THE GREATEST HITS OF THE HUMAN GENOME
`
`A tour through the most studied genes in biology reveals some surprises.
`
`BY E L I E D O L G I N
`
`P eter Kerpedjiev needed a crash course
`
`in genetics. A software engineer with
`some training in bioinformatics, he
`was pursuing a PhD and thought
`it would really help to know some
`fundamentals of biology. “If I wanted
`to have an intelligent conversation with some-
`one, what genes do I need to know about?”
`he wondered.
`Kerpedjiev went straight to the data. For
`years, the US National Library of Medicine
`(NLM) has been systematically tagging almost
`every paper in its popular PubMed database
`that contains some information about what a
`gene does. Kerpedjiev extracted all the papers
`marked as describing the structure, function
`or location of a gene or the protein it encodes.
`Sorting through the records, he compiled a
`list of the most studied genes of all time — a
`sort of ‘top hits’ of the human genome, and
`several other genomes besides.
`Heading the list, he found, is a gene called
`TP53. Three years ago, when Kerpedjiev first
`did his analysis, researchers had scrutinized
`the gene or the protein it produces, p53, in
`some 6,600 papers. Today, that number is at
`about 8,500 and counting. On average, around
`two papers are published each day describing
`new details of the basic biology of TP53.
`Its popularity shouldn’t come as news
`
`to most biologists. The gene is a tumour
`suppressor, and widely known as the ‘guard-
`ian of the genome’. It is mutated in roughly half
`of all human cancers. “That explains its staying
`power,” says Bert Vogelstein, a cancer geneti-
`cist at the Johns Hopkins University School of
`Medicine in Baltimore, Maryland. In cancer,
`he says, “there’s no gene more important”.
`But some chart-topping genes are less well
`known — including some that rose to promi-
`nence in bygone eras of genetic research, only
`to fall out of fashion as technology progressed.
`“The list was surprising,” says Kerpedjiev, now
`a postdoc studying genomic-data visualiza-
`tion at Harvard Medical School in Boston,
`Massachusetts. “Some genes were predictable;
`others were completely unexpected.”
`To find out more, Nature worked with
`Kerpedjiev to analyse the most studied genes
`of all time (see ‘Top genes’). The exercise offers
`more than a conversation starter: it sheds light
`on important trends in biomedical research,
`revealing how concerns over specific diseases
`or public-health issues have shifted research
`priorities towards underlying genes. It also
`shows how just a few genes, many of which
`span disciplines and disease areas, have
`dominated research.
`Out of the 20,000 or so protein-coding genes
`in the human genome, just 100 account for
`
`more than one-quarter of the papers tagged
`by the NLM. Thousands go unstudied in any
`given year. “It’s revealing how much we don’t
`know about because we just don’t bother to
`research it,” says Helen Anne Curry, a science
`historian at the University of Cambridge, UK.
`
`IN AND OUT OF FASHION
`In 2002, just after the first drafts of the human
`genome were published, the NLM started
`systematically adding ‘gene reference into
`function’, or GeneRIF, tags to papers1. It has
`extended that annotation back to the 1960s,
`sometimes using other databases to help fill in
`the details. It is not a perfectly curated record.
`“In general, the data set is somewhat noisy,”
`says Terence Murphy, a staff scientist at the
`NLM in Bethesda, Maryland. There’s prob-
`ably some sampling bias for papers published
`before 2002, he warns. That means that some
`genes are over-represented and a few may
`be erroneously missing. “But it’s not awful,”
`Murphy says. “As you aggregate over multi-
`ple genes, that potentially reduces some of
`these biases.”
`With that caveat noted, the PubMed records
`reveal a few distinct historical periods in
`which gene-related papers tended to focus on
`particular hot topics (see ‘Fashionable genes
`through the years’). Before the mid-1980s, for
`
`2 3 N O V E M B E R 2 0 1 7 | V O L 5 5 1 | N A T U R E | 4 2 7
`©2017MacmillanPublishersLimited,partofSpringerNature.Allrightsreserved.
