Mystery Rays from Outer Space

Meddling with things mankind is not meant to understand. Also, pictures of my kids

June 29th, 2008

Side effects of tumor vaccination

Melanoma blood vessel
Melanoma blood vessel

Designing tumor vaccines presents a bunch of problems that anti-pathogen vaccines don’t. One of those problems is identifying an appropriate antigen. There’s been a lot of interest in finding tumor antigens that cytotoxic T lymphocytes will recognize, and in fact hundreds have been identified. The database of tumor antigens at Cancer Immunity lists some 750 of them, divided into various categories:

  • Unique antigens result from point mutations in genes that are expressed ubiquitously. They are unique to the tumor of an individual patient or restricted to very few patients;
  • Shared tumor-specific antigens are expressed in many tumors but not in normal tissues;
  • Differentiation antigens are also expressed in the normal tissue of origin of the malignancy;
  • Overexpressed antigens are expressed in a wide variety of normal tissues and overexpressed in tumors

Overall, shared tumor-specific antigens may be the ideal target. Because they’re found in multiple tumors, a vaccine can be pre-designed and go through a time-consuming optimization and validation process; because they’re only found in tumors, there’s less concern about safety. That is, the risk of the vaccine precipitating an autoimmune reaction to normal cells is low.

MHC expression in eye (Palmer 2008 Fig 4)
MHC expression in eye
before and after vaccination

The problem with this class is that there just aren’t all that many tumor-specific antigens. The database lists 20-odd such antigens, and many of them are only found in a limited subsets of tumors (mostly melanomas). What’s more, I think it’s not merely that the targets are out there yet haven’t been identified. More likely, there simply are not many shared tumor-specific antigens.

The next-best category, as far as safety and effectiveness is concerned, is the unique antigens. These may be great as far as safety and effectiveness are concerned, but there are major technical problems in identifying them in a clinical context. Because they’re unique, you can’t pre-design the vaccine; you need to customize the antigen to each patient. And (at least with present techniques) by the time there is enough tumor available to look for unique antigens, the disease is likely to be pretty far advanced (and advanced tumors are more likely to be resistant to vaccination, for several reasons). There’s a lot of interest in making preparations of tumors that would contain unique antigens, without the trouble of identifying the antigen, but as far as I know that hasn’t made it very far into clinical trials yet.

So that leaves overexpressed and differentiation antigens. These are both, by definition, found in normal cells, and that means either the immune system is already tolerized to the antigen, or that targeting these antigens with a vaccine risks triggering an autoimmune reaction.

In fact, clinical trials using these kinds of vaccines against melanomas have found that successful tumor attack is almost invariably associated with an autoimmune effect, usually manifesting as vitiligo (de-pigmented patches on the skin).

It is expected that immune responses to such peptides will be compromised by self-tolerance or, alternatively, that stimulation of effective immune responses will be accompanied by autoimmune vitiligo. 1

I believe the first record of this goes back to 1964.2 Vitiligo is not, as autoimmune diseases go, a terrible problem, and certainly one would be delighted to trade melanoma for vitiligo. However, there are more serious potential problems as well, and one of them was recently reported by Nick Restifo’s group. 3 In this case, a highly active anti-melanoma vaccine not only killed the tumor, it also triggered severe autoimmune destruction of the eye; and the more effective the vaccine, the worse the autoimmune disease:

Thus, in the present model, the efficacies of the antitumor immune therapies were directly correlated with the induction of autoimmunity in the eye. … Our data suggest that, as tumor immunotherapies improve, these autoimmune manifestations may become more prevalent.

(My emphasis.) The autoimmune disease in these melanoma patients is still probably manageable (they mention using 30 months of steroid treatment in one of the most severely affected patients), and even if not, again the tradeoff is one most people would probably take (blindness vs. death). But other tumors may make the decision much more difficult:

Although the autoimmune side effects of melanocyte/melanoma-targeted therapies have been manageable, the unintended autoimmunity of therapies targeting colorectal, brain, or lung cancer might prove more severe.


