TOLERANCE MECHANISMS

Increasing insights into mechanisms of self-nonself discrimination have emerged over the
past 6 decades in parallel with growing knowledge of the immune system. More than 50 years
ago, Burnet and Medawar advanced the critical concept that tolerance was imposed by clonal
deletion of self-reactive lymphocytes during early development (i.e., central tolerance). Later,
with the discovery that mature B cells undergo somatic hypermutation in the periphery, it was
hypothesized by Bretscher and Cohn that the production of autoantibodies might be impeded by
the need for both B and T cell compartments to breach tolerance. In 1975, while studying
allogeneic responses, Lafferty and Cunningham posited that activation of T cells involved the
passing of a second signal that need not be antigen-related, thereby implicating costimulation
from antigen-presenting cells as a critical factor in lymphocyte activation.66 In 1987, the nature
of the costimulation, or two-signal, requirement was further defined when Jenkins and Schwartz
showed that engagement of antigen receptor alone without a second signal resulted in functional
inactivation of T cells. A novel mechanism for self-nonself discrimination that incorporated the
innate immune system was then advanced by Janeway in 1989 when he hypothesized that
antigenpresenting cells critical for T cell activation remain quiescent unless activated by the
engagement of PRRs by microbial products. This concept was further extended by Matzinger in
1994 to a “danger model” that includes activation of the immune system by both foreign and
endogenous factors associated with tissue stress and damage. These models have laid the
foundation for the current, more complex, view of self-nonself discrimination in which tolerance
is imposed by both innate and adaptive immune systems through layers of mechanisms occurring
at various stages throughout lymphocyte development and activation.

Clone-Specific Self-Nonself Recognition

In contrast to innate immune cells, which are activated primarily by hardwired microbial
PRRs, lymphocytes are unrestricted in specificity and therefore self-nonself discrimination must
be implemented at the clonal level. To achieve this, T and B cells use several mechanisms that
can be grouped into three general strategies. First, the type of response is controlled by
developmental stage. For example, immature lymphocytes respond to strong antigen receptor
stimulation by cell death, whereas a similar signal in mature cells leads to activation. Through
this mechanism selfreactive clones are eliminated from the nascent lymphocyte repertoire before
they can cause injury. Second, activation of mature lymphocytes requires—in addition to antigen
receptor engagement—a second co-stmulatory signal that, if absent, results in anergy or cell
death. For the most part, this requirement limits reactivity to self because costimulatory signals
are largely provided by activated cells of the innate immune system. Third, lymphocytes are
finetuned in various ways by a fairly extensive list of modulating factors, which is necessary for
controlling self-reactive clones. For example, defects affecting a broad range of surface receptors
on lymphocytes, including those with prosurvival (IL-7R, BAFFR, IL-2R); proapoptotic
(TNFRs, FasL, TRAIL); costimulatory (CD28, CD40, TLRs); differentiating (IL-12R, IL-4R,
IFNγR, IL-23R, retinoic acid R, transforming growth factor (TGF)-βRs, SAP/SLAM family
members, OX40, ICOS/ICOSL); inhibitory (FcγRIIb, CD22, CTLA4, PD-1); antigen receptor
signal modulating (CD19, CD45); and activating (SAP/SLAM family members) activities have
been shown to influence the development of autoimmunity.9 Collectively, these selfnonself
recognition mechanisms provide the basic cellular means by which the innate and adaptive
immune systems render T and B cell clonotypes tolerant to self-antigens and resistant to
autoimmune disease.

