More Recent Comments

Wednesday, September 11, 2019

Gerald Fink promotes a new definition of a gene

This is the 2019 Killian lecture at MIT, delivered in April 2019 by Gerald Fink. Fink is an eminent scientist who has done excellent work on the molecular biology of yeast. He was director of the prestigious Whitehead Institute at MIT from 1990-2001. With those credentials you would expect to watch a well-informed presentation of the latest discoveries in molecular genetics. Wouldn't you?

It's worth watching the video because it gives us some insight into a troubling problem in the field of molecular biology/molecular genetics.1 The problem seems to be a lack of rigor and a lack of critical thinking when it comes to some fundamental issues. In this case, it's how we think about genes and junk DNA.

Here's how his lecture is described on the MIT website [The evolving definition of a gene].
More than 50 years ago, scientists came up with a definition for the gene: a sequence of DNA that is copied into RNA, which is used as a blueprint for assembling a protein.

In recent years, however, with the discovery of ever more DNA sequences that play key roles in gene expression without being translated into proteins, this simple definition needed revision, according to Gerald Fink, the Margaret and Herman Sokol Professor in Biomedical Research and American Cancer Society Professor of Genetics in MIT’s Department of Biology.

Fink, a pioneer in the field of genetics, discussed the evolution of this definition during yesterday’s James R. Killian Jr. Faculty Achievement Award Lecture, titled, “What is a Gene?”

“In genetics, we’ve lost a simple definition of the gene — a definition that lasted over 50 years,” he said. “But loss of the definition has spawned whole new fields trying to understand the unknown information in non-protein-coding DNA.” ...

At that time, scientists were operating with a straightforward definition of the gene, based on the “central dogma” of biology: DNA makes RNA, and RNA makes proteins. Therefore, a gene was defined as a sequence of DNA that could code for a protein. This was convenient because it allowed computers to be programmed to search the genome for genes by looking for specific DNA sequences bracketed by codons that indicate the starting and stopping points of a gene.

In recent decades, scientists have done just that, identifying about 20,000 protein-coding genes in the human genome. They have also discovered genetic mechanisms involved in thousands of human diseases. Using new tools such as CRISPR, which enables genome editing, cures for such diseases may soon be available, Fink believes.

“The definition of a gene as a DNA sequence that codes for a protein, coupled with the sequencing of the human genome, has revolutionized molecular medicine,” he said. “Genome sequencing, along with computational power to compare and analyze genomes, has led to important insights into basic science and disease.”

However, he pointed out, protein-coding genes account for just 2 percent of the entire human genome. What about the rest of it? Scientists have traditionally referred to the remaining 98 percent as “junk DNA” that has no useful function.

In the 1980s, Fink began to suspect that this junk DNA was not as useless as had been believed. He and others discovered that in yeast, certain segments of DNA could “jump” from one location to another, and that these segments appeared to regulate the expression of whatever genes were nearby. This phenomenon was later observed in human cells as well.

“That alerted me and others to the fact that ‘junk DNA’ might be making RNA but not proteins,” Fink said.

Since then, scientists have discovered many types of non-protein-coding RNA molecules, including microRNAs, which can block the production of proteins, and long non-coding RNAs (lncRNAs), which have many roles in gene regulation.

“In the last 15 years, it has been found that these are critical for controlling the gene expression of protein-coding genes,” Fink said. “We’re only now beginning to visualize the importance of this formerly invisible part of the genome.”

Such discoveries demonstrate that the traditional definition of a gene is inadequate to encompass all of the information stored in the genome, he said.

“The existence of these diverse classes of RNA is evidence that there is no single physical and functional unit of heredity that we can call the gene,” he said. “Rather, the genome contains many different categories of informational units, each of which may be considered a gene.”
This is a pretty good summary of what Fink said in his talk. You can watch it yourself to confirm this. He repeats the common myth that DNA → RNA → protein is the Central Dogma and the myth that this was thought to be the only function of DNA: hence genes make proteins [see Basic Concepts: The Central Dogma of Molecular Biology] [Thinking critically about the Central Dogma of Molecular Biology]. Then, at exactly one hour into his lecture, he comes up with his own re-definition of a gene based on very recent work on functional RNAs. Now, he says, we have to think of a gene as "a DNA sequence that is transcribed into an RNA molecule with a function."

