Just under your skin lie whole aqueous worlds, where trillions of cells spark and beat and wriggle and secrete, doing all the complicated tasks of keeping you alive. They all share the same genetic code. But what they do with it is the difference between a neuron and a twitching muscle fiber.
Starting about a decade ago, a group of scientists began conducting a cellular census of every tissue in the human body to find out what cells actually live there, using a powerful new technology called single-cell RNA sequencing. It illuminates which parts of the genome a cell uses to conduct its unique task. The international collaborative effort, called the Human Cell Atlas, has since grown to include more than 2,000 researchers from 83 countries. And on Thursday, they reported a major feat: the creation of detailed maps of more than a million cells across 33 organs.
The landmark tissue atlases were published in four studies in Science. “You can think of it as a Google Maps of the human body,” Sarah Teichmann, head of cellular genetics at the Wellcome Sanger Institute and co-chair of the Human Cell Atlas, told reporters Tuesday.
In one study, her team sequenced RNA from 330,000 single immune cells from across the adult body, and in another, they cataloged the development of immune cells in prenatal tissues. They discovered that as infection-fighting T cells develop, they learn just as much from talking to each other as they do from their parent tissues. Deciphering this molecular code could allow researchers to better engineer T cells to do things like fight cancer. “The insights have implications for therapies that enhance or suppress an immune response to fight disease and for designing vaccines,” Teichmann said.
A third paper, led by co-chair Aviv Regev, one of the pioneers of single-cell sequencing who now leads R&D at Genentech, described how researchers from the Broad Institute created a cross-tissue atlas of 200,000 cells from frozen tissues. Using machine learning, they scanned the atlas to identify the cell types associated with 8,000 genetic diseases. “We hope that by using maps like these, we will be better able to understand the precise place in the body where disease arises,” Regev told reporters. “That would allow us to develop more precise diagnostics and new treatments.”
Stephen Quake, president of the Chan Zuckerberg Biohub Network and a member of the Human Cell Atlas organizing committee, contributed an update from the Tabula Sapiens consortium, which unlike many of the other efforts, is gathering sequences from a single donor’s cells. So far, it provides a portrait of nearly 500,000 cells from 24 organs of 15 recently deceased individuals.
STAT spoke with Quake about the scientific milestone and what comes next. Excerpts from the conversation are below, lightly edited for clarity.
The consortium has now mapped more than a million individual cells across 33 organs, an important feat, a first draft, if you will, of the Human Cell Atlas. How are you feeling?
This is a big moment. Around 2011, 2012, there were four or five people in different corners of the world saying we should build a whole-organism cell atlas. So it’s nice to see it now all come to fruition. But yes, it’s absolutely a first draft. In that way, there’s a good analogy with the Human Genome Project.
When the first human genome was published, it was a draft genome. There were all kinds of gaps and things missing but it was incredibly useful nonetheless. Now, 20 years later, we’re seeing the first telomere-to-telomere really finished human genomes, which are adding value. And I think about these cell atlases in the same way. These are drafts. We’re not saying we found every cell type in the human body, or even every tissue, but boy it’s going to be so useful.
How are researchers starting to use them?
I have a colleague who wants to use it to study brain cancer. He was finding potential drug targets and wanted to look elsewhere in the body for unanticipated toxicity. And I think a lot of people have been doing that approach. They have a drug target of interest for a disease and they want to know where else is that protein expressed – what other cell types, what other tissues – because making a drug against that target can affect other tissues besides the one you want.
Another good example is a paper that’s already out by one of my students, Sevahn Vorperian, where she used the atlas to understand something about liquid biopsies. She realized she could use Tabula Sapiens to decompose the cell-free transcriptome and the cell types of origin. And that has generated a lot of interest in the diagnostic community.
With the idea being you could look at signatures of disease coming from RNA circulating in someone’s blood and trace it back to the specific cells where that dysfunction is occurring?
The Human Cell Atlas consortium has taken a sort of one-tissue-at-a-time approach, with different research groups working on their tissue of expertise. How is Tabula Sapiens different?
What we’ve brought to the table was figuring out how to do these multi-organ experiments. Which has been a big collaboration in its own right. You know the idea of taking all these organs from a single donor had never been done before. And because these are living donors, we really have to get everyone there, as these people are being operated on. [The Tabula Sapiens project worked with an organ procurement organization to preserve tissues while surgeons were harvesting organs for donation.] And that’s a big management challenge. It’s been a big lift for me personally because I’d always run a small lab. I had to learn how to do Big Science.
But one great advantage of looking across tissues from the same person is you can control for all kinds of things like genetic background, epigenetic effects, environmental exposure. That allowed us to do things like studying splicing. Each gene has different pieces that can be spliced in or out, depending on which piece is being used. What hasn’t been well known is whether splicing depends on cell type. And we were able to map that out here to find very interesting variations in splice usage according to cell type and we discovered a whole bunch of new splices that had never been seen before.
The Human Cell Atlas is a successor to the Human Genome Project, which you mentioned earlier. In what ways do you see it as carrying forward the tradition of Big Science as it was defined by that era and in what ways is it charting a new legacy?
It definitely shares some aspects of Big Science. It requires a lot of coordination between lots of groups, lots of people all over the world. And it’s taking on a problem that couldn’t really be done otherwise, because we need all that diverse expertise and contributions. But it’s also different in a couple of ways.
It’s more collegial. The Human Genome Project was sort of famously acrimonious.
To what do you credit that?
It’s probably a function of personalities. The genome project had some big personalities involved that didn’t really get along. Aviv and Sarah, the co-chairs on this project, and I have a much better relationship. Also in this case there’s no private effort, so there’s not a public-private competition going on.
Another difference is the cost. The first human genome cost $ 3 billion. We strategically made the decision to wait until the technologies were a little more affordable. If the genome project had waited even five years, it would have been a lot cheaper. I’ve talked to Craig Venter [who led the private effort to sequence the human genome] about this, and I asked him if it was worth it to do it earlier. “Oh it was definitely worth it. We learned so much, ”he told me. I’m not sure I agree with that assessment. But the Human Cell Atlas teams have all been on the same page about doing this when we felt the cost-benefit was right.
And that’s important because I think that what we’re trying to do is enormously more difficult than sequencing a genome. And the reason for that is sequencing a genome is this incredibly well-defined chemical problem. Here’s a test tube with some chemicals in it. Tell me what the chemicals are – the chemical being the DNA molecule. Whereas understanding the nature of these cells, they’re much more complicated. It’s not a chemical problem, it’s a biological problem. And it’s harder to abstract it to a simple measurement.
Because what you’re really doing is redefining the parameters of what it means to be this kind of cell or that kind of cell. It’s not just how a cell looks or where it lives, but these gene programs that each one is running. So how do you decide how far to drill down – where do you make these cutoffs?
That’s such a good question, and one that’s been open for a long time. What’s the difference between a cell state and a cell identity? From my perspective, I don’t think it’s a solved question yet. The community still wrestles with it. We’re still scratching around what the fundamental nature of these objects is.