  1. Antigens derived from melanocyte differentiation proteins: self-tolerance, autoimmunity, and use for cancer immunotherapy. Engelhard VH, Bullock TN, Colella TA, Sheasley SL, Mullins DW. Immunol Rev. 2002 Oct;188:136-46. []
  2. Vitiligo In A Case Of Vaccinia Virus-Treated Melanoma. Burdick Kh, Hawk Wa. Cancer. 1964 Jun;17:708-12.[]
  3. Palmer, D.C., Chan, C., Gattinoni, L., Wrzesinski, C., Paulos, C.M., Hinrichs, C.S., Powell, D.J., Klebanoff, C.A., Finkelstein, S.E., Fariss, R.N., Yu, Z., Nussenblatt, R.B., Rosenberg, S.A., Restifo, N.P. (2008). From the Cover: Effective tumor treatment targeting a melanoma/melanocyte-associated antigen triggers severe ocular autoimmunity. Proceedings of the National Academy of Sciences, 105(23), 8061-8066. DOI: 10.1073/pnas.0710929105 []
May 18th, 2008

Autoimmunity and CD1 (Part II)

Last week I talked some general issues about autoimmunity, and gave a brief background on NKT cells. Today I’ll talk about the paper that spawned that discussion.1

A common general model for autoimmune goes something like this:

  • If you have a genetic predisposition toward autoimmunity2
  • And you are exposed to a microbial antigen,
  • Sphingomonas

  • That is somewhat similar to one of your body’s own antigens
  • And the exposure involves inflammation, which sends a “Danger!” signal to the immune system,
  • Then immune cells that are normally tolerant to the self antigen
  • Become reactive toward the microbial antigen
  • And cross-react with the self antigen. This low-level self-reactive inflammation
  • Causes cell death, releasing more antigen in the the presence of cell-death “Danger” signals.
  • Causing a runaway feedback loop that results in outright autoimmune disease

But as I said, it’s been very difficult to track through a reaction from beginning to end, to support or refute this model.

Matter et al., Fig 3 (Inflamed bile duct)
Matter et al., Fig 3 (Inflamed bile duct)

Primary biliary cirrhosis (PBC) is an autoimmune disease3 of the liver characterized by inflammation of the bile ducts (here is the American Liver Foundation’s PBC information page). The immunity seems to be mainly targeted at mitochondrial antigens, which raises the question of why the liver is specifically involved — mitochondria are found in just about every cell type.

NKT cells recognize CD1, which binds to lipid-type antigens typical of bacterial cell walls. Bendalac’s group found that they could cause a PBC-like disease in mice by infecting them with a particular bacterium4 that is normally considered to be a fairly innocuous commensal. They tested this bacterium because it was previously shown to trigger antibodies that cross-react with the mitochondrial antigens that are targets in PBC. (Remember that mitochondria are historically extremely symbiotic bacteria, so the cross-reactivity doesn’t come completely out of the blue.)

Antibodies are produced by B cells. However, the disease could be blocked by preventing NKT cells from getting activated (by infecting mice lacking the NKT target, CD1). The rationale for doing this experiment was that innate immune responses to this particular bacterium are, a little unusually, normally driven by NKT cells.

Novosphingobium aromaticivoransThe autoimmune-type disease lasted in these mice long after they had eliminated the bacteria — months, compared to a week or two to eliminate the actual infection. What’s more, even though NKT cells were essential to get the disease going, once it had started up, the disease could be transferred to new mice by swapping across classical T cells only (i.e. T cells but no NKT cells) — even into mice that had never seen the bacteria and didn’t even have CD1, which were doubly protected against having the disease start on its own. In other words, NKT cells start the disease, but don’t keep it going.

So what seems to be happening is that the NKT cells recognize the bacteria and produce massive inflammation. Because NKT cells tend to home to the liver5, they are able to overcome tolerance of cross-reactive cells in the liver, making liver antigens more at risk. The cross-reactive T and B cells, enraged by the constant roar of inflammation the NKT cells produce, attack the cross-reactive self antigens, damaging the cells and causing a constant inflammatory trigger. At this point the disease has become self-perpetuating, and you don’t need the NKT cells any more (and indeed, they quiet down about this time, as the bacteria are eliminated).

These findings establish the missing connection between the microbial innate immune trigger and chronic effector T and B lymphocyte attack of small bile ducts observed in PBC. 6

This is probably not a universal effect in detail — NKT cells are likely not important in the majority of autoimmune diseases — but it does give support to the general concepts that have been floating around for a while now.