This is remarkably similar to the definition that I have been using, and teaching to undergraduates, for forty years. That definition is: "A gene is a DNA sequence that is transcribed to produce a functional product" [What Is a Gene?]. I didn't invent that definition—it's been in many textbooks for half-a-century.2 In fact, that definition is in some of the textbooks used in undergraduate courses at various universities in the Boston area, including MIT.

Imagine that Gerald Fink had given a different lecture. Here's what he could have said ...
In preparing for this lecture I read the literature on the Central Dogma of Molecular Biology and on defining a gene. I have discovered, much to my embarrassment, that my concept of the Central Dogma was wrong. The real Central Dogma, according to Crick, is:
The central dogma of molecular biology deals with the detailed residue-by-residue transfer of sequential information. It states that such information cannot be transferred from protein to either protein or nucleic acid. (F.H.C. Crick, 1970)
For many decades I believed that the Watson version of the Central Dogma (DNA makes RNA makes protein) was the correct version. Furthermore, I believed that this was practically the only thing that DNA did so the correct definition of a gene is that a gene is responsible for making protein. I now realize that neither Crick nor Watson, nor many other eminent scientists at the time, ever believed in such a definition. They all knew that genes could also produced functional RNAs other than mRNA. I was wrong to assume for the past 50 years that protein-coding genes are the only kind of genes.

Furthermore, for many decades, I have been under the mistaken impression that all noncoding DNA is junk—or at least I thought that was the consensus among experts. I now realize that that too, is wrong. Very few real experts ever thought that all noncoding DNA was junk.
Why isn't this the talk that Gerry Fink gave? Why didn't he check the literature before giving an important lecture on the definition of a gene? Why is it that none of his colleagues, students, or post-docs ever raised questions over the past several decades about the misconceptions that were (are?) rife within the Whitehead Institute?

Some of my colleagues (hi, Alex!) think it's wrong of me to ask these questions because they sound like personal attacks. Maybe they are, but that's not the point. The real point is how did we get to a position where the most prestigious scientists at a place like MIT have never questioned their fundamental assumptions? That's not how science is supposed to work. How are we going to fix this problem if we continue to ignore it?

There's one other issue that I want to bring up. Gerald Fink's group recently published a paper on the function of some introns in yeast. If you listen to his talk, beginning at 55 minutes, you'll see that he interprets his work to mean that all, or most, introns have a function. In other words, "introns are not junk." But that's extremely misleading. In fact, when you put his paper into the proper context, the evidence shows conclusively that the vast majority of yeast introns are junk! [Yeast loses its introns] His group discovered the exception that proves the rule!

I do not understand how Fink could misinterpret and misrepresent his own work so badly. If they believe him, his audience will now have a completely false view of introns and junk DNA. Other scientists are now forced to spend considerable time and effort correcting these misconceptions—a task that is much harder when you are challenging prestigious MIT scientists. Am I the only one who finds this upsetting?

1. This problem seems to be pervasive in all of science but I'm not knowledgeable enough about other fields to be certain.

2. I don't deny that many textbooks also state that a gene is a DNA sequence that encodes a protein even though they may also talk about ribosomal and tRNA genes. However, I expect scientists to be capable of recognizing which textbook definition is more accurate. My 1989 textbook defines a gene as a DNA sequence that's transcribed and it contains the following description of eukaryotic RNA polymerases; "RNA polymerase transcribes class I genes, those that code for large ribosomal RNA; RNA polymerase II transcribes class II genes, those that code for proteins and a few that code for small RNA molecules; and RNA polymerase III transcribes class III genes, those that encode a number of small RNA molecules, including tRNA and 5S RNA." It's clear that 30 years ago we were very familiar with genes for various small RNAs such as snRNA, snoRNA, and various others. The idea that there are noncoding genes is not a 21st century discovery as Fink implies.


