When the Human Genome Project was completed, in 2003, it opened
the door to a radical new idea of health—that of personalized medicine,
in which disease risk and appropriate treatment would be gleaned from
one's genetic makeup. As more people had their genomes sequenced,
disease-related genes would start coming into view— and while this is
true in many ways, things also turned out to be much more complicated.
Sixteen
years on, tens of thousands of people have had their genomes sequenced
yet it remains a major challenge to infer future health from genome
information. Part of the reason may be that genes interact with each
other to modify trait inheritance in ways that aren't totally clear,
write Donnelly Centre researchers in an invited perspective for the
leading biomedical journal Cell.
"All the genome sequencing data is highlighting the complexity of
inheritance for the human genetics community," says Brenda Andrews,
University Professor and Director of U of T's Donnelly Centre for
Cellular and Biomolecular Research and a senior co-author, whose lab
studies interactions between genes. "The simple idea of a single gene
leading to a single disease is more likely to be an exception than a rule," she says.
Andrews and Charles Boone, who is also a senior co-author, are
professors in U of T's Donnelly Centre and the Department of Molecular
Genetics, as well as Senior Fellows of the Genetic Networks program at
the Canadian Institute for Advanced Research, which Boone co-directs.
Genome wide association studies, or GWAS, which scan the genomes of
patient populations and compare them to healthy controls, have unearthed
thousands of mutations, or genetic variants, that are more prevalent in
disease. Most variants are found in common diseases that affect large
swathes of the world's population but their effects can be small and
hard to see. Instead of there being a single gene for heart disease or
schizophrenia, for example, there may be many combinations of subtle
genetic changes scattered across the genome that tune up or down a
person's susceptibility to these diseases.
Vast genetic diversity in the human population further influences
trait inheritance while environmental effects, such as diet and
upbringing, further complicate matters.
In some cases, a single gene
variant can be extremely potent and cause a disease, as seen in cystic
fibrosis, heamophilia and other inherited disorders. But even two people
with the same disease variant can experience a wildly different disease
severity which, presently, cannot be gleaned from their genomes. Even
more astonishing, sequencing studies have identified people who carry
damaging mutations but remain perfectly healthy, presumably protected by
other, as yet unknown gene variants within their genomes.
"It would be a simpler problem if one particular mutation resulted in
Disease X all of the time, but that's often not the case," says Michael
Costanzo, Senior Research Associate in Boone's lab and one of the
authors on the paper. "To understand the effect of combinations of
variants is really difficult. We suspect it's particular sets of
mutations that really impact what the disease outcome is going to be in a
personal genome" says Costanzo. "How genes interact with each other is
important and, given our current understanding of gene-gene
interactions, it's not a problem that's easily solved by reading
individual genome sequences."
It's a numbers game as most genome analysis methods lack the
statistical power to confidently uncover multiple genes behind a
disease. An often-cited calculation, by researchers at the Broad
Institute in Boston, states that to identify a single pair of genes
underlying a disease, on the order of half a million patients would have
to have their genomes sequenced, with another half a million of healthy
people as controls. "If most genetic diseases involve gene
combinations, collecting enough patient data to find these interactions
is a huge challenge," says Costanzo.
Genetic interactions—what are they and how can they be identified?
"The concept of genetic interaction is simple, but the physiological
repercussions can be profound," write the authors. Two genes are said to
interact if a combined outcome of their defects is bigger or lesser
than expected from their individual outcomes. For example, a person
carrying a mutation in either gene A or in gene B can be healthy, but if
both A and B don't work, disease occurs.
Research in simple model organisms—most notably yeast—has mapped genome-wide genetic interactions revealing how thousands of genes organize
into functional groups within a network. From this, basic principles
emerged, allowing researchers to predict a gene's function and its
relative importance for the cell's health based on its position in the
network. Studies also revealed the identity of so-called "modifier
genes" which can suppress the effect of damaging mutations and how genetic background influences trait inheritance.
These types of studies rest on the researchers' ability to switch off
genes in precise combinations to find the ones that work together. For
human genes, however, such tools did not exist until very recently.
That's all changed now thanks to the gene editing tool CRISPR with
which human genes can be turned off in any combination with ease.
Although no genome-wide map is yet available, early work indicates that
the same principles uncovered in model organisms also apply to human
genes. This is already helping reveal function of the less studied human
genes and how they relate to disease. And with new computational
approaches, it is becoming possible to integrate findings from model
organisms with incoming human data to achieve an emerging glimpse of
more meaningful insights about health from genome information.
Genetic interactions and cancer therapy
Freed from normal checks and balances, cancer cells stockpile
mutations in their genomes and this sets them apart from healthy cells
in a way that can be exploited for therapy. Knowing how genes interact
in cancer holds promise for the development of selective drugs that kill
only sick cells and leave healthy ones unharmed.
"Cancer is a genetic disease and ultimately the genetic wiring of a
cancer cell is a product of mutations that occur its genome and we want
to understand that," says Jason Moffat, a co-author on the paper and a
professor of molecular genetics in the Donnelly Centre whose lab uses
CRISPR to map genetic interactions in cancer cells. "With CRISPR, we can
start to systematically map how genes interact in cancer cell lines in a
similar fashion to how geneticists have mapped genetic interactions in
yeast," he says.
This work has the potential to reveal distinct drug targets for
different forms of disease. The goal is to find a drug that synergizes
with a mutation that's only found in a type of cancer. The drug would
then kill sick cells more precisely and with fewer side effects than
chemotherapy or radiotherapy.
The knowledge of genetic interactions will also help shed light on
why so many approved cancer drugs only work in some patients and not
others.
"We can't think about genes in isolation anymore," says Boone. "We have to start looking at variants of multiple genes
as a major component of genetic disease, because those combinations are
going to be different for different people and these specific
combinations may not only profoundly affect disease susceptibility, but
they will likely dictate new personalized therapies."
