The History, Basic Science and Biology of TNF

Special Report


Lionel B. Ivashkiv, MD

Lionel B. Ivashkiv, MD

Attending Physician, Hospital for Special Surgery
Associate Chief Scientific Officer, Hospital for Special Surgery
David H. Koch Chair in Arthritis and Tissue Degeneration
Professor, Medicine and Immunology, Weill Cornell Medical College
Director, HSS Genomics Center

  1. The Characteristics of TNF
  2. The History of  TNF
  3. The Biology of TNF
  4. Regulating TNF Expression
  5. What Does TNF Do to Activate Inflammation?
  6. TNF Function at the Cellular Level
  7. TNF Function in the Whole Organism
  8. How TNF Suppresses Autoimmunity

The Characteristics of TNF

TNF is one of the large number of cytokines, many of which are being implicated in the pathogenesis of rheumatic and inflammatory disorders and, like most cytokines, it is a protein that mediates communication between cells. The way that TNF works -- and all cytokines work -- is by binding to cell receptors with high affinity and therefore with high specificity. And I think that the high affinity of that interaction is really one of the mechanisms in which our ideas led to the development of soluble TNF receptors as a potential therapeutic agent for disease.

Something that is going to be a theme of my talk is that, also like most cytokines, TNF is pleiotropic, which means it can have very different effects on different cell types at different time courses during the evolution of an inflammatory disease. Many cytokines are known to promote inflammation, but there are also suppressive effects that TNF has on autoimmunity and that needs to be kept in mind when we consider the potential side effects of TNF blockade.

Finally, when work on cytokines in rheumatoid arthritis began in the early 80's, Mark Feldman and Ravinder Maini really faced a lot of skepticism, because it was thought at that time that there were so many cytokines, and the number of cytokines has continued to grow. Moreover, it was noted that the cytokines were redundant; so many different cytokines had the same biological action on the same cell type. It was really, I think, very courageous on their part to try to target cytokines as therapeutic targets, and it has been remarkably successful.

The way TNF is going to work is by regulating the activation, differentiation, and proliferation of cells which are important in inflammatory diseases. And also it helps to regulate their survival. Actually, TNF also has a potent activity in promoting apoptosis and killing cells. This also can be part of how it works as a double-edged sword, because it can actually promote the death of some of the inflammatory cells that it has activated.

The History of TNF

TNF has been implicated in numerous inflammatory and other immune diseases, and the list continues to expand. To try to trace back to the discovery of TNF over time, I think that I can go back to the 1860s, with the work of P. Bruns in France. What was noted at that time was that when patients became infected with bacteria, they often evidenced regression of tumors. This was picked up on by W. Coley, the third surgeon-in-chief at HSS, who actually extended this into attempting to treat cancers with bacterial extracts. In those attempts, he did meet with some success, even though there was toxicity as well, and these were known as Coley's toxins. I think this was one of the first approaches to treating tumors.

Then we have to move forward in time to the 1940s and 1950s, when further work defined that the component of the bacterial extract which was important for tumor regression was LPS, lipopolysaccharide, a component of the cell wall of gram negative bacteria. These investigators also detected that the mechanism of tumor regression was hemorrhagic necrosis.

From there the work began to evolve more rapidly, and by the 1960s and into the 1970s, involved work by several people, including Lloyd Old, who worked at Sloan Kettering Institute, one of the member institutes of our academic center. What these investigators discovered was that it wasn't that the LPS was directly causing tumor regression. LPS was working indirectly by inducing the expression of tumor necrosis factor. These results set off quite a bit of investigation to try to identify this factor. This was done when TNF was purified, cloned and then sequenced in the 1980s by several groups including those of Lloyd Old, David Goeddel, and Anthony Cerami, who was at that time at Rockefeller, also a member of this tri-institutional complex. So there is really a rich history of TNF research at this institution. Actually Tony Cerami's group had originally defined TNF as a bioactivity called cachectin. It was found in Africa, in cattle that were chronically infected with parasites, and TNF was a factor which mediated cachexia.

