Properties and natural occurrence of maternal-effect selfish genes (‘Medea’ factors) in the Red Flour Beetle, Tribolium castaneum

Abstract

Maternally acting selfish genes, termed ‘Medea’ factors, were found to be widespread in wild populations of Tribolium castaneum collected in Europe, North and South America, Africa and south-east Asia, but were rare or absent in populations from Australia and the Indian subcontinent. We detected at least four distinct genetic loci in at least two different linkage groups that exhibit the Medea pattern of differential mortality of genotypes within maternal families. Although each M factor tested had similar properties of maternal lethality to larvae and zygotic self-rescue, M factors representing distinct loci did not show cross-rescue. Alleles at two of these loci, M¹ and M⁴, were by far the most prevalent, M⁴ being the predominant type. M² and M³ were each found only once, in Pakistan and Japan, respectively. Although M¹ could be genetically segregated from M⁴ and maintained as a purified stock, the M¹ factor invariably co-occurred with M⁴ in field populations, whereas M⁴ usually occurred in the absence of other Medea factors. The dominant maternal lethal action of M¹ could be selectively inactivated (reverted) by gene-knockout gamma irradiation with retention of zygotic rescue activity.

Introduction

In 1992 we reported the discovery, in wild populations of Tribolium, of a class of selfish genes previously unknown in the animal kingdom (Beeman et al., 1992). These genes showed the unusual property of maternal lethality to hatchlings, combined with zygotic self-rescue. They were termed ‘Medea’ (=M) factors, an acronym for M aternal Effect Dominant E mbryonic Arrest. This mechanism ensures that progeny of a carrier mother survive only if they inherit a copy of the gene from either parent. Such a genetic element is predicted to spread in a population, even in the absence of any selective advantage conferred upon the host (Wade & Beeman, 1994). Subsequently, a genetic disease in mice was shown to be associated with a selfish gene directly analogous to Medea (Hurst, 1993; Peters & Barker, 1993), and Medea-like maternal and zygotic effects were observed for a region of mouse chromosome 1 bearing a high-copy long-range repeat cluster (Weichenhan et al., 1996). In the current work we (i) demonstrate the existence of at least 4 distinct Medea loci in wild populations of T. castaneum; (ii) confirm the absence of cross-rescue between nonallelic M factors; (iii) show that maternal lethal activity of a Medea factor can be selectively reverted with retention of rescue activity; and (iv) examine on a global scale the distribution of the two commonly occurring M factors in wild populations.

Materials and methods

Description of strains

Five standard laboratory strains, GA-1, M¹, M⁴ au, M¹ M⁴ and 3P1/au¹⁴, were used to screen field strains for the presence of Medea factors. GA-1 is a standard laboratory strain collected in a farmer’s corn bin in Georgia (Haliscak & Beeman, 1983) and is apparently devoid of Medea alleles. M¹ is homozygous for the M¹ factor (third linkage group) derived from the SP strain from Singapore (Beeman et al., 1992). M⁴ au is homozygous for the visible recessive marker aureate (au), as well as for the incidental markers, Abdominal-missing abdominal sternites (A^mas) and pearl (p), and during the course of this work was found to be also homozygous for M⁴, although it lacks M¹. M¹ M⁴ is homozygous for the visible recessive markers microcephalic (mc) and jet (j) (Sokoloff, 1962), both incidental to the present work, and during the course of this work was found to be also homozygous for both M¹ and M⁴. 3P1 and 3P2 au are 3rd linkage group (LG) balancers (Mocelin & Stuart, 1996) that carry the dominant visible marker, Blunt abdominal and metathoracic projections (Bamp) (Beeman, 1986; Beeman & Stuart, 1990). Both 3P1 and 3P2 au eliminate crossing-over in a region that includes the Bamp, au and M¹ loci on LG3. au¹⁴ is a lethal aureate allele, and was used to maintain both 3P1 and 3P2 au as balanced lethal stocks. These balancer stocks are apparently devoid of M¹ alleles. Most field strains used in this study were collected from farms, grain storage facilities, mills, warehouses and food markets in North and South America, Europe, Africa, the Indian subcontinent, south-east Asia and Australia between 1985 and 1995, and have been maintained in the laboratory on whole wheat flour fortified with 5% brewers’ yeast. Additional information is available at the Tribolium web site (http://bru.usgmrl.ksu.edu/beeman/tribolium.html).

