NEW YORK (GenomeWeb) – A Chinese team led by Chinese University of Hong Kong researcher and Sequenom advisor Dennis Lo has demonstrated the feasibility of using RNA sequencing to profile fetal and maternal RNA transcripts in maternal blood.
Lo and his colleagues outlined the maternal RNA sequencing strategy in a proof-of-principle study appearing online in Clinical Chemistry earlier this month. By sequencing RNA from maternal blood and other samples collected from women at early and late stages of pregnancy — and, in some cases, after birth — they showed that it's possible to differentiate between the fetal- and maternal-expressed versions of thousands of genes.
And while fetal- and maternal-specific transcripts seem to make up a fraction of RNA circulating in a pregnant woman's blood, the approach proved useful for finding new and known genes that are expressed specifically during pregnancy. That, in turn, has prompted enthusiasm about the prospect of developing RNA biomarkers to study typical pregnancy progress and non-invasively pick up on the presence of certain complications.
"In essence, this direct plasma examination method presents a powerful approach for the discovery of circulating pregnancy-associated RNA transcripts," Lo and his co-authors wrote, "without a priori knowledge of the transcriptomic profiles of the placental tissues and the blood cells."
Whereas efforts to sequence DNA from maternal plasma are largely focused on finding genetic or chromosomal conditions in the developing fetus, the ability to sequence RNA in maternal blood samples has potential for following both fetal and maternal transcriptome patterns over the course of pregnancy, Rossa Chiu told In Sequence in an email message.
Chiu, a co-author on the Clinical Chemistry study, is affiliated with the Centre for Research into Circulating Fetal Nucleic Acids at Li Ka Shing Institute of Health Sciences and with the Chinese University of Hong Kong's chemical pathology department.
By allowing a look at genes expressed in fetal and maternal tissues that make their way into maternal blood, for example, she explained, RNA sequencing can help in tracking pregnancy progression and potential complications that can arise.
"[I]t offers a means to monitor the fetal and maternal well-being," Chiu said. "One would expect the maternal plasma RNA profile to change with pregnancy-associated diseases such as preterm labor, intrauterine growth retardation, and preeclampsia."
Despite the potential advantages of being able to profile fetal and maternal RNA from a pregnant woman's blood sample, there have been several complications associated with past efforts to do so.
Because expression levels vary dramatically for different genes, for instance, it is tricky to see representatives across this dynamic range in maternal plasma, which typically contains far less RNA than other tissue types.
"Ideally, one would like to determine the relative abundance of the different RNA species across a wide spectrum of gene expression levels but within the low amount of total RNA pool," Chiu explained.
While some researchers have turned to real-time reverse transcriptase PCR to quantify levels of specific RNA transcripts in blood from pregnant women, she noted that it can be difficult to compare levels of one RNA transcript to another. The PCR-based approach also has limitations with respect to the number of different RNA transcripts that can be characterized simultaneously from the same sample.
In an effort to come up with a way of getting a broader look at fetal and maternal transcripts carried in an expectant mother's blood, Lo, Chiu, and their colleagues performed massively parallel RNA sequencing experiments informed by DNA sequence-derived maternal genotypes.
The general strategy used for sequencing is similar to that used to sequence DNA from maternal blood samples, Chiu explained, except that a reverse transcription step is used to generate complementary DNA from the RNA transcripts present in plasma prior to the other library preparation steps.
There are other differences, too. To decrease interference from non-transcript RNAs such as ribosomal RNA or globin genes, the team treated some of the samples during the library preparation process. For other samples, the rRNA was allowed to stick around during sequencing before being weeded out bioinformatically during the sequence analysis steps.
In their Clinical Chemistry study, the researchers started by doing RNA sequencing on blood cell and plasma samples collected from one woman in her first trimester of pregnancy and one in her second trimester. Chorionic villus or amniotic fluid samples were also obtained as a source of genomic DNA for the women in early stages of pregnancy.
For two more women, the team assessed blood samples taken during the third trimester of pregnancy and a day after delivery by cesarean section, along with placental tissue samples collected at the time of birth.