The Biology of TNF

We can go over a little bit of the specifics of the TNF molecule itself. TNF is a protein which is synthesized as a 26 kD precursor protein and a type 2 protein, meaning that it is inserted into the membrane with its carboxy terminus on the outside. The way that TNF secretion is regulated is that it is cleaved by an enzyme called TACE at the cell surface. And this enzyme will cleave TNF and release it as a bioactive trimer into the extra-cellular environment, where it can act on the same cell in an autocrine fashion or on adjacent cells in a paracrine fashion. It is actually unusual for there to be enough TNF to make it into the circulation, where it could act at a distance in an endocrine-type fashion.

TNF is expressed mostly by cells of the myeloid lineage, monocytes and macrophages, even though there are conditions of infections and also autoimmunity, where it can be expressed by other cell types as well, especially by T-cells and also natural killer cells, neutrophils, and other cells outside the immune system. T-cells can produce TNF beta, also termed lymphotoxin. There are many types of stimuli which can activate TNF expression, and we will go over this in a little bit more detail later.

Finally, TNF acts by binding to two different TNF receptors which are widely distributed on most cells throughout the body. These are receptor 1, also known as p55, and receptor 2, also known as p75. There is a little bit of controversy about the relative importance of these two different receptors, and also about the expression of TNF receptors; some people feel p75 is more highly expressed on hematopoietic cells. But when looked at a little bit more carefully, p75 is widely expressed as well as p55. So, both of these receptors will bind to TNF with high affinity, with some different biologic consequences.

It has been shown, especially when it is expressed on T-cells, that cell surface TNF alpha and especially TNF beta or lymphotoxin is a very potent activator of adjacent cells, even while it remains expressed on the cell surface. So here there is a potential mechanism for cell-cell contact as a way that TNF can activate cells.

Probably the more conventional way that we think of TNF activity is that it is cleaved from the cell surface and released as a trimer, which can then diffuse and stimulate receptors either on nearby cells or feed back and activate the macrophage even further. TNF binds to receptors on the surface cell, which generates a signal -- which will activate the cell and activate its immune and inflammatory effector functions. The TNF receptors again are known as p55 and p75, so they will have an extra-cellular region, which is the region that binds the TNF with high affinity, and also an intra-cellular region, which is important for generating the signals by which TNF activates cells and regulates their function.

It is interesting because it is not only TNF which is cleaved from the cell surface and can act on adjacent cells, but it has been described that these receptors also are cleaved from the cell surface, likely by the same enzymes which cleave TNF. These receptors can be released from the cell surface and exist in a soluble form. It was also shown that the soluble form of the TNF receptor is a naturally occurring sink which can bind up TNF in the circulation. This is really an endogenous mechanism for modulating and regulating TNF alpha. This system is exploited in the therapeutic approaches to blocking TNF, which I will tell you about later.

Regulating TNF Expression

So how is TNF expression regulated? Once again, the major producers of TNF alpha, especially in the joint in rheumatoid arthritis, are the monocytes and then the tissue macrophages, and the synthesis is stimulated by a large variety of agents. I think this is one of the issues in trying to treat rheumatic diseases. TNF is a great therapeutic target, and it would be nice to be able to turn off its production. One of the problems in trying to turn off its production is that there is a multitude of factors which could turn it on. This includes a variety of microbial products. Essentially, any type of bacteria or microbe can turn on the expression of the TNF gene.

Many cytokines will activate TNF production, including cytokines which themselves were activated by TNF. An example is interleukin-1. This has the potential to set up a positive feedback loop. T-cell surface molecules can induce TNF production, which is a mechanism by which T-cells and lymphocytes can drive TNF production. Some people have proposed this as an important way that TNF production is regulated during an autoimmune process. A large number of somewhat less specific stimuli including ischemia, trauma, radiation, and UV light can induce TNF production.

