How the behavior of an Arabidopsis gene could overturn the classical laws of genetics.
By Andrea Gawrylewski
1 She had found that a mutant Arabidopsis plant could "fix itself" back to the wild-type and take on the genetics of its grandparents. That seemed to contradict the laws of Mendelian inheritance.
Since the late 1990s, Lolle, then at Harvard University, had been collaborating with Purdue University's Robert Pruitt, to study how the plant cuticle, or epidermis, does its job of wax production, water regulation, and overall plant protection. Lolle and Pruitt bred Arabidopsis plants to have a mutation in each gene associated with regulating organ development and fusion. This allowed them to observe the phenotypic traits associated with each gene, specifically on the cuticle.
They reported in 1998 on the fused reproductive organs characterizing the morphology of an Arabidopsis plant with a mutation at a gene they called HOTHEAD (HTH).2 The HTH gene is responsible for cuticle wax production and plays an important role in the plants' interactions with the environment by helping protect from water loss, radiation, and pathogens. It is a member of the gene family that regulates postdevelopmental organ fusion.
In 1998, they also reported the fiddlehead mutant, whose reproductive parts curled into a tight coil like an unrolled fiddlehead fern leaf. They published another paper confirming the genetic region associated with reproductive organ mutations in 2003.3 Because mutation of the HTH gene causes the reproductive organs to be fused together into tight green balls, the mutant plants have difficulty reproducing naturally and would likely be selected against in the wild, says Lolle.
By 2000, Lolle had left Harvard and taken a research position at Purdue. Over the course of their experiments, Pruitt and Lolle saw that some of the hothead mutant plants did not have the associated fused reproductive organs - without flowers or exposed stamen - as did their mutant counterparts, which is a double recessive trait (hth). Instead, these individuals had normal looking flowers and leaves, the signifying morphology of the wild-type plants.
As Lolle and Pruitt ran through the classic Mendelian experiments to test their observations, they were stunned at the results they were seeing: F1 generation mutants, bred from parents homozygous for the hth allele, were displaying wild-type morphology and wild-type alleles. "We were looking at each other, saying 'is this really right?'" she recalls. The researchers examined the entire DNA sequence of the coding region of the HTH gene in three separate reverted mutants to rule out the possibility that the reversions were due to a high rate of random mutation.
To be sure they were seeing a "true" phenomenon and not a product of outcrossing or pollen contamination, they began to run experiments to rule out those possibilities. Their paper, published in Nature, described a reversion frequency (the frequency that the hothead mutant plants reverted to nonmutant genes) of approximately 10%. Lolle had seen reversion before in the hotheads when she was working with them in 1992, during a stint as an associate professor at Reed College in Oregon. She had observed the reverted alleles but could not pin down what was causing the instability. "It bugged me because I knew it was there, I knew I could reproduce it," Lolle says. "I thought maybe it's not all in the alleles, but I didn't get the full connection until Bob [Pruitt] and I collaborated," and found all the fusion mutants from running the genetic screens on the various mutants, one by one.
In every case the reverted allele matched the wild-type sequence exactly, suggesting that the high rate of reversion they were observing was not from another mutation, such as gene silencing. This also suggested to them that the changes they were observing were somehow regulated by a DNA template. Lolle and Pruitt used PCR and DNA blotting to look for such a template, to no avail.
That left them to consider an RNA-based mechanism. Pruitt and Lolle devised the theory of an RNA cache to explain how the plants might revert to their grandparents' allelic frequency. According to this theory, somewhere in the plant cells exists an RNA copy of ancestral DNA, sequences of which the plants can tap and randomly substitute into their own DNA, accounting for the reversion. Of course, Gregor Mendel's law of segregation states that offspring inherit two alleles for one trait, one allele from each parent. That means there's a linear relationship of inheritance between parent and child, and each offspring's alleles can only come directly from their parents' alleles. A non-Mendelian system of inheritance has "enormous implications," says Lolle, with the potential to unseat nearly two centuries of assumptions in genetic research.
"My first reaction was kind of disbelief," says Animesh Ray, genetics professor at the Keck Graduate Institute. "It was kind of incredible given the result they got runs counter to all notions of standard genetics."
"The paper was a really big bang," says Virginia Walbot, a maize geneticist at Stanford University. "I was so glad the paper got so much discussion." But, if this reversion is happening at such a high frequency, why hadn't other Arabidopsis researchers examined the phenomenon instead of chalking it up to a seed contamination or bad experimental technique, she wonders. "Maybe the Arabidopsis community is overlooking a gold mine, but it would seem odd to me, if something were popping up every time you grow these lines, that a good lab would say 'well, we're really crummy at crossing and that must account for what we're seeing'."
