Type 1 Diabetes: Cellular, Molecular & Clinical Immunology
Theoretical Essay F - Pathogenesis of Type 1 Diabetes: A Viewpoint
Matthias G. von Herrath*
*La Jolla Institute for Allergy and Immunology, Division of Immune Regulation, 10355 Science Center Drive, San Diego, CA 92121 USA, Phone: 1-858-558-3571, Fax: 1-858-558-3579, matthias@liai.org
Understanding the etiology and pathogenesis of human type
1 diabetes is a difficult task. The main issue is that we are probably dealing
with a highly localized, slowly progressive inflammatory reaction that is
mostly confined to the islets and pancreatic draining lymph node. Since
the human pancreas is located retro-peritoneally, it is inaccessible and
samples or live cells cannot be recovered easily and certainly not on a
routine basis. Non-invasive imaging does not have the necessary resolution
at this point to allow precise analysis of the cellular immunological process.
Thus, our immunopathological knowledge of the human disease is limited and
largely based on the assumption that animal models allow us to draw parallels.
Therefore, it is important to recapitulate here, what we really know with
certainty.
1. Autoantibodies: Autoantibodies to islet antigens can be detected
accurately in serum samples from pre-diabetic individuals (1) (2; 3). They
predict the risk to develop disease well and antibodies to multiple b-cell
antigens (insulin, GAD, others such as IA-2) are associated with a very
high likelihood of clinical T1D. However, it is still controversial, whether
such autoantibodies play a pathogenetic role (4). Plasmapheresis in humans
only temporarily clears islet-antibodies from the circulation and no effect
on T1D development was consequentially observed (5). Recent studies in the
NOD mouse model suggest an instrumental role for maternal antibodies in
initiating diabetes development in offspring (6). In humans, the role of
maternally transmitted islet antibodies is less clear, because no differentiation
between maternal antibodies generated as a consequence of insulin therapy
versus islet-autoimmunity was possible in previous studies (2). However,
no direct association has been established at this point and offspring from
diabetic fathers carry a higher diabetes risk than those from diabetic mothers
(7). One could assume that such maternally transmitted antibodies would
possibly encounter an immunologically active field, since lymphoid structures
are routinely seen in fetal pancreata during development (P. Leenen, et
al. Unpublished). Future investigations should focus on correlating antibody-isotypes
and avidity with disease risk. This will bring increasing clarity to their
possible pathogenetic role in particular in young children (8).
2. T lymphocytes: Predominantly based on animal experimentation in
the NOD mouse or transgenic models, T1D is thought to be a T cell mediated
autoimmune disorder (9). However, one has to acknowledge that infiltrates
found in human islets are less T-cell rich than those present in NOD and,
for example, RIP-LCMV mice (10). Clearly, T cells (in contrast to autoantibodies)
as well as clones with specificity for islet antigens can transfer disease
in animal models consistently (11). Comparable studies are of course impossible
in humans and we must admit that no definite proof in respect to the importance
of T lymphocytes has been obtained. However, indirect proof can be attributed
to temporary success of systemic immunosuppressive regimens such as antibodies
to CD4, CD3 and cyclosporine A (12). Interestingly, certain human islet
infiltrates exhibited a predominance of distinct Vb T-cell receptor families,
raising the possibility for an oligoclonal repertoire with defined antigen
or super-antigen specificity (13). Generally, autoreactive T cells are difficult
to reliably detect, in particular in human peripheral blood. Even in NOD
mouse spleens, there is a significant variability between colonies when
comparing spontaneously arising islet specific T cells (14). At this point
one can therefore assume that numbers of autoreactive lymphocytes in T1D
are low and that they are mostly localized in the pancreas and draining
lymph node (15). Improved tracking reagents such as tetramers and ELISPOT
assays might elucidate their pathogenetic roles, because they include a
more precise assessment of effector functions (16). These could then be
correlated with progression and severity of the diabetogenic process on
an individual basis. Current evidence from animal models indicates that
autoreactive lymphocytes can have destructive or regulatory effector functions
depending on which cytokines they produce (15; 17). One issue that is currently
just emerging is to determine, which factors or cells provide the continuous
drive for autoreactive T cells during a chronic autoimmune process that
can take place over many years. Studies using adoptive transfers or one-time
immunization with dendritic cells in animal models have shown that autoreactive
T cells can easily run out of steam and that local inflammation might be
required to initiate and maintain autoaggression (18). It is therefore important
to consider the role for antigen presenting cells in T1D.
3. Antigen presenting cells: Due to the relative paucity of lymphocytes,
human islet infiltrates are exhibiting a higher degree of monocytes and
possibly other APCs compared to mouse insulitis (19). Their role is unclear,
but it is an intriguing finding that activated APCs are detectable early
in human pancreata. In animal models, there is no doubt that they are of
critical importance for driving the local process and providing continued
opportunity for autoreactive T cells to become activated. In support, one
can cite numerous reports. For example, loss of APC or costimulatory function
reduces diabetes incidence in NOD and other diabetes models, unless generation
of regulatory cells are affected (20). Conversely, adding B7 to the pancreatic
equation’ aggravates disease and, if expressed on b-cells, can precipitate
spontaneous autoimmunity (21). Once, destruction of some islet cells has
occurred one has to assume that local APCs will increasingly present b-cell
antigens. This leads to antigenic and epitope spreading, if non-tolerant
T cells to the respective antigen are available in the periphery. Furthermore,
one could hypothesize that activated APCs might be required to provide a
‘fertile field’ for the arrival of autoreactive lymphocytes, without
which these would not be capable of causing chronic inflammation and disease.