`
`Alnylam Exh. 1031
`
`

`

`SOURCE: PETER KERPEDJIEV/NCBI-NLM
`
`(cid:14)
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`(cid:14)0
`
`(cid:14)(cid:14)
`
`(cid:14)2
`
`CD4
`Encodes a T-cell
`receptor protein
`that is a target of
`HIV.
`
`HBB
`Encodes
`haemoglobin
`subunit beta, one
`of the two types of
`protein that join
`together to make
`adult
`haemoglobin.
`
`2
`TNF
`Encodes tumour
`necrosis factor, an
`immune molecule
`that has been a
`major drug target
`for in(cid:30)ammatory
`disease.
`
`4
`VEGFA
`Encodes vascular
`endothelial growth
`factor A, a protein
`that promotes the
`growth of blood
`vessels.
`
`6
`IL6
`Encodes
`interleukin 6,
`an immune
`molecule that
`can both
`stimulate and
`suppress
`in(cid:30)ammation.
`
`3
`EGFR
`Encodes
`epidermal growth
`factor receptor, a
`membrane-bound
`receptor protein
`often mutated in
`drug-resistant
`cancers.
`
`8
`MTHFR
`Encodes
`methylene-
`tetrahydrofolate
`reductase, an
`enzyme that helps
`to process amino
`acids.
`
`Gene position on the chromosome
`
`9
`ESR1
`Encodes oestrogen
`receptor 1, a
`nuclear receptor
`protein that has
`been a focus of
`study in breast,
`ovarian and
`endometrial
`cancers.
`
`TOP
`GENES
`
`PUBLICATION DATA OFFE(cid:18) A GLIMPSE
`INTO THE MOST STUDIED GENES OF
`ALL TIME AND OF ANY TIME.
`In 2002, the US National Library of Medicine (NLM) began
`annotating papers in its popular PubMed database of
`biomedical literature. Articles are tagged if they contain
`information about the structure, function or location of a
`speci(cid:29)c gene or gene product. The e(cid:31)ort has recorded
`1.2 million descriptions of 27,000 human genes —
`including RNA genes and pseudogenes — in about
`565,000 articles. These data reveal trends in genetics
`research, as well as the list of most studied human genes.
`
`BY ELIE DOLGIN
`
`DESIGN
`BY JASIEK
`KRZYSZTOFIAK
`
`Number of
`studies
`describing
`each gene
`
`4 2 8 | N A T U R E | V O L 5 5 1 | 2 3 N O V E M B E R 2 0 1 7
`©2017MacmillanPublishersLimited,partofSpringerNature.Allrightsreserved.
`
`

`

`FEATURE
`
`(cid:14)3
`
`(cid:14)4
`
`(cid:14)5
`
`(cid:14)6
`
`(cid:14)7
`
`(cid:14)8
`
`(cid:14)9
`
`20
`
`2(cid:14)
`
`22
`
`X
`
`Y
`
`7
`TGFB1
`Encodes transforming
`growth factor beta 1,
`an extracellular
`protein that controls
`cell proliferation and
`di(cid:31)erentiation.
`
`5
`APOE
`Encodes
`apolipoprotein E,
`which has
`important roles in
`cholesterol and
`lipoprotein
`metabolism.
`
`1
`TP53
`Encodes the
`tumour-suppressor
`protein p53, which
`is mutated in up to
`half of all human
`cancers.
`
`GRB2
`Encodes
`growth factor
`receptor-bound
`protein 2, which
`connects
`membrane-bound
`receptors to
`internal signalling
`processes.
`
`10
`AKT1
`Encodes a signalling
`protein known as a
`kinase, which
`phosphorylates other
`proteins to activate
`them.
`
`THE
`TOP (cid:14)0
`
`The ten most studied genes of all time are
`described in more than 40,000 papers.
`
`FASHIONABLE GENES
`TH(cid:18)OUGH THE YEA(cid:18)S
`
`Di(cid:31)erent genes have dominated the research literature in di(cid:31)erent eras. The trends re(cid:30)ect new
`understanding about the development of genetic diseases such as sickle-cell anaemia (HBB),
`concerns about new infectious diseases (CD4), breakthroughs in cellular signalling (GRB2) and more.
