IT IS one of biology’s dirty little secrets: we don’t really know what we are made of. The human genome may have been decoded back in 2003, but we still can’t list all the types of cells in our bodies.
Everything we do, from moving to thinking, digesting food to sleeping, depends on a vast array of different cells: disc-like red blood cells, spindly nerve cells, stretched-out cells that make our muscles, the list goes on. These specialised units come together to form our tissues and organs, and make us the complex organisms we are. And yet so many of them remain mysterious.
Advertisement
Now, for the first time, we are set to make a comprehensive inventory – an aim every bit as ambitious as the Human Genome Project that decoded our DNA. The project plans to identify and locate every type of cell we possess, and so revolutionise our understanding of the body in the same way that the first atlases transformed our view of the world. The first results are already showing how this can help us find our way around healthy bodies, not to mention finding routes to new treatments for conditions like cancer that occur when cells turn bad.
“Knowing cells is knowing life,” says of the Broad Institute in Cambridge, Massachusetts. “If you do not know which cells are there, and understand how they operate, you can’t really say you understand biology.”
Easier said than done. For more than 150 years, researchers have categorised cells by all manner of methods: their size and shape, their location in the body, the way they react to dyes and, most recently, by the proteins they produce. Look in a textbook today and it will probably divide the body’s 37 trillion-odd cells into about 300 types (see “What is a cell anyway?”).
But each time a novel way to look at cells has come along, more differences have emerged, and the number of distinct types has continued to rise. The Human Cell Atlas will use an even more powerful way to classify cells, by the genes that are switched on inside them – their “transcriptome”.
Barring a few exceptions, all our cells contain identical copies of our genetic code. Each cell controls which of our 20,000 genes it expresses, and different types of cell express different selections of genes. Knowing which genes are active and at what levels should go beyond the promise of the Human Genome Project. This allowed us to compare people’s genomes and understand better than ever which genetic changes are involved in various diseases. But exploiting that knowledge means first knowing which cells these changes are expressed in.
To do that, biologists needed to be able to read the gene expression profiles of individual cells. Until recently, that process needed a fair-sized chunk of tissue that inevitably contained many different types of cell. These cells then got mushed together into what Regev describes as a fruit smoothie. What was really needed instead, she says, was a fruit salad in which every individual piece of fruit could be identified.
Unknown fruits
In 2011, that became a real prospect, when Regev and her colleagues managed to create . They used their method on – called dendritic cells – and immediately picked up differences between them that signalled unknown fruits in the salad. “We found two new subsets of cells we didn’t know existed before,” says Regev.
Since then, there has been a revolution in transcriptomics, with new and faster techniques emerging. Most rely on separating out individual cells in a sample and adding a unique barcode, in the form of a DNA tag, to each one’s genetic material. The tags allow the output of individual cells to be traced even when many cells are mixed together for sequencing. There has been an explosion too in the number of cells that can be profiled at once, from just 18 to a whopping 250,000. “These single-cell methods are really transformational,” says Ed Lein at the Allen Institute in Seattle, who specialises in mapping brain cells. “You can input a large number of cells from a tissue and get a molecular classification of the cell types in a way that supersedes 100 years of prior research.”
While Regev was busy in Cambridge, Massachusetts, Sarah Teichmann of the Wellcome Sanger Institute in Cambridge, UK, had also seen the potential of profiling the gene expression of single cells. “I would really love to see a periodic table of our cells,” she says. In 2013, she helped to found a centre within the Sanger Institute to develop the field. “The obvious question in a place like this is how big can we take it? Can we scale it to a whole organism?”
In 2014, Regev started to think about a systematic project to sequence all the cells in the body. She began to evangelise about the idea in talks and seminars. In 2016, the two researchers spoke. They decided to gather specialists from a range of fields – medics, pathologists, cell and computational biologists, genomics technology developers and big data analysts – to gauge their interest. A hundred people came to the resulting meeting and they were enthusiastic about the proposal. The idea of the Human Cell Atlas was born, with Teichmann and Regev at the helm. “There was that zeitgeist thing,” says Regev. “It was an idea whose time had come.”
Over 1000 scientists are now involved in the project, representing 584 institutes from 55 countries including Ecuador, Kenya, Nigeria and Russia. Atlases of the cells of the immune system, skin and lungs are already under way.
The effort faces many technical challenges (see “Single cell, big data”). Uncertainty still surrounds a second pillar of the project, that of locating cells in the body. Regev’s analogy of its structures being like a fruit salad isn’t entirely accurate. “Biology is beautifully structured. It’s actually a fruit tart,” she says. “It’s not everything mixed together or we would look kind of messy.”
“There was that zeitgeist thing. It was an idea whose time had come”
So to really understand how our bodies work, we need to be able to pinpoint individual cells within tissues. The past five years have also seen huge advances in these location technologies. Cells in a slice of tissue can be flooded with fluorescent probes that latch onto a specific gene sequence. Or the details of each cell’s location can be encoded into its genetic material using DNA tags before tissue is broken up for profiling. In June, researchers at Stanford University in California described a technique for sequencing single cells embedded in a 3D tissue sample.
