1. Introduction
The ciliate
Paramecium tetraurelia is one of a few traditional objects of genetic studies of unicellular eukaryotes [
1]. Sex occurs through one of two different processes: conjugation, where two partners exchange gametic pronuclei to produce a pair of genetically-identical heterozygous progeny, and autogamy, a self-fertilization process resulting in entirely homozygous progeny. These features are highly advantageous for genetic analysis. Like all ciliates,
P. tetraurelia contains two types of nuclei in each cell. The germline, diploid micronucleus (MIC) acts as a silent transmitter of chromosomes across sexual generations, while the somatic, highly polyploid macronucleus (MAC) is responsible for all gene expression. The extensive programmed genome rearrangements that occur during the development of a new MAC from the zygotic nucleus in each sexual generation have led ciliates to become recognized models for epigenetics [
2,
3].
In this work, we studied the Spinning Top mutant of
P. tetraurelia strain 51, which appeared spontaneously in a back-cross of the unrelated mutant
mtGa-1 [
4] to the wild type. This mutant has a remarkable phenotype (
Figure 1), which is characterized by short pear-shaped cells quickly spinning around their pointed anterior tip when swimming (
Video S1 and Video S2). Sometimes such cells lose the ability to complete fissions (
Figure 1C).
In a summary of all available data on
P. tetraurelia mutants, Sonneborn [
5] mentioned mutations at two or three loci that result in very similar “screwy” phenotypes. The
sc1 and
sc66 loci were shown to be unlinked; the
sc66 locus was found to segregate independently from
sc1, but no complementation or linkage test was performed with
sc64, so it may or may not represent a third locus. At least 10 alleles of the best-documented
sc1 locus result in “cells of variable body shape: round or pear-shaped and extremely twisted like a cork-screw, when grown with excess food; when starved, shape approaches normal, but is somewhat bent, like cashew nut. Corkscrew-shaped cells rotate rapidly on their long axis during locomotion” [
5]. Mutant alleles were classified in 4 groups.
sc1-a,
sc1-b and
sc1-d alleles yield normal carrot-shaped trichocysts, while alleles Sonneborn grouped as
sc1-c have cigar-shaped (but functional) trichocysts, thinner and often longer than normal trichocysts, and do not taper to the same degree [
6]. Most screwy mutants are characterized by slow growth and the tendency to produce monsters due to incomplete fissions [
5,
7,
8]. Thus, many screwy mutations are pleiotropic.
Several
sc1 alleles obtained independently in different studies (
Table 1) were available from the stock collection of the Centre de Génétique Moléculaire in Gif-sur-Yvette, France. In this work we figured out that the Spinning Top mutation is allelic to
sc1 mutations. All alleles examined showed rather large MAC deletions encompassing several genes, which appear to reflect similar deletions in the germline MIC genome. We further identified the one gene at this locus that is responsible for the screwy phenotype, and analyzed the patterns of inheritance of the discovered deletions.
4. Discussion
In this work we studied the Spinning Top mutation in
P. tetraurelia and identified the gene leading to the screwy phenotype, which we named
Spade. The Spinning Top mutation was found to be a rather large (≥28.5 kb) MIC deletion that removes 18 genes, including
Spade. Several screwy mutants have been described earlier [
5], but the affected loci had not been identified molecularly. We analyzed 5 different mutant alleles of the
Sc1 locus which had been independently obtained in previous studies, and found that all of them show deletions of the
Spade and neighboring genes in their MAC and MIC genomes, though the deletion breakpoints differ among them. The high frequency of deletions observed at this locus suggests that it is intrinsically unstable.
The real function of Spade protein remains elusive. It appears to be conserved in sibling species of the
P. aurelia complex (amino acid sequence identity with the
P. biaurelia ortholog is 88%, and with that of
P. sexaurelia ~79%). Judging by the presence of zinc-binding domain, the best conserved part of its sequence, and a coiled-coil domain detected in the encoded protein, it can be predicted to be a transcription factor [
20]. However, coiled-coil domains are present in many proteins [
21], including some involved in subcellular infrastructure maintenance [
22]. One could speculate that the Spade protein may be a moonlighting protein, i.e., a protein which has a primary catalytic function but may also acquire a secondary non-enzymatic role, for example aldolase in
Plasmodium [
23] or transcription factors in mammals [
24]. Although
Spade expression increases during autogamy [
9], when many genes become differentially expressed [
25], it is evident that it plays an important role during vegetative growth. However, since
Spade gene knockdown leads to quick changes in aberrant cell shape, and the recovery of a normal cell shape requires several divisions, it is conceivable that the protein is involved, at least indirectly, in submembrane cytoskeleton formation. Moreover, like all screwy mutants, Spinning Top is prone to form multinuclear monsters, i.e., cells which cannot accomplish cytokinesis, which also points to some acute problems with cytoskeleton or cortex. The phenomenon of cortical inheritance is well known in
Paramecium [
26,
27]. It is thought to reflect the fact that aberrant association of monomers in multimeric proteins, leading to malformation of cytoskeletal filaments (or simply the absence of normal cytoskeletal structures), can be maintained during cell divisions despite continuous production of the proteins involved, much like prions. In the case of
Spade knockdown and recovery, cells regain the normal shape only after several complete cell cycles. Determination of the intracellular localization of the
Spade protein would help understand whether it may have a role in cortical or subcortical structure maintenance. As for the other screwy loci identified by previous genetic analyses (
sc66 and possibly
sc64) [
5], they might encode other proteins involved in cortical structures; if the
Spade protein simply acts as a transcription factor, they may be genes that are transcribed in a
Spade-dependent manner.