  1. He finds it remarkable that evoltions effects on genes has been so conservative.
    A mouse and man very alike.
    A reason for this is likely that genes are based on a blueprint and there has been no evolution affecting genes as thee should be.
    He was funny and interesting. I understand he invented using yeast to help vaccines and medicine but i think others actually did those things.
    However the use of yeast seems to be a real accomplishment as I understand it.
    I didn't know it was questioned what a gene is.
    i think a gene is a tiny memory byte. Just a tiny reflection of our own working memory.
    Probably a trivial insight of mine.

  2. I wonder how much of this is due to the desire to have one's own work be very, very important. Find one functional intron, that's nice. But wouldn't it be more exciting to generalize that to all introns? Hype sells.

    1. We all want to think that our work is very, very important and in most cases it is! In Fink's case, his discovery of an important function for some excised introns is novel and exciting. However, the problem I'm addressing is deeper than that. It's similar to the inability of the ENCODE workers to put their results into the proper context.

      I just don't understand why there are so many prominent scientists who have never thought very deeply about the problem they are investigating. Did Fink really think that the experts in genome studies and molecular evolution actually thought that all noncoding DNA was junk? Did nobody in his lab, or at any of his seminars, ever correct him? Why not? Is this kind of superficial dogma pervasive at places like MIT and the Whitehead?

    2. It's hard to believe that Fink had never heard of rRNA or tRNA. Did he think they didn't count? If so, why? It's a mystery.

    3. He makes a big deal of the human genome project and the idea that computers could only detect protein-coding genes. Here's a quote from the International Human Genome Consortium paper on the draft sequence published in Nature in February 2001.

      "Although biologists often speak of a tight coupling between 'genes and their encoded protein products,' it is important to remember that thousands of human genes produce noncoding RNAs (ncRNAs) as their ultimate products. There are several major classes of ncRNA. (1) Transfer RNAs (tRNAs) ... (2) ribosomal RNAs ... (3) small nucleolar RNAs (snoRNAs) ... (4) small nuclear RNAs (snRNAs).

      Other ncRNAs include both RNAs of known biochemical function (such as telomerse RNA and the 7SL signal recognition particle RNA) and ncRNAs of enigmatic function (such as the Xist RNAs implicated is X dosage compensation, or the small vault RNAs ...).

      ... novel ncRNAs cannot be readily found by computational gene-finding techniques .... We can, however, identify genomic sequences that are homologous to known ncRNA genes using BLASTN or, in some cases, more specialized methods."

      The Consortium then identifies 497 tRNA genes, dozens of copies of each of the three ribosomal RNA genes, 97 snoRNA genes, 21 snRNA genes, 3 genes for 7SL RNA, 1 gene for RNAse P, 1 gene for telomerse RNA, and several dozed other genes for ncRNAs. They are listed in Table 20 in the paper.

      One wonders whether Fink or any of his students and colleagues ever read the paper.

  3. I prefer to use "locus" rather than "gene" when teaching genetics. As Larry describes well, the term "gene" carries alot of historical baggage, including, at minimum, that the "gene" must be transcribed to have an effect on macro or molecular phenotypes. "Locus" bears no such baggage-- I use it to refer simply to a DNA sequence where polymorphisms exist that can have a discernable effect on phenotypes. For example, a mutation/polymorphism that alters a distant, protein-binding enhancer element could have substantial effects on phenotype but surely would not need to be transcribed-- even spuriously-- to exert that effect.

    1. I see where you're coming from but I still think we need to use the word "gene." It's kind of awkward to be talking about the locus for ribosomal RNA production or the locus for production of triose phosphate isomerase. I think we can distinguish between different loci such as genes and regulatory sequences or genes and centromeres.

      Also, your definition seems to be restricted to loci that have a discernible effect on phenoptype but traditionally a genetic locus can also refer to neutral alleles such as those used in DNA fingerprinting. A locus can also describe the position of a pseudogene or a fragment of a transposon.