Glomerulonephritis: Recent Advances in Understanding of

Pathogenesis: Introduction
real progress in the prevention and treatment of glomerular disease depends
heavily on a comprehensive understanding of the underlying pathogenetic
mechanisms of glomerular injury. Although animal models of disease certainly have
contributed importantly to our current comprehension of glomerular diseases in
human beings, they cannot substitute for knowledge of mechanisms relevant to
the diseases as they occur in real patients. In the recent past, significant
achievements concerning the pathogenesis of glomerular diseases have emerged.
It is the purpose of this issue of Seminars in Nephrology to review this new
information, primarily from
the perspective of human disease. Nine state-of-the art reviews encompassing the
most commonly encountered primary and secondary glomerular diseases are
included. The evolving views on the mechanisms of minimal change disease are
reviewed by Takuji Ishimoto and colleagues. Long held to be a disorder of T cell
biology, we are now gaining a much better glimpse of the potential antigenic
targets in this disease. This information promises much in the way of newer
diagnostic and treatment approaches. The lesion of focal segmental
glomerulosclerosis is addressed by Alain Meyrier. The pathogenetic heterogeneity
of this morphologic expression of glomerular injury is emphasized with particular
focus on damage to the podocyte as a central unifying concept. Idiopathic
membranous nephropathy is discussed by Sudesh Makker and Alfonso Tramontano.
This disease has been well studied in its closest animal model, Heymann nephritis,
and these studies have gleaned many insights relevant to the human disease.
However, major breakthroughs have developed in our understanding of the human
disease that likely will bring new advances in diagnosis and treatment. The
confusing area of membranoproliferative glomerulonephritis is addressed in a
compelling and novel way by Sanjeev Sethi and Fernando Fervenza. Building on
recent advances in understanding of the diverse biology underpinning this pattern
of glomerular injury, they present a new classification schema that should prove
useful in the clinic. Jonathan Barratt and John Feehally review the two hit
hypothesis for IgA nephropathy and comprehensively analyze the newer
information of the role of aberrant galactosylation of IgA1 in this very common
disease. Ruth Tarzi and colleagues provide a detailed and erudite examination of
the mechanisms underlying crescentic glomerulonephritis. Perhaps more than any
other glomerular disease, this group of immune-mediated diseases has benefitted
from a more complete understanding of underlying pathogenesis, but much work
still needs to be performed. Tibor Nadasdy and colleagues give an update on
mechanisms of glomerular injury in infections, emphasizing the changing patterns
of disease. Although many infectious diseases have been reduced in prevalence in
developed countries, this is still an important cause of morbidity and mortality from
kidney-related complications. Finally, lupus nephritis is reviewed by Johan van der
Vlag and Jo Berden. The role of auto-antibodies to nucleosomes is elegantly
reviewed. The concepts derived from this analysis have direct implications for both
diagnosis and treatment of this often devastating disease. It is hoped that
collecting the accumulating wisdom on the pathogenesis of glomerular diseases
into a single accessible site will enhance the knowledge base of nephrologists in
this important branch of the discipline. As modern biology advances, many of the
issues and unresolved areas exposed by this critical review will yield to clarity.
Translation of these advances into meaningful improvements in prevention,

diagnosis, and treatment of glomerular disease remains a major challenge. Finally,

I am deeply indebted to the contributors to this issue for their outstanding and
timely efforts. I am also very grateful to Joseph Bonventre for inviting me to serve
as Guest Editor. It has been an enjoyable and enlightening exercise.

Minimal Change Disease: A CD80 podocytopathy?

Summary: Minimal change disease is the most common nephrotic syndrome in children. Although the etiology of minimal
change disease remains to be elucidated, it has been postulated that it is the result of a circulating T-cell factor that causes
podocyte cytoskeleton disorganization leading to increased glomerular capillary permeability and/or changes in glomerular
basement membrane heparan sulfate glycosaminoglycans resulting in proteinuria.
Minimal change disease has been associated with allergies and Hodgkin disease. Consistent with these associations, a role
for interleukin-13 with minimal change disease has been proposed. Furthermore, studies evaluating podocytes also have
evolved. Recently, increased expression of CD80 (also termed B7-1) on podocytes was identified as a mechanism for
proteinuria. CD80 is inhibited by binding to CTLA-4, which is expressed on regulatory T cells. Recently, we showed that
urinary CD80 is increased in minimal change disease patients and limited studies have suggested that it is not commonly
present in the urine of patients with other glomerular diseases. Interleukin-13 or microbial products via Toll-like receptors
could be factors that induce CD80 expression on podocytes. CTLA-4 appears to regulate CD80 expression in podocytes, and
to be altered in minimal change disease patients. These findings lead us to suggest that proteinuria in minimal change
disease is caused by persistent CD80 expression in podocytes, possibly initiated by stimulation of these cells by antigens or
cytokines. Semin Nephrol 31:320-325 2011 Elsevier Inc. All rights reserved.
Keywords: Minimal change disease, CD80, podocyte, proteinuria

Minimal change disease is the most common nephrotic syndrome in children,

accounting for 70% or more of cases. The disease can be dramatic in presentation,
resulting in marked weight gain, disfiguring edema, and severe increases in serum
cholesterol level (Fig. 1). Minimal change disease,when untreated, is associated
with increased early mortality owing to a higher risk of infections and thromboses.
A few decades ago, mortality rates were reported to be as high as 30% by 2.5
years. Persistent nephrotic syndrome also may increase the risk for late mortality
from coronary artery disease.
The introduction of adrenocorticotropic hormone and corticosteroids in the early
1950s resulted in a remarkable decrease in mortality, but also has been associated
with stunted growth in children as well as other well-known consequences of
corticosteroid use. Thus, identifying the underlying pathogenesis of this important
condition will lead to more targeted and effective treatments. In this article we
review current concepts on the pathogenesis of minimal change disease.

HISTORICAL ASPECTS
Although nephrotic syndrome was recognized by Richard Bright in the 1820s, its
distinction as a separate manifestation of kidney disease did not occur until the
early 1900s. By the 1930s, however, the clinical description of nephrotic syndrome
in both children and adults was established. In these early days the nephrotic
syndrome often was referred to as lipoid nephrosis because of the creamy lipidladen serum and the large white kidneys that contained cholesterol esters.
Although the various features of nephrotic syndrome were all noted in early
reports, it was not until the early 1960s that the syndrome was defined in adults by
George Schreiner as the well-familiar pentad of proteinuria greater than 3.5 g/d,
serum albumin level less than 3 g/dL, cholesterol level greater than 300 mg/dL,
lipiduria, and pitting edema. Only a few years later Drummond et al showed that

the nephrotic syndrome of childhood, associated with corticosteroid sensitivity, was

