By Virginia Pascual, MD – Baylor Institute for Immunology Research – Dallas, Texas
Dr. Pascual has no significant financial interest or relationships to disclose.
Release Date: September 4, 2008
Expiration Date: September 4, 2010
We’re going to talk today about the work we have been doing to understand two rheumatic diseases: systemic lupus and systemic onset juvenile arthritis. We’ll look at the role of two cytokines in these diseases, and our efforts to not only understand the pathogenesis of these diseases but also to develop tools that will help rheumatologists follow these patients in the clinical setting and try to find isolated targets to treat these diseases.
I’ll start with lupus — a disease that needs no introduction. Because immune complexes play such an important pathogenesis in the disease, most of the research that has been done in lupus has concentrated on auto-reactive T and B cells.
At the Baylor Institute for Immunology Research, we decided to take a different approach a few years ago. I was to look at a different cell type, which is the dendritic cell. The dendritic cell is fundamental in immune responses.
They are the most potent antigen-presenting cells; in this way, they command B- and T-cells. Dendritic cells also secrete amazing cytokines that have a lot to do with innate immunity. They are a true link between adaptive and innate immunity.
It was interesting for us to see that there was almost nothing done on these cells in most rheumatologic diseases. But we decided to look at them for another reason as well. At the time we were initiating these studies, the concept emerged that dendritic cells were not only the most potent antigen-presenting cells but thus the most potent immunogenic cells. Through the work of different groups, it was becoming evident that one of the most important roles of dendritic cells is to maintain tolerance. These cells capture antigens in the peripheral tissues. They drive them to the lymph nodes, and instruct lymphocytes how not to react to them. Thus, immunity and tolerance were two very important reasons to look for them in lupus.
In our initial studies, published in 2001, we looked at dendritic cell precursors — monocytes that circulate in the blood, that by themselves have no good antigen-presenting function. They cannot activate naïve T lymphocytes. What we found was that with the majority of children with lupus, the monocytes were very good at provoking immune reactions.
Without any ex vivo manipulation, they were able to induce amazing proliferation of naïve T-cells. That meant there was something driving these monocytes to behave like dendritic cells. That could be an intrinsic alteration in the monocytes, or it could be a factor in the serum that induces the differentiation of these cells in a dendritic cell manner. We looked for the second possibility and found that healthy serum does nothing to monocytes; in vitro these cells will remain monocytes and will die like those in the tissue culture. The serum of lupus patients, however, will induce the differentiation of these monocytes into these cells very quickly.
They form clusters first, then they make these dendrites, and, upon closer look, these looked morphologically like dendritic cells.
What was in the lupus serum that would induce this differentiation? We tried to block it with every possible anti-cytokine antibody, and the only thing that could abrogate this effect was a polyclonal anti-interferon alpha antibody.
We concluded that this cytokine was driving dendritic cell differentiation in lupus. There was a lot of literature showing that patients with lupus have elevated levels (at least as a group of patients) of this cytokine. In a way, we established a link between the presence of this cytokine and the maturation of dendritic cells in these patients. The hypothesis was that mature dendritic cells, which normally keep tolerance, are exposed in lupus patients to interferon alpha.
Questions that we need to address include:
- Is interferon alpha a universal mediator of lupus?
- Or, knowing how heterogeneous lupus can be chemically, does this explain only a subset of patients?
To answer these questions, we went for an easy answer: to perform blood micro-array analysis. And the answer, which we published in 2003, was very straightforward.
We took peripheral blood from 30 lupus children, a few healthy controls and a few children with arthritis, and very blindly looked at blood signatures. We found that every child with active disease would have expression of these genes that we show were induced by interferon alpha. Since then we have looked at many more patients, adults and children alike. We found these in vitro interferon signatures, and exposed healthy PBMNCs for 6 hours in vitro to interferon alpha. We did up-regulation and down-regulation with more than 250 genes. When we looked at 100 patients with lupus with active disease as well as a group of patients who were ANA positive (such as fibromyalgia patients who did not fulfill the criteria for lupus), we found that everyone with active disease expressed at least a core of this interferon signature.
