NEW YORK – Researchers from the Broad Institute have developed a method that combines in situ sequencing with next-generation sequencing to enable spatial mapping of sequences in 3D.
The researchers described their in situ genome sequencing (IGS) method and proof-of-concept studies in a paper published Dec. 31, 2020 in Science.
"In a cell, there are these three billion bases of DNA that are something like two meters long. And somehow that gets packed into a sphere one-tenth of the width of a human hair. It's not packed randomly and how it's packed is important for the cell's function," said Fei Chen, a researcher at the Broad Institute and a senior author of the paper. "This is the first time we're able to visualize that by sequencing the DNA directly inside the cell."
The method combines fluorescent in situ sequencing (FISSEQ), a method originally developed for in situ RNA analysis by George Church's lab at the Wyss Institute and Harvard University, with Illumina paired end sequencing. Church coauthored the new paper along with Chen, Ed Boyden of the Massachusetts Institute of Technology, and Jason Buenrostro of the Broad. NGS libraries are generated in situ with the help of hairpin DNAs that simultaneously barcode the in situ read and the NGS read.
The authors said they were able to resolve spatial features between 400 nm to 500 nm.
Using IGS, the authors performed proof-of-concept studies in early mouse embryos.
"I think it's a very exciting conceptual breakthrough and very nice proof of principle. They show that their previous method, fluorescent in situ sequencing (FISSEQ), developed in the Church lab, can be extended to study DNA, rather than RNA," said Longzhi Tan, a postdoc at Stanford University and an expert on methods to study genome organization. Tan helped develop the Dip-C method, a chromosome conformation capture-based technique.
"Integrating DNA data inside the nucleus as well as other spatial information from inside the cell will be very useful," Tan said. "Overall, it's a very comprehensive study and very beautiful work."
"The framework we've developed for generating barcoded in situ amplicons and then sequencing that genomic DNA using NGS, that's the big picture," said Paul Reginato, a graduate student in Church's lab and another co-first author of the study.
"The important part is the mapping and not necessarily the specific sequencing chemistry," added Andrew Payne, a graduate student in Chen's lab and another co-first author. "The DNA hairpins are elegant in that they give you a way to amplify genomic DNA and bring in the barcode in one step." And even using sequencing by ligation "was a design choice that actually has a lot of flexibility," he said. "We chose seq by ligation because everything can be done isothermally."
The researchers have patents pending and declined to comment on whether they were pursuing commercialization of their method. They suggested their method could be applied to early embryo development and cancer research.
The project began about four years ago. "We first tried to make a genomic DNA library in situ and amplify it without necessarily sequencing it," Payne said. He and Reginato, along with Zachary Chiang, a graduate student in Buenrostro's lab at Harvard, and the third co-first author, got together by 2018 to take the next steps.
Their method is one of many spatial genomics methods introduced in 2020, but unlike most others it focuses on DNA rather than RNA. It joins several other molecular methods for probing the structure and organization of the genome in 3D, such as Hi-C, Dip-C, and DNA fluorescence in situ hybridization.
IGS is very accessible, said Chiang, who did much of the bioinformatics work for the study. "Before this, I had very little wet lab experience. Even I was able to do a library preparation," he said.
During the in situ amplification, the method incorporates an 18- to 20-base barcode that is sequenced in situ with the microscope. That barcode helps map the Illumina reads back to the tissue.
The most challenging part is the in situ sequencing, which requires "a lot of time" on a dedicated microscope," Reginato said. It produces terabytes of imaging data, "but the nice thing is that after you do all the image processing is it really does look like a normal sequencing data set, with exception of the x, y, and z coordinates," Chiang said.
Computational analysis takes on the order of days, Chiang said and is not a limiting factor.
While Tan found the study to be comprehensive, he said he would have liked to seen the authors integrate the DNA local density with their in situ sequencing results. "I would be interesting to see which parts of the genome are locally compacted," he said.
While the team had optimized their method by 2018, they spent "a long time on demonstrating we could do it in early embryos, where you have a rearrangement of the DNA of maternal and paternal genomes," Chiang said.
"You can look at the first two cells and see how they compare to each other," he said. At the four-cells stage, the method can tell if any two cells are more similar to each other than to the other cells. What the team found was that the first two cells "have some kind of memory of how the genome was organized in the zygote," Reginato said. And after the cells divide again, "you could tell there were two pairs of sister cells," he said. In turn, those pairs of sister cells are more similar to each other than to cells from other embryos.
Tan said these studies highlighted one advantage of IGS compared to Dip-C: the ability to study cells that aren't diploid. Cells with aneuploidy, including cancer patient samples, would "be very interesting" to investigate with this method, he said.
"Cancer involves these kind of series of mutations and rearrangements or amplifications of sequences and it would be really exciting to follow the patterns of those mutations in space in a tumor," Reginato said. "We think you might be able to find different clonal lineages within tumors. Studying the mutational patterns in cancer could be very interesting."
Moreover, the method will be useful for "anything where spatial context is important, like stem cell niches, or where there's complex anatomical structure," such as the mammalian brain or gut, Tan said.
One speculative application would be in in vitro fertilization, if it could be shown that genome structure influences embryo viability, Chiang said.
Whatever the method's future, this is just the beginning. In the paper, the authors wrote that they expected to extend IGS to a broader selection of cell types. "Further, because nuclear volume is the primary constraint on the amplicon yield of IGS, we anticipate a many-fold improvements in yield and resolution, either through smaller amplicons, or preferably through integration of IGS with expansion microscopy," they wrote.
"One reason we're so excited is that when doing everything in situ, you can layer in whatever you want," Chiang said. "It could be scaled up to measure many different proteins, RNA, or even combine them all in the same assay."