characterized histologically by minimal glomerular abnormalities and absent
immune globulin or complement deposition. Thus, minimal change disease as a
unique entity causing nephrotic syndrome was first identified. One of the first
observed associations of nephrotic syndrome, and particularly with lipoid
nephrosis, was with allergies. Nephrotic syndrome occasionally was observed in
subjects with allergic diatheses, including pollen sensitivity, allergic rhinitis, or hay
fever. Rarely, nephrotic syndrome would be precipitated by an allergic reaction,
such as a bee sting or after contact with poison oak. Furthermore, subjects with
minimal change disease often have increased IgE levels in their plasma, even in
the absence of identifiable allergies. Although there does appear to be an
association between minimal change disease and allergies, the studies that have
evaluated this link were not always well controlled and the associations were
relatively weak.
Another early observation was that infection with measles could commonly induce
remission in minimal change disease patients, especially after the resolution of
fever or during the early convalescence period. These observations even led to
pilot studies in which there was purposeful inoculation of nephrotic children with
measles, the majority of whom showed some improvement in their renal disease.
However, measles infection occasionally can result in meningitis and other severe
complications, so this form of treatment was abandoned.
A final observation was that minimal change disease could occur in subjects with
Hodgkin lymphoma. In these subjects the nephrotic syndrome sometimes preceded
the diagnosis of lymphoma and often resulted in greater steroid resistance or
dependence than the typically corticosteroid-sensitive minimal change disease
observed in children. Moreover, effective treatment of Hodgkin lymphoma was
associated with resolution of nephrotic syndrome in these subjects.
Early Clues to Pathogenesis: T Cells and the Shalhoub Hypothesis
The first major hypothesis for the pathogenesis of minimal change disease was
advanced by Shalhoub in 1974. Shalhoub proposed that minimal change disease
resulted from a disorder in cell-mediated immunity, and particularly of T cells. He
based his argument on several observations. First, there was an absence of
immunoglobulin and complement in the glomeruli observed in renal biopsies,
suggesting an absent humoral response. Second, the rapid response to
corticosteroids suggested a T cell disorder because corticosteroids were
particularly effective at blocking cell-mediated immune responses. Furthermore,
Hodgkin disease was thought to be a T cell derived lymphoma, raising the
possibility that the lymphomatous cells might be producing a factor that could
increase glomerular permeability. Finally, it was recognized that cell-mediated
immune responses often were blunted after measles infection, which, as described
earlier, led to remission of the nephrotic syndrome in some patients. After
Shalhoub presented his hypothesis, evidence for activation of T cells in minimal
change disease during relapse was reported. Increased levels of interleukin (IL)-2 in
lymphocyte supernatants of minimal change disease subjects have been described
as a marker of T-cell activation. Additional studies suggested a relative
predominance of a T helper 2 cytokine response, with relatively higher levels of IL-4
expression by peripheral T cells. However, some studies reported activation of T
helper 1 cytokine responses as well, with increased interferon. When looking at the
same cytokine profiles, other investigators have found opposing or negative

results. The often confusing and conflicting data have made it difficult to sort out
the role of a specific cytokine in the pathogenesis of minimal change disease.
Nevertheless, the observation that proteinuria can be induced by the infusion of
peripheral blood mononuclear supernatants from subjects with minimal change
disease has maintained interest in this pathway as a potential pathogenic
mechanism. Furthermore, Koyama et al reported that supernatants from
immortalized T cell hybridomas from patients with minimal change disease could
induce massive proteinuria with foot process fusion in rats. Regarding this last
observation, Koyama et al have not been able to duplicate their initial report. The
mechanism of proteinuria induced by cytokines has been postulated to be
mediated by changes in the negative charge of the heparan sulfate
glycosaminoglycans at the level of the glomerular basement membrane and/or the
disruption of the slit diaphragm by changes induced at the level of the podocyte. In
this regard, two lymphocyte cytokines appear to be increased persistently in
minimal change disease subjects and hence have been considered candidates for
causing the increased urinary protein excretion. IL-8, made by macrophages and
some subsets of T cells, is increased in the sera and peripheral blood mononuclear
cell supernatants from subjects with minimal change diseases. Furthermore,
infusion of IL-8 in the renal artery of rats causes increased turnover of heparan
sulfate proteoglycan in the basement membrane associated with low-grade
proteinuria. These effects can be blocked by antiIL-8 antibodies. Nevertheless,
massive proteinuria did not occur, which makes this cytokine an unlikely candidate
for being the only factor causing nephrotic syndrome in children with minimal
change disease. In addition, although reduced heparan sulfate glycosaminoglycan
in the glomerular basement membrane of subjects with minimal change disease
has been reported, there is debate over whether these changes are responsible for
the proteinuria observed IL-13 is another lymphokine considered a strong
candidate for mediating minimal change disease proteinuria.
IL-13 consistently has been shown to be increased in the serum and T cells of
patients with minimal change disease. Moreover, the overexpression of IL-13 in rats
can induce nephrotic proteinuria with histologic features suggestive of minimal
change disease. Receptors for IL-13 have been shown in podocyte cell culture and
isolated glomeruli of human beings and rats, providing a direct mechanism by
which this cytokine might induce podocyte cytoskeletal changes and proteinuria. IL13 also is increased in patients with atopy, providing a linkage with the reported
association of allergy with minimal change disease. Moreover, IL-13 is an autocrine
growth factor for Reed-Sternberg cells in Hodgkin lymphoma. Genetic
polymorphisms in IL-13 also have been associated with both the onset and course
of minimal change disease in children from Indonesia and Singapore, respectively.
However, these associations were not shown in subjects with minimal change
disease from Europe. Another vexing problem is that IL-13 levels remain increased
in the sera of patients with minimal change disease even after remission is
induced. Thus, a search for other mechanisms to account for nephrotic syndrome
has continued.
CD80 (B7.1) and the Role of the Podocyte in Minimal Change Disease
Although early studies focused on the role of lymphocytes and humoral factors in
the pathogenesis of proteinuria in minimal change disease, a shift in interest
toward the podocyte has occurred. This shift follows the recognition that nephrotic
proteinuria may be caused by an alteration in slit diaphragm integrity. In this

regard, a major breakthrough occurred when Reiser et al identified a role for CD80
expression on podocytes as a mechanism for proteinuria. CD80 (also termed B7-1)
is a transmembrane protein expressed on antigen-presenting cells (APCs), natural
killer cells, and B cells. CD80 provides a co-stimulatory signal for T cells by binding
to its receptor CD28. In the context of antigen recognition, the interaction of CD80
on APCs with CD28 is required for T-cell activation. Inhibition of T-cell activation
also occurs by the binding of CD80 by CTLA-4, which is a protein expressed on
Foxp3_ regulatory T cells (Treg). CTLA4 also down-regulates CD80 and CD86
expression on APCs such as dendritic cells. Reiser et al showed that podocytes can
express CD80 under certain conditions, such as after exposure to
lipopolysaccharide (LPS). In vivo CD80 also was induced not only by LPS, but also
by podocyte toxins such as aminonucleoside. Most importantly, the expression of
CD80 by podocytes resulted in podocyte shape change and transient mild
proteinuria. CD80 knockout mice were protected from LPS-induced podocyte shape
change and proteinuria, documenting that CD80 was responsible for the
proteinuria. Furthermore, SCID mice still developed proteinuria in response to LPS,
thus showing that the proteinuria can occur independently of T cells. Clinically, the
investigators showed that glomeruli from patients with lupus nephritis also
expressed CD80. Although they did not examine patients with minimal change
disease, they did propose that CD80 expression by podocytes might be involved in
its pathogenesis. Our group has confirmed that CD80 is abnormal in minimal
change disease. Our first observations showed that urinary CD80 is increased in
subjects with minimal change disease. The source of the urinary CD80 appears to
be the podocyte, as determined by Western blot showing that the CD80 observed
in the urine was the 53-kd intact molecule and not soluble CD80 (molecular weight,
23 kd) present in the circulation. In addition, limited biopsies of patients with
minimal change disease documented the presence of CD80 in the podocytes, with
no expression of CD80 in other parts of the kidney. In other glomerular diseases the
urinary CD80 excretion is low. Hence, increased CD80 urinary excretion is relatively
specific for minimal change disease. For example, in a series of 22 subjects with
focal segmental glomerulosclerosis (FSGS), only 1 subject had increased urinary
CD80 excretion. Additional studies in other glomerular diseases are necessary
before the overall specificity of this new assay can be verified. So, if podocyte
CD80 expression is involved in the pathogenesis of the proteinuria of minimal
change disease, which factor could be driving its expression? One possibility is IL13. In rats, IL-13 overexpression was associated with increased CD80 in their
podocytes and was manifested by proteinuria. The observation that LPS could
induce CD80 expression also raised the potential role of Toll-like receptors (TLRs)
because LPS activates cells via TLR-4. Consistent with this possibility, we have
found that CD80 expression can be induced in human podocyte cultures using TLR
ligands for TLR-3 and TLR-4 in association with cell shape change and actin
skeleton rearrangement (Shimada et al, unpublished data). We also have been able
to induce transient proteinuria in mice via injection of the TLR-3 ligand,
polyinosinic-polycytidylic acid (polyIC), in association with transient glomerular
CD80 expression and increased urinary CD80 excretion (Ishimoto et al, unpublished
data).
These studies suggest that CD80 expression could be induced by either T-cell
cytokines or TLR ligands. TLR ligands are commonly microbial products, and would
provide a linkage with the well-known association of viral infections as a
precipitating factor for minimal change disease. If CD80 is induced by TLR ligands
or T-cell cytokines, then which is the normal mechanism for terminating this
response? In this regard, CTLA-4 is both expressed and secreted by Treg, and, as

