By: Keith Elkon, MD
Disclosures: Dr. Elkon has no financial disclosures.
- Explain how antibodies can induce inflammatory cytokines in the brain and how this may impair brain function
Neuropsychiatric (NP) variables in the 1982 revised classification criteria for SLE
In 1982, the NP variables seemed simple. There were different well-documented complications (see table above) that occurred in a variable number of people, but we had very little idea about what was going on. Work had begun on what autoantibodies in lupus patients had reacted with in the cytoplasm, but it wasn’t known what the cytoplasmic antigens were.
Anti-P and neurospsychiatric systemic lupus erythematosus (NPSLE)
Through several biochemical techniques, anti-ribosomal P autoantibodies in lupus sera were identified. They were found to have a cytoplasmic immunofluorescence, and were binding to three proteins in the ribosome. The anti-cytoplasmic antibodies in lupus are binding to three proteins on the 60S subunit of the ribosome. Those three proteins (phosphoproteins) are called P0, P1 and P2. They form a pentameric complex sitting on the stalk on the 60S subunit. The antibody binds to all three proteins because there is a shared epitope in the carboxyl terminus of about 22 amino acids.
About five years later it was discovered that there were antibodies that bind to an RNA epitope and, using RNAs mapping, the sequence of that RNA epitope was also identified. Interestingly, the autoantibody is also part of what is called the GDPa subunit on the 60S part of the ribosomal stalk and functions together with the P0, P1 and P2 complex. So, it was a functionally related group of autoantigens that comprises protein and RNA.
This discovery led to the study of the clinical associations of these autoantibodies. Firstly, these anti-P antibodies could be detected in the CSF of patients. Through the development of an immunoassay (synthesis of the peptide corresponding to the 22 amino acids of the shared epitope), very sensitive tests were developed to pick up anti-P antibodies.
The connection between Anti-ribosomal P and NPSLE has been controversial.
- From the mid-80’s to the 2000’s, approximately half of the studies indicate an association between anti-p and NPSLE.
- High frequency of anti-P in CSF of NPSLE patients (Yoshio et al)
- Only antibody showing positive association with onset of NPLSE (Hanley et al., SLICC study, Arthritis and Rheumatism 2007)
- Anti-P antibodies are cross reactive with neuronal cell surface antigen (NSPA), which induces neuronal apoptosis (Matus, JEM, 2007)
Neuropsychiatric syndromes in SLE
In 1999, the ACR research committee defined 19 neuropsychiatric lupus subsets. Although this makes sense from a clinical perspective, it makes research much more difficult. Additionally, several common non-lupus specific problems are included in the definition (headache, anxiety disorder and cognitive dysfunction), resulting in very low specificity for neuropsychiatric lupus.
- Aseptic meningitis
- Cerebrovascular disease
- Demyelinating syndrome
- Movement disorder
- Seizure disorders
- Acute confusional state
- **Anxiety disorder
- **Cognitive dysfunction
- Mood disorder
- Guillain Barre´ syndrome
- Autonomic neuropathy
- Myasthenia gravis
- Cranial neuropathy
The most common features in NPSLE are cognitive dysfunction, headache, mood disorder and then cerebrovascular disease. These same features are also the least specific. The features with greater specificity for lupus are seizures, neuropathy, anxiety and psychosis.
Pathogenesis of NPSLE
Neuropsychiatric lupus remains the least understood manifestation of lupus. However, cerebrovascular disease (a major complication of long-term disease) is a reasonably understood subcategory of neuropsychiatric lupus and will be left out of the discussion below.
Associations between distinct autoantibodies and NPSLE, as well as presence of cell debris and certain cytokines, have been published over the last 20-30 years.
There are more than 20 different autoantibodies associated with lupus. They are subdivided into anti-non-neuronal and anti-neuronal antibodies. Anti-neuronal antibodies will react with neurons, either cell lines or authentic tissue. Many of these have been shown to be cytotoxic.
- There are antibodies to GFAP and antibodies to glutamate receptor NMDA.
- There is a subset of antibodies associated with CNS lupus (ribosomal P).
