Type 1 Diabetes: Cellular, Molecular & Clinical Immunology
Theoretical Essay A - The Development of Autoimmune Diabetes: Theoretical Aspects
Kevin J. Lafferty
The nonobese diabetic (NOD) mouse develops type I diabetes spontaneously and the disease is associated with lymphocytic infiltration of the pancreatic islet and eventual destruction of islet b-cells. This destructive process is quite specific, with a-cells and other cellular components of the pancreatic islets initially being spared. The disease process represents a form of tissue-specific autoimmunity. It is associated with the development of antibodies specific for islet components such as insulin glutamic acid decarboxylase (GAD) and other islet-cell antigens (see Chapter 2). Although such antibodies become elevated with the development of disease, there is no evidence that antibodies of themselves are pathogenic.1 The disease process appears to represent a form of cell-mediated immunity dependent on CD4 T cell activity, and it has been demonstrated that disease develops in mice whose capacity to produce humoral immunity has been blocked.1
Initiation of Disease
The specificity of the disease
process resulting in damage to only the b-cells of the islet suggests that the
pathogenic process may involve a cytotoxic T cell attack directed at some b-cell-specific
antigens. The development of disease is dependent on the activity of both CD4
and CD8 T cells2,3 and it is reasonable to suggest that the CD4 cell is behaving
as a helper cell required for the activation of the cytotoxic CD8 T cell that
in turn initiates islet damage. However, studies carried out at the Barbara
Davis Center demonstrated that although both CD4 and CD8 T cells are required
for the initiation of disease, the immunological specific effector cell is the
CD4 T cell (see Chapter 5).3,4 Pancreatic islet-reactive
T cell clones have been isolated from the NOD mouse and in all cases these cells
have proved to express the CD4 phenotype (see Chapter 4). Such cloned CD4 T
cells have the capacity to initiate disease in the absence of CD8 T cell function.5,6
Thus, while there is no question that the CD8 T cell plays an essential role
in the development of disease, this cell is not appear to be immunologically
specific effector cell. Ann Cooke has shown this cell to be required early in
the process of disease development, indicating that the CD8 T cell has an accessory
or helper function in the activation of the CD4 effector of the autoimmune response.7
There are examples were CD8 T cells can modify the activation of CD4 T cells
and this process is mediated via active involvement of the antigens-presenting
cell (APC).8 There is also evidence9 that the early activated CD8 T cell produces
high levels of interferon gamma (IFN-g) he and in this way forces the CD4 T
cell response toward the production of IFN-g interleukin-2 (IL-2) (Th1 phenotype).
The balance between Th1 and Th2 CD4 T cell activation plays a critical role
in determining whether or not to the autoimmune response generated will be destructive
(see below).
There has been considerable speculation concerning factors responsible for the
initiation of the disease process. Susceptibility to diabetes both in human
and in the animal models is determined by the major histocompatibility complex
(MHC) genotype; the disease-prone character is determined by genes in the class
II region of the MHC complex (see Chapter 3).
However, although susceptibility to the development of diabetes is under genetic
control, the clinical disease itself is not solely determined by genotype. Thus,
in the case of identical twins there is only a 30-40% concordance for disease.
In the NOD mouse model, where all individuals are genetically identical as the
results of inbreeding, disease develops in only 20-40% of male NOD animals in
approximately 80% of females. The observation that the incidence of disease
increases when animals are maintained under specific pathogenic-free conditions
indicated an environmental contribution to the development of disease.10 The
negative correlation with infection was an unexpected finding; one notion relating
to the development of this disease was that the process resulted from a cross-reactive
immune response to some environmental pathogen.11 If this were the case, one
would expect the disease incidence to fall when animals were spared the immunological
challenge of environmental pathogens.
Another interesting characteristic of the disease process is that although clinical
disease is seen only in a proportion of animals in the NOD colony, all animals
of the NOD genotype have pathology in their pancreas. That is, all animals express
autoimmunity and show the development of lymphocytic infiltrates around islet
tissue within the pancreas.12 There is, however, a difference between animals
that develop clinical disease and those that have insulitis but do not go on
to develop overt diabetes. In the latter case, the lymphocytes accumulate around
the outside of the islet and rarely penetrate within the islet itself. In the
former situation there is extensive invasion of islets by mononuclear cells
and associated destruction of islet b-cells.12 It is worth noting that following
destruction of islet b-cells the lymphocytic infiltrate disappears from the
pancreas in one is left with small pseudo-islets made up predominantly of a-cells
with little or no lymphocytic accumulation around the tissue. We must conclude
therefore that the disease process is dependent on recognition of some b-cell-associated
antigen that is not expressed by other cells within islet.
