Research Overview

Autoimmunity & Transplantation Division

This Division conducts research centred on the function of the immune system in conditions of health and disease. The immune system defends the body against the foreign "non-self" (such as viruses and bacteria) and "self" (such as tumours). However, sometimes the system malfunctions and the body's own tissues are mistakenly identified and marked for destruction. This results in autoimmune disorders, including type 1 (insulin dependent) diabetes, multiple sclerosis (MS) and rheumatoid arthritis. Important aspects of our work include the identification of environmental and genetic factors and the cellular and molecular mechanisms that lead to autoimmune disease; and designing and testing ways to engineer the immune system to protect against type 1 diabetes and the rejection of transplanted tissue. Recent highlights include positive follow-up results from the Intranasal Insulin Trial in children at risk for type 1 diabetes; further evidence that rotavirus is involved in triggering type 1 diabetes; improved ways of preventing autoimmunity by administering antigens to the mucosal immune system; and new insights into the mechanisms of arthritis.

Background

Our daily health - our very lives - depend on our immune system's ability to discriminate between the inner world of 'self', and an external universe of 'non-self', populated by potentially deadly viruses, bacteria, fungi, toxins and other potentially hazardous agents.

The immune system is one of nature's most wondrous inventions, but it is not perfect - occasionally, components of the body's own tissues stray into the twilight zone between 'self' and 'non-self', and find themselves marked for destruction. The immune system mounts a sustained attack on the body's own tissues, and an autoimmune disorder ensues.

Autoimmune disorders are among the major causes of lifelong disability in the community. They are also among the most enigmatic of all human diseases - their causes are obscure, and while drugs or other therapies can, in some cases, alleviate their symptoms, they have traditionally been regarded as incurable.

However, recent, rapid progress in understanding the molecular mechanisms of autoimmune disorders promises a revolution in the first decade of the 21st century. The promise is that such diseases, which collectively affect 10% of the population, will be preventable and, in the longer term, even curable. The Hall Institute is at the forefront of international research into autoimmune disorders, and scientists in the Autoimmunity and Transplantation Division recently began trialling an experimental preventative therapy for insulin-dependent diabetes mellitus (IDDM).

IDDM is one of two major autoimmune disorders in the Division's research portfolio - the other is the joint disease rheumatoid arthritis.

Insulin Dependent Diabetes Mellitus

The concordance rate for IDDM in identical twins is only about 40 per cent - that is, if one identical twin develops diabetes, the risk that the other will also develop diabetes is only 40 per cent.

This statistic suggests that chance operates at two different levels to cause diabetes.

The luck of the genetic draw predisposes certain individuals to develop diabetes. Nevertheless, if IDDM was due to genetic factors alone, the concordance rate in identical twins would be 100 per cent, and the rate in families with a history of diabetes would conform to Mendelian laws of inheritance.

In actual fact, only a minority of those at high genetic risk will actually develop IDDM, typically in infancy or childhood. The majority will remain free of diabetes for life. Such individuals are lucky enough to avoid the 'X factor', the event that triggers the onset of the autoimmune response in diabetes. Alternatively, these individuals have also inherited genes which render them resistant to the effect of the 'X factor'.

In IDDM, the immune system response is directed against antigens in beta cells of the islets of Langerhans, specialised structures in the pancreas that supply the body with the hormone insulin. The immune response causes inflammation that progressively destroys the beta cells, resulting in insulin deficiency.

Insulin regulates blood sugar (glucose) levels. In the absence of insulin, the blood glucose level increases, causing the symptoms and complications of diabetes. People with IDDM require insulin injections two to four times daily to replace their own, missing insulin, and must follow a strict diet, to maintain their glucose levels within normal bounds. Insulin injections cannot replicate the precision with which beta-cells match insulin supply to the body's demands. Long-term complications of inadequately controlled glucose levels include damage to small blood vessels, causing kidney disease and blindness, damage to large blood vessels, resulting in heart attack, stroke, blockage of blood flow in the limbs, and peripheral nerve damage.

Screening for IDDM

Over the past decade, Professor Len Harrison, Head of the Autoimmunity and Transplantation Division, and Dr Peter Colman, of the Royal Melbourne Hospital, have screened thousands of first-degree relatives of people with IDDM - mainly children who have a parent or sibling with the disorder.