Innate System and Tolerance

Given its vital role in initiating and modulating adaptive immune responses, it is not
surprising that the innate immune system strongly influences both tolerance and autoimmunity.
Indeed, although its contributions have yet to be fully explicated, several ways in which self-
tolerance is influenced by the innate arm have been defined. First, as noted previously, activation
of the innate immune system under normal circumstances typically requires engagement of
microbial PRRs, which endows the immune system with a direct and simple way to distinguish
foreign- from self-antigens. The importance of this mechanism is illustrated by the finding that
overexpression of TLR7 by spontaneous gene duplication or transgenic approaches promotes
systemic autoimmunity. This occurs because certain PRRs, such as TLR7 and TLR9 that equally
sense both foreign and self-nucleic acids, avoid significant exposure to endogenous nucleic acids
by virtue of their location in subcellular compartments. However, in the case of TLR7,
overexpression makes it possible for normally substimulatory amounts of endogenous RNA to
activate immune cells, thereby bypassing the usual requirement for microbial exposure. Second,
some cells of the innate immune system actively suppress adaptive immune system activation
under certain conditions. For example, immature and, under some circumstances, even mature
dendritic cells have been shown to promote tolerance by inducing CD4 T regulatory (Treg) cells
and other mechanisms. Third, another critical function of the innate immune system is the rapid
noninflammatory clearance of apoptotic cells. Failure can result in an increased supply of self-
antigenic material including nucleic acids, secondary necrosis, and release of proinflammatory
factors, leading to systemic autoimmunity. Accordingly, deficiencies in several key apoptotic
cell clearance molecules are associated with autoimmunity including (1) the Tyro3-Axl-Mer
receptors on phagocytes that bind through Gas6 or protein S, the exposed phosphatidylserine
(PS) on apoptotic cells (2) the milk fat globule-EGF factor 8 (MFG-E8) protein that bridges the
αvβ3 integrin on phagocytes and PS on apoptotic cellsand (3) natural IgM or C1q that binds to
and enhances clearance of apoptotic cells. Fourth, several complement components have been
directly implicated in autoimmunity. For example, SLE is associated with deficiencies of
proximal components of the classical pathway including C1q, C4, and C2. The mechanism for
this is not certain, but both defective clearance of apoptotic material/immune complexes and a
shift in the activation threshold of lymphocytes have been suggested.84 As another example,
deficiency of CD55 (or decayaccelerating factor), a cell surface protein that restricts complement
activation, is associated with enhanced T cell responses and exacerbation of neuroinflammation
and lupus in animal models. Thus at many levels, the innate immune system plays a critical role
in maintaining tolerance and controlling autoimmunity.