Studies carried out in transgenic animals where particular viral antigens are
expressed in islet b-cells demonstrate that neither the expression of foreign
antigens nor the possession of T cells specific for such antigens is a sufficient
requirement for the precipitation of disease.13 The initiation of disease requires
the "appropriate" presentation of antigen to an immune system that
has potentially reactive T cells. We do not fully understand what is involved
in "appropriate" antigens presentation. What we know is that infection
of animals with a particular virus can lead to islet destruction when some viral
antigens are expressed on islet b-cells. However, the development of a destructive
process is not always seen in such situations. Variability appears to depend
on the genetic background of the animal and the nature of the virus used in
such studies.13
What is now becoming clear is that two distinct forms of autoimmune response
can be observed. There is the response that leads to islet damage and the development
of diabetes, which we can define as destructive autoimmunity. There is also
a form of response that leads to pathology and associated lymphocyte accumulation
around pancreatic islet but that is a nondestructive process and does not lead
to clinical disease. The coexistence of the potential for either destructive
or nondestructive autoimmunity in animals of a defined genotype suggests the
development of diabetes is a stochastic process. That is, although we can precisely
define the kinetics of the disease process and the proportion of animals in
any group that will express overt diabetes at any given time (provided we are
dealing with a colony maintained under controlled conditions), we are unable
to specify beforehand exactly which animals will become diabetic. That is, we
are unable to predict which animals will undergo the destructive process and
which will lead to develop nondestructive autoimmunity before the process is
well under way. Clearly, environmental conditions have a large influence on
the proportion of animals that fall into either of these groups. This is why
animals of the same genotype held under conventional laboratory conditions have
a much lower incidence of disease than those maintained under specific pathogen-free
conditions.
We do not precisely know what is responsible for the development of the disease
process leading to clinical diabetes. We know that the genotype-specifically
the class II MHC antigens type-in both animals and humans is a major factor
controlling the disease-prone character. However, possession of the disease-prone
genotype is not in itself sufficient requirement for the development of disease.
Moreover, T cells with the capacity to recognize and respond to islet autoantigens
can exist animals that do not develop clinical diabetes. All we can say with
any confidence is that disease develops and disease-prone animals and individuals
following "appropriate" stimulation of the immune system. The problem
is to define what is meant by the term "appropriate stimulation."
Pathogenesis of the Disease Process
When it comes to
understanding the basis of the pathogenic process involved in b-cell destruction
in the development of IDDM we are on somewhat firmer ground. As we mentioned
above, the disease process is a T cell-depended phenomenon in which both CD4
and CD8 T cells are required for the initiation of disease but where the CD4
T cell is the immunologically specific effector cell. CD4 T cells of the disease-prone
individual are probably not interactions directly with antigen expressed on
islet b-cells . Such antigens would be presented in association with class II
MHC antigens; that is, in a form which does not favor CD4 T cell recognition.
The CD4 T cells are most likely to be interacting with antigen derived from
b-cells and processed by class II MHC-bearing antigen-presenting cells. This
process would lead to the development and expression of cell-mediated immunity
in which cytokine production by the inflammatory cells is a major pathogenic
event leading to islet destruction.
There is evidence that free radical production is involved in this process.
Both superoxide production by activation macrophages and nitric oxide production
by islets in response to IL-1 appear to be involved in the pathogenic process (see Chapter 5). The Okamoto hypothesis provides
the most useful model for understanding the involvement of three radical damage
in the disease process (see Chapter 5). According
to this hypothesis free radicals initiate DNA damage, which in turn activates
the DNA repair enzyme (poly ADP-ribose synthetase) that is involved in DNA repair.
This enzyme uses cellular NAD in the process and as a result depletes the b-cells'
radical scavenging capacity. The cycling to this process results in b-cell death.