Detection and differentiation of Medea loci

M factors were initially detected in a subset of all field strains by crossing females from each strain with GA-1 males, then testcrossing F₁ females with GA-1 males in single pairs. In the presence of Medea, offspring of such F₁ families will segregate 50% Medea heterozygotes (viable) and 50% non-Medea homozygotes (inviable). Thus, the presence of an M factor in the original field strain was indicated by the death of ≈50% of the hatchlings from the testcross. Dead hatchlings were never found in control crosses or in crosses involving non-Medea segregants. M factors diagnosed in this way were then tested for genetic linkage and for cross-rescue. All M factors were initially tested for linkage to M¹ on the third linkage group (LG) using the closely linked visible marker aureate (Beeman et al., 1992). Additional genetic mapping was carried out using backcrosses of multiple heterozygotes to the multiple homozygous recessive as described (Beeman et al., 1992). Apparent linkage of a new Medea with a recessive marker in trans was confirmed by separate linkage tests in cis. After initial detection and mapping, each apparent Medea factor was confirmed by demonstrating that hatchling kill was strictly maternal, and that it occurred only in non-Medea progeny. This was accomplished by testing Medea-bearing males for progeny kill, and by testing surviving progeny of segregating Medea females for the presence of M factors. Cross-rescue was tested by assessing the zygotic rescue activity of a paternally derived M factor in an embryo under the lethal maternal influence of a different M factor.

Reversion of M¹ lethality

To confirm that the lethal and rescue activities of M factors are controlled by separate (but tightly linked) genetic elements, we tested whether maternal lethal activity of M¹ could be selectively deleted by knockout mutation with retention of intact zygotic rescue activity. Screens for loss of maternal lethal activity were conducted after gamma irradiation of spermatozoa using a Co⁶⁰ source. One-hundred homozygous M¹ au males 1–2 weeks of age were irradiated at a dose of 4 kR, then immediately crossed en masse to 200 3P2 au/au virgin females. After 3 days the males were discarded and the females were allowed to oviposit for 1 month. F₁ virgin females heterozygous for 3P2 au and for the treated M¹ au chromosome were testcrossed to 3P1/au males in single pairs. Revertants of maternal lethality were recognized by the presence of phenotypically Bamp, au beetles (presumably 3P2 au/au) among the progeny. Because 3P2 au/au progeny did not inherit M¹, they would not express rescue activity, and thus would normally be killed by maternal M¹. Putative revertants were tested to rule out false positives derived by recombination between M¹ and the balancer chromosome. Testcross progeny that were phenotypically Bamp, non-au (presumably M^1R au/3P1, where R = revertant) were used to establish balanced lethal stocks, or to determine whether the revertant was homozygous viable.

Geographical distribution of Medea factors

To survey wild populations more extensively for the presence of M factors, separate screenings were conducted for each of three Medea categories, namely M¹, M⁴, and ‘all othe rs’. Initially, two individuals were tested for each Medea type in each of 123 strains (total =738 beetles tested). These strains represented 25 source countries in Europe, the Middle East, North and South America, Africa, south-east Asia, Australia and the Indian subcontinent. For most strains, follow-up tests were carried out using 2–5 additional beetles for each Medea category (total ≈1400 beetles tested). Diagnostic tests were as follows.

(i) To detect M¹, field strain males were crossed in single pairs with standard M¹/3P1 virgin females and the adult progeny were scored for the 3P1 (=Bamp) phenotype. If the field strain male was either heterozygous or homozygous for M¹, the maternally derived 3P1 chromosome would be rescued from maternal lethality by a paternal M¹ chromosome. Thus, the presence or absence of Bamp progeny indicated the presence or absence, respectively, of a paternal M¹ allele. M¹ was confirmed in females from a subset of positive field strains by demonstrating the presence of M¹-linked maternal lethal activity. Test females from field strains were crossed to standard M⁴ au males, and F₁ females (potentially heterozygous for M¹ in trans with au) were backcrossed to M⁴ au males. The presence of an M¹ allele in trans with au in the F₁ female would result in the death of almost all homozygous au progeny, because au is closely linked to M¹ (Beeman et al., 1992). Note that M⁴-derived maternal lethality is completely suppressed in this backcross, as all progeny inherited an M⁴ chromosome from their father, resulting in zygotic rescue.