T Cell Tolerance

T cells are critical players in both achieving and fine-tuning tolerance to a high degree of
specificity. Researchers have identified several mechanisms and divided them into three main
areas: central tolerance wherein T cells first acquire their antigen receptor, peripheral tolerance
wherein T cells encounter self-antigens not present in the thymus, and postactivation regulation
wherein activated and expanded T cell clones are returned to their resting state. Central
tolerance, as alluded to earlier, is imposed on T cells with self-reactive specificities during
thymocyte development by mechanisms using primarily deletion, to a lesser extent anergy, and
possibly receptor editing of the TCR α-chain. This process, although highly effective, is not
completely efficient, and T cells with autoreactivity, primarily those with intermediate to low
affinity to self or recognizing selfantigens poorly expressed in the thymus, emigrate in significant
numbers. This leakiness is probably necessary to generate a broad repertoire, but it then creates
vulnerability to autoimmunity and necessitates peripheral tolerance mechanisms. The latter is
important because T cells are typically first activated in secondary lymphoid organs and
subsequently migrate to target organs, where reactivation by local antigen-presenting cells and
the production of proinflammatory factors leads to tissue damage. Anatomic sequestration alone,
however, is not sufficient for tissues to support immunologic privilege and, as discussed later,
such sites employ a host of additional local mechanisms. Another peripheral mechanism is the
aforementioned two-signal paradigm wherein T cell activation requires both TCR engagement
and a co-stimulatory signal usually provided by CD28. Because the two ligands for CD28, CD80
and CD86, are primarily expressed at high levels on activated professional antigen-presenting
cells, presentation of self-antigen by quiescent antigen-presenting cells would lead to tolerance.
Indeed, immature DCs promote tolerance in this manner by constitutively presenting low doses
of self-antigen on MHC, resulting in cell death or anergy of the corresponding T cells. Peripheral
tolerance of T cells is also maintained by active suppression via immunoregulatory cells of the
immune system, among which CD4+ Tregs are the best characterized. They constitute a distinct
αβ T cell subset generated in the thymus (natural Treg, nTreg) or in the periphery from naïve or
mature CD4+ T cells exposed to TGF-β (induced Treg, iTreg). Both are developmentally
induced by the Foxp3 (forkhead boxP3) transcription factor. They typically express high levels
of the IL-2 receptor component, IL-2Rα (CD25), and require IL-2 for survival. Tregs participate
in every adaptive immune response, are critical for maintaining the proper level of immune
response, and are activated at the same time as conventional T cells. They are thought to
suppress the magnitude of the response by (1) initially downregulating DC function and then
inhibiting T cell activation by competing for IL-2, (2) producing immunosuppressive cytokines
such as TGF-β, IL-10, and IL-35, and (3) by cell-cell inhibitory interactions that lead to cell
killing or the induction of negative signals.95 These inhibitory actions suppress T cells that are in
proximity to Treg cells regardless of their antigen specificity. Researchers have described other T
cells with regulatory activity in autoimmunity including Tr1, CD8 Treg, Qa-1/HLA-Erestricted
CD8 T cells, and γδ T cells, but they are less well characterized. Tissues themselves also employ
mechanisms that suppress self-reactivity and contribute to establishing immune privilege. These
comprise three general categories. First, certain tissues are decorated with cell surface inhibitory
molecules such as the proapoptotic FasL and TRAIL, lymphocyte inhibitory and Treg-promoting
PD-L2, and complement regulatory proteins, CD55 and CD46, that can potentially eliminate or
impede the activation of autoreactive T cells. Second, soluble inhibitors of inflammation and
immune activation are secreted by particular tissues. Notably, in the aqueous humor of the eye,
there is a broad spectrum of such factors that include TGF-β, α-melanocyte-stimulating hormone,
vasointestinal peptide, calcitonin gene–related peptide, somatostatin, macrophage inhibitory
factor, and complement inhibitors. Third, lymph node resident stromal cells have been shown to
induce tolerance of CD8+ T cells recognizing peripheral tissuerestricted self-antigens. Thus, it
has been proposed that stromal cells in lymph node and tissues may provide a means to eliminate
T cells that bind to tissue-restricted antigens not expressed in the thymus. Fourth, the anterior
chamber of the eye elicits a unique type of altered immune response through a complex multistep
process termed anterior chamber-associated immune deviation (ACAID) that leads to a
dampened and less tissue-destructive response. Although ACAID was long thought important for
tolerance, it was recently argued that its primary function may be to modulate the immune
response so that the eye can respond to infection without damaging its integrity. Another
possible peripheral mechanism is immune deviation wherein polarization away from a
predisposing cytokine pattern, such as from a Th1 to a Th2 profile, suppresses the development
of autoimmune disease. Similarly, activation of NKT cells with α-GalCer, which induces IFN-γ
production, is associated with dampening of the adaptive Th1 and Th17 effector responses and
protection from experimental autoimmune uveitis. In these models, autoreactive T cells are
activated but do not produce the proinflammatory factors necessary for tissue damage. In
addition to central and peripheral tolerance, the immune system must also avert autoimmunity by
suppressing or eliminating T cells following their activation or expansion. This regulation is
mediated by several processes, including upregulation of inhibitory receptors such as CLTA4,
expression of proapoptotic receptors like Fas, and release of intracellular proapoptotic factors
such as Bim. Deficiencies of such mediators that control the magnitude of T cell responses are
associated with severe expansion of lymphocyte populations and with varying degrees of
autoimmunity.