These individuals are at increased genetic risk of developing diabetes. Of the first-degree relatives screened in the Melbourne Pre-Diabetes Family study, 3-4 per cent had tell-tale antibodies or T cells targeted to their own islet antigens, pre-clinical signs that they are in the early stages of developing diabetes. Many of those followed since 1988 have gone on to develop diabetes.

In the complementary Oz BabyDiab study, babies who have a parent or sibling with IDDM are tested at six-monthly intervals from birth, to determine the sequence of immunological and biochemical events involved in the onset of diabetes, and to monitor their exposure to common childhood viruses.

Some 15 per cent of these children develop antibodies to one or more of their islet antigens - particularly insulin - in the first two years of life.

Three targets for 'unfriendly fire'

What, precisely, is the immune system firing at in the insulin-secreting beta cells?

Professor Harrison says research (in an animal model - the non-obese diabetic (NOD) mouse - and in humans) has shown that three main antigens are involved:

• Insulin - or more specifically, its parent molecule, proinsulin. Insulin consists of two peptides, both derived from the same proinsulin precursor. An enzyme cleaves proinsulin into two pieces, which bind to each other by sulphur atoms to form insulin.
• Glutamic acid decarboxylase (GAD), an enzyme most abundant in the brain, that converts glutamic acid to gamma amino butyric acid (GABA), a major inhibitor of nerve transmission. In the pancreatic islets, GABA probably inhibits the secretion of other, locally made hormones like somatostatin and glucagon.
• Tyrosine phosphatase IA-2, another enzyme that cleaves phosphorus atoms off proteins - its function in the islets of Langerhans is unknown.

Of the three antigens, only proinsulin occurs exclusively in beta cells. Both GAD and tyrosine phosphatase IA2 are components of other tissues throughout the body.

'Our hypothesis is that proinsulin is the main target, and that the initial T cell attack on proinsulin is somehow exposing GAD and tyrosine phosphatase IA-2 to attack as well,' Professor Harrison said.

'We have direct evidence for the central role of proinsulin from studies of the NOD mouse.'

In 1996, Dr Michelle French, a postdoctoral fellow in the Division, developed a transgenic strain of the NOD mouse which expresses proinsulin in cells that are normally specialised to present antigens - macrophages, dendritic cells and antibody-secreting B-cells. Proinsulin, remember, is normally expressed exclusively by pancreatic cells, not by these immune-system cells.

Unlike humans, all NOD mice are genetically predestined, rather than merely pre-disposed, to develop diabetes. But to the surprise and delight of the Hall Institute researchers, transgenic NOD mice expressing proinsulin that could be seen as 'self' by T cells during embryonic development did not suffer inflammation of the islets of Langerhans, and remained free of diabetes.

This historic experiment showed that it might be possible to develop an immune therapy to prevent at-risk people with pre-clinical signs of IDDM from developing the disease.

But gene transplantation is not a practical therapy for humans - so Professor Harrison and his colleagues instead decided to use a different experimental therapy to reprogram the immune system away from proinsulin

IDDM - a sniff of victory

The therapy that the Hall Institute researchers chose was an aerosol inhalation or intra-nasal spray containing insulin, and later, small fragments (peptides) from proinsulin.

The mucous membranes of the respiratory tract, and of other tissues such as the digestive and reproductive tracts, are constantly exposed to everyday antigens from the outside world.

They contain a special, mucosa-associated immune system whose function was not studied in detail until the 1980s. The mucosal immune system contains specialised T cells, called gamma-delta T cells - 'They're Cinderella cells, looking for a role,' says Professor Harrison.

The role of gamma-delta T cells turns out to be crucially important - 'They are the first immune-system cells to see antigens from the external environment that we ingest or inhale, for example. Instead of reacting in a destructive way, they may actually induce immunological tolerance.

'When we gave aerosol insulin to NOD mice, we were thrilled to discover that the incidence of IDDM more than halved, and that CD8 gamma-delta T cells transferred from the treated mice could protect untreated mice against diabetes.

'We now call these CD8 gamma-delta T cells 'regulatory' or 'protective' T cells. Furthermore, we have good evidence that introducing peptides derived from proinsulin directly into the nose can induce a second type of regulatory T cell called CD4 alpha-beta.

Late in 1996 Professor Harrison and Dr Peter Colman began the first trial of intra-nasal aerosol insulin in humans

Why does the immune system attack proinsulin? The underlying mechanism is unclear, but Professor Harrison's team has implicated a small peptide bridging the site where proinsulin is normally cleaved to create the two peptides that make active insulin.

