Systemic lupus erythematosus (SLE) is a disease of multiples. This extends from the clinical subsets and the likely genetic predisposition to the animal models that appear to have striking similarity to components of the human disease. Similarly, every year, the number of autoantibodies associated with lupus increases, although their provenance and role in disease pathogenesis remain unclear. Theoretically, autoantibodies could be `blamed' for many of the manifestations of disease; a recent report suggests that antibodies directed against erythropoietin might be responsible for the anaemia observed in some patients . It is, however, antinuclear antibodies that have received most attention and research has focused on whether (and how) they directly cause disease (in particular nephritis), and their origins. Significant advances have been made in these areas in recent years.
It was ∼40 yr ago that the association between anti-DNA antibodies and lupus was first described. Subsequently, it has become clear that this association reflects the fact that a subset of antinuclear antibodies can directly cause overt nephritis, one of the major causes of morbidity and mortality in SLE. The evidence for this contention has come from many sources. Thus, a variety of antinuclear antibodies can be eluted from diseased glomeruli [2, 3] and in many patients high titres of anti-DNA antibodies correlate with (and predict exacerbations of) nephritis . Furthermore, a proportion of both mouse and human monoclonal anti-DNA antibodies can cause nephritis when transferred into non-autoimmune mice [5–7]. Different monoclonal autoantibodies, when administered to mice, resulted in various patterns of immunoglobulin deposition and histological profiles. Some high-affinity IgG anti-double-stranded (ds) DNA antibodies did not cause nephritis. The clear implication from these experiments is that the different clinical and histological presentations of lupus nephritis may in part be due to differences in the specificity of the autoantibodies produced.
What is less clear are the defining characteristics that govern the nephritogenicity of a particular antibody; it does not appear simply to reflect the affinity for DNA. Indeed, lupus nephritis can occur in patients without any evidence of circulating anti-DNA antibodies and, conversely, persistent high serum titres of IgG anti-dsDNA antibodies can occur without clinical evidence of renal injury . In one study, the pathogenicity of a monoclonal anti-DNA antibody was lost after site-directed mutagenesis of its antigen binding region despite retaining DNA binding activity . One explanation for the imperfect association between binding to DNA and pathogenicity is that the nephritogenic antinuclear antibodies are cross-reacting with glomerular antigens . An alternative mechanism whereby antibodies can cause nephritis is through the deposition of immune complexes, i.e. antibody plus nuclear antigens . To what extent direct binding to glomerular antigens or immune complexes contribute to nephritis in lupus is still unclear. Despite these controversies, it is the measurement of high-affinity anti-DNA antibodies that remains the best autoantibody marker of lupus nephritis in the majority of patients.
One intriguing claim that has arisen (or perhaps been revisited) from studies of monoclonal anti-DNA antibodies is that some of these immunoglobulins can penetrate living cells and react directly with the nucleus . It has been proposed that this subset of antibodies bind to myosin 1 on the cell surface, become internalized, cause glomerular hypercellularity and subsequently proteinuria . Indeed, a peptide derived from one of these antibodies has been shown to be able to act as a vector for the intranuclear delivery of macromolecules . These autoantibodies are not specific for glomerular cells, but have the ability to penetrate a wide variety of cells. It has been speculated that after these antibodies have been internalized, they modulate the process of apoptosis through an interaction with DNase I . This modulation of apoptosis could have a major impact on inflamed tissues and perhaps explain the glomerular hypercellularity observed when these antibodies are administered to normal mice.
The observation that some antinuclear autoantibodies can penetrate cells and bind in the nucleus is unlikely to be the main mechanism that accounts for the pathology observed in lupus nephritis. Once the self-reactive antibody has found its way to the glomerulus, it has two main effector pathways which could lead to damage: through the activation of complement and/or via the ability of IgG antibodies to bind to Fc receptors, thus recruiting other cellular effector mechanisms. Convincing evidence has demonstrated that the latter pathway can account for much of the injury observed. When NZB/W mice, which develop a nephritis very similar to that seen in lupus patients, were crossed with mice deficient in the Fc receptor (gamma chain) for IgG, renal injury was prevented and the mice had a prolonged survival . These mice still had immunoglobulins deposited in their glomerulus along with complement, but showed no severe pathological response. Thus, this pathway represents a potential target for therapy of lupus nephritis.