Assuming that indeed the researchers had observed reversion, what could have accounted for it other than an RNA cache? Other plant geneticists got busy coming up with alternative theories that would explain the Lolle team's observations. In a communication to Nature, Ray proposed that DNA heteroduplexes had conserved the wild-type alleles; fragments of chromosomes carried on the two independent parental DNA chains were used in a form of mismatch repair to restore the mutants to wild-type, and could account for some of the reversion.4
Ray's colleague and friend, Abed Chaudhury, a researcher at Australia's Commonwealth Scientific and Research Organization, suggested another mechanism, based on two phenomena. The first was DNA homology: He suggested that small homologous sequences of Arabidopsis genome could be translated into RNA. The second, the plant cuticle (waxy layer), was characteristically weaker in the mutant plants. In a comment accompanying the paper, he suggested that a porous embryo sac might allow heterozygous alleles from a degraded heterozygote spore to get into the mutant embryo and get expressed.5
In November, 2005, researchers at the University of Washington and the University of Georgia proposed that, rather than a novel genetic mechanism causing the reversion, the HTH gene product might be a metabolic enzyme, which when mutated produces toxic and mutagenic compounds that act on the DNA. In particular, they suggested that the HTH protein may be in a class of proteins that, given a deleterious mutation, revert precisely to the wild-type from mutagenic effects.
Lolle and her colleagues, however, rejected these explanations.6 For one, the frequency at which some of the phenomena that Chaudhury and Ray proposed might occur were too small to account for the 10% reversion frequency that she and Pruitt had observed. "In fairness, they don't have all the data available. They wrote their interpretations based on the Nature paper, which is perfectly valid, but we know a lot more now," she says. "We know that instability persists for multiple generations, then explanations for the porous embryo or DNA heteroduplexes don't seem like viable explanations because it has to happen over numerous generations," to demonstrate that it's more than a one-time random event, but, rather, a discrete mechanism of action.
The RNA cache, on the other hand, Lolle proposes, is tapped when plants are under stress and is perhaps more than a one-time, random event. It is maintained outside the normal genomic context, inherited, and copied back into the DNA. There is a strong precedent for this mechanism to reside in RNA action, she says; Nobel laureate Craig Mello and colleagues have shown in Caenorhabditis elegans that RNA interference can be transmitted from generation to generation. RNA interference works by fooling the organism's cells into destroying a gene's messenger RNA before it can express a protein. Researchers speculate this mechanism evolved as a way to evade viruses that often create double-stranded RNA as they spread in the host system.7
"Work on other systems has supported RNA being able to be a template during, for example, DNA double-stranded break repair," says Rebecca Lamb, a plant geneticist at Ohio State University, citing the 2007 paper in Nature by Michael Resnick's group at the National Institute of Environmental Health Sciences.8 "Certainly it's not beyond normal possibility that there is, in fact, a cache of RNA, but definitely more experiments need to be done to figure out what contribution [it] makes."
There was one explanation, however, that couldn't be easily disproven. Suppose the reversion results could simply be explained by pollen contamination, or outcrossing? Although Arabidopsis is known to self-pollinate in virtually every case - the near clonality is what makes it such a useful model organism - the hothead mutant's fused reproductive organs make the plant more susceptible to being fertilized by another plant's pollen carried on the wind or by an insect, otherwise known as outcrossing. "Outcrossing is the bane of my existence. It haunts me," Lolle says. "It is possible, we know it happens. Now it's just [a question of] how much of the instability is outcrossing. It's a fraction. There are genome changes going on here that cannot be explained by outcrossing" because of the high frequency of reversion.
Raphael Mercier and his colleague Fabien Nogué, both senior scientists at the French National Institute for Agricultural Research in Paris, started out thinking that chimerism was responsible for the reversions, but ended up concluding that outcrossing accounted for the phenomenon entirely.
In September 2006, Steve Jacobsen's group at the University of California, Los Angeles, published what happened when they tried to reproduce the phenomenon that Lolle and Pruitt had observed.9 They found that the hothead mutants had a strong tendency to outcross, rather than self-fertilize, and that plant populations grown in isolation were stable. They did not see any reversion. Jacobsen's findings deflated the sails of non-Mendelian inheritance, not to mention the RNA cache. Lolle's phone stopped ringing.