- There have been studies on anti-Sm and anti-Ro antibodies being associated with NPSLE.
Anti-NMDAR and NPSLE
- An R4A anti-DNA mAb was identified that causes nephritis in SCID mice
- A screened phage display library revealed that the R4A antibody recognizes the Asp/Glu-Trp-Asp/Glu-Tyr-Ser/Gly (DWEYS) consensus sequence and that this sequence is shared by NMDAR in the brain
Above: (Left) NMDAR—different subunits can associate and form different channels—some of the receptors will bind glutamate and some glycine and depending on the combination of neurotransmitters released, there will be different responses. (Right) NR1 and NR2 are enriched in the hippocampal region of the brain but they are present in different combination of subunits in different parts of the brain.
Anti-NMDAR antibody (brain injection)
The antibodies bind to the NMDAR and the receptor, and have a very interesting effect. After brain injection, the antibodies induced apoptosis or cell death in the hippocampus resulting in neuronal cell death.
CSF anti-peptide antibody (brain injection)
Antibodies that cross-react with DNA and the NMDAR peptide by screening serum or CSF, will bind to the glutamate receptor after brain injection and induce cell death.
Anti-NMDAR and NPSLE
If the antibody is injected peripherally (not into the CNS) and the blood-brain barrier is damaged, the antibodies can also exhibit neurotoxicity.
At this point in the research, the question became: if neurotoxicity through autoantibodies cross-reacting with the NMDAR is the cause of NPSLE, what is the correlation between these antibodies and clinical features? A number of studies and papers have tried to answer this question, but the majority of studies failed to find an association.
There has been a fairly consistent association between certain cytokines and NPSLE, although different cytokine studies show different cytokine profiles. The field started with some early reports on interferon alpha (IFN-α). Furthermore, Interleukin- 6 (IL-6) is a fairly consistent cytokine that has been found in the CSF. Additionally, a variety of chemokines (IL-8, MIP-1α, MCP and IP-10 or CXCL-10) are associated.
The neuron keeps our patient awake and thinking, and is important in evaluating pathogenesis of NPSLE. The macrophage light cell in the brain is a microglia and astrocytes are the support cells that have foot processes connected to the blood vessels and provide nutrition and other support. They are the most abundant cells in the brain. The oligodendrocytes make the complement AMS that make the myelin and keep the neurons firing.
Cell debris in CSF
The third finding of importance in NPSLE is the presence of cell debris in the CSF. Estelle Trysberg has shown evidence that the release of damaged material in the CSF patients causes a threefold increase in GFAP, which is a protein derived from astrocytes, and a sevenfold increase in neurofilament triplet protein (NFL). This is biomarker evidence of damage to neuronal elements in support tissue in the brain.
In an attempt to link the autoantibodies, cytokine and cell debris into a unified hypothesis, we will consider how immunocomplexes are potent inducers of IFN-α and some other cytokines. While autoantibodies are frequently thought to be an epiphenomena and not too relevant for disease pathogenesis, it has become clear that immunocomplexes between autoantibody and antigen can induce inflammation through activation of Toll-like receptors (TLRs).
The “Toll Hypothesis”
Above: (1) The autoantibodies to nuclear proteins bind to particles; (2) they get ingested by Fc receptors; (3) the plasmacytoid dendritic cell; (4) they go into endosomes; (5) these endosomes fuse with endosomes containing toll-like receptors. (6) The RNA component of the immunocomplex then stimulates the toll-like receptors 7 and 8 those then activate IRFs, which are transcription factors that will induce the transcription type 1 interferon.
- Antineuronal antibodies are present in many patients that cause neurocytotoxicity. This induces cell death and the release of antigens and the expression of the antigens on cell surface.
- The epitotic bodies are broken up and become antigenic and the antibodies found in the CSF to intracellular proteins bind to these particles forming new complexes. These complexes will bind to plasmacytoid dendritic cells and induce INF-α.
- They would also bind to macrophages and microglia in the brain and potentially induce some other cytokines.
This model can explain how all of the three major components (antibodies, cytokines and cell debris) can be found in neuropsychiatric lupus.
Possible role of IFN-α in NPSLE
What is the role of INF-α in neuropsychiatric lupus?