  1. Mattner, J., Savage, P., Leung, P., Oertelt, S., Wang, V., Trivedi, O., Scanlon, S., Pendem, K., Teyton, L., Hart, J. (2008). Liver Autoimmunity Triggered by Microbial Activation of Natural Killer T Cells. Cell Host & Microbe, 3(5), 304-315. DOI: 10.1016/j.chom.2008.03.009[]
  2. Usually the mechanism is unknown[]
  3. Probably. There us still some uncertainty, but that is the best bet[]
  4. Novosphingobium aromaticivorans[]
  5. For reasons that are not, as far as I know, understood[]
  6. Invariant Natural Killer T Cells Trigger Adaptive Lymphocytes to Churn Up Bile. Sebastian Joyce and Luc Van Kaer. Cell Host & Microbe (15 May 2008) 3:275-277[]
May 14th, 2008

Autoimmunity and CD1 (Part I)

Dr. Kilmer's Swamp Root Kidney Liver & Bladder CureWe walk a fine line between death due to immune deficiency, smothered under the weight of pathogens and parasites, and death by hyperimmunity, eaten alive by our own defenses. It’s amazing that our immune system can be tuned so precisely as to recognize anything foreign, yet ignore the vast antigenic universe of our own normal self.

Of course, sometimes the immune system fails, in both directions. We often hear about deaths from pathogens, and autoimmune diseases in general are pretty common. There are many ways by which (it’s believed) the immune system can become self-reactive, but a very common observation is that there are both genetic and environmental predisposing causes to autoimmunity. That is, you may have the genetic makeup to be autoimmune, but until you’re exposed to some environmental trigger, autoimmunity never develops. So, for example, if your identical twin has an autoimmune disease, you are much more likely than someone in the general population to develop the disease; but you still have a good to excellent chance of never getting the disease.

Liver blood vesselsIn many cases the neither the environmental triggers nor the genetic factors are well understood. The most likely environmental trigger, though, is some kind of microbe. In some cases, this may be because of “molecular mimicry” — the microbe has an antigen that looks like self antigen; the self antigen is normally ignored, because the immune system needs some kind of “danger” signal before it becomes activated; the microbial antigen is seen in the context of microbial “danger” signals; an immune response forms against the microbial antigen; the immune response cross-reacts with the self antigen; self cells are damaged by this immune response; the dead cells release more danger signals along with self antigen; and a positive feedback loop drives a full-fledged autoimmune disease.

That’s the model, but there aren’t many, if any, diseases where the whole process has been tracked through step by step; in fact, I think that there has been so much difficulty getting clear molecular connections between microbes and autoimmunity that there’s a robust search for other mechanisms. However, in the latest issue of Cell Host and Microbe, Albert Bendelac’s group shows a series of links between bacterial infection and the autoimmune disease human primary biliary cirrhosis (PBC).1 (There’s also a helpful, if rather dry, commentary2 by Sebastian Joyce and Luc van Kaer in the same issue.) Rather than trying to cover everything today I’m going to give background here, and then talk about the specific findings in a few days.

CD1 - top view with ligandOne interesting thing about Bendelac’s paper is that they link CD1 to the disease, through NKT cells. CD1 is an MHC class I family member; I talked about it back here, and that’s its mug shot to the left here (click for a larger version). CD1, like many members of the MHC class I family, has a “groove” in its “top” side. MHC class I proper binds peptides in that groove, but CD1 has a much more hydrophobic groove that binds to greasy things like lipids, glycolipids, and lipopeptides. These kinds of molecules are typically found in some kinds of bacteria — especially mycobacteria, like tuberculosis and leprosy, but also other kinds of bacteria such as the commensal microbe Sphingomonas.

MHC class I molecules, with their peptides, are recognized by cytotoxic T lymphocytes (CTL),3 but CD1 molecules and their lipids are recognized by a specialized subset of T cells, “natural killer-like” T cells (NKT cells). The function of this CD1/NKT system really isn’t all that clear. The early guesses that this was a branch of the immune system specialized for dealing with mycobacteria has been weakened as NKT cells have been linked to resistance to various viruses, and also as various viruses have been shown to block CD1 — suggesting that CD1 and NKT cells would otherwise eliminate them.

OK, enough for now. In my next post I’ll talk more about the disease itself, and then try to spell out the process by which, according to Bendelac, NKT are central to the autoimmune reaction; as well as how this abnormal reaction suggests some of the normal functions of NKT and CD1.