(ii) M⁴ alleles were detected by testing the ability of field strain males to rescue the maternal lethality associated with standard M⁴/+ virgin females. The latter were generated by crossing GA-1 males with M⁴ au virgin females, then collecting virgin F₁ females. Field strain males that lacked M⁴ alleles failed to rescue the maternal lethality associated with M⁴. Crosses involving such males were expected to result in ≈50% survival of progeny. Crosses involving males heterozygous for M⁴ should show ≈75% progeny survival, whereas those employing homozygous M⁴ males should give rescue of all progeny.

(iii) All other M alleles were detected by testing field strains for the presence of maternal lethal activity that was not rescuable by standard alleles of either M¹ or M⁴. F₁ females derived from either of the testcrosses (i and ii) described above were in turn crossed in single pairs with standard M¹ M⁴ males. Mortality of 50% of the hatchlings indicated the presence in the F₁ female of a field-strain-derived Medea factor other than M¹ or M⁴, because the presence in all zygotes of paternally derived M¹ and M⁴ factors would preclude any mortality associated with those elements. All standard strains either lack M factors (other than M¹ or M⁴) or are fixed for the same set of these ‘other’ M factors, because all hatchling mortality associated with hybrids between standard strains used here is fully accounted for by the effects of either M¹ or M⁴.

Results

Medea alleles occur at four loci

The M¹Medea factor has been previously described and mapped to the far ‘right’ end of the 3rd linkage group, within one map unit of the au locus (Beeman et al., 1992). The M¹ allele examined in that work was derived from a strain collected in Singapore. New Medea factors characterized in the present work were first tested for linkage to au to determine whether they were potentially allelic with M¹. Preliminary tests revealed that several additional strains from south-east Asia carried M factors that were closely linked to au. One of these (M²), collected in Peshawar, Pakistan, was unique in that it did not show cross-rescue with M¹ (Table 1). When females doubly heterozygous for M¹ and M² in trans were testcrossed to males carrying various combinations of M alleles, it became apparent that M¹ and M² were closely linked to each other and to au, but that neither M factor rescued the lethality of the other. A third M factor, designated M³ was also unique, being found in a single strain from Kukisaki, Ibaraki prefecture, Japan. This factor was mapped to the 8th linkage group, ≈11 recombination units from antennapedia, away from squint (Table 2). The fourth Medea gene recognized (M⁴) was found in numerous laboratory and field strains from many sources. We have not yet succeeded in mapping M⁴, but it assorts independently of the other three M loci (data not shown). As expected, M⁴ shows normal Mendelian segregation in heterozygous males and the unique ‘selfish’ mechanism (maternal lethal with zygotic self-rescue activity) that is the hallmark of this class of genetic elements (Table 3 and Thomson & Beeman, 1997). In addition to the evidence presented above for the absence of cross-rescue between M¹ and M², we also have extensive evidence that M¹ and M⁴ do not show cross-rescue (data not shown). However, cross-rescue among M², M³ and M⁴ has not been tested.

Table 1 The closely linked Medea factors M¹ and M² do not cross-rescue

Table 2 Assignment of M³ to linkage group 8, and three-point mapping of M³

Table 3 Segregation of Medea factor M⁴

Selective reversion of M¹ lethality

We screened ≈1000 irradiated M¹ au chromosomes for loss of maternal lethal activity and detected three apparent revertants. Two of these proved to be false positives, possibly resulting from recombination between M¹ and the balancer chromosome. One true revertant (=M^1R) was confirmed. A female M^1R au/3P2 au testcrossed to a male 3P1/au produced phenotypically Bamp, au progeny (testcross 4, Table 4). The latter must have been genotypically 3P2 au/au, and would have been killed if the maternal M^1R au chromosome had intact lethal activity (see control testcross 3, Table 4). Testcrosses 1 and 2 in Table 4 clearly show that a paternally derived M^1R au chromosome, lacking its own maternal lethal activity, could still rescue progeny from the maternal lethal effect of an intact M¹ chromosome. The revertant chromosome was homozygous lethal and was maintained for several generations as a true-breeding stock over the 3P1 balancer. However, it was associated with severely reduced fertility and eventually died out in spite of careful husbandry.