'It's possible that some people make genetic variants of proinsulin, or that some environmental insult, such as a common viral infection, may disrupt the normal processing of proinsulin in beta cells,' Professor Harrison said.

'We know that T cells from people with signs of pre-clinical diabetes are reacting to this peptide fragment, and we have recently found that if we administer just the peptide, rather than the entire proinsulin molecule, into the nose, it prevents diabetes in NOD mice.'

It's not clear how the peptide fragment is involved, but Professor Harrison says one possibility is that if the fragment is generated in beta cells, for example by abnormal cleavage of proinsulin, it could find its way onto major histocompatibility complex (MHC) molecules that present antigenic peptides to receptors on T cells, thereby activating the T cells.

Environmental agents and IDDM

Scientists have long suspected that certain common virus infections of childhood may trigger IDDM in individuals who are genetically predisposed to develop diabetes.

Dr Margo Honeyman, Professor Harrison's long-time collaborator, takes credit with two other Sydney researchers, Margaret Menser and Jill Forrest, for establishing the first link between a common virus, genetic susceptibility and diabetes.

In the 1970s, while working at the Royal Children's Hospital Research Foundation in Sydney, the three researchers showed that children of mothers who had suffered congenital rubella (German measles) infections, and who had one of the first-identified MHC genes, called HL-B8, were significantly more likely to develop IDDM than children whose mothers had not had rubella. So the rubella vaccine, developed soon after, not only protected children against blindness and deafness, the most common complications of rubella during pregnancy - it significantly reduced their risk of diabetes.

Dr Honeyman has also implicated rotavirus, another common virus that causes diarrhoea, as a risk factor in IDDM - she suggests that rotavirus epidemics in the past few decades may account for the rising incidence of IDDM. Like the rubella vaccine, a vaccine against rotavirus, now under development, could significantly reduce the incidence of diabetes in children in future.

A third common virus infection of childhood - Coxsackie B virus - has been previously isolated by other investigators from pancreatic tissue of two children diagnosed with diabetes. In one case, the virus was shown to cause diabetes in laboratory mice.

Why should some viruses cause diabetes? Apart from direct infection of the islets, evidence has strengthened for the viral mimicry hypothesis, which proposes that autoimmune disease originates with infections by common viruses carrying antigens that mimic antigens in the host's tissues. The infection elicits an immune response that targets both the viral antigen and the 'self' antigen it resembles. For reasons that are unclear, the immune system continues to attack the 'self' antigen even after the virus has been eliminated, damaging or destroying the tissues in which it occurs.

Dr Honeyman's studies of antigens from two of the suspect viruses supports the hypothesis. One of the rotavirus antigens is similar to part of the human tyrosine phosphatase IA2 enzyme, and a region of Coxsackie B virus resembles a region of human GAD enzyme. But as yet, says Professor Harrison, no virus has been identified with significant sequence homology to proinsulin.

Summary: the IDDM triangle

The emerging story on IDDM points to a triangle of circumstance:

• Certain genes, notably certain alleles of MHC genes, predispose some individuals to diabetes, perhaps by coding for MHC molecules that bind peptides like the proinsulin peptide, triggering beta-cell autoimmunity. But most of these individuals will never develop diabetes.
• Certain viruses may infect beta cells and disturb the normal processing of proinsulin, generating immunogenic peptide antigens. Alternatively, viruses with antigens that mimic 'self' antigens in the host's beta cells, including proinsulin, GAD and tyrosine phosphatase IA2, may provoke the immune system into mistakenly attacking 'self' beta-cell antigens.
• Chance brings susceptible individuals into contact with these viruses, usually in infancy or early childhood. Diabetes develops as a sequel to one or more common childhood viral infections that provoke an autoimmune attack on islet cells in the pancreas.

Solving a murder mystery: Who - or what - killed the beta cell?

The cause of IDDM is the death of insulin-secreting beta cells in the islets of Langerhans in the pancreas, which come under fire from the patient's own immune system.

But what immune-system cell types actually destroy the beta cells? Dr Tom Kay is studying the cell-to-cell interactions that result in their destruction.

Dr Kay says his team has been using transgenic mice to address clinically relevant questions in IDDM. 'We want to know which particular cells to target with diabetes therapy - do we target CD4 helper T cells, or CD8 cytotoxic T cells?'.