It is not only expression of genes by itself but also how these genes correlate with disease activity that leads us to believe that interferon is an important pathogenic mediator of lupus. When we get very sick patients, we treat these patients with solumedrol pulses very aggressively. The patients get better after these pulses. One of the things we found was that the interferon signature that is present at the beginning was extinguished after three days of high-dose solumedrol
We have data in vitro that show that one of the effects of these high levels of solumedrol — not low levels — is to completely abrogate the generation in the bone marrow of plasmacytoid dendritic cells, which are the main interferon-producing cells in the body. They also induce the apoptosis of any circulating plasmacytoid dendritic cells that remain in the blood. For two weeks after these pulses, these patients have no plasmacytoid dendritic cells. At three weeks, everybody has the signature again.
The expression of some of these interferon-related or -induced genes correlates with disease activity
However, what we have found is that this is true for patients while untreated. Some of the genes are wonderful markers. When we looked at patients who have been on medication for awhile, even if they are active, these correlations with single gene expression do not work as well.
Expression of these interferon-induced genes not only occurs at the gene of the RNA level. We have been able to find interferon-induced proteins that previously we did not know were interferon-induced, and these turned out to be very nice markers.
Interferon not only has an effect on monocytes and myeloid dendritic cells but on many other cells of the immune system
One of them is B-cells. We wanted to understand this better because, of course, one of the hallmarks of lupus is the excessive production of antibodies, among them auto-antibodies. The blood of lupus patients has significant expansion of these cells, which are CD20 negative, CD19 positive as opposed to conventional reserves, which have double positive
These cells are not unique to lupus patients. They are plasma cell precursors. They do not look morphologically like differentiated plasma cells, but they are loaded with immunoglobulins. These cells are very interesting; we all can have them after immunizations or infections, but they get under control immediately, and they go down to less than 5,000 cells per milliliter. But in lupus patients, these cells are expanded most of the time.
In vitro experiments showed that when we expose mature memory cells to the supernatants from the plasmacytoid dendritic cells activated with virus in a way that they will make interferon-α, these mature memory B cells differentiate into a phenotype that looks identical to the cells in the blood of lupus patients.
These are plasma cell precursors loaded with immunoglobulins with good secretory capacity, but not as good as terminally differentiated plasma cells. A second cytokine can come from virally activated plasmacytoid dendritic cells: interleukin-6. The combination of these two cytokines makes beautiful plasma cells. It is clear that there is another aspect of lupus where interferon may play a very important role.
The question is: What is first? Is it immune complexes? Is it interferon?
We know that immune complexes can activate plasmacytoid dendritic cells, especially chromatin-containing immune complexes through Fc receptors and Toll receptors that can drive these cells to secrete interferon-α. What is first? That is a very important question. I do not know, but we have some pieces of evidence that come from the studies we are doing on cutaneous lupus. When we look at patients with discoid lupus, subacute cutaneous lupus who do not have the serological markers of systemic lupus, we still find that patients with no evidence of systemic lupus but cutaneous lupus already have an interferon signature. They do not have other signatures of lupus that we see in systemic patients, but they have interferon to a very large extent. We believe interferon may be there early. This is preliminary and indirect evidence; however, it may be helpful.
Interferon has an effect on T-cells as well.
This is an important concept that my collaborators are looking at very actively. It is very obvious that the cytokine environment where dendritic cells mature is going to lead to very different T-cell responses.
When myeloid dendritic cells mature or differentiate in the presence of interleukin-15, they make T-cells that are very efficiently cytotoxic. Where they mature in the presence of TSLP, they drive amazing TH-2 responses. When they mature in the presence of interferon-α, they give rise to very peculiar dendritic cells, which give rise to very peculiar T-cells. One of the ways we are looking at this is with micro-array. We take monocytes that have been exposed to interferon-α, have been made into myeloid dendritic cells driven by interferon-α, and we expose either CD-40 or CD-80 cells to these interferon-made dendritic cells. We looked at the gene expression profile of the CD-40 cells. We compared these T-cells exposed to interferon DCs to T-cells exposed to interleukin-4,DCs, tumor necrosis factor (TNF) DCs,, and TSLP DCs,, and we saw very different signatures.
