By Julia Karow
In one of the first examples of a study applying exome sequencing to common disease, researchers at the Radboud University Nijmegen Medical Centre in the Netherlands have found that a large fraction of currently unexplained mental retardation appears to be caused by rare de novo point mutations in protein-coding genes.
The findings, published in Nature Genetics this week, could have implications for other common diseases, in particular neurological disorders such as autism or schizophrenia. Many researchers believe, however, that it's unlikely that rare mutations play an equally large role in common diseases with late onset, such as type 2 diabetes, heart disease, or Alzheimer's, where natural selection has had a different effect.
"Our paper is the first evidence that these de novo mutations are a major cause of common disease, and now we have to see how this also works out for other diseases," said Joris Veltman, an associate professor of genomic disorders at the Nijmegen Centre for Molecular Life Sciences and a senior author of the study.
"It's a nice, well-done study that clearly supports the role of rare variants in central nervous system disorders," said Hakon Hakonarson, director of the Center for Applied Genomics at the Children's Hospital of Philadelphia, who has been studying the genetic causes of a variety of common diseases in children. "I think you are going to see a lot of these types of papers coming out shortly."
Mental retardation affects about 2 percent of the population and has some overlap with other neurodevelopmental disorders, such as autism. The Nijmegen researchers have been studying it for years, and have found in previous work that de novo copy number variations explain the disease in about 15 percent of patients. "For us, the whole idea of looking for de novo [point] mutations in the genome was a very logical next step," Veltman told In Sequence.
Mental retardation differs from many other common diseases in two regards: patients rarely have children, and up to 1,000 genes are estimated to be involved. Both features mean that rare de novo variants with large effects are expected in this type of disease.
Because of the decrease in reproductive fitness, such disease-causing variants are strongly selected against and do not spread in the population. "Hemophilia is a great example: one-third of all cases are new mutations," said Eric Lander, director of the Broad Institute, in an e-mail message. That is different for common diseases of late onset, like type 2 diabetes, Crohn's, or Alzheimer's, he said, where there are only "mild effects on selection."
Also, the large number of genes involved in mental retardation represent a huge target for de novo mutations to score a hit, unlike a Mendelian disease, where only a single gene is involved. "Lots of genes affect the brain and can likely cause the phenotype of mental retardation," Lander said.
According to recent studies, each person carries on the order of 100 de novo mutations that he or she did not inherit from their parents, and the Nijmegen researchers wanted to test what role they play in causing mental retardation. "And of course, nowadays, with new sequencing techniques, we could for the first time really look at that," Veltman said.
Until now, he said, scientists thought that mental retardation is either caused by common variants — although there has been little evidence for that — or that it may be mostly a recessive disease, which "has definitely shown to be true in some cases." But now, "it looks that these de novo mutations may actually be a much more common cause" of mental retardation, especially in populations with few consanguineous marriages.
[ pagebreak ]
7 Out of 10
For their study, the Dutch researchers selected 10 patients with moderate to severe mental retardation that had no family history of the disease, and for which they had been unable to identify another cause.
They sequenced the patients' exomes, as well as those of their parents, using the SOLiD 3 Plus platform, generating a median exon coverage of 42-fold, where 90 percent of the targets were covered at least 10 times. To enrich the exome, they used the Agilent SureSelect in-solution kit, which at the time targeted about 18,000 genes, or 37 megabases of DNA.
On average, the researchers identified more than 20,000 high-confidence variants per individual and applied various filters to reduce this number.
After excluding most nongenic or intronic variants as well as variants that did not change amino acids, they were left with about 5,500, and after eliminating known, likely benign variants from dbSNP and an in-house database, the number came down to about 140.
Using the parental exome data as a filter to exclude inherited variants, they were left with between two and seven candidate de novo mutations per patient, a total of 51. Sequencing both patients and their parents was "very important" to reduce the amount of validation required, according to Veltman, and his team will continue to use this approach forwarding the future.
The researchers then validated the 51 candidate mutations using Sanger sequencing and were able to affirm 13. This low validation rate was not unexpected because they initially called variants with low stringency for fear of missing some. Mutations they could not validate, they found, had only five valid sequence reads, whereas those they could validate had 17 valid reads on average. In future studies, they will therefore be more stringent in their variant calling, Veltman said.