Table 4 Selective reversion of M¹ lethality

World distribution of Medea types

We tested 113 strains representing 23 countries for each of three Medea ‘types’: M¹, M⁴ and M^X (=all others). These 113 include only unique strains whose geographical origin is well-documented. The results, summarized in Table 5, reveal that M¹ and M⁴ are the two predominant types worldwide, M⁴ being by far the most prevalent. Of the 113 strains, 42 (=37%) representing 14 countries contained M⁴, whereas 11 (=10%) representing nine countries had M¹ and 13 (=12%) representing six countries appeared to carry other Medea factors, based on the presence of maternal lethal activity that was not rescuable by either M¹ or M⁴. Most strains tested from Africa, South and Central America and south-east Asia contained these new Medea-like factors, whereas they were absent from North America, Europe, Australia and the Indian subcontine nt. However, they appeared to have variable penetrance, and their existence could not be consistently confirmed in subsequent tests. Thus, they are not included in Table 5. Major geographical differences in Medea distribution are evident from the data in Table 5. North America and Europe have a high incidence of M⁴ and a low incidence of M¹; Australia and the Indian subcontinent have a low incidence of Medea factors of either type; whereas South America, Africa and south-east Asia have a high incidence of both types. If the 55 strains from Australia and the Indian subcontinent are excluded, M⁴ occurs in 38 of 58, or 66% of all remaining strains. M¹ occurred in 11 of 28 (=39%) of all strains tested from South America, Africa and south-east Asia, but in none of 52 strains tested from North America, Europe (including the Middle East), Australia and the Indian subcontinent. In North America we observed a distinct regional nonuniformity in M⁴ distribution (data not shown). M⁴ factors were found in all 11 strains originating from the midwest or northern plains, but in none of the five strains originating from the deep south.

Table 5 World distribution of Medea types

M¹ always co-occurs with M⁴

It is notable that M⁴ was present in all 11 M¹ strains, whereas only 30% of the non-M¹ strains carried M⁴. Although M¹ was never detected in the absence of M⁴ in nature, the two could be readily separated by genetic segregation in the laboratory. Purified M¹ strains appear to be fully viable and the isolated M¹ factor is stable and functional.

Discussion

Although we first reported the existence of Medea factors as long ago as 1992, the detailed workings of this unprecedented mechanism for self-propagation of parasitic DNA remain obscure. The present work poses or leaves unanswered a number of intriguing questions. What could explain the patchy distribution of Medea elements in nature? Why is M¹ never found in the absence of M⁴? Are M factors of recent evolutionary origin? Are they foreign elements or endogenous genes? Are they dispensable? Are they unique to the genus Tribolium?

We hoped that the question of whether M genes are dispensable would be answered by reversion analysis. If independently derived revertants of a particular M gene are lethally noncomplementary, then that gene is probably vital. As we obtained only one revertant, this question must be revisited in the future with additional reversion analysis.

The fact that the M¹ revertant chromosome had lost maternal lethal activity while retaining zygotic rescue activity confirms the bifunctional nature of this locus. A Medea locus could be two separate but closely linked genes (encoding a maternal poison and a zygotic antidote, respectively) that selection favours to be maintained in linkage disequilibrium. Such a gene pair might best thrive in a recombination-suppressed region, for example, near centromeric heterochromatin. Although cytological positioning of Medea loci has not been accomplished, M¹ and the closely linked M³ have been recombinationally mapped to one extreme end of metacentric LG3, and thus are probably much nearer a telomere than a centromere.

Until recently all evidence suggested that each M locus is fully independent of all other M loci, i.e. each M factor expresses maternal lethal and zygotic rescue activity independently of other M factors, and rescue activity of a given M factor does not protect against the maternal lethal activity of a different M factor. In 1996, however, we discovered that the hybrid incompatibility factor H (Thomson et al., 1995) interacts lethally with both M¹ and M⁴. That is, H/+ heterozygotes die prior to adult development if they also carry a copy of either M¹ or M⁴ (Thomson & Beeman, 1997). More detailed examination of this phenomenon could give new insight into the Medea mechanism.

In geographical regions where Medea is at or near fixation, such as M⁴ in the midwestern United States, its selfishness is silenced and the element remains hidden. Similarly, it has been reported that sex-ratio distorter is widespread but phenotypically silenced in natural populations of Drosophila simulans (Atlan et al., 1997). These observations lend support to the idea that genetic conflicts in general might be more common than often considered (Hurst & Hurst, 1996). It would be remarkable if maternal-effect selfish genes, represented by at least four different loci in the only species to be carefully scrutinized, were not much more widespread than is now apparent.

A full understanding of the Medea system will not be achieved without cloning and sequencing at least one M locus. Because the Medea mechanism appears to be unprecedented, as no homologues are known in other species, and as we have no DNA sequence information, it would seem that positional cloning might be the easiest approach. The feasibility of such a strategy is now under investigation.