`
`1 TP53
`2 TNF
`3 EGFR
`4 VEGFA
`5 APOE
`6 IL6
`7 TGFB1
`8 MTHFR
`9 ESR1
`10 AKT1
`
`8,479 citations
`
`5,314
`
`4,583
`
`4,059
`
`3,977
`
`3,930
`
`3,715
`
`3,256
`
`2,864
`
`2,791
`
`3
`
`2
`
`1
`
`0
`1980
`
`Proportion of citations by year (%)
`
`HBB
`
`CD4
`
`GRB2
`
`APOE
`
`TP53
`
`1984
`
`1988
`
`1992
`
`1996
`
`2000
`
`2004
`
`2008
`
`2012
`
`2016
`
`2 3 N O V E M B E R 2 0 1 7 | V O L 5 5 1 | N A T U R E | 4 2 9
`©2017MacmillanPublishersLimited,partofSpringerNature.Allrightsreserved.
`
`NEWS
`
`

`

`BEYOND HUMAN
`BEYOND
`HUMAN
`
`The US National Library of Medicine has
`tracked references to genes from dozens
`of species, including mice, flies and other
`important model organisms, as well as
`viruses. Looking at genes from all species,
`more than two-thirds of the 100 most studied
`genes over the past 50 years have been
`human (see ‘The gene menagerie’). But non-
`human genes do appear quite high on the list.
`Often, these have a clear link to human health,
`as with mouse versions of TP53, or env, a viral
`gene that encodes envelope proteins involved
`in gaining entry to a cell.
`Others became foundational to broader
`genetic studies. A gene from the fruit fly
`Drosophila melanogaster known simply
`as white has been the focus of about
`3,600 papers — dating back to when
`biologist Thomas Hunt Morgan, working
`
`at Columbia University in New York City,
`peered through a hand lens one day in
`1910 and saw a single male fly with white
`eyes instead of red11. Because its product
`causes an easily observable change in the
`fly, the white gene serves as a marker for
`scientists looking to map and manipulate
`the fly genome. It has been involved in
`many fundamental discoveries12, such as
`the demonstration that large stretches of
`DNA can be duplicated because of unequal
`exchange between matching chromosomes.
`The most popular non-human gene
`of all time is actually a spot in the mouse
`genome whose normal function remains
`poorly understood. Rosa26 comes from
`an experiment published13 in 1991, in
`which cell biologists Philippe Soriano and
`Glenn Friedrich used a virus to insert an
`engineered gene randomly into mouse
`embryonic stem cells. In one cell line,
`dubbed ROSA26, the engineered gene
`
`seemed to be active at all times and in
`nearly every cell type. The discovery served
`as a building block for the creation of tools
`to make and manipulate transgenic mice.
`“People starting using it like crazy,” recalls
`Soriano, who is now at the Icahn School of
`Medicine at Mount Sinai in New York City.
`So far, the genetic locus known as Rosa26
`has been involved in some 6,500 functional
`studies. It is second only to TP53. E.D.
`
`THE GENE MENAGERIE
`Of about 1.3 million publications pertaining to genes
`in any species, nearly half are on human genes.
`
`1,268,788 PUBLICATIONS
`
`Human
`565,357
`
`Mouse
`281,400
`
`Fruit (cid:31)y
`42,145
`
`Rat
`146,586
`
`Other
`233,300
`
`SOURCE: PETER KERPEDJIEV/NCBI-NLM
`
`Lifson, director of the AIDS and Cancer Virus Program at the US
`National Cancer Institute in Frederick, Maryland.
`An even bigger part of CD4’s popularity had to do with basic
`immunology. In 1986, researchers realized that CD4-expressing T cells
`could be subdivided into two distinct populations — one that elimi-
`nates cell-infecting bacteria and viruses, and one that guards against
`parasites such as worms, which cause illness without invading cells.
`“It was a fairly exciting time, because we really understood very little,”
`says Dan Littman, an immunologist at the New York University School
`of Medicine. Just the year before, he had helped to clone the DNA that
`encodes CD4 and insert it into bacteria5, so that vast quantities of the
`protein could be made for research.
`A decade later, Littman also co-led one of three teams to show6 that to
`enter cells, HIV uses another receptor alongside CD4: a protein identi-
`fied as CCR5. These, and a second co-receptor called CXCR4, have
`remained the focus of intensive, global HIV research ever since, with
`the goal — as-yet unfulfilled — of blocking the virus’s entry into cells.