But these methods cannot yet cope with the sheer number of cells and genes the Human Cell Atlas needs to analyse. Instead, some samples – which come from a range of sources, from operating theatres to tissue banks – will be stored and examined later once the technology catches up. Eventually, the atlas will bring together data from many labs relating to one organ or point within it. This will enable researchers to run analyses that encompass all the latest information. That is where its real power lies, says Lein.
Despite it being early days, the project is making groundbreaking discoveries. Single-cell profiling has already been used to investigate the earliest stages of pregnancy, as the fetal placenta implants into the mother’s uterus. It revealed the locations of fetal and maternal cells across the interface and the protein signals they use to communicate. These findings should shed new light on the causes of miscarriage and disorders such as pre-eclampsia.
Teichmann has also been working with Sam Behjati at the Sanger Institute and others on the gene-expression profiles of kidney cells. In August, they showing that cells from Wilms’ tumour, the most common childhood kidney cancer, resemble cells that haven’t matured properly. Today, Wilms’ is treated with toxic chemotherapy. Behjati suggests a better alternative would be to give the cancerous cells the correct protein signals to nudge them into becoming mature kidney cells. “This could lead to an entirely new model for treating childhood cancer,” he says.
Perhaps most surprising of all, in August, two teams discovered a previously unknown cell type in the lining of the lung that may well be at the root of cystic fibrosis. This is the most common single-gene disorder in people of northern European descent, and develops when the body doesn’t make enough of a protein called CFTR. The deficit causes sticky mucus to build up in the lungs, repeated infections and early death.
For 30 years, we thought CFTR was made in the lungs mostly by ciliated cells. Existing drugs and gene therapies for cystic fibrosis were designed assuming this to be true. Yet the new research shows that the lion’s share of CFTR is made by the newly discovered cells, now named ionocytes. These findings suggest a way to boost the number of ionocytes, which should help some people with cystic fibrosis.
Improved treatment for cystic fibrosis and Wilms’ tumour, and better understanding of why pregnancies fail are significant steps in their own right. But this is to miss the broader value of the Human Cell Atlas, reckons Sten Linnarsson at the Karolinska Institute in Sweden, one of the project’s organising committee. “The atlas is not enabling one specific thing,” he says. “It’s a foundation for exploring all human biology.”
What is a cell anyway?
Cells don’t exactly broadcast their importance. Most are invisible to the naked eye, yet without them there would be no life.
It was the English scientist Robert Hooke, looking at slices of cork through his home-made microscope in the 1660s, who first saw cells and named them. It took another 200 years for scientists to realise their true significance. The cell is the most basic unit of life – all living organisms are made of one or more of them.
Our cells contain an astonishing array of tiny structures, such as the nucleus, which is home to the genome; the mitochondria, responsible for generating energy; and the endoplasmic reticulum, where proteins and fats are made. All this paraphernalia is surrounded by the cell membrane, which acts like a national border, allowing only certain commodities in and out.
Human cells are nothing if not diverse. Among the smallest is the 50-micrometre-long sperm. Compare that with the egg. With a diameter of 0.1 millimetres – or 100 micrometres – it is just visible to the naked eye. A nerve cell that runs from the lower spinal cord to the big toe can reach more than a metre in length.
There is similar diversity in their roles. Nerve cells receive, process and transmit information through electrical and chemical channels. Hair cells in the ear turn vibrations into impulses that we perceive as sounds. Natural killer cells carry deadly chemicals inside them, with which they eliminate infected or cancerous cells. This is to name but a few. Each type depends on others to sustain them. Nearly all cells, for example, depend on red blood cells to bring them oxygen.
The types emerge in the embryo, where cells in the various regions are exposed to different cocktails of proteins, driving them to express a particular combination of genes and hence to develop into different cell types.
The number of active genes vary even in mature cells. “If you have a rapidly proliferating cell, you’re talking 5000 to 6000 genes,” says Sarah Teichmann of the Wellcome Sanger Institute in the UK. “But if you have a quiescent cell, that’s just minding its own business and not doing much, you can be talking a couple of thousand genes.” Cells can also change their outputs in reaction to signals from their surroundings. Teasing apart all these signals and reactions is a key aim of many Human Cell Atlas researchers.
Single cell, big data
When it comes to profiling the genes within a cell, the numbers are mind-boggling.
Different cell types express different genes, and to quantify the activity of each could mean 20,000 scores per cell. The Human Cell Atlas project expects to profile billions of cells. Making sense of data at this scale is going to need methods that don’t yet exist.
The project has received funding from the Chan-Zuckerberg Initiative, set up by Facebook owner Mark Zuckerberg and his wife Priscilla Chan. The money will go towards creating software tools and a platform where researchers can upload results and analyse data. The project released its first tranche of data, from more than 500,000 cells, this April.
One technique popular with the project’s biologists is a data visualisation tool, called t-SNE (pronounced tisnee), which groups cells according to similarities between their thousands of gene-expression scores. It then displays them in two dimensions.
This tool played a crucial role in the discovery of a new kind of lung cell. When all the different cell types from the lining of the lung were plotted using t-SNE, the graph showed a blob distinct from all the other cells. This led researchers to a previously unknown cell type, which they called an ionocyte, that has now been implicated in the disease cystic fibrosis.
This article appeared in print under the headline “You, deconstructed”
Article amended on 23 November 2018
Correction: We have corrected the diameter of a human egg.
