Hammer, 1988; Korn et al. 1988), that nature has devised ways to power actin-based motility other than by bipolar filaments of myosin II. The results obtained from the subsequent cloning and sequencing of Acanthamoeba myosin IB (Jung et al. 1988), as well as the partial sequence of a third isoform, are completely consistent with all of these conclusions.

One of the important observations to come from determining the sequences of several Acanthamoeba MIHCs is that, in addition to the high degree of similarity between head sequences, tail domain sequences are conserved throughout (Jung et al. 1988). Specifically, the tail domains contain in every case three distinct and conserved regions of sequence, which we refer to as tail homology regions 1, 2 and 3 (numbered from the N terminus). Tail homology region 1 (TH-1) spans approximately 220 residues, is about 60 %t conserved between the isoforms, and is characterized by clusters of basic residues. Tail homology region 2 (TH-2) spans approximately 180 residues and is characterized by being extremely rich in glycine, proline and alanine residues (GPA-rich; these residues appear in part as irregular short repeats) and by having a strong net positive charge. While the GPA-rich regions from different isoforms cannot be aligned in any unique way, they are conserved in their unusual amino acid composition and net positive charge. Tail homology region 3 (TH-3) spans approximately 50 residues, is about 75 % conserved between the isoforms, and is also found in a wide range of other proteins that associate with the submembraneous cytoskeleton, including all members of the nonreceptor tyrosine kinase family (such as pp60-src, in which this sequence motif is referred to as src homology region 3, or SH3), phospholipase cy, brain spectrin, and a yeast actin binding protein (Rodaway et al. 1989; Drubin et al. 1990).

What gives these conserved tail sequences particular significance is that their positions correlate very well with the locations of the two apparent functional domains within the tail, as defined in vitro using proteolytic fragments of the heavy chain and/or heavy chain fragments expressed in E. coli. These studies have shown that the myosin I tail domain contains two potential anchoring sites for the Sl-motor domain: a membrane binding site and a second actin binding site. The interaction site for anionic phospholipid membranes has been mapped to the N-terminal half of the tail, i.e. within TH-1 (reviewed by Adams and Pollard, 1989), while the second actin binding site has been mapped to the C-terminal half of the tail, i.e. within the sequences corresponding to TH-2 plus TH-3 (reviewed by Korn et al. 1988) . While mapping studies may further refine the positions of these two interaction sites (for example, whether both TH-2 and TH-3, or just one of these regions, contribute to formation of the second actin binding site), the current data provide a valuable correlation between tail domain sequence and function. This correlation should prove to be of benefit in classifying additional MIHCs as they are sequenced (see below), and in exploring further the relationships between structure and function in myosin I.

One striking exception to the results described above is a high molecular weight form of Acanthamoeba myosin I (HMWMI) (Horowitz and Hammer, 1990). The deduced sequence of this protein reveals a polypeptide of approximately 177 ×10³M_r composed of an Sl-like domain fused to an approximately 800 residue tail domain, which shows essentially no homology to the tail domains of either myosins I or II. The only exception to this is the last ∽ 50 residues of the HMWMI, which is 50 % similar to the TH-3 region of the smaller myosins I. Because the tail domain of HMWMI clearly cannot form a coiled-coil structure, we predict that the protein will be single headed and nonfilamentous. For this reason we have tentatively classified it as a high molecular weight form of myosin I (high molecular weight because its heavy chain is —50 % bigger than that of Acanthamoeba myosins IA, IB and IC; this difference is due to an additional 350 or so residues in the HMWMI tail domain relative to the smaller amoeba myosins I). While immunological evidence that the protein exists in cells has been obtained, and the protein has been partially purified, biochemical studies directed at correlating function and sequence of the HMWMI tail domain remain to be done.

Are there additional MIHC genes in Acanthamoeba?

The answer appears to be yes, as Southern blots probed at moderate stringency using the TH-3 region of Acanthamoeba myosin IB are consistent with there being up to six MIHC genes. Furthermore, the partially sequenced MIHC gene (J. Gu, G. Jung and J. A. Hammer III, unpublished data) does not contain the peptide sequence obtained from myosin IA, indicating that it represents a fourth MIHC isoform (fifth, counting HMWMI).