`
`FIFTEEN MINUTES OF FAME
`By the early 1990s, TP53 was already ascendant. But before it climbed
`to the top of the human gene ladder, there were a few years in which a
`lesser-known gene called GRB2 was in the spotlight.
`At the time, researchers were starting to identify the specific protein
`interactions involved in cell communication. Thanks to pioneering
`work by cell biologist Tony Pawson, scientists knew that some small
`intracellular proteins contained a module called SH2, which could
`bind to activated proteins at the surface of cells and relay a signal to
`the nucleus.
`In 1992, Joseph Schlessinger, a biochemist at the Yale University
`School of Medicine in New Haven, Connecticut, showed7 that the pro-
`tein encoded by GRB2 — growth factor receptor-bound protein 2 — was
`that relay point. It contains an SH2 module as well as two domains that
`activate proteins involved in cell growth and survival. “It’s a molecular
`matchmaker,” Schlessinger says.
`Other researchers soon filled in the gaps, opening a field of study
`in signal transduction. And although many other building blocks of
`cell signalling were soon unearthed — ultimately leading to treatments
`for cancer, autoimmune disorders, diabetes and heart disease — GRB2
`stayed at the forefront and was the top-referenced gene for three years
`in the late 1990s.
`In part, that was because GRB2 “was the first physical connection
`between two parts of the signal-transduction cascade”, says Peter van
`
`example, much genetic research centred on haemoglobin, the oxygen-
`carrying molecule found in red blood cells. More than 10% of all studies
`on human genetics before 1985 were about haemoglobin in some way.
`At the time, researchers were still building on the early work of Linus
`Pauling and Vernon Ingram, trailblazing biochemists who pioneered the
`study of disease at a molecular level with discoveries in the 1940s and
`1950s of how abnormal haemoglobin caused sickle-cell disease. Molecular
`biologist Max Perutz, who won a share in the 1962 Nobel Prize in Chem-
`istry for his 3D map of haemoglobin’s structure, continued to explore how
`the protein’s shape related to its function for decades afterwards.
`According to Alan Schechter, a physician-scientist and senior
`historical consultant at the US National Institutes of Health in Bethesda,
`the haemoglobin genes — more than any others at the time — offered “an
`entryway to understanding and perhaps treating a molecular disease”.
`A sickle-cell researcher himself, Schechter says that such genes were
`a focus of conversation both at major genetics meetings and at blood-
`disease meetings in the 1970s and early 1980s. But as researchers gained
`access to new technologies for sequencing and manipulating DNA,
`they started to move on to other genes and diseases, including a then-
`mysterious infection that was predominantly striking down gay men.
`Even before the 1983 discovery that HIV was the cause of AIDS, clini-
`cal immunologists such as David Klatzmann had noticed a peculiar
`pattern among people with the illness. “I was just struck by the fact that
`these people had no T4 cells,” recalls Klatzmann, who is now at Pierre
`and Marie Curie University in Paris. He showed2 in cell-culture experi-
`ments that HIV seemed to selectively infect and destroy these cells, a
`subset of the immune system’s T cells. The question was: how was the
`virus getting into the cell?
`Klatzmann reasoned that the surface protein (later called CD4)
`that immunologists used to define this set of cells might also serve as
`the receptor through which HIV entered the cell. He was right, as he
`reported3 in a study published in December 1984, alongside a similar
`paper4 from molecular virologist Robin Weiss, then at the Institute of
`Cancer Research in London, and his colleagues.
`Within three years, CD4 was the top gene in the biomedical literature.
`It remained so from 1987 to 1996, a period in which it accounted for
`1–2% of all the tags tallied by the NLM.
`That attention stemmed in part from efforts to tackle the emerging
`AIDS crisis. In the late 1980s, for example, several companies dabbled
`with the idea of engineering therapeutic forms of the CD4 protein that
`could mop up HIV particles before they infected healthy cells. But
`results from small human trials proved “underwhelming”, says Jeffrey
`
`4 3 0 | N A T U R E | V O L 5 5 1 | 2 3 N O V E M B E R 2 0 1 7
`©2017MacmillanPublishersLimited,partofSpringerNature.Allrightsreserved.