  1. Mattner J, Savage PB, Leung P, Oertelt SS, Wang V, Trivedi O, Scanlon ST, Pendem K, Teyton L, Hart J et al. (2008) Liver Autoimmunity Triggered by Microbial Activation of Natural Killer T Cells. Cell Host & Microbe 3:304-315.[]
  2. Joyce S, Van K, Luc (2008) Invariant Natural Killer T Cells Trigger Adaptive Lymphocytes to Churn Up Bile. Cell Host & Microbe 3:275-277.[]
  3. And natural killer cells, but let’s not go into that now[]
February 4th, 2008

Autoimmune hypotheses

Autoimmunity is surprisingly common, and amazingly complex. About 5% of people will develop some form of autoimmune disease — that’s tens of millions of people in North America alone — yet the causes underlying the diseases are still not known. It’s clear that there are both genetic and non-genetic factors, because if one identical twins has an autoimmune disease the other is much more likely to also develop it; but even with identical twins the concordance is far from perfect, usually being well under 50%.

In this genomic age, the genetic side of the equation is becoming more accessible, and large-scale genomic scans for gene variants that affect risk of autoimmune disease are becoming routine. But again — apart from outright mutations that cause autoimmune disease with very high penetrance, such as AIRE deficiency1 — there are clearly environmental factors that are at least as important as all the genetic factors put together. What’s the rest of the story?

The usual assumption has been that the environmental factors are infections. If you’re genetically predisposed to autoimmune disease, this argument goes, then exposure to a particular pathogen may tip you over the brink. This might be because of cross-reactivity — the pathogen contains epitopes that look like self epitopes; the self epitopes are usually ignored because they are not seen in an inflammatory (“danger”) context; but the pathogen epitope is seen in such a context and drives the activation and expansion of cross-reactive T cells, that then return home and attack the self antigen, causing autoimmune disease that can then expand and become more severe.

There are some problems with this argument, at least as a universal explanation. In particular, at least some animal models of autoimmune disease do show the typical long latency and sporadic disease that is expected with an infectious trigger, yet the disease continues to appear in pathogen- and germ-free animals. 2 In fact, it’s even argued that autoimmune disease, like allergy, is associated with reduced exposure to pathogens — an extension of the “Hygiene Hypothesis”.3

New ideas

Perhaps springing from the general sense of dissatisfaction with the present explanations for autoimmune disease, there are two recent papers that propose a new, conceptually similar, but different explanations; and a letter to the editor that notes some weaknesses in one of the hypothesis.

The first I ran across4 was
GOODNOW, C. (2007). Multistep Pathogenesis of Autoimmune Disease. Cell, 130(1), 25-35. DOI: 10.1016/j.cell.2007.06.033

To make a long story short5 Goodnow draws parallels between cancer (especially lymphoid cancer) and autoimmune disease:

the potential for relentless growth of self-reactive lymphocytes is normally blocked by a series of checkpoint mechanisms that also prevent lymphoid neoplasia and … autoimmune outgrowths only develop when multiple checkpoints are eventually bypassed.

Autoimmune corneal ulcerHe proposes that, in fact, autoimmune disease and lymphoid cancer share a common pathogenesis: Somatic mutations. Lymphocytes have a relatively high mutation rate (perhaps because they undergo somatic mutation as part of their normal maturation process), and perhaps some forms of this mutation cause dysregulation of the lymphocytes without causing cancer — a dysregulation presenting as abnormal self reactivity rather than uncontrolled growth. He puts forward a number of ancillary arguments and evidence supporting this, as well.

By comparison, I think that most models of autoimmune disease postulate that the self-reactive lymphocytes are broadly normal; that everyone carries potentially self-reactive lymphocyte, but they aren’t activated by self antigen either because they are actively suppressed, and/or because they don’t receive adequate support during the self recognition process, or because the self antigen is normally hidden and not presented to the immune system.

The second paper, by Pederson,6 addresses the third of these possibilities. His is also a somatic mutation hypothesis, but it’s quite different from Goodnow’s. He suggests that the somatic mutation may hit the target cell, rather than the lymphocyte.

It is here postulated that mutant proteins are formed randomly in non-malignant cells and can influence the immune system towards priming against self-antigens. If mutations are the result of somatic mutations in a stemcell such proteins will persist, but possibly cause no immediate harm … However, such mutant proteins might be pivotal in the break down of tolerance and induction of autoimmunity.