One production of this model is that iron-chelating agents such as desferrioxamine,
which block the conversion of super oxide to the more damaging hydroxyl radical,
and specific inhibitors of the DNA repair enzyme such as nicotinamide, may be
used to regulate this destructive process. Experimental evidence is consistent
with such predictions.14
It has been suggested that nicotinamide is then agents that need be used to
prevent the development of diabetes in a clinical situation. Certainly this
agent is effective in animal models. However, the Okamoto hypothesis indicates
that the effect of nicotinamide is to block the DNA repair enzyme. This process
made lead to the development of oncogenic change because of interference with
the DNA repair process. Earlier studies that examined the effect of nicotinamide
on streptozotocin-induced diabetes indicated that nicotinamide could prevent
the induction of diabetes by this agent.15 However, animals protected in this
way were tumor prone and a significant proportion developed b-cell adenomas
later in life. Any use of nicotinamide for the control of autoimmune diabetes
in a clinical situation should be approached with considerable caution.
Exposure to environmental pathogens could have a marked effect on the development
of diabetes in disease-prone animals. Environmental stimulation of the immune
system leads to a decrease in the proportion of animals that go on to develop
clinical disease. Immunostimulation of disease-prone animals with agents such
as completes Freund's adjuvant or the vaccine BCG has been shown to block the
development of diabetes in disease-prone NOD animals.16,17 Immunostimulation
with either of these agents blocks the development of disease both when administered
early, that is, soon after weaning, or quite late in the pathogenic process.
Animals vaccinated at 85 days of age (clinical disease is first expressed from
90 days on) do not go on to develop clinical disease.17 Such immunostimulation
does not inhibit the development of autoimmunity. It does, however, force the
autoimmune response along the nondestructive pathway.17 Although such animals
are protected from the development of overt diabetes they remain disease prone,
and administration of cyclophosphamide can acutely result in the development
of diabetes in these animals.17
These observations emphasize the need to view the immune system as an integrated
network regulated by both positive and negative influences. Although "appropriate"
stimulation of the immune system can lead to the development of diabetes in
disease-prone animals, it is now also clear that other forms of immunostimulation
can negatively regulate this process. This positive and negative regulation
of immune function is correlated with the nature of cytokine production in either
destructive or nondestructive lesions.12 In the former case, local sites within
destructive lesions are producing IFN-g in high proportion and they contain
relatively low numbers of IL-4-producing T cells (high proportion of Th1 cells).
The nondestructive lesion, on the other hand, has a smaller proportion of IFN-g
producing cells in the higher ratio of IL-4-producing cells (high proportion
of Th2) cells. More direct evidence for cytokine involvement in this process
comes from studies in which antibodies to IL-10 and IL-4 were used in an attempt
to reverse the protective effect of immunostimulation. In the studies antibody
treatment was shown to reverse the protective adjuvant effect (Calcinaro et
al, unpublished data).
Although the mechanism of the adjuvant effect is not fully understood, the fact
that quite marked effects have been obtained in animal models using both complete
Freund's adjuvant and the more benign BCG vaccine has prompted clinical testing
of BCG vaccination in newly diagnosed diabetic individuals. A preliminary clinical
study of this kind has provided evidence that BCG vaccination may alter the
natural history of the disease process in humans.17 These studies were carried
out in an unblinded fashion and use historical controls and must, therefore,
be considered as preliminary observations. Further studies carried out in a
blinded fashion are required to establish whether or not BCG vaccination can
alter the pathogenesis of diabetes in humans. If such studies are positive it
may be possible to prevent the development of clinical disease in humans by
immunostimulation as is the case in the experimental model.
The development of spontaneous diabetes as seen both in animals and humans is
an autoimmune process dependent on the development of islet specific cell-mediated
immunity. b-cell destruction is the result of inflammatory tissue damage and
the specificity of this process appears to reflect the sensitivity of b-cells
to radical damage. The class II MHC antigen genotype of both animals and humans
regulate their disease-prone status. However, whether or not clinical disease
develops in such individuals is the result of a balance between positive and
negative regulation within the immune system. Such regulation is associated
with differential cytokine production. Thus, appropriate stimulation of the
immune system can lead either to the development of diabetes or to the development
of nondestructive autoimmunity. This latter observation may provide the means
for safe regulation of diabetes in disease-prone individuals.
Reference List - links to PubMed available in Reference List.
For comments, corrections or to contribute teaching slides, please contact Dr. Eisenbarth at: george.eisenbarth@ucdenver.edu