Researchers have long regarded CD4 T cells as the prime suspect. Not so Dr Kay.

'It's been a fairly unfashionable idea, but we have been exploring the possibility that CD8 T lymphocytes, whose classical physiological role is the elimination of viruses, might be the real culprit,' Dr Kay said.

'Although CD-8 cells are present in the infiltrate from inflamed islet tissue, when researchers used monoclonals directed against T cells, the more impressive effect was against CD4 cells.

'But experiments with mice showed that, to transfer diabetes from a diabetic to a healthy mouse, both CD4 and CD8 cells were needed.'

For several years, Dr Kay and his colleagues worked on a hunch: CD8 lymphocytes typically direct their cytotoxic attack at virus-infected cells displaying MHC class 1 antigens. Viruses hide away inside cells, subverting their protein-synthesis machinery to make new virus particles, and the only way to for the immune system to eliminate the virus is to kill off all infected cells. Cytotoxic T cells recognise infected cells because they display viral antigens on their MHC Class 1 bound to the MHC Class I molecules at the cell surface.

Normally, beta cells express low levels of MHC Class 1, but when the pancreatic islet tissues become inflamed, there is a rapid, pronounced increase in the expression level. Could this provoke CD8 cytotoxic T cells to attack?

'We thought the increased MHC Class 1 antigen expression in beta cells might be significant - it might allow CD8 cytotoxic cells to interact directly with beta cells and kill them, which would accelerate the onset of diabetes.'

Dr Kay and his colleagues explored this possibility by crossing a knockout mouse that was unable to express Class 1 MHC, with the non-obese diabetic (NOD) mouse, which is genetically predestined to develop diabetes.

The result: a mouse predisposed to diabetes, but lacking the MHC class 1 molecules suspected of provoking CD8 cytotoxic T cells to attack its beta cells.

The MHC class 1-deficient NOD mice defied their genetic destiny: they did not become diabetic! In fact, their islet tissue showed no sign of inflammatory infiltrate - in fact, there was no trace of any of the pathology of diabetes.

'The experiment showed that CD8 cytotoxic T cells are early players in establishing the hostile immune response,' Dr Kay said.

The Hall Institute researchers were able to confirm the result by transferring functional MHC class 1 genes back into the beta cells of the diabetes-free mouse to restore normal class 1 MHC expression. The resulting beta cells were then destroyed, just as happens in the normal NOD mouse.

CD8 cytotoxic T cells kill virus-infected cells with a potent toxin called perforin, which punches holes in the target cell's membrane, causing it to collapse. Is perforin the lethal weapon in the attack on insulin-secreting beta cells?

Dr Kay says other researchers have made perforin-deficient NOD knockout mice. They do not develop diabetes. Perforin thus appears to be the murder weapon. But why do inflamed beta cells increase their expression of MHC class 1 antigens, and does this provoke attack by cytotoxic T cells?

Dr Kay says that expression levels of MHC Class 1 antigens are usually up-regulated in response to biochemical signals, including interferon gamma, and sometimes in response to viral infections.

To explore this question, the Hall Institute researchers developed a transgenic NOD mouse whose beta cells lacked the receptor for interferon gamma. NOD mice lacking interferon gamma in all tissues - not just beta cells - do not develop diabetes. 'Furthermore,' says Dr Kay, 'if beta cells are exposed in vitro to interferon gamma, they show the classic symptoms of T cell attack, and die. If gamma interferon is a driving force in diabetes, one that biases the immune response towards CD8 cytotoxic T cells, we need to know it works.'

When they analysed a mutant transgenic mouse with interferon gamma-unresponsive T cells, Dr Kay and his colleagues found that its beta cells maintained normal levels of expression of MHC class 1 antigens - yet the mice still developed inflamed islets and diabetes, just like the unmanipulated NOD mouse.

This result established that interferon gamma is indeed responsible for the increased expression of MHC Class 1 on beta cells. Very importantly, however, it also demonstrated that cytotoxic T cells can still attack beta cells with normal MHC Class 1 antigen expression.

The strong inference, according to Dr Kay, is that, there must be other important targets of interferon gamma that promote diabetes. Perhaps interferon gamma acts on macrophages to increase antigen presentation.

Current research in his lab is focussed on solving this fascinating and important conundrum.

Xenografted Islets: Potential Cure?