One of our principle goals has been to use MIHC genes as tools to define the functions of the myosins I in vivo. Because our efforts to do these experiments in Acanthamoeba have been unsuccessful, we have recently switched our emphasis to the protozoan Dicytostelium discoideum. Like Acanthamoeba, Dictyostelium is a highly motile amoeboid cell. Unlike Acanthamoeba, Dictyostelium is also haploid, easily transformed and demonstrates high frequencies of homologous recombination (reviewed by DeLozanne, 1989). This latter characteristic allows one to target genes for disruption and, based on analyses of the phenotype of the resultant cells, to infer the in vivo functions of the protein. Furthermore, Dictyostelium has already been shown to contain myosin I as well as myosin II (Coté et al. 1985) and to retain many motile functions, including the abilities to locomote, chemotax and phagocytose, when only myosin I is expressed (reviewed by Spudich, 1989). We set out, therefore, to clone the gene or genes for MIHC in Dictyostelium and to determine their physiological roles in the cell using gene targeting techniques. Pilot experiments like those described above indicated that cloning Dictyostelium MIHC genes using heterologous probes (e.g. Acanthamoeba MIHC gene fragments) would not work, almost certainly because of the very unusual codon bias in Dictyostelium (very strong preference for A and T over G and C). Instead, we screened a λgtll cDNA expression library using an antibody prepared against the whole heavy chain of purified Dictyostelium myosin I (Fukui et al. 1989). The partial cDNA clone obtained was used to identify two genomic clones which together spanned the entire 124 ×10^sM_rDictyostelium MIHC (Jung et al. 1989). This gene turned out to encode a bonafide MIHC, as evidenced by the fact that its entire amino acid sequence was similar to that of the Acanthamoeba MIHCs.

When we probed Southern blots of Dictyostelium DNA at moderate stringency using a fragment that encodes the ATP-binding site region from this MIHC gene, we found evidence for up to three additional MIHC genes (Jung and Hammer, 1990). Using this probe, we were able to clone from a Dictyostelium genomic library three additional MIHC genes which corresponded to the additional bands seen in the Southern blot (Jung and Hammer, 1990). In parallel, Titus et al. (1989) cloned three MIHC genes using as a probe a fragment encoding the ATP-binding site of Dictyostelium myosin II. Comparisons of our results with theirs indicated that there are at least five MIHC genes in Dictyostelium, which we have named myosins IA, IB, IC, ID and IE (with IB corresponding to the gene described by Jung et al. 1989). The complete sequences of IA (Titus et al. 1989) and IB (Jung et al. 1989) have been published.

As mentioned above, the deduced 124×10³M_r (approx.) heavy chain sequence of Dictyostelium myosin IB is very similar throughout to the Acanthamoeba MIHCs and contains all three tail homology regions described above (TH-1, TH-2 and TH-3). Interestingly, the sequence of Dictyostelium myosin IA reveals a somewhat truncated heavy chain (approximately 113×10^sM_r) which terminates just after tail homology region I, i.e. it lacks the sequences that correlate with the second actin binding site (TH-2 plus TH-3) (Titus et al. 1989). These observations suggest that Dictyostelium myosin IA would be limited to driving movements of membranes relative to actin, while myosin IB could power the movement of one actin filament relative to another as well as the movement of membranes relative to actin. Preliminary results from our laboratory indicate that Dictyostelium myosin ID (G. Jung, R. Urrutia and J. A. Hammer III, unpublished data) is very similar to IB, while IE (R. Urrutia, G. Jung and J. A. Hammer III, unpublished data) is very similar to IA, suggesting that these multiple isoforms may indeed fall into at least two groups based on tail domain sequence. While we can only speculate for now, it seems possible that these two groups of isoforms might be largely, if not solely, responsible for supporting different motile functions in the cell. If so, then it would probably be best when creating cells carrying more than one disrupted gene to target members of the same group first. This approach of prioritizing gene targeting experiments will be necessary if efforts to block the expression of all myosin I isoforms simultaneously are unsuccessful.