In the end, they found that nine of the validated mutations had occurred de novo, affecting nine genes, and were absent in 1,664 control chromosomes. Five patients carried one mutation each, while two patients had two mutations.
Two of the nine genes were already known to be involved in mental retardation, and four have a role in neuronal functions. The six mutations found in these genes also occurred at highly conserved positions and are likely to have a large effect on protein function, and thus, to be causing the disease.
The other three genes, however, did not have a link to brain function and appear to be "background mutation," according to Veltman.
In one male patient, the researchers found a mutation in a known X-linked mental retardation gene that was also present in his unaffected mother. Testing his maternal grandparents showed that the mutation had appeared de novo in the mother but only showed an effect in the next generation.
Overall, the researchers were able to explain mental retardation in seven out of the 10 cases they tested, although they are still conducting validation studies to make sure the mutations they identified are actually disease-causing. These include analyzing the exomes of more patients — some hopefully with mutations in the same genes — and making sure the same mutations are not found in healthy individuals. In collaboration with other groups, they also plan to conduct functional studies in model organisms.
The remaining three cases may have eluded them for various reasons. For example, the first-generation exome capture kit they used might have missed some de novo mutations, and in future studies, they are planning to use Agilent's recently updated SureSelect kit.
Also, causative mutations may lie in non-coding regions of the genome, which exome sequencing cannot address. Next year, the Nijmegen group will have its three SOLiD 4 machines replaced by new 5500xl SOLiD instruments, and if these enable whole-genome sequencing at a high enough throughput and "relatively low" cost, Veltman and his colleagues plan to sequence some entire genomes.
Their current goal for the coming year is to sequence the exomes of at least 100 mental retardation patient-parent trios. In addition, they plan to apply exome sequencing to a host of other inherited diseases — including rare syndromes with a few cases each — in order to define their genetic causes. Earlier this year, the researchers published a study in which they identified mutations leading to Schinzel-Giedion syndrome (IS 6/1/2010).
[ pagebreak ]
For some of these, the results will have direct clinical implications by improving diagnosis, Veltman said. In addition, the findings could point to new targets for therapy development.
Veltman also believes the concept of disease-causing de novo mutations could apply to common diseases other than mental retardation. "There is no reason to think that this would only work in mental retardation, because these mutations occur randomly in the genome," he said. "They can hit all kinds of genes, not only genes resulting in a brain disorder."
"Now that we can start sequencing genomes or exomes of patients, I think it would be very interesting to look at these [other diseases.]"
Others are already doing so. CHOP's Hakonarson and his team, for example, are currently sequencing the exomes of several hundred autism patients, and early results from the first 20 or so samples indicate that de novo mutations play an important role in that disease as well. "We see that many of the variants that we are identifying are in gene networks that we have previously published," he said. "We know about these genes … and now we are capturing the causative variants in these genes, which seem to be quite heterogeneous from one family to the next."
Overall, he said, it appears that common variants play a larger role in autoimmune or inflammatory-type diseases such as type-2 diabetes, heart disease, and inflammatory bowel disease than in central nervous system disorders, based on results from genome-wide association studies.
Mutations in genes affecting the brain, he said, have dire consequences, so natural selection keeps them rare, whereas variants involved in inflammatory diseases, "have by nature been advantageous to you to survive infections," he said. "And because people lived only so long until recently … the bad consequences of getting heart disease, or diabetes, or these other conditions just did not exist."
Another team that has been studying autism by exome sequencing — in particular, sporadic cases of the disease — is Jay Shendure's and Evan Eichler's groups at the University of Washington (IS 10/19/2010).
Like in mental retardation, de novo copy number variants also play a role in autism and schizophrenia, and it is thus likely that de novo point mutations are also important. "It's a similar set of diseases where you might expect the same model to hold," Shendure said. "Point mutations, essentially, give you much higher resolution, but you are basically looking at the same thing, which is large-effect, highly deleterious mutations occurring de novo."
"I think this model will prove fruitful in more contexts than just mental retardation."
Have topics you'd like to see covered in In Sequence? Contact the editor at jkarow [at] genomeweb [dot] com.