Early clinical trials indicate that IDDM may be prevented by aerosol insulin therapy. But this does not help diabetics who have already lost their beta cells, because beta cells have no natural capacity for regeneration. The Division is investigating ways of curing, as well as preventing, IDDM. Just as many people who have suffered kidney failure receive kidney transplants, people with IDDM (islet failure) would receive islet transplants.

Dr Andrew Lew, another researcher in the Autoimmunity and Transplantation Division, says there are two major hurdles to islet-cell transplants. First, the powerful drugs used to prevent the host's immune system rejecting the grafted tissue involve a risk of infection and toxicity that outweighs any risks associated with insulin injections. Second, there is a poor supply of human islet donors.

A promising alternative is xenografting: transplanting islet tissue from another mammal species into diabetics, restoring their ability to make insulin. Given that pig insulin was the standard treatment for diabetics for more than half a century, until recombinant human insulin became available in the 1980s, the pig is the most suitable species.

Unfortunately, the human immune system would rapidly reject unmodified pig islet tissues as 'non-self' - it would treat it as it would a raging infection. The solution, says Dr Lew, is to used recombinant DNA technology to develop transgenic pigs whose islet cells make factors that the human immune system normally uses to regulate itself.

Dr Lew says graft rejection is normally initiated by professional antigen-presenting cells, called dendritic cells, which dissect potential alien antigens and display on their surface, bound to specialised molecules called MHC molecules.

The dendritic cells then migrate to lymph nodes where they 'dock' with T cells that orchestrate the immune response - the T cell determines whether the displayed antigen is 'self' or 'non-self'.

Docking is accomplished by a receptor on the surface of the T-cell that binds (adheres) to the MHC/antigen complex on the surface of the dendritic cell. But successful docking also involves two other molecules, called co-stimulatory factors.

 

The protein CTLA4-Ig blocks the docking of the potentially aggressive T cell to the antigen presenting cell and prevents it from being activated.

 

One factor, CD28, resides on the surface of the inspecting T-cell. It participates in a lock-and-key reaction with a complementary molecule, called B7, on the dendritic cell. Without this reaction, docking fails and the suspect antigen is ignored.

Dr Lew and his colleagues conceived the idea of fusing a gene for molecule called CTLA4Ig, which binds B7 even better than CD28, with part of an antibody molecule called Ig, so that the hybrid molecule, delivered into the pig islet cells, would be secreted into the microenvironment of the graft. The CTLA4Ig molecules should bind to the B7 co-factor on any antigen-presenting dendritic cells in the immediate locality, via the normal lock-and-key reaction, making it inaccessible to T cells.

The result would be that, even though the dendritic cells would still display pig islet antigens, T cells would not be able to detect them, so they would not be rejected.

Dr Lew and his colleagues have trialled this approach in NOD mice, using a mouse cell line called an insulinoma, a cell that, like normal beta cells, secretes large amounts of insulin - but which, unlike beta cells, grows readily in tissue culture. The insulinoma cells also express a virus antigen called SV40 large T antigen, a potent trigger that would normally cause mice to rapidly reject the insulinoma cells.

After putting the CTLA4Ig gene into the insulinoma cells, by a process called transfection, they transplanted the cells beneath the skin of NOD mice - a procedure equivalent to transplanting pig islet cells into human diabetics. The transfected insulinoma cells, instead of being rejected, grew aggressively and formed tumors. By blockading the normal co-stimulatory response, the CTLA4Ig molecules effectively rendered the insulinoma cells invisible to the mouse's immune system. This exciting result, Dr Lew says, suggests that genetically-modified pig islets could be successfully grafted into human diabetics.

But not yet - scientists involved in xenografting research are proceeding with great caution, because of the possibility that proviruses - dormant retroviruses in the pig genome - could be activated and cause trouble when pig tissues are transplanted into humans. All mammals carry a suite of proviruses, the genetic legacy of millions of years of encounters between retroviruses and their hosts.

Dr Lew says that although the risk is probably very low - no pig retrovirus is known to infect humans, despite thousands of years of contact between pigs and humans - the current outbreak of mad cow disease in humans illustrates the potential hazards that animal infectious agents pose to humans.

The consequences of introducing a new retroviral disease into the human population via xenografting are potentially so serious that clinical xenografts will not be attempted until the risks have been thoroughly assessed.

In the meantime, Dr Lew has begun in vitro experiments with cultures of foetal islet cells from pigs, using adenovirus as a vector to introduce the human CTLA4Ig gene.