mentioned, CTLA-4 is known to down-regulate CD80 expression in dendritic cells

and other APCs. We have found that Tregs are abnormal in subjects with minimal
change disease. Likewise, minimal change disease has been reported in subjects
with mutations in FoxP3, which is a key transcription factor expressed by Treg.
Nonsteroidal agents, which occasionally can precipitate minimal change disease,
also can impair Treg function, whereas rituximab, which occasionally can benefit
minimal change disease, is associated with an improvement in Tregs. The induction
of Tregs also has been reported to induce remission of the nephrotic syndrome of
the Buffalo rat, which is thought to be an animal model of minimal change disease
or FSGS. Finally, we have some evidence that podocytes themselves can express
CTLA-4 and that the expression appears to regulate podocyte CD80 expression (C.
Rivard, unpublished data). Urinary CTLA-4 also tends to be low in patients with
minimal change disease during relapse. Hence, CTLA-4 appears to be a mechanism
for regulating CD80 expression in podocytes that may be altered in subjects with
minimal change disease. Based on the earlier-described findings, we postulate that
minimal change disease is a two-hit podocyte immune disorder. The first hit is the
induction of CD80 by a microbial agent and/or T-cell cytokine, and the second hit is
the ineffective censoring of CD80 expression owing to an inadequate CTLA-4
response (Fig. 2).
Minimal Change Disease and the Graft-Versus-Host Response
The concept that the podocyte may be induced to express CD80 suggests that in
essence it is functioning as a dendritic cell reacting to some allergen or antigen.
Because the podocyte cannot directly engage the T cell owing to the intervening
basement membrane, the teleologic function of this response is unclear. One
possibility, which has been proposed by Reiser and Mundel, is that increasing
glomerular permeability facilitates the excretion and removal of the toxin or
antigen. The possibility that the podocyte expresses CD80 as a reaction to antigens
could explain the long-standing association of minimal change disease with the
graftversus- host reaction in subjects undergoing bone marrow transplantation.
This also may account for the rare development of minimal change disease in the
renal transplant patient. Indeed, we have preliminary evidence that glomerular and
urinary CD80 are expressed in the nephrotic syndrome associated with xenografts
in studies performed in collaboration with David Sachs at Massachusetts General
Hospital in Boston (unpublished data). In these cases the podocyte likely is
responding to host antigens.
Other Potential Mechanisms and the Relationship of Minimal Change Disease to
FSGS
Another interesting model of minimal change disease has been reported in which
CD34_ stem cells from subjects with minimal change disease have been injected
into severely immunocompromised (NOD/SCID) mice. Interestingly, the injection of
CD3_ cells in these mice is known to result in a complete expression of human
dendritic cell subsets. This suggests that engraftment of dendritic cells from
subjects with minimal change disease might be able to induce nephrotic syndrome,
and again switches the focus to the role of CD80. Other potential mediators of
proteinuria are emerging in nephrotic syndrome, including angiopoietin-like 4 (S.
Chugh, personal communication) and circulating urokinase receptor (J. Reiser,
personal communication). Additional studies are necessary to identify their role in
minimal change disease and how they might interface with the IL-13 and CD80
hypotheses. However, it is becoming more and more evident that
corticosteroidsensitive minimal change disease is likely distinct in mechanism from

idiopathic FSGS. Although there is no doubt that the latter is owing to a circulating
factor, the lack of urinary CD80 in these subjects argues for another pathway. In
addition, Savin et al, when describing the circulating factor in FSGS, could not find
such a factor in patients with minimal change disease. In conclusion, we suggest
that proteinuria in minimal change disease is associated with increased expression
of CD80, by podocytes activated by foreign antigens, or cytokines. We suggest that
a key inability to censor the CD80 response has a major role in causing persistent
proteinuria and nephrotic syndrome. It will be interesting if stimulation of CTLA-4
may represent a novel way to control this disease.

Lupus Nephritis: Role of Antinucleosome Autoantibodies

Summary: The discovery of autoantigen clustering in blebs at the surface of apoptotic cells boosted research on the role
of apoptosis in systemic lupus erythematosus (SLE) and led to the discovery of autoantigen modification during apoptosis.
Normally, apoptotic cells are cleared efficiently and swiftly. However, it became clear that in SLE insufficient removal of
apoptotic material leads to the release of these modified autoantigens. This creates the danger that these modified
autoantigens are recognized by the immune system. Indeed, dendritic cells, the professional antigen-presenting cells,
phagocytose these modified autoantigens, which leads to maturation and induction of a proinflammatory state of these
dendritic cells. As a consequence, they present these modified autoantigens to T cells in an immunogenic way, which
become activated and stimulate autoreactive B cells to secrete autoantibodies. In this review the currently available evidence
for the sequential steps in the pathogenesis of SLE is discussed. Furthermore, the mechanisms responsible for the
nephritogenicity of antinucleosome antibodies are reviewed. This will reveal that nucleosomes are not only a major driving
force in the formation of antinuclear antibodies, but also play a pivotal role in the development of tissue lesions by mediating
binding of autoantibodies to basement membranes as exemplified for the kidney. Semin Nephrol 31:376-389 2011
Published by Elsevier Inc. Keywords: Systemic lupus erythematosus, lupus nephritis, antinucleosome antibodies, antidsDNA antibodies, apoptosis, apoptotic cell clearance, autoantigen modification, dendritic cells.

Central in the development of lupus nephritis (LN) is the formation of immune

deposits within the
glomerulus. The in situ formation or deposition of circulating immune complexes
incites the glomerular inflammation, which depends on the localization and the
amount of the deposits. Although the new International Society of
Nephrology/Renal Pathology Society (ISN/
RPS) classification is based only on light microscopic evaluation of the glomerular
lesions, each class has its distinct localization of immune deposits. In mesangial LN
(classes I and II), deposits are localized in the mesangium without (class I) or with
(class II) mesangial proliferation. In the proliferative forms of LN (classes III and IV),
the immune deposits also are localized along the capillary loop, and are most
pronounced in the subendothelial space. Because of this specific localization
immune effector mechanisms (such as complement, leukocytes, macrophages, Fcreceptor ligation, and so forth) are activated, leading to an influx of inflammatory
cells, endothelial damage, loop necrosis, and extracapillary proliferation.
Sometimes these immune deposits are extensive, thereby creating wire loops, a
characteristic feature of LN classified as class IV-global (IV-G). If the immune
deposits mainly are localized subepithelially then membranous LN ensues, without
an inflammatory response. It is still poorly understood which mechanisms dictate
the site and extent of localization of the immune deposits. It is conceivable that
initially immune deposits are formed in the mesangium with subsequent, if clearing
is overflowed, localization at the inside of the capillary loops. Next to the amount of
immune complexes delivered over time, the size, composition, and charge are
additional factors that are important. This is illustrated by 2 key features of
antibodies involved in the development of membranous LN: they are of low avidity
and belong predominantly to the IgG4 subclass. Because of the low avidity

antibodies can dissociate more easily from an immune complex, which allows
passage as a single molecule through the glomerular basement membrane (GBM)
and subsequent binding to a planted antigen or in situ immune complex formation.
IgG4 subclass antibodies are more cationic, which also enhance its passage
through the GBM. These insights in the mechanisms of immune deposit formation
and their association with renal phenotypes originate mostly from experimental
animal work, especially from the laboratory of Wilson and Dixon, including
experimental serum sickness, murine models of lupus and LN, and Heymann
membranous glomerulopathy. For the formation of these deposits in LN
autoantibodies are of crucial importance, as outlined in the following section.