B Cell Tolerance

B cells are required not only for antibody production but also serve as potent antigen-
presenting cells for T cells and follicular DCs, and can act in regulation as well. Moreover,
depletion of B cells with rituximab has shown promise even in autoimmune diseases considered
to be mediated by T cells such as T1DM112 and multiple sclerosis. Therefore, there is
substantial interest in defining both mechanisms of B cell tolerance and the specific role of B cell
tolerance in autoimmune disease. Before discussing specific tolerance mechanisms, however, it
should be emphasized that the fate of B cells after antigen receptor (BCR) cross-linking is highly
dependent on developmental stage, context and strength of signal, and the nature of the antigen,
probably more so than for T cells because B cells are subjected to less stringent selection during
central tolerance. Several checkpoints considered important for controlling autoreactive B cells
and for maintaining self-tolerance have been identified and include many central and peripheral
mechanisms similar to those described for T cells, as well as a few additional ones. Central
tolerance of B cells takes place in the bone marrow during preB to immature B cell transition as
they express rearranged immunoglobulin (Ig) genes on their surface.114 It appears that the
dominant mechanism for B cells with high affinity to membrane-bound self-antigen is receptor
editing (replacement of L-chain) and, to a lesser extent, deletion, while soluble self-antigens
induce both receptor editing and anergy. Anergic B cells are detectable in the periphery as an
IgD+ IgM− population or in mice as splenic transitional 3 (T3) B cells. They are shortlived at
least in part because they downregulate the BAFF receptor, which is required for their survival,
putting them at a competitive disadvantage with other immature B cells, and they are less able to
enter B cell follicles. In the periphery, the earliest tolerance checkpoint occurs at the transitional
1 (T1) B stage over a 2-day interval before maturation to T2 and later naïve B cell subsets. T1 B
cells are the immediate bone marrow emigrant population, retain an immature phenotype, and
are BAFF-dependent for survival. Importantly, they undergo apoptosis and not activation when
stimulated, which results in deletion of B cells recognizing peripheral self-antigens not expressed
in the bone marrow compartment. Thus, this mechanism, which is unique to B cells, represents
in essence an extension of central tolerance to the periphery. Other peripheral tolerance
mechanisms are achieved through many of those described earlier for T cells but differ
qualitatively due to distinct differentiation pathways and differences in antigen recognition by B
cells and T cells (i.e., BCRs can bind to virtually all tertiary structures while TCRs are restricted
to recognizing self-MHC/peptide complexes on host cell surfaces). Accordingly, B cells can be
ignorant of their corresponding self-antigen because of insufficient quantity or access19 or can
undergo anergy and ultimately cell death if there is engagement of the BCR without co-
stimulation (i.e., two signals). Another notable checkpoint occurs during T cell– dependent
immune responses as B cells undergo affinity maturation in germinal centers (GCs) and acquire
new specificities, which may include self-reactivity. Evidence suggests that tolerance at this
juncture is often defective in autoimmune diseases because most autoantibodies have acquired
autoreactivity through somatic hypermutation and are class-switched, both indicative of GC
maturation. Although recent studies have provided significant insights into GC processes
involved in the selection of B cell clones with high affinity to foreign antigens, how tolerance of
class-switched autoreactive B cells is achieved remains to be clarified. Nevertheless, the
strongest evidence suggests (1) autoreactive B cells compete poorly for the cognate T cell help
essential for GC B cell survival because the autoreactive BCR would bind less well to the
original antigen, resulting in less internalization of antigen for processing and presentation to T
cellsand (2) B cells that acquire BCRs with high affinity to membrane antigens are deleted by a
Fas-dependent mechanism.
Autoimmunity Caused by Activation of Intolerant or Partially Tolerant T Cells

Another theory is that autoimmunity develops through the conventional activation of self-
reactive T cells that have not been deleted in the thymus and remain oblivious to the
corresponding self-antigen after emigration. Such ignorant T cells, commonly found in the
periphery of both normal humans and animals, can be activated by antigen presented by
professional antigen-presenting cells in the context of an innate inflammatory milieu. Once
activated, T cells can gain access to virtually all tissues and, when activated again locally, can
elaborate proinflammatory factors causing tissue damage. Breach of tolerance by this mechanism
is most often associated with organ-specific diseases, presumably because tissue-specific
antigens are less likely to be expressed in the thymus. This theory is supported by the finding that
type 1 diabetes is prevented in BB rats and NOD mice by intrathymic transplantation of
pancreatic is lets or inhibited in NOD mice by intrathymic administration of GAD, a major β-
islet autoantigen in this model. Similarly, EAE can be prevented by intrathymic administration of
either the immunizing antigen, myelin basic protein, or its major encephalitogenic epitope, and
autoantibody production can be delayed in lupus-prone mice following intrathymic injections of
polynucleosomes.154 Taken together, these data support the concept that autoimmunity can be
caused by the activation of lymphocytes to selfantigens for which there was incomplete central
tolerance. This is similar to the mechanism underlying the APECED syndrome caused by AIRE
deficiency, but in this instance central tolerance is not known to be defective. Breach of tolerance
by ignorant lymphocytes has been shown to depend on many factors including the (1) nature of
the self-antigen, (2) extent of exposure to antigen, (3) antigen receptor affinity, (4) frequency of
autoreactive lymphocytes, (5) types and levels of co-stimulatory molecule expression, (6)
cytokine and chemokine profiles, and (7) presence of inflammation. It should also be emphasized
that despite the presence of lymphocytes that recognize self-antigen, peripheral tolerance
mechanisms, under normal conditions, are difficult to overcome. Consequently, although the
experimental autoimmune models are highly reproducible, they require supraphysiologic
amounts of self-antigen, strong adjuvant, specific MHC haplotypes, and a susceptible
background to break tolerance. Nonetheless, based on largely experimental evidence, researchers
have identified a wide range of mechanisms that can lead to loss of tolerance, activation of
autoreactive lymphocytes, and autoimmune disease. Although their specific contributions to
human diseases are not yet known, they provide a framework for understanding how
autoimmunity can develop.

