Deciphering the DNA sequence of the human fetal genome, without penetrating the womb, was considered impossible. But in a recent breakthrough
, University of Washington geneticists Jacob O. Kitzman and Jay Shendure, working with colleagues in the United States and Italy, not only successfully sequenced the fetal genome, but they did so using genetic clues teased from the readily accessible reservoirs of maternal blood and paternal saliva.
Although the genetic clues consisted of only fragments of fetal DNA found in maternal blood and genetic variations identified in maternal and paternal DNA, the team stitched the pieces together using modeling and statistical analyses and ultimately predicted, with 98 percent accuracy, the genetic sequence of an 18.5-week-old fetus. They confirmed the accuracy of their prediction by using traditional whole genome sequencing technology to elucidate the complete DNA sequence of the fetus after it was born.
The research, detailed in the June 6 issue of Science Translational Medicine, represents a major landmark in the conquest for genetic omniscience that is embodied by the advance of DNA sequencing technology. It has also brought renewed interest to that most fundamental of biological features, heredity.
Capitalizing on genetic variation
As Kitzman explained, “We each carry two copies of most genes, and we obtain one copy from each of our parents.” But random processes, such as independent assortment and crossing over, which occur during the form of cell division known as meiosis, decide which genes we inherit from which parent, ensuring that each of us is genetically distinct from our parents and siblings (identical twins excepted).
The random nature of inheritance explains why a fetus’s DNA sequence cannot be reconstructed by simply merging the sequences of each of the genes carried by its parents. In fact, to sequence the human fetal genome, Kitzman and colleagues had to find a way to capitalize on genetic variation. They did so by first defining what variants each of the parents carried, which entailed sequencing each of the parent’s genomes, using blood from the mother and saliva from the father.
The researchers then built upon the previous observation
by geneticist Dennis Lo that, in a pregnant woman, a small but measurable quantity of fetal DNA fragments circulate among maternal DNA fragments in the mother’s plasma (the liquid portion of blood). The fragments enter the plasma when DNA is released from broken down cells.
Kitzman and colleagues performed deep sequencing (in which bits of DNA were read out many times over) on the free-floating DNA. They anticipated that the high redundancy of deep sequencing, together with information on variants from the maternal and paternal genome sequences, would enable them to predict which genetic variants the fetus had inherited and possibly to identify new (“de novo”) mutations that were not present in either parent.
Before they could move ahead with their predictive analyses, however, they had to find a way to distinguish maternal DNA from fetal DNA. While in theory this could be accomplished by measuring the small increase in variant copy number associated with the presence of fetal DNA in the mother’s plasma, copy number provides only a weak signal for identifying which variants the mother passes to the fetus. So, to more reliably detect fetal variants, Kitzman’s team decided to combine information about genetic variants passed on from the father with information on maternal haplotype.
Help from haplotypes
Haplotypes are groups of alleles (alternate forms of genes) that are located at different points on the same chromosome and that typically are inherited together. “Haplotypes are passed down mostly whole from generation to generation,” Kitzman said. “For instance, if you look at all the variants I inherited from my mother along a particular stretch of my DNA, those variants are likely to have come either all from my maternal grandfather or all from my maternal grandmother.”
The inability to resolve parents’ haplotype information, however, has been a major obstacle to noninvasive fetal genome sequencing. In fact, as Kitzman explained, haplotype information is neglected by even the newest generation of whole genome sequencing technologies.
Thus, to successfully capture haplotypes for their study, the researchers employed a new technique of their own invention that allowed for the identification of genetic variations known as single nucleotide polymorphisms (SNPs) lying within “haplotype blocks”—stretches of DNA consisting of about 300 kilobase pairs (or about 300,000 A-T and C-G rungs in the DNA ladder). SNPs embedded within the blocks flagged sites in the DNA from maternal blood where genetic constitution was unique, differing from both maternal and paternal sequences.
The technique, which the team spent two years developing, greatly simplified the task of separating out fetal from maternal DNA. “When looking at the DNA sequences from the maternal plasma, rather than having to infer whether she passed a given variant on, and then carry out the same inference separately for each successive variant, we were able to infer whether whole haplotypes were passed down,” Kitzman said. “This gave a much more robust signal—in a sense, many more ‘votes’ contributing to the inference—and allowed us reach over 99 percent accuracy when predicting what variants the fetus inherited from the mother.”
The final step in sequencing the fetal genome involved the detection of new mutations in the fetal DNA. “Finding these is a major needle-in-a-haystack problem, as there are only expected to be around 50 of these in each offspring, out of 6 billion possibilities,” Kitzman explained. In the fetus whose genome was sequenced, 44 new mutations were detected in sequencing following birth; the team successfully predicted 39 of these using the maternal samples collected at 18.5 weeks of pregnancy. The predictions were accompanied by a large number of false positives, a problem that likely could be overcome, at least in large part, by careful computational filtering.
Leveraging the human fetal genome
The techniques used for sequencing the fetal genome could be applied to research on fetal development. For instance, fetal RNA sequencing or detection of epigenetic marks (chemical modifications) present on fetal DNA fragments could offer new insight into gene regulation in fetal cells.
In addition, the sequencing breakthrough has opened up new possibilities in the realm of prenatal genetic screening. A noninvasive test based on parental blood and saliva samples could eliminate the risk of miscarriage associated with existing invasive prenatal genetic tests, which require the collection of fetal cells by amniocentesis or placental tissue by chorionic villus sampling.
Moreover, as Kitzman explained, “[Fetal whole genome sequencing] would greatly broaden the number of disorders that could be screened to any of the approximately 3,000 single-gene (or ‘Mendelian’) disorders, where the underlying genetic cause is known. In some cases, this could help expectant parents plan for appropriate care and support, and in the more distant future, could help guide prenatal therapies.”
Related Science Friday links: