My work in innate immunity began in the early 1980s, during which time I identified and purified a molecule that we now know as tumor necrosis factor (TNF). I also was the first to show that TNF is an inflammatory protein. My interest in TNF turned then to what caused its production. I knew that TNF was made in large quantities in response to infection and that it could mediate many of the effects of an infection. The most powerful stimulus for TNF induction was lipopolysaccharide (LPS), and I began to wonder what the receptor for LPS was. I became obsessed with finding this receptor and that led directly to the 1998 paper in Science.

“There are now known to be 12 Toll-like receptors in mice; 10 in humans, and together they protect mammals very strongly against infection.”

In 1985, although, to be precise, what I reported was the purification of mouse TNF. Human TNF was purified independently by workers at Genentech and they completed that a little before I did. It was also in 1985 that I showed that TNF is able to mediate the lethal effect of LPS in mice. That was strongly controversial at the time. People conceived of TNF as a non-toxic factor for treating cancer, when, to the contrary, I said it was really very toxic and does much of what LPS can do.

LPS was originally known by the term "endotoxin," a word coined around 1890 by Richard Pfeiffer, who was a colleague of Robert Koch. These researchers had discovered something intrinsically toxic about gram-negative bacteria; Pfeiffer referred to this toxic essence as endotoxin. People then managed over the decades to show that endotoxin was equivalent to LPS, an important structural component of all gram-negative bacteria.

For some reason, LPS is extremely toxic to animals and humans. Inject it into either and they develop fever and shock and look very much as though they have a real infection. At least a large part of that effect has to do with the ability of LPS to induce TNF, which is a toxic cytokine. So TNF is the mediator of many of the effects of LPS. Yet it was also shown that animals that can’t respond to LPS as the result of a mutation in the LPS receptor are highly vulnerable to gram-negative infections. This established that sensing LPS is essential to an effective immune response. The big question then came to center on how we detect LPS. In effect, how do we really know when we have an infection? How do we know when we’re sick? How do we discriminate self from non-self?

Yes. I tried various approaches. There had been a mouse known since the 1960s that could survive any amount of LPS you could give it. That mouse did not make TNF in response to LPS, and in fact, did not show any detectable response of any kind if you injected it with pure LPS. People suspected that it had a mutation in the LPS receptor. I began working with that mouse extensively.

I first tried to use biochemical methods to find the LPS receptor and that got me nowhere. Then I tried expression cDNA cloning and that also went nowhere. Finally, in 1993, with a growing density of markers in the mouse genome, I decided to try to positionally clone the mutation.

We mapped the mutation to very high resolution and proved it was in a very small region of mouse chromosome 4. We then cloned all the genomic DNA across that region, and began looking for genes. Finally we found a candidate gene that was homologous to the interleukin-1 receptor. It was also similar to—and named after—a Drosophila protein called Toll. And before I go on I should tell you a little bit about Toll in the fly.

Dr. Bruce Beutler's most-cited paper with 2,424 cites to date: Poltorak A, et al., "Defective LPS signaling in C3H/HEJ and C57BL/10SCCR mice: mutations in TLR4 gene," Science 282(5396): 2085-8, 11 December 1998. Source: Essential Science Indicators. Related:   Dr. Bruce Beutler is featured in ISIHighlyCited.com

The Drosophila Toll story was extremely exciting to us at time. In the fly, the Toll protein had been known since the mid-1980s to participate in development and to be involved in differentiation of dorsal structures from ventral structures that ultimately form different elements in the adult fly. The Nobel Prize was awarded to Christiane Nüsslein-Volhard for her remarkable genetic studies in this area. .

Of great importance, in 1996, Jules Hoffmann and his colleagues in Strasbourg showed that Toll is required in adult flies to fight off fungal infections. If flies have a mutation in Toll or its endogenous protein ligand, then the flies will be wiped out if they are injected with fungal spores. And so the Toll protein is used by flies to detect a fungal infection, while in mice, we had identified a paralogue of the family that was used to detect a gram-negative infection.

This was the paper that told what Toll-like receptors do. This paper clearly showed that TLR4 was a non-redundant sensor for LPS, and from that, it could be inferred that other TLRs probably detect other conserved molecules of microbial origin. LPS had been the object of keen interest for over 100 years—people had known a receptor for LPS must exist, but nobody had known what it was. We cloned it and opened the door to understanding much more about microbial sensing. That’s what this paper was all about and why it has been so highly cited.

As it turned out, TLR4 is one of a family of very similar proteins with similar functions: another member of the family detects lipopeptides, which is something many different microbes make. Another member of the family detects double stranded RNA; another member detects unmethylated DNA; another member detects flagellin, which is the protein in the whip-like structures that bacteria use to propel themselves through the media they’re in. Similar versions of flagellin are very common among both gram-positive and gram-negative bacteria and evidently we have evolved a TLR to detect it. Another member of the family detects single-stranded RNA.

There are now known to be 12 Toll-like receptors in mice; 10 in humans, and together they protect mammals very strongly against infection. They’re probably the main form of protection that we have against viral, bacterial, and fungal infection. They’re not totally unique in that respect, but we now know that if mice don’t have TLR signaling, they are severely compromised. So this is how we become aware of infection. The first events in sensing microbes had been a fundamental question in immunology.

Well, we always knew what we were after. The process of positional cloning is one in which you more and more tightly confine a mutation to a part of a chromosome. We knew that eventually we would find the mutation, so there was nothing serendipitous about that. Although I guess you could say that the fact that those mice existed was serendipitous. The phenotype of LPS unresponsiveness was caused by a spontaneous mutation—someone just happened to notice in the 1960s that C3H/HeJ strain mice don’t respond to endotoxin, knew that this was an important observation, and published their findings.

At the time, we were surprised to see how very similar the strategies used by mammals and insects for host defense were. If you were a Drosophila person at the time and you were wondering what the TLRs do in mammals, you might have guessed that they were involved either in development or in immunity, but you wouldn’t make a very specific guess about what any of them did. Our discovery told us that they really have very, very similar functions, although the mammal has no need for TLRs in development—that’s a unique requirement in the fly.

It’s been a real explosion in immunology. For starters, innate immunity gained immense credibility from this discovery. And now I think people are heading in several different directions. Some people are interested in the possibility that autoimmune diseases and chronic inflammatory diseases in general might be caused by signaling through this family of TLRs. Lots of companies are spending considerable effort trying to either neutralize TLRs. That’s a practical application that may have enormous fallout.

Another possible application might be to use TLRs to drive immune responses when you want them to occur. Some molecules like LPS have long been known to be adjuvants—that is, they help to drive an adaptive response to an antigen; they help to make antibodies. We’ve done a fair amount of work in that area; we know that TLRs can indeed signal an adaptive response. We also know that they’re not the only way to get an adaptive response—there might be some ways that prove to be even better.

In hindsight, we see that certain useful drugs actually work by stimulating TLRs. I am thinking of a drug called imiquimod that has been around for a long time—the trade name is Aldara. This drug is quite effective at treating warts, but as more recent clinical experience has shown, it’s also good for treating basal cell carcinomas. We didn’t know the mechanism of action until quite recently. As Shizuo Akira and his colleagues showed, imiquimod turned out to be a ligand for TLR7, which normally senses single-stranded RNA. This suggests that one might potentially harness TLR signaling to treat other kinds of cancer as well.

On a more basic level, there are still a lot of questions about just how TLRs signal, and a lot of people are still working on that.

I have become enamored of the forward genetic approach, through which we made the first assignment of function to a mammalian TLR. You start with a phenotype—for instance, a mouse that doesn’t respond to endotoxin—and you track down the mutation that is responsible. We’ve continued to do that since about 2000. But we don’t rely upon spontaneous phenotypes; what we do is mutagenize the mice at random. We screen them for defects of TLR signaling or other aspects of immunity. We might look, say, for mice that can’t handle an ordinarily trivial virus or a mouse that’s highly resistant to a particular pathogen. Any time we find such a phenotype, we track it down and find the cause.

The key is that things are much faster in positional cloning than they used to be. Where it took five years to clone one mutation a decade ago, we can now clone 20 or 30 mutations in a year. In this way, we’ve found new molecules involved in TLR signaling pathways and we’ve found many surprising genes that are required for mice to combat viral infections. We’ve worked very closely on this with Jules Hoffmann and Shizuo Akira. We share with them a grant that is dedicated to about the study of host defense in viral infection.

I’ll preface my answer by saying that we do live in a golden age of genetics and, in fact, we are limited only by resources. The long-term goal is to find out what all genes do. We focus on the area of innate immunity and we would like to pinpoint all the genes required for a robust immune response, and understand how they operate to give us resistance to infection. We’re getting there. We could get there faster if we had infinite resources.

I think the day will be here fairly soon when we can see a strange phenotype in the morning, sequence the genome of that mouse for a few dollars, know the sequence by the afternoon, and then identify the mutation responsible for that phenotype. That time is coming.

I guess if I had infinite resources, I would try to make it come sooner. When that day arrives, we’ll soon know the essential function of any and every gene required for an immune response. Then we should be able to figure out, although not quite so easily, how they all fit together.

Bruce Beutler, M.D.
The Scripps Research Institute
La Jolla, CA, USA