ROLE OF B CELLS IN LN
In systemic lupus erythematosus (SLE) many autoantibody specificities are formed,
but those directed against nuclear components are the most characteristic. These
include autoantibodies against double-stranded (ds) DNA (regarded as the
serologic hallmark of the disease), nucleosomes, and histones. A number of clinical
observations support the pathogenic role of these antinuclear autoantibodies for
the development of LN. These antibodies are present in 70% of SLE patients and in
more than 90% of patients with LN. Furthermore, onset or flares of LN often are
preceded by a significant increase in the titer of these antibodies, especially of
anti-ds DNA antibodies. Elution of immunoglobulins from glomeruli from patients
and mice with LN revealed antichromatin specificities. Based on these observations
LN has been proposed as the prototype of an immune complex disease. The
decisive role of autoantibodies for the development of LN was supported further by
experimental studies in which B cells were silenced in several ways. First,
introduction of the xid gene in MRL/lpr and (NZBxW)F1 lupus mice. This xid
mutation blocks downstream signaling of the B-cell receptor by inactivating
Brutons tyrosine kinase. These xid-lupus mice did not develop anti-dsDNA
antibodies and proteinuria, and had a normal survival. Second, the strongest
support was provided by knocking out the Jh region in MRL/lpr mice. If this deletion
is present homozygously, B-cell development is totally blocked. In these mice
without B cells no autoantibodies were formed and there was a total absence of
renal and vascular lesions. However, besides their role in antibody secretion, B cells
may contribute to the development of nephritis via other properties such as
antigen presentation to T cells and secretion of cytokines and chemokines. The
contribution of these B-cell functions, apart from antibody secretion, was illustrated
by creating a MRL/lpr mouse in which B cells could not secrete antibodies because
of a deletion in the Ig heavy chain. Despite the lack of autoantibodies these mice
develop lupus glomerulonephritis, although less severe. In this review, we discuss
the specificity of antinuclear antibodies, the relation of antinucleosome antibodies
with LN, current insights in the autoimmune response leading to the formation of
these antibodies, and mechanisms responsible for their nephritogenicity.

ANTINUCLEAR AUTOANTIBODIES IN SLE

Traditionally, anti-dsDNA has been regarded as the characteristic serologic feature
of this disease. However, how these anti-dsDNA antibodies arise remained unclear
for a long time because attempts to induce these antibodies by immunizations with
dsDNA, in all kinds of forms and approaches, failed. Two important observations led
to a breakthrough in our understanding. The first observation was the
determination that the autoimmune response in SLE was a T-cell dependent

autoantigendriven response. This conclusion was based on a number of

observations on pathogenic anti-dsDNA antibodies, as follows: (1) they were of the
IgG class; (2) they were somatically mutated with a restricted use of heavy chain
variable regions (VH) and light chain variable regions (VL) genes; (3) they were of
high affinity; and (4) they shared common idiotypes between individuals, even
across species. These facts indicated that dsDNA could not be the driving
autoantigen because for a T-cell dependent response the antigen needs to be
presented as a peptide in the context of major histocompatibility complex class II
on antigen-presenting cells. Subsequently, several lines of evidence pointed to the
nucleosome as a major autoantigen, which also is responsible for the induction of
anti-dsDNA antibodies. The nucleosome is the basal building block present in
chromatin and is composed of two dimers of the core histone proteins H2A and
H2B, and a tetramer of H3 and H4. dsDNA is wrapped twice around this octamer of
histones. Histone H1 is positioned at the outside similar to a finger on a knot to
tighten the complex.
What supports the central role of nucleosomes as major autoantigens for the
induction of the antinuclear autoantibody response in SLE? First, nucleosome (ie,
histone peptide)-specific T-helper (Th) cells can be found both in lupus mice and
SLE patients. Through epitope spreading the Th cells not only induce
antinucleosomespecific antibodies, but also anti-dsDNA and antihistone antibodies.
Second, antinucleosome-specific antibodies can be detected in murine lupus
models before the onset of other autoantibody specificities, such as anti-dsDNA
and antihistone antibodies. Finally, antinucleosome antibodies were detected with
high prevalence in lupus mice and SLE patients, and with a higher sensitivity and
equal specificity than anti-dsDNA antibodies, especially in LN, as discussed later.
Also, in certain patients who are anti-dsDNA negative, antibodies to nucleosomes
can be detected. Later we summarize the prevalence of antinucleosome antibodies
in SLE and/or LN and their sensitivity and specificity. The second important
breakthrough was the finding by Casciola-Rosen et al that chromatin omponents
and other autoantigens appear at the surface of apoptotic cells. This observation
provided a clue as to why in this disease antibodies were formed against
autoantigens deeply hidden within the nucleus of the cell. Furthermore, this finding
boosted research into the sequence of events leading to the autoimmune response
in this disease. This is discussed in the Origin of Antinucleosome Antibodies
section.

PREVALENCE OF ANTINUCLEOSOME ANTIBODIES IN SLE AND LN

Before reviewing the literature on antinucleosome antibodies in SLE and LN, it is
necessary for a good understanding to define the nomenclature and the definitions
of the various antinuclear antibodies. The entire complex of compacted DNA and
associated proteins (both histones and nonhistones) is termed chromatin.
Antinucleosome antibodies comprise antibodies to all components of the
nucleosome including anti-dsDNA and antihistone antibodies. Furthermore,
antibodies can be formed against conformational epitopes in the nucleosome.
These latter antibodies have no or a very low reactivity against individual histones
or native dsDNA, and therefore are called nucleosome-specific antibodies.
Currently, there are no diagnostic tests available to measure specifically these
nucleosome-specific antibodies. In the commercially available tests, all three major
antinucleosome autoantibody
specificities are measured (ie, anti dsDNA, antihistone, and nucleosome-specific
antibodies).17 As we discuss later, antibody binding to nucleosomes is a major

determinant for nephritogenicity, therefore, these tests are useful in patients with
LN. In the past decade several studies have been performed to analyze the
sensitivity and specificity of antinucleosome antibodies in SLE and LN. The most
important studies are listed in Table 1. The major conclusions from these studies
are as follows:
Analyses were performed worldwide in different countries and in different ethnic
groups, except for the United States;

The majority of the studies evaluated unselected SLE patients and 2 studies
specifically addressed patients with LN;

In unselected SLE patients the sensitivity ranged from 45% to 100% and in LN it
was about 90%;

With one exception, the sensitivity for antinucleosome antibodies was higher
than for antidsDNA in the same patient groups;

Specificity for antinucleosome antibodies was analyzed by using different control

groups (systemic autoimmune diseases, systemic sclerosis, Sjgren syndrome,
inactive SLE, infectious diseases,
and healthy controls);

Regardless of the studied control cohorts, the specificity for antinucleosome

antibodies was very high (85%-100%) and comparable with the specificity for antidsDNA;

In a number of studies antinucleosome antibodies were correlated positively with

LN, with odds
ratios between 3.4 and 25.8.