This consideration is relevant to the next topic discussed here. On the
other hand, however, one must acknowledge that NOD mice as well as diabetic
patients have distinct defects in APC activation and stimulatory capacity
(19). How can this be reconciled with the notion that they are required
to drive autoaggression? This issue is unclear at this point, but it will
be important to distinguish between local inflammation and systemic dysregulation
to better understand experimental results.
4. Genetic and environmental triggers: The genetics of T1D are quite
well understood. It is clear at this point that multiple genes can have
predisposing as well as protective function resulting in a complex overall
interaction. This is true for human (22; 23) as well as NOD diabetes (24;
25). The MHC class II complex for sure plays a central role and it is intriguing
that the NOD MHC class II that is required for spontaneous disease bears
structural similarities to the human susceptibility allele HLA-DQ8 (26;
27). Importantly, overall contribution of genetic factors will not explain
the etiology of disease alone, since there is a significant discordance
in monozygotic twins (28). Other environmental factors are therefore important,
and many potential candidates are under evaluation, among them dietary cow’s
milk in young infants and viral infections. Viruses have been shown to influence
the diabetic process in a positive or negative way in animal models (29;
30). In humans, no association has been proven, although Coxsackie and rota-viruses
have been reported to precede autoantibody sero-conversion in young infants
(31). Positive association could occur through mimicry, enhanced local inflammation
and bystander activation. Negative association could involve apoptosis of
aggressive lymphocytes and immune modulation. Very precise epidemiological
studies will be required to establish a firm cause, it is well possible,
however, that viruses contribute to T1D pathogenesis in a multifactorial
way similar to genes.
5. Dysregulation: Does a possible systemic defect in immune regulation
underlie or enhance the development of T1D? There is relatively good evidence
from human studies that NK T cell activity, APC function as well as generation
of CD25+ regulatory lymphocytes are all reduced in human diabetic patients
(32). This finding coupled with strong evidence from animal models that
enhancing NK T cell activity (33), systemic infections leading to immune
activation or transfer of CD25+ cells (32) can prevent T1D, makes generalized
immune dysregulation a likely underlying cause for T1D. Genetic factors,
exposure to infections and other environmental influences might all be able
to enhance or alleviate this systemic dysregulation. It is intriguing that
most defects that have been found constitute a lack of certain immune functions
including generation of molecules such as interferons by NK T cells that
are known to be detrimental to islets. However, lack of NK T cell activity
also results in less IL-4 production and lower degrees of APC activation
might result in lower levels of CD25+ or TH2-like regulatory cells than
needed to maintain a healthy immune equilibrium. Therapeutically, this situation
can possibly be exploited, but one will have to proceed with caution, because
any type of systemic correction will affect other immune functions such
as host defense.
A hypothesis for T1D pathogenesis: A primary requirement for the
development of islet autoimmunity is a non-tolerant peripheral repertoire
containing lymphocytes that bear T cell receptors that can react with islet
antigens (34). This is not an uncommon occurrence, because thymic negative
selection is not complete (35) and naïve autoreactive lymphocytes are
present in peripheral lymphoid organs. This situation does normally not
pose any problems and will not lead to autoimmunity, unless these cells
become activated and gain access to the target organ (34). I would like
to propose that local events are more likely than systemic events to trigger
T1D, because systemic activation of autoreactive lymphocytes requires large
numbers of activated cells that are unlikely to be generated to a sufficient
extent via molecular mimicry or bystander effects (36). As a possible local
precipitator, viral infections, APC dysfunction, or even maternal antibodies
could all play a role. It is well possible that T1D is etiologically a quite
heterogeneous disease and that all of the above will occur in conjunction
with a varying genetic predisposition. Once a sufficient momentum of APCs
that drive autoreactive lymphocytes has been reached in islets and draining
lymph node, development of T1D is well on its way. As islet destruction
progresses, more antigens will become targeted reflected in increased levels
and varieties of islet autoantibodies. It is important that not all autoreactive
lymphocytes will have detrimental effector functions and certain cytokines
such as IL-4 or IL-10 can have positive effects (15), if expressed constitutively
at the right level and time prior to massive islet loss. This can be therapeutically
exploited and maybe the ideal intervention would do a combination of both,
augment regulatory responses and curb aggressive anti-islet activities.
This could be achieved by combining systemic modulators such as anti-CD3
(37) or Vitamin D3 analogs with islet-antigen specific DNA vaccines that
induce regulatory lymphocytes (38). Regardless of the etiology, the crucial
effector pathways in islet destruction are well known and exhibit a significant
amount of plasticity and redundancy. Cytotoxic T cells, type 2 interferon
(39), TNFa, IL-1b ad nitric oxide generation are all detrimental to b-cells
(40). Since these can be produced by various effector lymphocytes including
APCs and B cells, the above-mentioned redundancy would be expected. These
considerations make effective late interventions harder to achieve. However,
one must not forget, that successful islet destruction will rely on a critical
mass of inflammation. Thus, acutely ‘resetting’ the system can
brake the vicious cycle (for example by anti-CD3 or similar agents). Long
term tolerance, however, should be maintained in an antigen specific way
avoiding prolonged systemic immunosuppression (38; 41).
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