ABOVE: Cancerous prostate tissue NATIONAL CANCER INSTITUTE, OTIS BRAWLEY
In a study published August 10 in Nature, scientists have charted the genetic landscape across a human prostate in high resolution and discovered how additional or missing chunks of chromosomes, known as copy number variations, thought to be unique to cancer are often present in seemingly healthy tissue.
“This was surprisingly and completely unexpected,” says study coauthor Alastair Lamb, a urologist at Oxford University’s Nuffield Department of Surgical Sciences. “We thought that these kinds of changes defined prostate cancer. But they are present in [tissue] which is entirely benign.”
Far from a uniform mass of cells, tumors consist of a patchwork of malignant, benign, and healthy tissues. Understanding how normal cells become cancerous requires that scientists plot the genetic changes within this complex ecosystem. In collaboration with gene technology researcher Joakim Lundeberg and colleagues at the KTH Royal Institute of Technology in Sweden, Lamb’s team used a technique called spatial transcriptomics to map copy number variations to precise locations within a prostate.
Unlike methods such as bulk DNA sequencing, where cells are ripped apart and any information about their physical location is lost, spatial transcriptomics preserves the sample’s 3D structure by genetically tagging RNA molecules with location data before they’re sequenced, enabling each sequence to be matched to its original location within the organ.
This involves placing a tissue section on a glass surface containing a grid of dots, each of which anchors short nucleotide strands containing a spatial barcode and a string of thymine residues that bind to mRNA molecules. The tissue sample is permeabilized just before being placed on the glass so that mRNA will spill out of the cells, allowing it to attach to the grid, thereby marking each molecule with its location within the tissue.
In the study, the researchers wanted to create a map of copy number variations within a prostate that had been surgically removed from an elderly patient, so they trawled tissue cross-sections for transcripts using a grid containing 30,000 spots. Each dot, corresponding to an area of just ten cells, captured around 3,500 mRNA molecules released from the cells nearest it, allowing the team to map gene expression. Then, using a computational approach, the team used the RNA they sequenced to predict copy number variations in the cells’ DNA and identified clusters of genetically identical cells, or clones. They then assembled the clones into a phylogenetic hierarchy to chart how the cells’ genetic composition had changed over time.
Meanwhile, the team visualized the intact tissue under a microscope to annotate regions of cancerous and benign tissue based on morphology. Combining the imaging data with spatial genetic profiles revealed a striking amount of heterogeneity within a single tumor, says Lamb, with low-grade tumor cells (those that more nearly resemble healthy cells) nestled among both healthy and further-progressed, high-grade tumor cells. As expected, cancer cells contained copy number variations in oncogenes such as MYC and the tumor-suppressor gene PTEN. Surprisingly, though, the same changes occurred in nearby healthy tissue.
“What’s so fantastic here is the two-dimensional snapshot. We see the very early events, the intermediate events and that the tumor has diverged,” says Lundeberg. He contrasts the technique to histological sampling, in which a pathologist identifies a tumor under the microscope and dissects it using a laser. This can provide detailed insight into the tumor but ignores the cellular environment around it, Lundeberg says. “But with spatial transcriptomics, you capture the early events that aren’t evident when you look under the microscope.”
This sort of whole-organ approach to cancer genetics is something that Elana Fertig, a cancer researcher at Johns Hopkins University who did not take part in the study, is particularly excited about. “I think it’s incredible and that’s something we increasingly need to do. We want to make sure we’re not missing the molecular markers of recurrence which might not be in the predominant lesion,” she says.
To investigate whether similar patterns occur in other cancer types, Lundeberg and colleagues repeated the procedure using tissue sections of skin cancer, breast cancer, and glioblastoma. As before, they found chromosomal alterations in both tumor cells and nearby healthy tissue.
These findings suggest that copy number variations occur before, not after, benign cells transform into malignant tissue, says Francesca Ciccarelli, a cancer geneticist at King’s College London who was not involved in the study. “The very observations that seemingly normal cells bear tumour-associated [copy number variations] implies that they are not sufficient for transformation,” she writes in an email to The Scientist.
This raises the question as to what genetic events, if not these copy number variations, drive tumorigenesis. DNA changes not investigated in the study, such as epigenetic modifications, may play an important role, Lamb suggests. “There is also the potential for [chemical] signaling between the cancer clones and the surrounding stromal tissue,” he adds.
Nevertheless, capturing early genetic events that aren’t microscopically visible could allow researchers to predict whether an area of benign tissue may give rise to lethal cancer, with important implications for early diagnosis and targeted therapy. “That is so, so powerful,” Lundeberg says.