Despite the obvious limitations of this comparison in Table 1, it can be concluded

that antinucleosome testing is clinically useful. In many studies a positive
correlation was found between antinucleosome titer and several measures of
disease activity (Systemic Lupus Activity Measure [SLAM], European Consensus
Lupus Activity Measure [ECLAM], Systemic Lupus Erythematosus Disease Activity
Index [SLEDAI], and British Isles Lupus Assessment Group [BILAG]). Initially, it was
reported that the prevalence in systemic sclerosis and mixed connective tissue
disease was about 45%. Later, it was shown that the used nucleosomes for these
analyses were contaminated with topoisomerase, the major autoantigen in
systemic sclerosis. In all other systemic autoimmune diseases the prevalence was
very low. Data from several studies on antinucleosome antibodies in anti-dsDNA
negative SLE patients are given in Table 2. These studies show the additional value
of antinucleosome testing in those patients because 54% were antinucleosome
positive. Moreover, in one study it was found that anti-dsDNA negative patients
with LN were all antinucleosome positive.
Can antinucleosome measurement also be used for disease monitoring? There are
only very limited data available. In a follow-up study of 102 patients with lupus,
there was a high prevalence of antinucleosome antibodies (86.1%), but no
correlation with disease activity during follow-up evaluation. However, it should be
mentioned that the majority of patients had stable disease. We also prospectively

evaluated antinucleosome reactivity in 87 patients with proliferative LN. At

baseline, 81% of the
patients were antinucleosome positive. During treatment, antinucleosome
reactivity declined over time. Renal flares were not heralded by increases of
antinucleosome titers. Therefore, we had to conclude that antinucleosome analysis
is helpful for the diagnosis but of limited value for monitoring. However, more
prospective studies are needed to draw definitive conclusions.

ORIGIN OF ANTINUCLEOSOME ANTIBODIES

As already outlined previously, the discovery of autoantigen clustering in blebs at
the surface of apoptotic cells boosted research on the role of apoptosis in SLE, and
led to the discovery of autoantigen modification during apoptosis. Normally,
apoptotic cells are cleared efficiently and swiftly. However, it became clear that in
SLE insufficient removal of apoptotic material led to the release of these modified
autoantigens. This creates the danger that these modified autoantigens trigger the
immune system. Indeed, dendritic cells (DCs), the professional antigenpresenting
cells, phagocytose these modified autoantigens, which leads to maturation and
induction of a proinflammatory state of these DCs. As a consequence, DCs present
these modified autoantigens to T cells in an immunogenic way, which become
activated and stimulate autoreactive B cells to secrete autoantibodies. This
sequence of events is depicted in Figure 1.
Later we review the evidence for the subsequent steps in Figure 1. In the final
section we discuss the mechanisms responsible for the nephritogenicity of
antinucleosome antibodies. This will reveal that nucleosomes are not only a major
driving force in the formation of antinuclear antibodies, but also play a pivotal role
in the development of tissue lesions as exemplified for the kidney.
Dysregulated Apoptosis
The first notion that abnormal apoptosis could be relevant for lupus came from the
discovery that MRL/lpr lupus mice had a functional Fas deficiency. Binding of the
Fas-ligand (FasL) to Fas induces apoptosis. Also, deficiency of the FasL as in gldmice leads to a lupus phenotype. Transgenic insertion of a normal Fas gene in
MRL/lpr mice prevents autoimmunity. In human beings deficiency for Fas or FasL
leads to an autoimmune lymphoproliferative syndrome, but not to lupus. Other
examples in experimental animals on the association between dysregulated
apoptosis and SLE are transgenic overexpression of Bcl-2 (a pro-apoptotic
molecule) and DNAse-1 deficiency. In both models antinuclear antibodies are
formed and an immune complex glomerulonephritis develops. At present, several
other molecular defects leading to abnormal apoptosis have been identified in
association with human and murine SLE. It goes beyond the scope of this review to
analyze these in depth, but we published an extensive review recently. Although
subtle changes in apoptosis have been found in lupus patients, the contribution of
abnormal apoptosis in human SLE is less clear than in experimental animal models.
Nevertheless, it is obvious that deregulation of apoptosis plays a role in the
development of SLE.
Clearance of Apoptotic Cells in SLE

Normally, the removal of apoptotic debris is secured through a redundancy of

molecules, acting as chemotactic agents, recruiting phagocytes (come and get
me), eat me signals on apoptotic cells to identify the bait, bridging molecules
(including C1q, _2GP1, serum amyloid protein P) connecting apoptotic cells to
phagocytes and receptors on these phagocytes, responsible for the uptake and
subsequent degradation. These processes are discussed and beautifully illustrated
in a review by Savill et al.37 The relevance of an adequate removal of apoptotic
waste came from a number of studies in mice with a targeted disruption of
molecules involved in this waste disposal. Deficiency for C1q, serum amyloid
protein P, DNase I, IgM, and the cytoplasmic tail of the mer receptor on
macrophages (to which the bridging molecule GAS6 binds) is associated with the
formation of antinuclear antibodies (directed either against histones and/or
nucleosomes and/or dsDNA) and immune complex glomerulonephritis.
In some of these models the removal of apoptotic cells was analyzed and found to
be defective.
Also, in human beings a C1q deficiency is strongly associated with SLE (93% of
affected patients develop lupus). In addition, in SLE and especially in LN antibodies
are formed against C1q in a high frequency. These autoantibodies can prevent the
binding of C1q to apoptotic cells and thereby block removal. We analyzed
phagocytosis of apoptotic cells in MRL/lpr and New Zealand BWF1 lupus mice. We
did not find a constitutive defect in premorbid mice, but in mice with clinical
disease the removal of apoptotic cells was impaired. More importantly, defects in
apoptotic cell removal were documented in SLE patients in several ways: in vitro by
using macrophages, in skin biopsies after ultraviolet irradiation, and in the
germinal centers of lymph node biopsies.
Taken together, these observations in murine and human lupus indicate that
inadequate removal of apoptotic cells leads to and is associated with lupus. This
defect is an important step in the genesis of the autoimmune response in SLE. As a
consequence of this insufficient disposal, the process of apoptosis continues with
extensive blebbing and finally release of the contents of these blebs. Indeed,
nucleosomes, which can be generated only through apoptosis, can be found in the
circulation of SLE patients and lupus mice. As outlined in the next paragraph these
released autoantigens contain apoptosis induced modifications that provide a
danger signal to DCs.
Autoantigen Modification During Apoptosis
During apoptosis several demolishing circuits are set in motion including caspases
and reactive oxygen species, to dismantle the cell. This process not only alters
several proteins but also nucleosomes. For proteins it has even been suggested
that susceptibility for modification during apoptosis predisposes this target to
become an autoantigen.
For proteins a number of modifications have been described including caspase and
granzyme Bmediated cleavage, (de)phosphorylation, transglutamination,
ubiquitination, and citrullination.This latter modification is targeted predominantly
in patients with rheumatoid arthritis (anti-cyclic citrullinated peptide an tibodies)
and indicates that apoptosis-induced modifications are not only relevant for SLE,
but are important triggering events in many, if not all, autoimmune diseases.
Detection of apoptosis-induced nucleosome modifications