`
`

`

`der Geer, a biochemist at San Diego State University in California.
`Furthermore, “it’s involved in so many different aspects of cellular
`regulation”.
`GRB2 is something of an outlier in the most-studied list. It’s not a
`direct cause of disease; nor is it a drug target, which perhaps explains
`why its moment in the sun was fleeting. “You have some rising stars
`that fall down very quickly because they have no clinical value,” says
`Thierry Soussi, a long-time TP53 researcher at the Karolinska Institute
`in Stockholm and Pierre and Marie Curie University. Genes with staying
`power usually show some sort of therapeutic potential that attracts
`funding agencies’ support. “It’s always like that,” Soussi says. “The
`importance of a gene is linked to its clinical value.”
`It can also be linked to certain properties of the gene, such as the levels
`at which it is expressed, how much it varies between populations and the
`characteristics of its structure. That’s according to an analysis by Thomas
`Stoeger, a systems biologist at Northwestern University in Evanston, Illi-
`nois, who reported this month at a symposium in Heidelberg, Germany,
`that he could predict which genes would garner the most attention, simply
`by plugging such attributes into an algorithm.
`“THE
`Stoeger thinks that the reasons for these
`associations largely boil down to what he calls
`IMPORTANCE
`discoverability. The popular genes happened to
`OF A GENE IS
`be in hot areas of biology and could be probed
`LINKED TO
`with the tools available at the time. “It’s easier
`to study some things than others,” says Stoeger
`ITS CLINICAL
`— and that’s a problem, because vast numbers
`VALUE.”
`of genes remain uncharacterized and under-
`explored, leaving major gaps in the understand-
`ing of human health and disease.
`Curry also points to “intertwined technical, social and economic
`factors” shaped by politicians, drugmakers and patient advocates.
`
`RIGHT PLACE, RIGHT TIME
`Stoeger has also tracked how the general features of popular genes have
`changed over time. He found, for example, that in the 1980s, researchers
`focused largely on genes whose protein products were found outside
`cells. That’s probably because these proteins were easiest to isolate and
`study. Only more recently did attention shift towards genes whose
`products are found inside the cell.
`That shift happened alongside the publication of the human genome,
`says Stoeger. The advance would have opened up a larger percentage
`of genes to enquiry.
`Many of the most explored genes, however, don’t fit these larger
`trends. The p53 protein, for example, is active inside the nucleus. Yet
`TP53 became the most studied gene around 2000. It, like many of the
`genes that came to dominate biological research, was not properly
`understood after its initial discovery — which may explain why it took
`several decades after the 1979 characterization of the protein for the
`gene to rise to the top spot in the literature.
`At first, the cancer-research community mistook it for an oncogene —
`one that, when mutated, drives the development of cancer. It wasn’t until
`1989 that Suzanne Baker, a graduate student in Vogelstein’s lab, showed8
`that it was actually a tumour suppressor. Only then did functional stud-
`ies of the gene really begin to pick up steam. “You can see from the spike
`in publications that go up essentially at that point that there were a lot of
`people who were really very interested,” says Baker, now a brain-tumour
`researcher at the St. Jude Children’s Research Hospital in Memphis,
`Tennessee.
`Research into human cancer also brought scientists to TNF, the
`runner-up to TP53 as the most-referenced human gene of all time, with
`more than 5,300 citations in the NLM data (see ‘The top 10’). It encodes
`a protein — tumour necrosis factor — named in 1975 because of its
`ability to kill cancer cells. But anticancer action proved not to be TNF’s
`main function. Therapeutic forms of the TNF protein were highly toxic
`when tested in people.
`The gene turned out to be a mediator of inflammation; its effect on
`
`tumours was secondary. Once that became clear in the mid-1980s,
`attention quickly shifted to testing antibodies that block its action.
`Now, anti-TNF therapies are mainstays of treatment for inflammatory
`dis orders such as rheumatoid arthritis, collectively pulling in tens of
`billions of dollars in annual sales worldwide.