Steve Mount, a molecular geneticist and member of the Arabidopsis Research Initiative at the University of Maryland (ATRIUM), says that outcrossing could explain most of the results in the Lolle group's 2005 paper. But two results, particularly the Lolle group's reporting of a double wild-type embryo in mutant progeny and a female reverted mutant, can't be outcrossing. Those phenomena need to be explored in further experiments. Still, he says, "I think there is a chance this is an artifact of the experiment. I don't think she's faking anything, but exceptional claims require exceptional evidence."
Elliott Meyerowitz, a plant geneticist at the California Institute of Technology, told The New York Times when the original paper was published that "it looks like a marvelous discovery." He declined to comment further on these findings, instead writing in an E-mail that he was awaiting more data on the phenomenon.
Plant geneticists may not be convinced of Lolle's observations until they see them expanded upon and reproduced repeatedly. Walbot, at Stanford University, says that the 2005 paper lacked the robust molecular analysis of the mutants and their progeny, which might show indisputably where the instability was arising. For example, further research needs to establish whether the alleles are transmitted through the maternal or paternal side; parental imprinting can affect gene expression, and is carried through generations. Also, says Walbot, researchers must establish that the two alleles are not influencing each other's expression, and thereby indeed adhering to Mendel's second law of independent assortment, which states that alleles operate independently and explains why he saw only either purple- or white-flowered progeny when crossing purple and white flowers.
As for the genetics community's reaction: "The stronger the people's genetics skills, the stronger skepticism about this paper," says Walbot. "There is just missing data on transmission and molecular markers. So the maize community was suspicious, wondering is this something about fertilization biology or some [mechanism of] reproductive development" giving rise to the observations.
Lolle was already on the path to determining just what effect outcrossing was having on the mutant populations when Jacobsen's group published their brief communication. She had done a short rotation as an NSF reviewer and then taken a post as an associate professor at the University of Waterloo. While she was discouraged by Jacobsen's assertion that outcrossing accounted for any reversion she and Pruitt may have observed, Lolle was bolstered by similar observations in other plant species, and aspects of Jacobsen's paper that she considered major holes in his findings. She was relieved when Jacobsen's results appeared: "I said, 'Now I have the time to nail this.'"
Lolle says that Jacobsen's failure to mention the size of the founder population on which they ran their experiments in his group's communication may be significant. Lolle and her colleagues had found that a large founder population was important - on the order of several hundred - because the rate of instability for one particular individual was quite variable (the 10% reversion they reported is a population rate). Jacobsen did not report that they had performed molecular genotyping on any individuals; for mutants grown in complete isolation, Jacobsen's group just looked at the plant phenotypes, or their flowers and reproductive organs, to garner whether they had inherited the mutation or not, rather than dissect embryos, or genetically type each plant, incontrovertibly showing that outcrossing caused the phenotype.
Lolle has asked him about this omission, but says she's never received a straight answer. Jacobsen declined any further comment on his work on hothead, saying only that his group stopped studying it since they determined that outcrossing was at the root of the phenomenon.
In early 2007, Nature received another comment about the original 2005 paper and asked Lolle if she wanted to issue a response, or issue a correction in light of the new comment. Lolle declined and was surprised that Nature editors felt the new comment called for a correction of the original paper; nothing had been fabricated, she says. "I think I'm a good scientist, an observant scientist, and you can't be a good scientist unless you're skeptical of yourself," she says.
Lolle maintains that outcrossing does contribute to reversion frequencies but it doesn't account for a sizeable portion of the instability they're seeing. But for many members of the scientific community, Jacobsen's results have been the final word for nearly a year and a half.
So instead of going after the mechanism of reversion, either by finding the RNA cache or something else, Lolle has spent the past year trying to pin down exactly what proportion of her results is explained by outcrossing and what proportion isn't. She plans to publish her results sometime this year on a hothead experiment involving 200,000 plants that she hopes will put an end to the outcrossing debate.
Lolle cites 1950s work on flax as further evidence that there is some mechanism in plants that works outside the traditional Mendelian system. In 1958, Alan Durrant, a plant researcher at the University of Wales, was trying to determine how to maximize seed output and seed weight from individual flax plants. He planted flax varieties in varying levels of fertilizer, bred the plants' progeny and measured the progeny seeds' weight. Durrant observed that certain flax plots that had been treated with heavy fertilizer produced plants of greater weight, but more surprisingly, their progeny were also of greater weights.10 "With different environmental conditions, you see genetic differences arising, and these can be transmitted to generations," says Christopher Cullis, plant molecular biologist at Case Western Reserve University. "That's not traditional Mendelian inheritance; the environment is not supposed to influence" the genome so directly.