It was conceivable that similar alterations also could occur within nucleosomes.
The first indications were found rather serendipitously. We discuss the first
description of apoptosis-induced nucleosome modification in detail because by
using this approach we discovered a number of these modifications within
nucleosomes. During screening for epitopes of several lupus-derived monoclonal
antibodies by a 15-mer random peptide phage library, one of the monoclonal
antibodies, KM2, revealed a motif that localized in the N-terminal part of histone
H4. All amino acids from position 5 until position 18 in the consensus motif were
identical to the native sequence of H4, but the lysines at positions 8, 12, and 16
were absent. These lysines are potential acetylation sites. Therefore, the reactivity
of KM2 to peptides acetylated at these positions was analyzed. The binding in
enzymelinked immunosorbent assay (ELISA) to acetylated peptides was higher
than with the unmodified H4 (122) peptide. This was confirmed in inhibition
ELISAs using unmodified H4 (122) peptide as a coated antigen, in which
acetylated H4 peptides showed better inhibition than the unmodified peptide. To
analyze whether acetylated H4 also was recognized in vivo, Jurkat cells were
treated with the histone deacetylase inhibitor trichostatin A (TSA). KM2 showed a
better nuclear binding to TSAtreated Jurkat cells than to untreated cells. This higher
reactivity also was found by Western blot with histones purified from TSA-treated
Jurkat cells, compared with histones purified from untreated cells. A similar reaction
pattern was found with commercial antibodies against acetylated lysines at
positions 5, 8, 12, or 16. To investigate whether hyperacetylation occurs during
apoptosis, Jurkat cells were made apoptotic with either camptothecin or anti-Fas.
Apoptosis induction was confirmed by cleavage of caspase-3 and Poly-ADP-Ribose
Polymerase (PARP), and by annexin V staining. Again, KM2 showed a higher binding
to histones isolated from apoptotic Jurkat cells in Western blot and with nuclear
immunofluorescence of apoptotic cells when compared with normal cells. The
staining of KM2 by immunofluorescence was confined to terminal deoxynucleotidyl
transferasemediated deoxyuridine nick-end labeling (TUNEL)positive areas. The
mechanism behind this hyperacetylation was a significant increase in the activity
of histone acetyltransferases and a significant decrease in the activity of histone
deacetylases during apoptosis. Hence, based on these observations it could be
concluded that hyperacetylation of histones (including H4 and H2A) occurred
during apoptosis and that these modifications are a target for this lupus-derived
monoclonal antibody KM2.
Pathogenicity of apoptosis-induced nucleosome modifications
Do these modifications have clinical relevance? To evaluate this, plasma from
MRL/lpr mice (age, 69 weeks and _20 weeks) and from patients with proliferative
LN were tested for their reactivity in ELISA with either the unmodified H4 (122)
peptide (non-AcH4) and the H4 peptides (122) acetylated at positions 8, 12, and
16 (tri-AcH4). A significantly higher reactivity was found toward tri-AcH4, both in
mice and patients. In addition, the majority of plasma from patients showed a
higher nuclear reactivity with camptothecin- or TSA-treated Jurkat cells. Next,
prediseased 8-week-old MRL/lpr mice were given either tri-AcH4, non-AcH4, or
phosphatebuffered saline, subcutaneously every 2 weeks. Treatment with tri-AcH4
significantly enhanced mortality, proteinuria, and skin lesions. Subsequently,
delayed-type hypersensitivity was measured toward tri-AcH4 versus non-AcH4. In
MRL/lpr mice a significantly higher delayed- type hypersensitivity reaction was
found, in contrast to MRL_/_, B10.B2 (H2-haplotype control), or BALB/c mice.
Remarkably, this higher reactivity also was found in MRL/lpr mice not injected with
the tri- AcH4 peptide. Apparently, this higher reactivity was caused by endogenous

antigens. To evaluate the immunogenicity of tri-AcH4 versus non-AcH4, bone

marrow derived DCs from MRL/lpr mice were exposed to either one of these
peptides. We did not find an effect on DC maturation (CD40 and CD86 expression)
or DC cytokine production (interleukin [IL]-6 and tumor necrosis factor [TNF]).
However, if we added hyperacetylated nucleosomes or normal nucleosomes to
immature MRL/lpr DCs, we found a significant up-regulation of CD40 and CD86
expression with hyperacetylated nucleosomes but not with normal nucleosomes.
Also, the production of IL-6 and TNF-_ by DCs was enhanced significantly by
hyperacetylated nucleosomes but not by normal nucleosomes. Finally, in a
syngeneic co-culture of DCs and splenocytes from MRL/lpr mice, IL-2 production
was increased significantly by DCs matured by hyperacetylated nucleosomes
compared with DCs exposed to normal nucleosomes.
By using a different, but also lupus derived, monoclonal antibody, LG11-2, and
using similar techniques, we found that the target of this antibody was also an
apoptosis- induced epitope, namely, acetylated lysine at position 12 in histone
H2B. Based on these findings, we argued that selection of monoclonals on
apoptotic nucleosomes would provide an approach to detect relevant immunogenic
epitopes within the nucleosome. Indeed, these modifications were found. Recently,
we generated a panel of nucleosome-specific monoclonals, recognizing acetylated
conformational epitopes composed of both dsDNA and histones. To date, the
identified apoptosisinduced modifications on nucleosomes were caused by
acetylation. It is not yet known what the physiological function of chromatin
acetylation is during apoptosis.
Because acetylation leads to a more open accessible conformation of chromatin, it
is conceivable that this makes chromatin more susceptible to enzymes involved in
the degradation of chromatin. Besides acetylation, we very recently found, by
using the monoclonal BT 164, selected on apoptotic nucleosomes, trimethylation
on lysine at position 27 in histone H3 as an epitope for autoantibodies in SLE. Taken
together, these data indicate that nucleosomes also are modified during apoptosis,
generating immunogenic epitopes. In face of the decreased removal of apoptotic
debris, these neo-epitopes are noticed by the immune system, in particular by DCs,
as discussed in the subsequent section.
Activation of DC by Apoptotic Debris
Because apoptotic blebs are loaded with autoantigens, which are modified during
apoptosis, and clearance of this apoptotic material, including blebs, is insufficient
in SLE, we wondered whether these blebs are able to activate DCs. Therefore, we
analyzed the effects of late apoptotic blebs on mouse DCs cultured in vitro from
bone marrow in the presence of granulocyte-macrophage colony-stimulating factor,
which generates myeloid DCs. We presented immature DCs with either late
apoptotic blebs or with apoptotic cell bodies (ACBs). These ACBs are the cellular
remnants after the blebbing process has ended. Co-incubation for 24 hours of DCs
with equal amounts of either blebs or ACBs showed that blebs were taken up more
efficiently (2-3 times higher) than ACBs. Ingestion of blebs led to maturation of DCs
as evidenced by a significantly enhanced expression of the activation markers and
co-stimulatory molecules CD86 and CD40 compared with activation after ingestion
of ACBs. The activation of DCs induced through blebs had a similar magnitude as
with lipopolysaccharide, used as a positive control. Furthermore, ingestion of blebs
led to significant secretion of the proinflammatory cytokines IL-6 and TNF-_, which