Something that may be of interest is that the CD-40 cells that are made in the presence of interferon dendritic cells express a very interesting cytotoxic gene program. This is at the RNA level, but there is suggested skewing toward these cytotoxic protein-expressing CD-40 cells.
The same thing happens at the CD8 T-cell level. The CD8 T-cells exposed to interferon dendritic cells expressed more granzyme A and more granzyme B than any other cytokine exposed T-cell or any other dendritic cells in the presence of other cytokines in vitro. Others have shown how the CD8 T-cells make very nice nucleosomes in vitro. We believe this is very relevant to the pathogenesis of lupus. CD8 T-cells are very good at killing. For instance, we found that these cells kill a breast cancer cell line target much more efficiently than any other CD8 T-cell made in the presence of any other dendritic cell. They are very good cytotoxic cells.
We have found that the adult lupus population expresses or have expanded populations of CD8 T-cells, which express more intracytoplasmic granzyme A and granzyme B than in healthy people.
So, I hope I have convinced you that interferon-α may be a very relevant cytokine to the pathogenesis of lupus. It can work at different levels, not only at the level of dendritic cells, and it surely drives the maturation of myeloid dendritic cells into very peculiar antigen-presenting cells.
It drives mature B-cells into antibody-secreting cells. It directly induces cytotoxic responses, but also through the making of these interferon dendritic cells. It also drives indirectly the expansion of cytotoxic cells.
One question that remains — and this is a true work in progress — is this a lupus monocyte?
We do not understand it completely. In many cases, it induces alloreaction. It already works like a dendritic cell. However, when we look at its phenotype, it looks like a regular monocyte. We have looked at an amazing number of phenotypic markers and cannot find obvious differences in terms of Class-II expression of stimulatory molecules. There is only interferon. Additionally, we have done micro-arrays, and do not have the answer. This can be very complex.
For instance, in a micro-array expression profile of purified monocytes from five newly diagnosed, completely untreated children compared with lupus versus healthy children’s monocytes, we found close to 1,000 genes that go up and down, straight from the blood. Many are interferon-induced. In a signature of monocytes exposed in vitro to interferon-α, 600 genes out of the almost 1,000 could be explained based on interferon. However, there were 300 to 400 that could not be explained based on interferon. When we compared these genes with the gene profile of bona fide myeloid dendritic cells, those genes are expressed in the myeloid dendritic cells. Therefore, there is a program of differentiation at the genetic level in these lupus monocytes already that we cannot completely explain through interferon. We are looking at what other pathways may exist to explain this dendritic cell-like differentiation of the monocytes.
Again, I hope I have convinced you that interferon is important, that this is one of the main pathogenetic loops in the disease, and that blocking interferon-α may be of relevance to our patients. We have now developed a monoclonal antibody that sees most of the species of interferon-α. There are many, unfortunately, in the human. This antibody blocks the ability of lupus serum to induce this effect. It completely abrogates the differentiation of the monocytes into dendritic cells. We hope we will be able to take this antibody to the clinic soon.
At this point, I’d like to discuss how we are using some of the tools that we have developed to assess clinical activity. What we have done is a bit away from conventional micro-array analysis. We were not going to be restricted to autoimmune diseases such as lupus and juvenile arthritis; rather, we have a large database of infectious diseases — patients with liver transplants, diabetes, graft versus host disease, even fibromyalgia — which allows us to have a lot of data to compare with our autoimmune diseases. Basically, we look at the blood arrays of different diseases, and a lot of in vitro data. We exposed blood to many different cytokines, many different stimuli, and we saw what each of these cytokines does to the cells in vitro. Rather than a “regular signature” type of analysis, we are doing modular analysis. We have been able to select what we believe are biologically relevant genes from a variety of diseases and identify disease-specific transcriptional fingerprints. From a conventional signature, it’s not possible to get an idea of what is going on; but after lots of analysis and rounds of selection of signature, the fingerprint emerges. Visually, they are much more attractive. These fingerprints are composed of the analysis of modules. We looked at modules first by identifying sets of transcriptionally co-expressed genes within individual signatures. These genes go in the same direction in individuals all the time. In the second round of selection, an algorithm does the data mining. We go into groups of genes and try to find evidence in the literature that they are biologically linked. Sometimes we can do that, sometimes we cannot. However, we still believe — because of the third round of selection — that they are truly co-expressed and they truly mean something together. In the third round, we look at healthy versus patients and see that these modules behave differently in disease compared with healthy individuals. Almost 5,000 genes can be classified into 28 modules. We’re amazed at how these modules behave. The way we visualize this is with little circles: We take all the genes expressed within a module, and we represent the number or percentage of genes in red or blue that are either up or down in a particular disease. We can see this in a slide and can see what lupus would look like. We have a fingerprint of lupus, and we can compared it with the fingerprint of other diseases and conditions — the flu, another disease that is mediated in big part by interferon-α; melanoma, liver transplant patients going into rejection. They are all very different.