`“This is an example where the knowledge of the gene and the gene
`product has relatively quickly changed the health of the world,” says
`Kevin Tracey, a neurosurgeon and immunologist at the Feinstein
`Institute for Medical Research in Manhasset, New York.
`TP53’s dominance was briefly interrupted by another gene, APOE.
`First described in the mid-1970s as a transporter involved in clearing
`cholesterol from the blood, the APOE protein was “seriously consid-
`ered” as a lipid-lowering treatment for preventing heart disease, says
`Robert Mahley, a pioneer in the field at the University of California, San
`Francisco, who tested the approach in rabbits9.
`Ultimately, the creation of statins in the late 1980s doomed this strategy
`to the dustbin of pharmaceutical history. But then, neuroscientist Allen
`Roses and his colleagues found the APOE protein bound up in the
`sticky brain plaques of people with Alzheimer’s
`disease. They showed10 in 1993 that one particu-
`lar form of the gene, APOE4, was associated with
`a greatly increased risk of the disease.
`This generated much wider interest in the
`gene. Still, it took time to move up the most-
`studied chart. “The reception was very cool,”
`recalls Ann Saunders, a neurogeneticist and
`chief executive of Zinfandel Pharmaceuticals
`in Chapel Hill, North Carolina, who collabo-
`rated with Roses, her late husband. The amyloid hypothesis, which states
`that build-up of a protein fragment called amyloid-β is responsible for
`the disease, was all the rage in the Alzheimer’s-research community at
`the time. And few researchers seemed interested in finding out what a
`cholesterol-transport protein had to do with the disease. But the genetic
`link between APOE4 and Alzheimer’s risk proved “irrefutable”, Mahley
`says, and in 2001, APOE briefly overtook TP53. It remains in the all-time
`top five, at least for humans (see ‘Beyond human’).
`Like other popular genes, APOE is well studied because it’s central
`to one of the biggest unsolved health problems of the day. But it’s also
`important because anti-amyloid therapies have mostly flamed out in
`clinical testing. “I hate saying this, but what helped me were the failed
`trials,” says Mahley, who this year raised US$63 million for his com-
`pany E-Scape Bio to develop drugs that target the APOE4 protein.
`Those failures, he says, forced industry and funding agencies to rethink
`therapeutic strategies for tackling Alzheimer’s.
`There’s the rub: it takes a certain confluence of biology, societal
`pressure, business opportunity and medical need for any gene to
`become more studied than any other. But once it has made it to the
`upper echelons, there’s a “level of conservatism”, says Gregory Radick,
`a science historian at the University of Leeds, UK, “with certain genes
`emerging as safe bets and then persisting until conditions change”.
`The question now is how conditions might change. What new
`discoveries might send a new gene up the chart — and knock today’s
`top genes off their pedestal? ■
`
`Elie Dolgin is a science writer in Somerville, Massachusetts.
`
`1. Mitchell, J. A. et al. AMIA Annu. Symp. Proc. 2003, 460–464 (2003).
`2. Klatzmann, D. et al. Science 225, 59–63 (1984).
`3. Klatzmann, D. et al. Nature 312, 767–768 (1984).
`4. Dalgleish, A. G. et al. Nature 312, 763–767 (1984).
`5. Maddon, P. J. et al. Cell 42, 93–104 (1985).
`6. Deng, H. et al. Nature 381, 661–666 (1996).
`7. Lowenstein, E. J. et al. Cell 70, 431–442 (1992).
`8. Baker, S. J. et al. Science 244, 217–221 (1989).
`9. Mahley, R. W. et al. J. Clin. Invest. 83, 2125–2130 (1989).
`10. Strittmatter, W. J. et al. Proc. Natl Acad. Sci. USA 90, 1977–1981 (1993).
`11. Morgan, T. H. Science 32, 120–122 (1910).
`12. Green, M. M. Genetics 184, 3–7 (2010).
`13. Friedrich, G. & Soriano, P. Genes Dev. 5, 1513–1523 (1991).
`
`2 3 N O V E M B E R 2 0 1 7 | V O L 5 5 1 | N A T U R E | 4 3 1
`©2017MacmillanPublishersLimited,partofSpringerNature.Allrightsreserved.
`
`FEATURE NEWS
`
`