And the regulation of TNF expression is also very finely tuned at multiple levels. The first of these is transcriptional control of the gene. The TNF gene is usually not active in a quiescent cell, and transcription is turned on in response to a number of these stimuli. In general, at least in macrophages, this goes up to a 50-fold increase in transcriptional rate of the gene promoter. It has been well-defined that probably the most important transcription factor for regulating TNF production is NF kappa B, putting it very firmly on the list as a potential therapeutic target. Also, AP1 transcription factors regulate the TNF promotor. This is step one, but that is not all that happens.

TNF messenger RNA, once made, is unstable, and it is prone to being rapidly turned over. Under certain circumstances, the half-life of the TNF RNA can be very short -- on the order of 30 to 45 minutes. The RNA stability is also regulated, so the message can be stabilized and the pathways leading to the stabilization of the message involve the mitogen-activated protein kinase cascade, which is also activated by a large number of inflammatory stimuli, including most of those which I have mentioned above.

Finally, the TNF message, even though it is expressed, is not translated unless translation is activated. The translational efficiency of the TNF messenger RNA can be induced up to 100-fold. As far as it is known at this point, the major regulator is a kinase called p38, which is known as a stress kinase, a member of the larger family of these kinases.

So if we start to multiply these out just between the transcriptional control and regulation of translation, there is already the potential for 5000-fold induction in the amount of TNF production. On top of that you have to factor in the regulation of stability of message. In addition to turning TNF on, there are many mechanisms for turning off the expression of TNF. Among those are corticosteroids. Probably an important mechanism of action of corticosteroids is to down-regulate the production of TNF, especially transcription. Prostaglandins have a negative effect on the transcription of TNF and, finally, down-regulatory or immunosuppressive cytokines play a role. Interleukin-10 is probably the most potent factor in terms of turning off TNF production during the course of the normal immune response or a self-limited inflammatory response.

What Does TNF Do to Activate Inflammation?

One of the issues that I am interested in is why factors like IL-10 seem to be extremely inefficient in regulating TNF production during the course of a disease such as rheumatoid arthritis. So just to look at the molecular level, what are the things that TNF does which can potentiate inflammation? I tried to show the steps which are clearly pro-inflammatory, which include the induction of expression of chemokines, such as IL-8, which attract neutrophils to the sites of inflammation, and MCP-1, which is a strong chemokine for white macrophages, and other chemokines which will attract lymphocytes as well. The induction of proinflammatory cytokines such as IL-1, IL-6 and IL-18 will contribute to this inflammatory cascade. The induction of expression of adhesion molecules on endothelial cells and on leukocytes themselves will promote the trafficking of leukocytes from the blood into the site of the inflammation, and induction of protease expression, for example, collagenase and stromelysin, will contribute to tissue degeneration which occurs in rheumatoid arthritis. Activation of production of reactive oxygen and nitrogen intermediates with cytotoxic properties will contribute to tissue damage.

The activities of TNF could have both a pro- and an anti-inflammatory effect, including the induction of COX-2 and prostaglandin production. And I think that, at least to rheumatologists, prostaglandins are mostly thought of as pro-inflammatory factors. But it is known that, during a chronic inflammatory response, a variety of prostaglandins is produced, some of which have an anti-inflammatory and suppressive effect as well. There is also the ability of TNF to activate caspases, which leads to apoptosis of cells, and this can also be either pro-inflammatory or immunoregulatory. For example, if TNF activation begins killing normal cells of the tissue where the inflammatory response is going on, this obviously contributes to tissue damage. However, in contrast, if TNF begins to trigger the apoptosis of inflammatory cells, for example, lymphocytes or myeloid cells, then this will have a homeostatic effect. So, again, as with most cytokines, there is a good side and a bad side.

Another thing that TNF does is to inhibit lipoprotein lipase, which is a mechanism by which it induces cachexia. It will also induce the production of a variety of hormones, including epinephrine, and this is actually part of a feedback of a homeostatic mechanism as well. And finally, TNF also induces the expression of the IL-1 receptor antagonist and TGF beta, which are a blocker of IL-1 and an inhibitory cytokine. So it is again a double-edged sword and the pro-inflammatory effects appear to be dominant in many diseases. We really need to keep in mind the beneficial effects of TNF as well.