Cullis, who began researching flax in 1971, has been trying to pin down at a molecular level what accounts for such rapid genetic alteration in the plants. In 2005, he published a report demonstrating that a single copy 5.7 kilobase DNA fragment is inserted into stable lines of flax, as a result of environmental conditions, and is heritable.11 Cullis has shown that this DNA fragment is not present in any previous genotypes until it appears in the experimental plants. His group is trying to pin down where the fragment may have come from; it is unlike any known transposable element or retrovirus, he says, but may be a programmed rearrangement of the DNA. A similar mechanism of DNA-programmed rearrangement may be operating in hothead, he says. "I've worked with this for 36 years, and the reaction has almost always been 'this can't happen.'" The common response to his findings is that something is wrong with the original plant lines he's been given. "There's a strong reluctance to admit that genetics might not always work in the way we think it does," he says.
Evidence for these genetic alterations in flax have existed for nearly three decades but has received relatively no attention, wrote Steven Henikoff, epigeneticist and Howard Hughes Medical Investigator at the Fred Hutchinson Cancer Center, in Plant Cell in November 2005.12 "The evidence for some kind of massive programmed rearrangement upon environmental induction in flax is unequivocal," he writes, "but inheritance of acquired changes has been an anathema to evolutionary biologists ever since Darwin's time."
Lolle says flax isn't the only example. Reid Palmer, a soybean researcher for the USDA's Agricultural Research Service based at Iowa State University, has been working on a phenomenon called allele switching. Gordon Lark, a former, now retired, collaborator with Palmer at the University of Utah originally observed in soybean tissue culture restricted fragment length polymorphism (RFLP) allelic differences at various loci in the soybean DNA - allelic combinations not present in any parental tissues. Lark surmised that the soybean has developed a mechanism, in response to stress and population homogeny, for generating genetic diversity by replacing alleles at various loci through, what he would later theorize, is the result of specific recombination events.
Palmer and Lolle have recently started a collaboration, and a postdoc in Lolle's lab, Marianne Hopkins, went to Puerto Rico at the end of last year to collect the first generation of soybean progeny. She's now doing a molecular analysis to see if there's a reversion similar to that seen with hothead. Palmer has already done some crosses and observed allele frequencies that, by classical Mendelian rules of inheritance, should not be possible. So far he has seen allele switching in 4-5% of progeny in aconitase 4 gene. Palmer also believes that Lolle's observations with the hothead mutant may be operating by some similar mechanism to what is occurring in soybean. Suggestions that outcrossing or contamination account entirely for the phenomenon are just not enough based on the frequencies of switching that he has observed. "Like [Lolle], those type of possibilities of contamination et cetera are not the explanations for what we have," Palmer says.
Meanwhile, Hopkins just finished her last round of outcrossing experiments on hothead. Over the past year she and Lolle have bred, grown, counted, dissected, processed, and frozen some 200,000 plant individuals, with tissue samples filling six large freezers in their modest University of Waterloo lab. They replicated experiments, in some cases five times, to ensure they were seeing "true" results. In particular, isolation experiments using hothead mutant plants were performed two floors and an entire building away from any wild-type Arabidopsis, just to be sure that there was no chance that seeds of plants of different genetics and any other plant, vector, or breeze that might carry pollen, could reach the mutants.
So far, they've found nothing to discount Lolle and Pruitt's original findings. Indeed, even when growing the mutants in isolation, they've observed reversion at a rate as high as 40%. Genetic analysis shows that reverted mutants display the wild-type genes that weren't present in their parents' genes. In other preliminary results that Lolle and Hopkins hope to publish later this year (and which they think will further convince people that this phenomenon is real), hothead isn't the only gene that experiences restoration to the wild-type.
Do Lolle's findings, if they turn out to be true, overturn Mendel's laws of inheritance? Some researchers say probably not, but they acknowledge that there are likely many undiscovered phenomena in transmission genetics that don't operate on a large enough scale to warrant throwing out the old rulebook. "Epigenetics has already shaken Mendel's laws," says Lamb, from Ohio State. These mechanisms are complementary to the laws, she adds. "Clearly these processes, if they exist and they certainly do, have been going on all the time. It's just harder to see them."
"We're trying to discover patterns in what seems to be chaos," says Charles Fenster, evolutionary biologist at the University of Maryland. "This interesting example is a departure from the known laws, but my guess is that it doesn't explain the vast majority of how genes are transmitted from generation to generation - the reason why individuals resemble each other. That's really the key issue, why individuals that are relatives resemble each other. That's why people are interested in genetics in the first place."