Basically, Pederson’s hypothesis is that a mutated protein might be misfolded and/or degraded faster and this will result in increased recognition even of normal (“self”) epitopes in the mutant protein. This is a fairly plausible suggestion in itself — however, I’m not quite convinced. We’re talking here, I think, about quantitative increase in presentation, whereas most models of autoimmunity consider it to be somehow qualitatively different. (Of course, if you get quantitative enough, things become qualitative, if you see what I mean. Still, it seems like a relatively conservative hypothesis, somehow.) As Pederson points out himself, testing this hypothesis might be extremely difficult, because you’d have to identify a very small population of mutated protein swimming in the pool of normal unmutated protein. (Also — my own addition to the difficulty — by the time the disease is apparent, the mutated cells would have probably been destroyed by the initial autoimmune response.) Overall, I have a hard time getting very excited by this hypothesis — not by any means because it is self-evidently wrong, but it’s not clear to me how to test it, or where to take the next step in using it in a model for pathogenesis or treatment.

Questions resurface

Back to Goodnow’s more radical, and exciting, hypothesis. In his paper Goodnow adds a note crediting F.M. Burnet for a previous iteration of the idea:

Since submitting this manuscript, my colleague Dr. Stephen Daley has drawn my attention to F.M. Burnet’s extraordinary 1972 book (Burnet, 1972). This monograph assembled the limited information about lymphocytes at that time to arrive at the same hypothesis developed here-that the stochastic onset of autoimmunity reflects “a conditioned malignancy” caused by emergence of “forbidden clones” through a combination of germline and somatic mutations that disrupt the normal mechanisms for eliminating or inactivating self-reactive lymphocytes.

… which brings me to the letter to the editor, by Thomas Brodnicki:7

I was reminded of a graduate lecture course given by Professor Roderick MacLeod in 1995 at the University of Illinois. While introducing F.M. Burnet’s clonal selection theory to our small class, Professor MacLeod mentioned Burnet’s idea that germline and somatic mutations provide the inherited and stochastic mechanisms that disrupt tolerance and result in autoimmunity.

Brodnicki points out four flaws in Goodnow’s model, none of which strikes me8 as being overwhelming, but which together make me say “Hmm”. His four points are that Goodnow’s (or Burnet’s) somatic mutation hypothesis doesn’t account for:

  1. age (cancer hits older people, as you’d expect with somatic mutations that should become more common over time; autoimmune disease often hits fairly young people more than old);
  2. sex (autoimmune disease often affects females far more often than males — or in some cases affects males more often than females — which is hard to explain by somatic mutations);
  3. epitope spreading (how does the autoimmune reaction “spread” from the epitope recognized by the initial hypothetically-mutated lymphocyte clone, to recognize the wide range of targets that are typical of autoimmune diseases); and
  4. infectious disease (why is increased incidence of autoimmune disease linked with reduced incidence of infectious disease?)

As I say none of these points is itself insuperable; there are reasonable explanations that could account for all of them. But as far as I can think of there is no single explanation that will account for all of them, so we are multiplying William of Occam’s entities to an uncomfortable point.

Nevertheless, I think Goodnow’s hypothesis is a very interesting — and testable — one, that also offers conceptual avenues toward treatment and prevention; even if it’s wrong, I think it’s wrong in an interesting way. I’m looking forward to seeing what happens to it in the next couple years.


  1. For example, Pereira, L.E., Bostik, P., & Ansari, A.A., 2005. The development of mouse APECED models provides new insight into the role of AIRE in immune regulation. Clinical & developmental immunology, 12(3), p.211-6.[]
  2. In some other autoimmune disease models, exposure to pathogens, or to normal intestinal flora, does seem to increase the incidence of disease, as predicted. So it’s complicated.[]
  3. Bach, J., 2002. The Effect of Infections on Susceptibility to Autoimmune and Allergic Diseases. N Engl J Med, 347(12), p.911-920. []
  4. Actually, to tell the truth, the first I ran across was the letter to the editor, which intrigued me so that I looked again at Goodnow’s paper, and then while following up on Goodnow I found Pederson’s paper. Obviously I don’t follow the autoimmune literature as closely as I should. I hereby resolve to turn over a new leaf.[]
  5. Actually, it’s not all that long. In fact Goodnow’s paper is well worth reading if you’re interested — it’s well written and easy enough to follow[]
  6. Pedersen, A., 2007. The potential for induction of autoimmune disease by a randomly-mutated self-antigen. Medical Hypotheses, 68(6), p.1240-1246. []
  7. Brodnicki, T.C., 2007. Somatic Mutation and Autoimmunity. Cell, 131(7), p.1220-1221.) []
  8. Or Brodnicki, for that matter[]
|