- Previously reported in CSF in small series (Shiozawa, 92; Ronnblom 5/28 (IL-6 3/28)
- Therapeutic administration (CA, HVC) – 30% depression, psychosis, confusion, seizures
- Transgenic mice that express increased IFN in the brain have NPS syndrome
- Aicardi Goutieres Syndrome – increased IFN in CSF
- Inherited – several clusters including native Americans (Cree encephalitis with systemic immune abnormalities – thrombocytopenia, hypergamma, autoAb)
- Presents in early childhood
- Features include
- Mental retardation
- Cerebral atrophy
- Calcification of the basal ganglia
- CSF pleocytosis
- Elevated CSF IFN
- Molecular Pathogenesis
- Mutations in 3-5 exonuclease TREX1
- Mutations in RNase H2
- 2% of SLE patients have AGS
Does NPSLE+ serum induce greater amounts of IFN-α?
Testing was performed with 20 different NPSLE lupus patients, lupus patients without neuropsychiatric lupus and normal controls to see whether the sera were potent inducers of INF-α (maybe related to the neuropsychiatric features).
The first study showed no difference between anti-P positive and negative patients, disproving the initial hypothesis. This led to the hypothesis that it is not anti-P, but other autoantibodies that are very potent inducers of interferon.
Using different extracts that would include all the usual antigens of Sm/RNP and chromatin were used. And again, there was no difference.
The next step was to test CSF instead of the serum. Other autoimmune disease controls that had been tapped for some neurological manifestation (a couple of Sjogren’s, some RA patients with various disease of Behcet’s disease, etc) were used.
Asking the same question: Do the CSF from these patients induce more interferon in a bioassay using peripheral blood mononuclear cells as the cell type that will produce interferon? The NPSLE patients induced more interferon than the control.
Does NPSLE+ CSF induce greater amounts of IFN-a with nuclear antigen?
Above: The baseline levels are shown without the addition of antigen. The interferon inducing activity causes the LOG scale to go up dramatically.
CSF induces chemokines reported in NPSLE
The next step is to look at the level of induction of other cytokines implicated in CNS lupus as well as the chemokines that are associated. The control set was three different cytokines that were not associated with neuropsychiatric lupus: IL-10, TNF and IL-12p40. This step showed that INF-α and IP-10, IL-8 and MCP-1 all induce very large amounts of cytokine and chemokine whereas there was no significant difference in the control set of cytokines.
Correlation between IFN-α and IP10
It is very difficult to detect interferon alpha in the blood. It is easier to measure interferon signature, which is the effect of interferon on the cells. The same is true in the CSF, part of the reason for that is that interferon is a small module that’s excreted rapidly and has a very short half-life in the blood.
There is a commercial company that measures interferon at a great sensitivity. This company was engaged to measure and low levels of interferon were detected in the CSF patients. There was a correlation between the amount of interferon that could be detected and the level of IP-10. IP-10 was also found with some of the other chemokines (IP-10 is a chemokine that is induced almost exclusively by either type 1 or type 1 interferon).
Based on the tests outlined above:
- There is no difference between interferogenic activity between NPSLE+ or NPSLE- patient SERUM
- NPSLE+ CSF induces significantly more IFN-α compared to NPSLE- or other disease control patients
- NPSLE+ CSF induces most of the chemokines reported in NPSLE and there is a strong correlation between IFN-α and IP-10.
Immunoglobulin G (IgG)
An interesting finding that has some clinical relevance: CSF has very low levels of IgG—close to 1,000 fold less than serum.
Specific activity normalized for IgG is 800-fold higher than serum:
In an effort to explain this, we diluted the serum until the IgG levels would be comparable to what there was in CSF. As the serum is diluted, the amount of interferon induced gets higher. The same was true for IP-10. These findings reaffirm the link between some of the chemokines and the interferon, but it wasn’t true for all the cytokines and cheomokines of interest. For example, IL-8 decreased most of the time when the serum was diluted out.
The same experiment with CSF (a sera values of CSF) did not show this. The amount of interferon induced actually went down.