Are there more than five MIHC genes in Dictyostelium? The answer is probably yes, since low stringency Southern blots using a battery of ATP binding site probes from the various cloned MIHC genes indicate that there may be as many as four more MIHC genes (G. Jung, R. Urrutia and J. A. Hammer III, unpublished data). Interestingly, none of these additional putative MIHC genes (nor myosin IC) cross hybridize at low stringency with high specific activity probes containing the tail homology 3 regions of IB and ID. This finding suggests that IB and ID may be the only isoforms in their group. For this reason, our current efforts have been directed at disrupting the IB and ID genes in Dictyostelium. Cells lacking the IB isoform, which is known to localize to the leading edge of migrating cells, have been created (Jung and Hammer, 1990). These cells show delayed and somewhat inefficient chemotactic aggregation, impaired uptake of bacteria, three-fold slower growth rate on bacteria and abnormal fruit morphology (Jung and Hammer, 1990). Efforts are now underway to create a IB^-/ID∽ double mutant, with the expectation that this mutant will show even greater impairment of chemotaxis and phagocytosis.

One approach for identifying MIHC genes in other organisms would be to use portions of the tail domains of the Acanthamoeba and/or Dictyostelium MIHCs as heterologous probes. If such probes were to work, they would have detected only myosins of the type I class. While this approach might work well when searching for additional members of a family of MIHC genes within one organism, it unfortunately may not work well when crossing species because of (1) differences in codon bias, (2) the large number of conservative substitutions that occur between isoforms within tail homology region 1, and (3) the lack of direct sequence homology within the GPA-rich TH-2 region. Furthermore, the two known examples of MIHC genes in higher eukaryotes, Drosophila nina C (Montell and Rubin, 1988) and brush border myosin I (Hoshimaru and Nakanishi, 1987), share very little tail sequence homology with the protozoan myosins I (the best match is between the TH-1 region of Acanthamoeba myosin IB and the last approximately 20×10’³M_r of bovine brush border myosin I, which are only about 45 % similar; Jung et al. 1989). Nevertheless, this approach might represent a reasonable starting point, especially since pilot experiments in which heterologous probes are checked against Southern and Northern blots for the organism of interest are simple to do. Of all the regions within the protozoan MIHC tail domains to try, probes encoding tail homology region 3 may be the best.

An approach which is in essence the opposite of the approach described above is again to use heterologous probes, but ones encoding the most highly conserved regions within the myosin globular head domain, i.e. regions that are conserved between type I and type II myosins. This approach worked quite well for Acanthamoeba (in which a nematode muscle myosin head probe identified both amoeba myosins I and myosin II), although a similar codon bias is again a prerequisite for the success of this approach. Once pure clones are obtained, Northern blots can be used to quickly sort myosin I clones from myosin II clones, based on the large difference in their transcript sizes. As a variation on this approach, oligonucleotides made against these highly conserved head sequences could be used to clone MIHC genes by polymerase chain reaction (PCR) from genomic DNA or first-strand cDNAs. Owing to the large degree of divergence seen between the tail domain sequences of MIHCs from protozoans and metazoans, this approach of targeting highly conserved head sequences (e.g. ATP binding site, reactive thiol region), whether via heterologous probes or PCR, may be the best approach available.

Antibodies can of course be used in combination with cDNA expression libraries to clone MIHC genes. In many cases, however, sufficient myosin I cannot be purified from the organism of interest in order to generate specific antibodies (lack of the pure protein also precludes determination of protein sequence, from which specific oligonucleotides could be made for use as probes or in PCR reactions). Antibodies (both monoclonal and polyclonal) against the protozoan MIHCs are available, however, and their efficacy can be checked rapidly by screening a Western blot of a whole cell extract from the organism of interest. One should also be aware that the number of recombinant MIHC clones within a cDNA expression library may vary considerably, depending on the state of the cells or their developmental stage at the time the RNA was extracted for library construction.

Once a MIHC clone has been obtained, an ATP-bindingsite probe from the gene may work well for finding other members of the family within that organism. This approach has worked very well for Dictyostelium (Jung and Hammer, 1990; Titus et al. 1989). Again, myosin I clones can be distinguished from myosin II clones early on by back screening first-round positives with a myosin II-specific probe or by probing Northerns with clone inserts.

We thank Edward D. Korn for advice on the manuscript and for his support of this work.