The pleiotropic effect of screwy mutations at the
sc1 locus can now be explained by the fact that these MIC deletions remove several genes. The
sc1-a mutant has normal trichocysts, while the
sc1-c mutants are characterized by an unusual, cigar-like shape of trichocysts [
5,
6] (
Figure S4). This phenotype may thus be due to one or more of the 5 genes that are deleted in the
sc1-c mutants, but retained in the
sc1-a mutant. Consistently, the Spinning Top mutant, which also lacks these genes, has cigar-shape trichocysts.
It has long been known that alternative rearrangement patterns in the MAC, such as the deletion of a cellular gene in one cell line and its retention in another, are maternally transmitted across sexual generations in wild-type cells through both conjugation and autogamy. These maternal effects have led to the discovery of the scnRNA pathway, which regulates genome rearrangements during MAC development [
2,
3]. In this study we have shown that in crosses between screwy mutants and the wild-type cells, the absence of the
Spade gene region in the MAC of the mutant parent does not result in the same deletion being reproduced on the wild-type allele that is introduced by conjugation in the mutant’s cytoplasmic lineage. This may seem surprising, but is in fact fully consistent with the current model for scnRNA action. Indeed, scnRNAs are initially produced from the entire MIC genome during meiosis in each conjugating partner, and are then thought to scan the parental MAC genome in each cell by pairing interactions with nascent transcripts; only those that cannot find a match, i.e., those produced from MIC-limited sequences, will then be licensed to target the elimination of homologous sequences in the developing zygotic MAC [
28]. In the present case, the mutant parent cannot produce scnRNAs from the
sc1 locus during meiosis of the MICs, since these sequences are absent from its MIC genome. Thus, after introduction of the wild-type
sc1 allele from the wild-type partner, there are simply no homologous scnRNAs in the mutant cell to target deletions in the wild-type allele, which is therefore fully retained in the heterozygous F1 MAC. A very similar case has already been described, though this was long before the discovery of the scnRNA pathway. The d12 mutant carries a MIC deletion of the gene encoding surface antigen A, but when it is crossed to the wild type, the lack of that gene in the d12 parental MAC does not induce the developmental deletion of the wild-type allele in the MAC of its heterozygous cytoplasmic progeny [
29,
30]. This effect has so far remained mysterious, but can now be explained exactly in the same way.
The induction of MAC deletions at the
sc1 locus by dsRNA feeding during autogamy always resulted in a broad spectrum of deletions of different sizes in each postautogamous clone, and most clones contained at least a fraction of wild-type MAC copies. The phenotypes of cells in each clone varied accordingly, from wild-type to pronounced screwy phenotype, and the ratios of phenotypes were probably revealing the ratios of wild-type and deletion-bearing MAC copies and their random fluctuation during amitotic divisions of the MAC (
Figure 8). When the wild-type copy number increases over some threshold, the
Spade gene dose becomes sufficient to produce enough protein for wild phenotype recovery. During vegetative growth, such clones always gave rise to phenotypically wild-type subclones, even when mutant phenotypes were selected for daily isolation.
Phenotypic reversion also occurred at a high frequency after autogamy of phenotypically mutant clones. Nevertheless, we were able to maintain them without selection for almost a year in stock tubes, and in all cases, we were still able to find both phenotypically mutant and wild-type cells in the cultures. Thus, induction of MAC deletions by dsRNA feeding during autogamy allowed us to obtain epimutants, as has been shown for other genes [
19].
The appearance of screwy mutants has often been described in mutagenesis studies. Whittle and Chen-Shan [
8] reported that cortical mutants had appeared under nitrosoguanidine treatment with a frequency of 1:1000, and two of eleven mutants they obtained were screwy mutants. Of all mutagenic factors which have led to screwy mutants in
P. tetraurelia (see
Table 1), only X rays can directly induce single and double strand breaks in DNA. Ultraviolet radiation induces formation of thymine dimers which later are repaired by nucleotide excision repair enzymes or by photoreactivation system; nitrosoguanidine is a well-known chemical mutagen which adds alkyl groups to guanine and thymine, which later can be repaired by the DNA mismatch repair system [
31]. Furthermore, the Spinning Top mutant seems to have arisen spontaneously. Of course, we cannot exclude the possibility that this was somehow triggered by the heterozygous
mtGa-1 mutation in its parental clone [
4], especially since similar phenotypes appeared on two other occasions during the genetic analyses of the
mtGa-1 mutant (S. Bhullar, unpublished data). Spinning Top deletion determines conspicuous phenotype, and it might be that other stealth deletions also occur under the influence of the
mtGa-1 mutation. However, whether the
mtGa-1 mutation may have a mutator effect remains unknown.
Such frequent occurrences of screwy-like mutations in different conditions, and the absence of the
Spade gene in all available
sc1 mutants, suggest that the
sc1 locus is intrinsically unstable in the strain 51 MIC genome, and is a hotspot for some enigmatic chromosome breakage event. We have found at least 3 different deletion breakpoints among the mutant
sc1 alleles examined, but did not find any consensus or sequence motif which could be readily suspected to serve as a breakage site at these positions. The so-called common chromosome breakage sites in mammalian cells are often found in regions of the genome that are particularly sensitive to replication stress [
32], increasing the probability of DNA damage in each cell cycle, and genome instability. Common chromosome breakage sites are known to be associated to loci containing many AT dinucleotides or minisatellite repeats which are also very AT-rich [
33]. The
Paramecium MIC genome is extremely AT-rich [
34], as are the sequences around the mapped breakpoints. It may be also important that these sites are surrounded by IESs (
Figure 4). The possible roles of IES density and AT richness in determining chromosome breakage hotspots in
Paramecium remain to be elucidated; this would require identifying other examples of similar chromosome fragile sites.