Rheumatoid Arthritis

Rheumatoid arthritis (RA) is one of the most common and debilitating autoimmune diseases, but in spite of over a century of investigation, the cause or causes of RA are unknown.

Symptoms arise from joint inflammation and there is also progressive destruction of bone and cartilage. Loss of normal joint structures restricts mobility, which impacts on all normal daily activities. In more severe cases, RA can be a crippling and deforming disease.

Dr Ian Wicks, a Rheumatologist who heads the Reid Rheumatology Laboratory at the Hall Institute, says RA can strike at any age, but typically targets the 20 - 50 age group. In addition to complications of the disease, patients can suffer serious side effects from treatment, including bone fractures and stomach ulcers, from the cortico steroid and non-steroidal anti-inflammatory drugs (NSAIDs) used to inhibit the immune system.

Clues from cancer

RA shares many features of a classic autoimmune disease - Dr Wicks says the synovial tissue and fluid adjacent to cartilage and bone in affected joints swarms with inflammatory and immune-system white blood cells that have left the blood stream and invaded the joints. However, a feature of RA is that new blood vessels proliferate within the affected joint, a process called angiogenesis. Angiogenesis normally ceases after the rapid growth phase of foetal development and childhood. In adults sustained angiogenesis is usually a pathological condition, apart from the special circumstances of pregnancy and the female menstrual cycle. Angiogenesis is often associated with tumours and RA shares many features of a localised form of malignancy. Dr Wicks says a naturally occurring protein called endostatin, which recently came to light in the US for its ability to starve tumours by inhibiting blood vessel growth, could also help RA patients. Inhibition of angiogenesis should be safe and relatively free of side effects, because people do not need to grow new blood vessels once development is complete. Cells in the inflamed joints of RA patients also secrete large amounts of vascular endothelial growth factor (VEGF); VEGF molecules dock with VEGF receptors on the surface of cells lining the vascular endothelium, reactivating long-dormant genes for cell growth and division. Dr Wicks says it may be possible to inhibit angiogenesis by designing synthetic molecules to bind VEGF, or to block the VEGF receptors. Inhibiting the growth of blood vessels into inflamed joints could literally starve the inflammatory process.

Cell adhesion molecules

The WEHI team is exploring a similar approach with another class of cell-surface molecules that guide white blood cells into the joints. The invading cells must first pass a check point in the blood vessel wall. Dr Wicks and his colleagues are focussing on a class of molecules called cellular adhesion molecules (CAMs) which are expressed by the vascular endothelial cells. Drugs or blocking antibodies might quarantine lymphocytes and other inflammatory cells to the blood stream. 'Control of expression of CAMs on blood vessels is crucial because the whole process of chronic inflammation in the joint requires these cells to be constantly recruited from the blood stream' Dr Wicks said. Two CAMs - and their receptors - have emerged as candidates for antibody therapy, or for new 'designer' drugs-intercellular adhesion molecule (ICAM) and vascular cellular adhesion molecule (VCAM).

From mouse to man

Genetically-modified mice are providing valuable information in the search to understand the chain of events that leads to arthritis. Knockout mice are particularly informative - by knocking out specific genes, researchers can determine their roles in signalling pathways and simulate the effects of novel drug or antibody therapies. The logic of the approach is simple: any gene deletion that prevents the onset of arthritis or lessens its severity in the mice identifies that gene product as a candidate for drug or antibody therapy.

Bone marrow transplantation - a second chance?

Cancer therapists have observed that RA patients who receive bone marrow transplants after undergoing aggressive chemotherapy or radiation therapy for cancer, often experience an improvement in their RA. Before administering chemotherapy or radiation, doctors extract and preserve bone-marrow stem cells for an autologous ('self') transplant after therapy is complete. Bone marrow contains a self-renewing population of stem cells, the precursor cells from which all immune-system and blood cells are derived. Dr Wicks says doctors are now considering autologous bone marrow grafts as an experimental treatment for RA. When the immune system is destroyed by cancer therapy, and reconstituted by bone-marrow stem cells, it must recapitulate the process of learning to discriminate between 'self' and 'non-self' antigens, including the antigens in joint tissues that were formerly under auto-immune attack. If the immune system learns to tolerate these antigens, the auto-immune attack on the joints should cease, and the patient should then be at no greater risk of re-developing RA during the rest of their lives than an identical twin whose sibling has RA.