Genome Cartography
Genome Cartography
The genome is doubly mapped: once digitally, and now physically
When the original Nintendo console appeared in my life, it was in the living room of a friend of mine. Naturally I became much closer with my friend after this point. Although video games to me were auspicious, magical worlds, my mind often returned to one question: where did these worlds exist physically? How were all those Bubble Bobble levels conjured by the Nintendo to appear on my friend’s TV? At the time I concluded that every world, level, and character must exist in miniature somewhere inside the console. There just had to be some symmetry between the guts of the Nintendo console and the screen.
I was not alone in this pleasing oversimplification, as it is precisely the operating fiction in the film TRON. The movie is essentially a melodramatic depiction of how computers work according to a 6 year old. A steamy young Jeff Bridges plays a rebellious game designer trapped inside a computer by the renegade Master Control Program. Once inside, Bridges is challenged to partake in the very games he designed, this time orchestrated to kill him. He teams up with a few righteous programs who “believe in the users,” and together they defeat the Master Control Program.
TRON, released in 1982, turned computers into the very frame in which a quite human drama unfolded. But besides recasting computer programs as characters, the movie also created a landscape from a computer’s inner workings. It gave a bold answer, however inaccurate, to my question about Bubble Bobble: the glittering, smooth geography over which every computer function is executed.
TRON implicitly enacts a mapping. Here is the Gaming Grid, here is an accounting program, there is the Input/Output Tower. The film is a pleasure to watch in part because it offers a coherence of space with function. Tear apart a real computer, and the green and grey collage of metal and silicone gives no evidence of the computer’s workings. But TRON lets its viewers fantasize that computer functions occur across a particular landscape. Suddenly these functions are anchored, moored in 3D space. Cyberspace, the non-space of digital computation, is embodied into a particular landscape with a particular geography. It is this embodiment that builds the world of TRON, and animates its narrative.
A set of biologists are embarked on a similar geographical adventure, also linking spatial relationships with function, and also building new narratives. But their space is quite small: the 3D tangle of 46 chromosomes squished inside the nucleus of human cells. Recently, this tangle has become a mappable terrain. Using improved microscopy, chemical tricks, and better sequencing techniques, biological explorers are making spatial, not digital, maps of the human genome.
Genome mapping asks what regions of the genome make physical contact with one another, perhaps bound by this or that specialized DNA-binding protein. Very often the question is posed in terms of contact frequency (and thus is couched as the 4D map of the genome). This mapping considers the genome less as a series of long stretches of A, T, C, and G’s, and more as chromatin, i.e. DNA along with the huge mess of proteins that bind DNA and twist and curl it into its native, highly condensed state.
An oft-cited 2009 paper by Erez Lieberman-Aiden and colleagues offers a nice introduction to physically mapping the genome. The authors, at Harvard and UMass Worcester, devised a method called Hi-C to map all the long-range interactions in the human genome in a comprehensive manner. This yielded a rather complicated picture of interactions: the genome wadded up into hairball. Then they asked how does our DNA, which is linear, fold itself into this particular hairball? The predicted folding pattern looked really weird, and the authors proposed that it folds like something called a fractal globule (see the original paper, video of methods, and dance representation of methods).
One might argue that this description is an unnecessary overreach, that the genome is the genome whether or not we call it a fractal globule. But the 46 chromosomes of the human genome are, among many other things, a material, and as such they invite investigation of their material properties. DNA is code, but chromatin is stuff. Fractal globules obey certain rules, and classifying the folded genome as such made very specific predictions about its structure. That is, precisely how sequence space is related to physical space, how this code relates to that stuff.
Through this and other studies, the genome appears to see and touch itself with enormous complexity. What biological functions are imparted by this touching? Let us examine both a simple and a complex case.
A recently fertilized human egg might find itself with two X chromosomes. Critical to its development into a female-bodied embryo is the inactivation of one of these X chromosomes (a fertilized egg with one X and one Y chromosome need not worry about this). A specialized glob of RNA and protein called the Xist complex appears at one of the X chromosomes. It spreads along its chosen chromosome, taking care to leave the other one unmolested. When the work of Xist is done, this X chromosome is forever inactivated, and will be inherited as such by every daughter cell stemming from this fertilized egg. Xist spreads only along contiguous DNA stretches, and never jumps to nearby strands. Thus a spatial arrangement accomplishes this crucial and heritable task.
Now take your sense of smell. Each sensory neuron in your nose makes one and only one odor receptor, picked at random during development. On its way to maturing into a sensory neuron, the immature neuron selects a single receptor gene among around 1,100 such genes encoded in the genome. Not only that, the cell selects only one allele, i.e. the maternal or paternal copy, of the one receptor gene. The other allele, and all the other receptor genes are permanently silenced. This ensures that the sensory neuron reports on just one odor, encoded by the one gene. How is this exquisitely specific choice made within the soup of DNA, RNA, and proteins within the cell’s nucleus?
Extensive analysis of the regulatory DNA sequences that flank genes, and the proteins that bind to and activate genes, yielded no clues that would explain this winner-take-all molecular scenario. However, using microscopy to visualize where in the nucleus these genes are located, researchers found that most olfactory receptor genes are brought together, from vastly different stretches of the genome, into tight little clusters. Meanwhile, the one chosen gene copy was spared from these clusters and kept in its own domain of the nucleus. Generating and maintaining this physical setup involves the physical interactions across these 1,100 locations of the genome, coordinated by DNA binding proteins that grab, loop, and bunch DNA (see this review).
Considering these examples together, we might think of these lively chromatin gymnastics as a means of communication within the genome. As a concatenation of sequences, the human genome enjoys the ubiquity of any digital code: it is copied, pasted, scanned, and saved. The functions of various sequences are quite well described with a kind of cyberspace glee. But if biologists define DNA sequence as a source code, the execution of its functions becomes a question of materials, of location in time and space. The part that matters most happens not in cyberspace but through the objective of a microscope.
A recently published video (above) shows how a single stretch of DNA gets looped into a lasso by the protein Condensin, a central player in organizing the genome.
Surveying the physical organization of the genome means tethering a disembodied code with the messy gunk — the chromatin — inside of cells. Just as in TRON, this mapping opens genomes to a new set of queries, animated by new metaphors and narratives for how the genome functions. The genome might eventually become a set of places, and its biological functions might cohere into rather simple place-based inquiries. Physical mapping could render a strange collapse of the genome’s enormous complexity (3 billion letters!!) into a more intuitive map (this gene is here and not there).
This is not to argue that we throw away the code. DNA is heritable as a code, and its study as such has already transformed our understanding of biology and human health. Rather, DNA cartography is slowly adding back our materials to the code, and bringing with it new ways of considering our biology.
[This piece first appeared on Medium in 2018: https://medium.com/@timisstuck/genome-cartography-29be1607f996]