Does serum contain ‘inhibitory factor(s)’ or CSF activating factors?
There are two obvious explanations of this:
- Serum either contains an inhibitory factor; or
- CSF contains an activating factor.
From what the dilution studies proved, one would suspect that there’s an inhibiting sera. To verify this we did a mixing experiment:
- Take the CSF that’s very potent at inducing interferon,
- Add normal serum and there is dramatic reduction interferon inducing activity.
- On the other hand, if you CSF is added to serum, it doesn’t actually have any effect.
Hypogammaglobulinemic sera show less inhibition of IFN-α
What is in the serum that’s inhibiting the interferon? Is IgG itself involved? Using common variable immunodeficiency patient serum we showed that hypogammaglobulinemic sera was very poor at inhibiting interferon induction.
IgG depletion of normal serum reduces the inhibitory effect
If the IgG is depleted, there is very poor inhibitory activity. And if the IgG is added back, the inhibitory activities are partially restored.
Purified IgG inhibits IFN-α production by Auto-Ab
IgG is one of the inhibitory factors in the serum that helps dampen down interferon responses.
Three models for high-dose intravenous immune globulin (IVIG) anti-inflammatory mechanism
IVIG has been studied exhaustively and remains a very controversial issue. However, recently there have been some discoveries from Jeffrey Abbott. IVIG is thought to either (1) help get rid of autoantibodies by flooding the body with IgG so the bad stuff is spit out; or (2) the IVIG is blocking Fc-activating receptor.
The reason that that doesn’t make too much sense is that when putting in almost 99% monomeric IgG, all the activating receptors pick up aggregates or immunocomplexes. An idea that Abbott has been pushing, based on experimental data, is that IVIG induces a percolation of an inhibitory of CSF—FcγRIIB. That is the only inhibitory receptor, all the other ones are activating.
The anti-inflammatory IgG glycoform has a terminal sialic acid
How can monomeric IgG be inhibitory? Based on Abbott’s work—using animal models—it is most likely the carbohydrate moiety that is present on the Fc, there’s only a single site at which the protein is glycosylated and the terminal group is acidic (sialic acid). Through a variety of experiments where the sialic acid was chopped off or the sialated IgG was affinity purified, Abbott showed that he could control inflammation inhibition through using the sialated form of IgG or not. The sialated form is inhibitory and the non-sialated form is not. This might explain why so much IVIG is needed in patients—because it’s only a small subcomponent of the IVIG that is actually functionally active.
Limitations of this model
There are limitations of this model.
- PRE-treated with IVIG (Prevention vs treatment)
- Arthritis model – measure footpad swelling at day 4-6
- Mouse vs Human – human IVIG in a mouse model
The IVIG receptor
In addition to the subsection of IVIG that’s biologically active, this group found the receptor to which the sialated IgG binds. The receptor was identified in mouse models: a C-type lectin receptor that’s expressed on macrophage in the marginal zone of the spleen that’s called SIGNR1. The human homologue of SIGNR1 is called DC-SIGN.
DC-SIGN is expressed on dendritic cell. It is a receptor that is involved in recognition of various foreign organisms. Interestingly, the foreign organisms may be able to hijack DC-SIGN and promote an anti-inflammatory response. For example, HIV hijacks this receptor, activates it and through sialylation of NF-κβ promotes the production of IL-10, which is an immunosuppressive cytokine.
How does Sialylated IgG suppress IFN-α?
- SIGNR1 (mice) binds sialylated IgG
- Required for IVIG antiinflammatory activity
- Serum contains potent inhibitors of IFN-α production induced by immune complexes
- One inhibitor is monomeric IgG
- In mouse models, SA+ IgG is the active moiety, acting through engagement of SIGNR1 (DC-SIGN)
- A two-step NPSLE model has been developed that can:
- Account for all of the abnormalities reported in the CSF; and
- Show a mechanistic connection.
- There’s a special role for INF-α and it could be altering brain function.
- Using this model, all of the cytokines and chemokines reported in NPSLE can be accounted for.
- Sites outside of serum (e.g. CSF, serosa) allow for large amplification of IFN-a by IC due to the lack of inhibitory factors.