PATHOPHYSIOLOGY OF AUTOIMMUNE RHEUMATIC DISEASES

Although the general principles underlying loss of tolerance and autoimmunity provide a
useful conceptual foundation, the actual pathophysiologies—suggested by the few betterdefined
autoimmune diseases—are likely to also employ specific and unique mechanisms. Two notable
examples of this are the pathophysiology of antinuclear antibody production in SLE and arthritis
in RA. Brief discussions in the context of their broader significance to autoimmune diseases
follow. Recent findings suggest a model of SLE pathophysiology that provides a mechanism for
why antinuclear antibodies are virtually always present in lupus despite substantial genetic and
clinical heterogeneity.36 First, autoreactive B cells activate when self-reactive BCRs bind to
nucleosomes or RNPs and internalize them into the endolysosome compartment, where the
released nucleic acids engage TLR7 and TLR9 and provide a second signal. Such activated B
cells act as potent antigen-presenting cells for T lymphocytes, and following class-switch
recombination they produce IgG autoantibodies. Next, immune complexes formed by the
binding of these autoantibodies to nucleic acid–containing material activate pDCs and DCs
following their internalization via FcγRIIA (FcγRIII in mice) and then engagement of TLRs by
nucleic acids. Elaboration of lupus-promoting cytokines such as type I IFN and BAFF by pDCs
and DCs, as well as enhanced antigen presentation, are thought to further drive loss of tolerance,
activation of autoreactive B cells, and autoantibody production, thereby resulting in an
amplification loop. Thus in lupus-susceptible individuals, confinement of nucleic acid–binding
TLRs to endolysomes is not enough of a barrier to block their activation by normally innocuous
amounts of self-nucleic acids. This mechanism provides an explanation for the high prevalence
of autoantibodies in SLE that either bind to nucleic acid or nucleic acid–complexed self-antigens
such as DNA, nucleosomes, RNP, and myeloperoxidase (ANCA) or that exhibit cross-reactivity
to such antibodies. Importantly, this also raises the possibility that other autoimmune diseases
might also be mediated by specific PRRs. The pathophysiology of RA also involves common
and specific pathways that together promote inflammatory synovitis in roughly two phases,
which has been best described for the HLA-DR1*401 (DR4) anti cyclic citrullinated peptide
(CCP)+ subset of RA patients with greater disease severity. The initial phase is postulated to
involve the activation of T cells and B cells and the production of autoantibodies, including RF
and anti-CCP. Targets for anti-CCP include citrullinated fibrinogen, vimentin, and α-enolase,
and these antibodies are thought to play a major role in disease pathogenesis, although the
mechanism remains controversial. The specific triggers of the anti-CCP response are also not
known, although it is postulated that release of peptidylarginine deaminase from apoptosing
granulocytes and monocytes may be a contributing factor. The resulting immune complexes and
activated T cells lead to stimulation of macrophages, synovial fibroblasts, endothelial cells, mast
cells, and osteoclasts, and the eventual production of proinflammatory factors such as TNF, IL-1,
IFN-γ, chemokines, matrix metalloproteinases, osteopontin, and many others that contribute to
synovitis, pannus formation, bone erosion, and cartilage destruction. This later inflammatory and
tissue damaging phase is largely mediated by activated fibroblast-like synoviocytes, which
produce a wide array of proinflammatory mediators that promote recruitment and activation of
circulating and resident immune cells. It has also been suggested that the spread of arthritis to
unaffected joints could, in fact, be mediated by transmigration of these activated fibroblast-like
synovial cells. The pathophysiology of RA provides an additional example of the collaboration
of the innate and adaptive arms of the immune system in autoimmune disease but also has two
unique features; first, the major autoantigen, citrullinated protein, is a neoantigen specific for this
disease, and second, tissue damage is ultimately mediated by fibroblast-like synoviocytes.