The story with complement appears to be more complex, but reveals some fascinating insights into the workings of this cascade. Deficiencies in the early complement components C1q or C4 exacerbate the tendency to develop lupus nephritis, whereas if one loses C3, a central player in the workings of complement, no effect is seen [16–18]. Thus, it appears that complement is not required to effect the tissue damage observed when immunoglobulins are bound to the glomerulus. Rather, a number of properties have been ascribed to some complement proteins that lead to protection against autoimmunity. These include removal of immune complexes, binding to and removal of apoptotic blebs containing potentially immunogenic autoantigens, and the regulation of tolerance against soluble autoantigens. In this last role, it has been proposed that early components of complement bind to self-antigens, boosting the signal delivered to immature self-reactive B cells, leading to their deletion.
The observation that C1q deficiency leads to an increase in the presence of apoptotic blebs in the glomerulus adds strength to an earlier observation that provided a framework to explain the presence of the most frequently reoccurring autoantibodies . Autoantigens such as DNA, histones, Ro, La and RNP were found to be present in structures on the surface of apoptotic cells. Thus, it was speculated that one of the first cellular events in SLE is an increase in apoptosis of certain cells, leading to the accessibility of antigens to trigger autoimmunity. This notion gains further support with the observations that lupus patients have high numbers of apoptotic cells and that these cells release large quantities of oligonucleosomes. In mouse models of lupus, antinucleosomal antibodies are present before anti-DNA and anti-histone antibodies .
What the initiating genetic and/or environmental triggers might be is still unclear. However, experiments where lupus is induced in non-autoimmune mice by manipulating their immune system demonstrate that a lupus-type reaction can be achieved in a variety of ways. These can be divided into two main groups. The first are experiments where lupus is induced by immunizing mice with a range of antigens and the second involves genetically altering mice to induce disease, usually by removing a component of the immune system. A recent example of the first illustrates how the initiating antigen might not contain nucleosomes. Putterman and Diamond  screened a peptide library with an anti-dsDNA antibody and isolated a peptide that, if immunized into BALB/C mice (a non-autoimmune strain), induced lupus nephritis. Here, they argue that a cross-reactive epitope induces antibodies against DNA and also results in the appearance of antibodies binding cardiolipin and histones, and the development of renal pathology. This last observation highlights the potential phenomenon of epitope spreading where the initiating antigen leads not only to antibodies reactive with itself, but also other, often related, antigens. This type of process could provide part of the explanation for the diverse set of autoantibodies observed in lupus patients and hinder the identification of the `initiating' antigen.
Examples of the second type of experiment, i.e. manipulating the genetic make-up of a mouse, are plentiful. The relevance of some of these experiments is reinforced by a parallel genetic predisposition in lupus patients. As mentioned above, deficiencies in the complement components C1q and C4 both predispose to lupus in humans and mice [16–18]. The list of other immune deficiencies that can lead to a lupus-like disease with antinuclear antibodies varies from hyperactivity in B cells to a more broad dysregulation of cellular responsiveness and proliferation (e.g. [22–23, 24]). The multiplicity of genetically manipulated mice that lead to a lupus-type reaction is reflected by the number of genes that have been proposed to contribute to lupus in man.
The range of new findings have provided a framework for discriminating consequence from subsequence. What is apparent is that lupus autoantibodies can occur by a variety of genetic and environmental causes, and their presence can directly lead to tissue injury. What is now required is precise information on the sequence of critical events that cause lupus in patients and how these can be manipulated to reduce disease.