Blood samples from 20 more women in their third trimester of pregnancy and two non-pregnant control women were sequenced for some of the team's additional analyses.
During sample preparation, the study's authors treated blood cell and placental tissue samples with Epicentre's Ribo-Zero kit to remove rRNA.
Plasma samples from one of the women during late-stage pregnancy were also treated to remove RNA during library prep, though the group opted for bioinformatics-based approaches for plasma from the other women to minimize GC biases associated with sequences rich in guanine and cytosine bases.
After removing reads attributed to rRNA, the team was left with 12 million reads per pregnant woman's plasma sample, on average — far more analyzable reads than were present after RNA sequencing on plasma from non-pregnant women, but fewer than those obtained by sequencing RNA in placental tissue or blood cells.
To distinguish between maternal and fetal transcripts in the resulting RNA sequences, the researchers relied on the presence of fetal polymorphisms not present in the expectant mother and vice versa, using genotyping information obtained by Illumina paired-end sequencing on exome-enriched DNA from blood, chorionic villus, amniotic fluid, or placental tissue samples.
In samples from each of the early-stage pregnancy cases, for example, the team identified more than 6,700 informative genes containing one or more SNPs that differed between the mother and the fetus. That jumped to almost 7,800 informative genes in the first two late-stage pregnancy cases considered.
The team also compared the transcripts detected in pregnant women with those found after pregnancy and in women who were not expecting, identifying fetal- and maternal-specific alleles that became more pronounced during later stages of pregnancy.
Together, these approaches made it possible to determine which RNA molecules made their way into maternal plasma and to begin tallying up transcripts in the mix that originated in fetal and maternal tissues.
The team estimated that just under 1 percent of circulating transcripts detected in maternal plasma in first or second trimester pregnancies were highly expressed by the fetus. That proportion climbed to more than 2.5 percent in plasma samples from third trimester women.
The fraction of transcripts from genes with high maternal expression was higher during both early and late pregnancy, the researchers reported, ranging from more than 42 percent to almost 51 percent.
Transcripts associated with fetal-specific alleles tended to disappear from maternal plasma after pregnancy, as did some of the transcripts with maternal-specific SNPs. For example, in one of the late pregnancy cases, the team saw both fetal and maternal transcripts of the pregnancy-specific gene PAPPA in the placenta and in a plasma sample taken prior to birth.
All told, the team saw more than 130 potential pregnancy-associated genes, including 15 genes with known ties to pregnancy and five more that were subsequently verified using real-time RT-PCR.
Such findings raise the possibility that at least some of the maternal and fetal transcripts in maternal blood might prove useful for following biological events that occur in pregnancy and uncovering instances in which they begin to go awry.
Still, Chiu noted that more work is needed to verify the reproducibility of the apparent pregnancy-associated transcripts and unravel their physiological significance, if any.
Likewise, she said that more tweaking, testing, and validation of these candidate markers — and the sequencing method itself — will be required before researchers make a move to introduce plasma RNA sequencing into a more clinical setting.
"[W]e are still in the early stage of exploring the clinical potential of maternal plasma RNA sequencing, Chiu explained. "As with the development of any new clinical assessment approaches, one would need to first refine the analytical protocol to improve cost-effectiveness, and then perform clinical studies of various target diseases."
While the current study was done using Illumina's HiSeq instrument, a similar maternal plasma RNA sequencing approach should be applicable with a range of sequencing chemistries, according to Chiu.
Moreover, she noted that there may be advantages to using long-read platforms in some cases, particularly for those interested in trying to characterize various transcript isoforms for given genes and/or determine the lengths of RNA molecules present in plasma.
The team is considering ways of improving the maternal RNA sequencing method further, including efforts to dial down interference by ribosomal RNA and globin RNA.
The reagent price tag for the current maternal RNA sequencing protocol is around $1,800 per plasma sample, not including the cost of related equipment, lab space, and staff.
The approach has been patented, though Chiu declined to comment on potential licensing plans, citing confidentiality agreements. In the past, Lo's team has licensed intellectual property to Sequenom related to other non-invasive prenatal testing methods, including approaches for finding fetal aneuploidies from cell-free DNA in maternal blood.