So how does the signaling cascade work? Basically TNF binds to a receptor on the cell surface. We show here the p55 receptor for TNF. This binding results in a conformational change in the homotrimeric receptor and the generation of a signal, and the signal is then communicated to the cell. And this results in the activation of these protein kinase cascades. This will go in one direction to the activation of a kinase that phosphorylates an inhibitor of NF kappa B. This results in the destruction of this inhibitor on NF-kappa B by the proteosome, with the resultant liberation of the NF-kappa B transcription factor, which will translocate to the nucleus, where it can turn on gene transcription. There is a variety of genes which are turned on, which, as I have mentioned before, include cytokines and cytokine receptors, the collagenases and other proteinases, growth factors, adhesion molecules, COX-2, nitric oxide and phospholipase. This is the predominant pro-inflammatory arm of TNF. But at the same time, within the p55 receptor there are these so-called death domains. And this is a domain that serves to recruit caspases to this receptor and can also result in apoptotic cell death. When TNF acts, there is a balance between these two different mechanisms such that, in the typical scenario, the activation of NF-kappa B, by mechanisms which are just being defined, induces expression of a molecule called Flip, which will block this apoptotic cascade, and therefore TNF will have mostly pro-inflammatory activity. However, under conditions where NF-kappa B is not able to suppress apoptosis, TNF is a very effective producer of cell death and potentially a down-regulator of immune responses.

TNF Function at the Cellular Level

So how does TNF function at the cellular level? As I mentioned, it activates cells to achieve immune effector function. I think a very important mechanism of how TNF works in inflammatory diseases is through the activation of endothelium, inducing expression of adhesion molecules and chemokines and thereby recruiting immune cells to inflammatory sites. And I think one of the important discoveries in the mechanisms of how drugs like etanercept work is that they prevent the migration of new cells into inflammatory sites.

TNF can cause toxicity, destruction and invasion of tissue, and it can promote Th1 responses by T-cells and cell proliferation. So, these again can be considered pro-inflammatory, and the induction of apoptosis would be homeostatic and potentially balancing.

TNF Function in the Whole Organism

So how does this work at the level of the whole organism? TNF has a lot of beneficial functions for the organism. It is very important, both in innate immunity and for the initial response to pathogens, but also in helping to promote Th1 immunity. Inflammation is mostly a beneficial process, for example, in wound healing and tumor surveillance. TNF is also important in the induction of the acute phase response and in regulating energy metabolism.

However, if there is too much TNF or TNF is inappropriately produced, it is implicated in the pathogenesis of a large number of inflammatory conditions. I will mention briefly rheumatoid arthritis and inflammatory bowel disease. But there is a large number diseases for which TNF blockade is an effective treatment, such as psoriasis and a number of genetic periodic fever syndromes, at least one of which has been shown to be caused by a mutation in the TNF receptor. Also, too much TNF may contribute to septic shock -- at least in animal models -- and also to cerebral malaria, which is thought to be related to a relative over-production of TNF.

However, there is a flip side to TNF, in that it can cause immunosuppression, especially during chronic autoimmune disease in animal models. But I think it is relevant if you think of some of the potential side effects of blocking TNF. The animal models where TNF has been shown to be suppressive include EAE (Experimental Autoimmune Encephalitis), which is a model of multiple sclerosis and the NOD (non-obese diabetic) mouse, the model of diabetes, and also models of systemic lupus.

I would like to start here with the cytokine network of TNF and try to bring across a couple of concepts. This is from Gary Firestein. The way people have been thinking about this is that there is a cytokine network in rheumatoid arthritis synovium, where macrophages, fibroblasts and other cells are making cytokines, which will be cross-regulating each other and cross-activating cells. What I would like to bring out is that if you look at this, there are a lot of cytokines, and TNF is sort of embedded in all of this. It is very hard to know which cytokines are the most important ones for pathogenesis. I think that the real contribution of Mark Feldman and Ravinder Maini was to try to dissect out which cytokines are going to really be the keys to regulating inflammation in RA. What they defined, in a series of experiments going back to the early 80s, is that TNF appears to be at the apex of this pro-inflammatory cascade, at least in rheumatoid synovium. The way they got this was by taking fragments of RA synovium and growing them in vitro, and showing that, if you neutralize TNF, you actually block production of a large number of cytokines. The way they see this TNF is at the apex and drives production of cytokines. They and others have supported these ideas using transgenic models, where it has been shown that the expression solely of TNF was sufficient to get chronic inflammatory arthritis. And one way of thinking about this is that TNF is at the apex of an inflammatory cascade which drives the activation of cells in RA synovium. This is countered by some of these homeostatic cytokines that I have discussed.