was not observed after phagocytosis of ACBs. The strength of the proinflammatory
stimulus provided by ingestion of apoptotic blebs was indicated by the fact that the
amount of cytokines produced was similar as observed after stimulation with
lipopolysaccharide, a strong proinflammatory stimulator for DCs. Then, we cocultured bleb or ACBs fed DCs with allogeneic T cells for 6 days. Bleb-fed DCs, but
not ACBs fed DCs, were able to activate T cells as indicated by the production of
the T-cell cytokines IL-2 and interferon, consistent with a Th1 proliferation. This was
evidenced further because none of the Th2 cytokines IL-4, IL-5, and IL-10 were
produced. The most interesting observation was that production of IL-17 also was
induced. The amount of IL-17 after exposure to bleb-matured DCs was 3-fold higher
than after stimulation with lipopolysaccharide matured DCs.
Based on these observations, we concluded that apoptotic blebs, but not ACBs, can
act as a danger-associated molecular pattern. These danger-associated molecular
patterns are recognized by Toll-like receptors (TLRs) on the surface or within
intracellular compartments of DCs.
We focused on TLR3, TLR7, and TLR9 because these TLRs can be triggered by
various endogenous nucleic acid containing compounds, which are present within
these blebs. These TLRs are inhibited by chloroquine, which inhibits the
acidification of the endosome. Addition of chloroquine, however, did not inhibit the
production of IL-6, whereas IL-6 production induced by a TLR9 ligand (a Cytosinephosphate-Guanine (CpG)-containing oligonucleotide) was inhibited.
Because apoptotic blebs represent a bag of autoantigens, it remains to be
determined which constituents(s) is/are responsible for the induction of DC
maturation. There are several candidate molecules within these blebs.
DC maturation has been described for nucleosomes, especially apoptosis-induced
hyperacetylated nucleosomes, DNA and RNA. Recently, it was found that another
nucleosome-associated protein high-mobility group box protein 1 is a strong
inductor of DC maturation, via TLR2 activation. The relevance of this mechanism
for the pathogenesis of SLE is illustrated by studies showing that administration in
both normal and lupus mice of bone-marrowderived DCs loaded with apoptotic
cells can induce the formation of antinuclear antibodies and glomerular immune
deposits. In lupus mice this DC vaccination leads to aggravation of the disease.
Furthermore, interference with several TLR functions, either deficiency or
transgenic overexpression, leads to, respectively, augmentation or enhancement of
disease manifestations.
However, the involvement of TLRs for human lupus still remains to be elucidated,
although relevant in this context is the therapeutic beneficial effect of
hydroxychloroquine, which inhibits activation of intracellular TLRs (reviewed by
Fransen et al). Another aspect of bleb-induced DC maturation needs to be
mentioned, namely their capacity to activate T cells. From the induced cytokine
profile it can be deduced that Th1 and Th17 cells were activated. Several recent
observations point to an important role for Th17 cells in many autoimmune
diseases, including lupus. In human SLE an overproduction of IL-17 and IL-23
(another Th17 cytokine) was observed, and an increased number of Th17 cells
were found during lupus flares, whereas the amount of regulatory T cells was
diminished. Once autoantibodies are formed, they can bind to released nuclear
antigens and these immune complexes can activate another subset of DCs, the
plasmocytoid DCs. On activation, these plasmocytoid DCs produce high amounts of

interferon-alfa and other type I interferons, which amplify the autoimmune

response in several ways. It supports maturation of myeloid DCs and autoantibody
production and isotype switching of autoreactive
B cells. Based on these observations, a pivotal role can be assigned to the various
DC populations in the
initiation (myeloid DCs) and progression (plasmocytoid DCs) of lupus.

NEPHRITOGENICITY OF ANTINUCLEOSOME ANTIBODIES

In the early 1980s several monoclonal anti-DNA antibodies were generated after
establishment of the technique by Khler and Milstein. An unexpected feature of
these anti-DNA monoclonals was the capacity to react with several other molecules
than DNA. This rather promiscuous reactivity was called cross-reactivity. Binding to
various structures has been described (reviewed by van Bavel et al), including
several components of the glomerulus-like actinin-4, type IV collagen, and laminin.
On the basis of these observations, it was postulated that anti-DNA antibodies bind
to the GBM through this cross-reactive binding. Indeed, intrarenal perfusion of
monoclonal anti-DNA antibodies or intraperitoneal inoculation of anti-DNA
producing hybridomas led to glomerular immune deposits.
When we were analyzing the binding properties of several murine lupus derived
monoclonal anti-DNA antibodies, we found binding to heparan sulfate
proteoglycan, isolated from the human GBM. Heparan sulfate proteoglycan (HSPG)
belongs to the family of proteoglycans that are heavily glycosylated proteins
present on many cells and in many extracellular matrices such as the GBM.
Because of the strong negative charge of heparan sulfate (HS), we wanted to
document the iso-electric point of the HS binding anti-DNA monoclonals. To this
end, we purified the monoclonal antibodies under high (3 mol/L) salt conditions on
a protein A column after DNAse I digestion. To our surprise the reactivity of the
purified anti DNA monoclonal to HS was lost, whereas the binding to DNA was
unaltered or even increased. Addition of the column eluate to the purified antibody
reconstituted the HS reactivity. It appeared that nucleosomes bound to the antiDNA antibodies were responsible for the binding to HS and the GBM. In retrospect,
this was not unexpected because during hybridoma culture-secreted anti- DNA
antibodies can form complexes with nucleosomes released from apoptotic
hybridoma cells. Therefore, we analyzed whether the binding to HS in ELISA of
purified monoclonal anti-DNA was reconstituted by nucleosomes. Indeed, addition
of nucleosomes, but not pure DNA or purified histones, could restore the binding to
HS. To extend this we tested the binding of purified anti-DNA monoclonals to GBM
loops isolated from human kidney. Again, addition of nucleosomes led to a strong
binding to the GBM in vitro. To block the anionic sites within the GBM loops, we
preincubated the loops with cationic ferritin, which prevented binding of
nucleosome/anti-DNA complexes. To explore whether such a binding also could
occur in vivo, we performed direct intrarenal perfusion by administration via the
renal artery at physiological perfusion pressures. Antibodies complexed to
nucleosomes showed strong granular binding along the GBM as illustrated in Figure
2A and induction of proteinuria. Perfusion of either nucleosomes alone, purified
monoclonal anti-DNA antibodies, or mixtures of nucleosomes and a nonrelevant
monoclonal antibody with the same iso-electric point showed neither glomerular
binding (Fig. 2B) nor proteinuria. On the basis of these experiments we concluded
that anti-DNA and antinucleosome antibodies complexed to nucleosomes can bind
to the GBM in vivo and induce proteinuria. To analyze whether HS was the ligand in

the GBM for these complexes, we removed HS in vivo by prior perfusion with
heparinase. By using the HS-specific monoclonal antibody JM403,70 we could
confirm the removal of HS from the GBM by heparinase, whereas the HSPG core
protein, laminin, and collagen IV were unaffected. After heparinase perfusion, the
binding of subse quently perfused complexes was decreased considerably, but not
totally. Apparently, some other ligands still remained, probably collagen IV, which
has been identified as a ligand for nucleosome-complexed lupus autoantibodies.
Based on these in vivo binding studies, we argued that circulating heparin, a
molecule with strong similarities with HS, could in theory prevent or inhibit the
binding of these complexes to HS in the GBM by covering and neutralizing the
positively charged histones, which we held responsible for the binding to the
strongly anionic HS. This idea was strengthened by the observation that heparin
could inhibit dose dependently the binding of nucleosome/antinucleosome
complexes to HS in ELISA. Also, in direct intrarenal perfusion experiments heparin
could prevent the binding of these complexes to the GBM. Therefore, we treated
MRL/lpr mice daily from the age of 8 weeks with subcutaneous heparin or
noncoagulant heparin derivatives. Heparin (oids) prevented the development of
proteinuria and glomerular lesions significantly. On immunofluorescence only
mesangial deposits were seen, whereas saline-treated controls had unaltered
extensive granular deposits along the glomerular capillary wall.
Further evidence for the nucleosome-mediated binding to the GBM originated from
our hybridoma inoculation studies in BALB/c nude mice. We inoculated three
different kinds of hybridoma cells: producing either antidsDNA antibodies,
antinucleosome antibodies, or antihistone antibodies. Our prediction was that in
mice in which anti-DNA or antinucleosome antibodies were produced, glomerular
deposits would develop in contrast to antihistone antibodies. This assumption was
based on the fact that the epitopes for antihistone antibodies reside in the
positively charged histones. Binding to these epitopes would reduce the capacity to
bind to HS. The production of autoantibodies after intraperitoneal inoculation was
confirmed by detection of the inoculated specificity in the ascites and circulation.
Also, the levels of nucleosomeautoantibody complexes were comparable in the 3
groups. For antihistone hybridomas we found immune deposits in 15% of the
glomeruli, whereas for the antidsDNA and antinucleosome groups we found
immune deposits in 60%. From these collective data in experimental animals it
may be concluded that nucleosomemediated binding to the GBM of complexes
consisting of antinucleosome or anti-DNA antibodies is a mechanism for
nephritogenicity of these autoantibodies, as depicted in Figure 3.
But what evidence is there that this also occurs in vivo in lupus mice and patients
with nephritis? By using several antihistone and antinucleosome-specific
(monoclonal) antibodies we detected nucleosome deposits in glomeruli in all
patients with proliferative LN and in 25% of the patients with membranous LN. 74 In
none of the nonlupus biopsies with membranoproliferative or membranous
glomerulonephritis deposits were found except for one patient with membranous
glomerulopathy. Our findings in human lupus are consistent with previous reports
on murine lupus. Because HS is expressed not only in the GBM, but also in other
extracellular matrices such as the dermal basement membrane, we analyzed
whether we could detect immune deposits at the dermal basement membrane in
patients with proliferative LN enrolled in a prospective randomized trial, assessing
the efficacy of azathioprine versus cyclophosphamide. In this study 30 patients
with active LN and 15 patients with inactive LN were included, whereas 14 patients
with parapemphigus and 10 patients with diabetic nephropathy were studied as

