A curious genetic phenomenon allows certain genetic instructions to be passed between generations without the gene variants involved being transmitted. Some spotty mice provide clues to how this might happen.
One of the benefits of sex is that our parents give us two copies of nearly all our genes. So if one copy, or allele, doesn't work, there is a good chance that the other, operating independently, will be sufficient. Exceptions exist, but even in those cases the two alleles usually function independently. In rare cases, however, the alleles interact intimately, even though they reside on separate chromosomes, and whether one allele is expressed depends on directions given by the other. One example of this phenomenon is paramutation, a process in which orders issued by an allele in one generation are remembered in subsequent generations, even if the allele issuing the order is not transmitted. How do chromosomes remember who their partners were in previous generations, and how do those partners continue to exert their control long after they have segregated away? On page 469 of this issue, Minoo Rassoulzadegan et al.1 provide evidence for a mechanism involving small RNA molecules that might underlie paramutation.
In 1956, R. Alexander Brink2 coined the term paramutation to describe the unexpected means of regulating purple pigmentation in maize kernels (see ref. 3 for a review). Pigmentation is controlled, in part, by the R gene locus. If all the alleles at this locus are recessive alleles (rr), the kernels are colourless, but when a single dominant allele (RR) is present, the kernels are typically purple. This is not always the case, however; it depends on what the RR allele was exposed to in the previous generation. In Brink's experiments, if the RR allele came from pollen of a plant that also had the Rst allele, which caused the kernels to be spotted, then the normally dominant influence of the RR allele was silenced and kernels showed reduced pigmentation. The untransmitted Rst allele exerted its effects in the next generation, overriding the dominance of the RR allele in violation of Mendel's rules of genetic inheritance. Brink called RR paramutable and Rst paramutagenic, and he showed that the influence of the paramutagenic allele could persist for many generations. Paramutation-like phenomena have been reported in mammals1,4,5,6 and might be of clinical relevance in humans, for example affecting certain forms of diabetes4.
The system examined by Rassoulzadegan et al. in mice involves an allele of the Kit locus called Kittm1Alf. This is a null allele; that is, it makes no functional Kit protein. Hetero-zygous mice — those with one Kittm1Alf allele and one normal ‘wild-type’ Kit allele — had white spotting at their tail tips and showed reduced expression of Kit (as measured by production of Kit messenger RNA from the gene). When these mice were crossed with wild-type mates, the wild-type progeny showed the same white spotting and reduced Kit mRNA levels as their heterozygous parent, even though they were fully wild type and lacked the null allele that caused spotting in their heterozygous parent. Moreover, these wild-type progeny also had an accumulation of a mixture of smaller RNAs with sequences that matched various parts of the Kit mRNA. Notably, sperm precursor cells from male mutants had aberrant Kit mRNAs, and their sperm showed an unexpected accumulation of RNA. It is possible that these RNAs are transmitted to the next generation on fertilization, even if the allele from which they arose is not passed on.
The nature of aberrant RNAs is unknown, but they have attracted attention for a variety of reasons. First, paramutation is one of several heritable ‘epigenetic’ phenomena, in which inherited trait differences arise by covalent modifications to DNA and DNA-bound histone proteins rather than by changes to the DNA sequence of the genes themselves. These modifications can be regulated by small interfering RNAs (siRNAs) generated by the RNA interference (RNAi) pathway, and can arise from transcribed DNA repeats7,8,9, which are structures found at some loci undergoing paramutation4,6,10. Second, siRNAs and the related microRNAs (miRNAs) can cause the degradation of mRNAs whose sequences they share, blocking production of the encoded proteins11. Finally, RNAs have been proposed to play a role in another non-mendelian system of inheritance12. In combination with Rassoulzadegan and colleagues' data, these observations raise the possibility that the aberrant RNAs arising from the Kittm1Alf allele include siRNAs, miRNAs or other regulatory RNAs that are packaged in sperm and cause paramutation on transmission to the next generation.
To test this possibility, Rassoulzadegan et al. injected total RNA from tissues containing the aberrant Kit RNAs into fertilized mouse eggs. This led to spotting in many of the progeny that came to term — and in their progeny. Injected miRNAs designed to degrade Kit mRNA had the same effect. Control miRNAs with no similarity to Kit unexpectedly also caused spotting, but at a lower frequency, and the spotted animals only rarely transmitted the spotted phenotype to their progeny. It is not certain that spotting caused by the Kittm1Alf allele arises by the same mechanism as spotting induced by injected miRNAs. However, what might be occurring in this system, and possibly in other paramutation models, is that small RNAs produced from a paramutagenic allele are acting on the corresponding paramutable allele or on its transcribed mRNA, effectively silencing it. Because RNAi-mediated degradation of mRNAs produces more siRNAs, the silencing might be propagated if these small RNAs are packaged into germ cells and carried into the next generation. This could allow successive generations to display a certain characteristic, even if the paramutagenic allele that caused it was not transmitted (Fig. 1). Rassoulzadegan and colleagues' proposal that RNAs are involved in paramutation is strongly supported by work from Vicki Chandler's group showing that paramutation requires an RNA metabolizing enzyme that is involved in other epigenetic phenomena13.
Rassoulzadegan and colleagues' model has yet to be validated, and several points need to be clarified. Notably, the mechanism by which aberrant RNAs from Kittm1Alf heterozygotes induce spotting is not known. Does it involve bona fide siRNAs or miRNAs emanating from the mutant allele? Also, are aberrant RNAs causing mRNA degradation or epigenetic modifications at the wild-type allele? Are these effects mediated by RNAi or by other pathways influenced by small RNAs? If siRNAs or miRNAs do emanate from the paramutagenic Kittm1Alf allele, how do they arise, and do similar RNAs arise in other paramutation models? A particularly intriguing possibility is that such RNAs regulate other non-genetic modes of inheritance, such as metabolic or behavioural imprinting. These have far greater consequences for human development than for spotty mice and maize, but we may learn about such mysterious processes by studying those mouse tales.