So the concept that a lot of rheumatologists think about is that the way cytokines regulate inflammation is really the balance in the synovium between pro-inflammatory cytokines, such as TNF and interleukin-1, and anti-inflammatory cytokines, such as TGF beta and interleukin-10. The question that comes onto the table is that if TNF is so important and its neutralization or blockade is so effective, what really regulates TNF production during inflammation? The answer to that is not known. NF-kappa B is thought to play an important role. Some people have proposed IL-15, and I think this is an interesting area of research.

So what actually turns on TNF - because if you could understand how to turn off production of TNF, this would obviate the need for continual and chronic suppression with neutralizing medications. There are lots of ideas. Some people think interferon gamma; this is very controversial. Could it be T-cell dependent? This is an idea under investigation, again being pioneered by Mark Feldman and colleagues. Is it cell-cell interactions between macrophages and T-cells and fibroblasts? No one is really quite sure. I think with the reemergence of immune complexes in the pathogenesis of rheumatoid arthritis, it could well be that immune complexes are an important contributor to TNF production, and this is felt to be dependent on NF-kappa B.

How TNF Suppresses Autoimmunity

Briefly, how does TNF suppress autoimmunity? There are lots of ways it can do it. As I discussed, TNF can induce apoptosis of immune cells; in some animal models TNF can actually induce T cell tolerance, for example in a model of diabetes. TNF actually turns off the production of interleukin-12, which is an important regulator of Th1 immunity. The TNF receptor p75 has some anti-inflammatory actions that actually appear in a transgenic model where, if this receptor is knocked out, the mice get arthritis that is more severe. TNF plays a role in activation-induced cell death, especially in CD8 positive T cells. In long term chronic autoimmune disease, chronic exposure of TNF actually desensitizes signals of the T-cell receptor, which can have effects both upon ongoing selection in the thymus and also on homeostasis in the periphery.

So examples of where TNF actually suppresses autoimmunity -- again in the model of diabetes, where giving TNF decreases incidence and severity of disease -- blocking TNF makes disease worse. There is a very similar pattern in multiple sclerosis. In lupus, there is evidence that TNF might suppress systemic lupus erythematosus. There is TNF deficiency in the NZBxNZW model of systemic lupus erythematosus. We see acceleration of disease when the TNF receptor is knocked out. Giving TNF makes mice with lupus better and I think we'll hear about this more -- some patients that are treated with TNF blockade develop antibodies against double stranded DNA and a lupus-like syndrome.

I think the cytokine network in rheumatoid arthritis and lupus and other immune diseases can be quite different. TNF deficiency potentially leads to immunosuppression, since it is important for both innate and acquired immunity. TNF deficiency perturbs cytokine networks, which can have beneficial effects on inflammation but some potentially deleterious effects on some autoimmune disorders and also lead to the reversal of some of the suppressive effects on the acquired immune system.

So to summarize, TNF is a potent inflammatory cytokine. It has been implicated in chronic inflammation and multiple disorders. It does, however, have the nature of both activating and suppressive effects. We need to think about its overall biological activity, which is determined by the cell type, the timing, the exact location where it is working and the balance between pro-apoptotic and inflammatory signals. It almost always enhances innate immune responses early during the initiation stage. However, during a chronic autoimmune process it can have suppressive effects as well, so that the consequences of TNF deficiency or blockade can be somewhat difficult to predict.

^ Back to Top
Request an Appointment