controls. Deposits of IgG were found in 87% of patients with active LN, in 33% with
inactive LN, in 71% with parapemphigus, and in 0% with diabetic nephropathy. By
using different antihistone antibodies, nucleosomes were detected in 87% of the
patients with active LN and in none of the inactive LN or diabetic nephropathy
patients. In parapemphigus patients 14% of the biopsies stained positive for
nucleosomes. In confocal laser scanning microscopy we found co-localization of IgG
and nucleosome deposits in the dermal basement membrane, as depicted in Figure
4. Therefore, it seems that nucleosome-mediated binding is not restricted to the
kidney, but also occurs at other basement membranes. To evaluate the specificity
of the glomerular-deposited antibodies we eluted IgG from glomeruli of 18 to 24
weekold MRL/lpr lupus mice. We divided the mice into 3 groups based on the
degree of proteinuria: none, mild, or heavy. The onset of proteinuria was associated
with a three-fold higher amount of eluted IgG. Antigen specificity of the eluted
antibodies showed that antinucleosome antibodies deposited first, whereas the
highest reactivity for anti-DNA antibodies was found in mice with heavy
proteinuria. The antihistone reactivity was low in all three groups and did not
increase with progression of disease. These elution studies suggest that
antinucleosome antibodies deposit first with subsequent deposition of anti-dsDNA
antibodies. Therefore, the specificity of the second component of the deposits, the
antibodies, was identified at least for murine LN. The presence of nucleosomecontaining immune complexes in kidney biopsies of murine and human LN was
found indirectly. When analyzing HS expression in various glomerular diseases
using monoclonal antibodies against HS and the HSPG-core protein, we found an
almost complete absence of HS staining in both mice81 and patients82 with LN,
whereas the expression of the HSPG-core protein was unaltered. In human kidney
biopsies the loss of HS was correlated negatively with the amount of histone
deposits. The finding in lupus mice allowed more detailed studies, which revealed
the following: (1) the decrease in HS was correlated inversely with the amount of
IgG deposits and the amount of proteinuria; (2) the amount of HS within the GBM
was not decreased; (3) elution of IgG from glomeruli restored the HS staining; and
(4) treatment with heparin(oids) prevented IgG deposition and, concomitantly, loss
of HS staining.
These data are consistent with masking of HS by nucleosome-containing immune
complexes.
Additional circumstantial evidence for the nucleosome mediated nephritogenicity
of autoantibodies comes from ELISA studies for HS reactivity in plasmas at onset or
exacerbation of LN in patients. In both studies a higher HS reactivity was found. We
interpreted these results to mean that the HS reactivity represented the presence
of nucleosome/autoantibody complexes. Indeed, in mouse studies we found a
correlation between nucleosome/autoantibody complexes and onset of proteinuria.
In a recent study of 16 patients with new onset of LN, the antinucleosome, antidsDNA, and anti actinin autoantibody titers were measured prospectively at regular
intervals and correlated with renal parameters.
Over time, antinucleosome and anti-dsDNA titers were associated significantly with
proteinuria and renal remission in contrast to anti actinin antibodies. Moreover,
anti-actinin antibodies did not significantly differ between healthy controls and
patients with LN. Recently, research from Rekvigs laboratory (reviewed in), using
co-localization immunoelectron microscopy, provided strong evidence for the
central role of nucleosomes in the development of LN. The involvement of
autoantibodies in LN has been discussed earlier, but the precise mechanism for the
binding was still a subject of much debate.

As outlined previously, two mechanisms have been proposed for the glomerular
binding of anti-dsDNA antibodies. Either they cross-react with intrinsic glomerular
antigens such as laminin, collagen IV, or _-actinin, or they bind via nucleosomes to
the GBM. In a first attempt by Rekvigs group to define the glomerular targets more
precisely, antibodies were eluted from glomeruli of (NZBxW)F1 mice with LN.
These eluted antibodies reacted with nucleosomes, dsDNA, and histone H1, but not
with actinin. Subsequently, the exact glomerular localization of in vivo deposited
antibodies was investigated in (NZBxW) F1 mice with LN. In vivo deposited mouse
IgG antibodies were detected by incubating sections with rabbit anti mouse IgG
and subsequently with protein A conjugated with 5-nm gold particles. With this
technique, antibodies bound in vivo were confined strictly to electron dense
deposits (EDD) present in the GBM of nephritic mice, but not in non-nephritic mice.
There was no localization of in vivo bound autoantibodies outside these EDD along
the GBM. This implies that the GBM itself was not a target for these autoantibodies
in vivo. To define further the composition of these EDD, colocalization
immunoelectron microscopy was performed using antibodies directed against
various chromatin constituents such as dsDNA, histone H1 or H3, or against the
transcription factor TATA binding protein, which is bound constitutively to
chromatin. These antibodies were labeled with 10-nm gold particles. In addition, an
antibody was used that was specific for GBM laminin. These experimental
antichromatin antibodies bound to the EDD and co-localized with the in vivobound
autoantibodies within these EDD. There was, however, no co-localization of
antilaminin antibodies and in vivobound autoantibodies.
These results indicate that EDD are composed of extracellular chromatin fragments
complexed to autoantibodies in vivo both in murine and human LN. Because one
could criticize the use of antichromatin antibodies because of their potential crossreactivity, an Antibody independent DNA specific assay was developed. By using
the terminal deoxynucleotidyl transferase mediated deoxyuridine nick end labeling
technique, it was shown that apoptotic nucleosomal material was present in these
EDD and again co localized with in vivobound autoantibodies. The results of these
studies are shown in Figure 5.

SUMMARY
In conclusion, in lupus and LN, we found the following: (1) experimental evidence
suggests a role for deregulated apoptosis, in SLE patients this is less clear; (2)
more important is a decreased removal of apoptotic cells; (3) autoantigens
including nucleosomes are modified during apoptosis and become more
immunogenic; (4) blebs and modified nucleosomal constituents can activate
myeloid DCs, which mature and achieve a proinflammatory state; (5) through these
functional changes DCs can activate T cells, Th1 cells, and Th17 cells; (6) released
nucleosomal material complexes with antinucleosome antibodies; and (7)
nucleosomeantinucleosome complexes are able to bind to the GBM, probably via
ligation to HS.