This is teaching us about pathogenesis, but can we use it to follow disease activity? Looking at individual gene’s correlations or even signature correlation with disease activity can work, but it is not good enough. Patients under heavy treatment do not behave as well as untreated patients, for example. If these fingerprints are of pathogenic importance, maybe by combining them we can come up with good activity markers. We used a U statistic method developed elsewhere in multi-varied analysis with our modules. After many rounds of module combination, we concluded that, by using five of these modules in a multi-varied U statistical method, we could come up with what we call a genomic score, which at least against the SLEDAI behaves very well, with nice correlations. This behaved very well in treated and untreated patients. It not only works cross-sectionally, but we can follow patients longitudinally. For example, with one girl who we have been following for a while, her SLEDAI went from 5 to 0, and we would think that she is doing very well. However, her genomic score was very high. It was telling us that she was at the end of the spectrum for disease activity. It turned out that she was diagnosed with pulmonary hypertension — which, of course, is not recorded with SLEDAI. Consequently, this child was very sick. Our score behaved very well.
What I have told you so far in terms of pathogenesis is that dendritic cells may play an important role in autoimmune diseases, but more than that, the cytokine environment where these dendritic cells mature is extremely important for the immune response that will follow. My message today is that plasmacytoid dendritic cells, or maybe other cells, produce an interferon-α, make a type of dendritic cell that is very peculiar, and drive a very peculiar immune response.
This explains a lot of what goes on in lupus, but there’s more. We know that inflammatory myopathies, dermatomyositis in particular, also have an interferon signature in the blood. We know that plasmacytoid dendritic cells infiltrate the muscles of these patients. We also know that, in the early stages of psoriasis, plasmacytoid dendritic cells and interferon-α infiltrate the skin of these patients. Interferon may be a mediator for different types of immune responses. What follows may be dictated by the genetic background of the patient. TNF drives different types of immune responses, which may explain other diseases.
We have been interested in the interplay between TNF and interferon-α because of the fact that blocking TNF drives some people to develop lupus symptoms. When the plasmacytoid dendritic cell, which is the main type-1 interferon reservoir, is hit by either viruses or immune complexes, it not only becomes a bona fide dendritic cell, but it also secretes type-1 interferon and large amounts of TNF-α and interleukin-6. It turns out that TNF-α is an autocrine maturation factor for these cells. When they mature, they can no longer make interferon-α. We thought that perhaps blocking TNF-α could keep these cells at an intermediate stage of maturation, where they could continue secreting type-1 interferon. We immediately began experiments, and they showed that that was true. When you treat plasmacytoid dendritic cells with an anti-TNF blocking agent, that cell keeps making interferon-α in response to viruses and immune complexes. If we look at an interferon signature in vitro, healthy children do not have that signature; lupus patients do (slide 26). Children with arthritis do not express this signature. However, when we attempted to control the disease by giving the anti-TNF agents, they did express the interferon signature. This seems to be very indirect evidence, but supported by the in vitro data, it looks as if one of the things we may be doing to patients treated with our anti-TNF agent is allowing them to make more interferon-α.
Now I’d like to focus on a different disease — systemic onset juvenile arthritis.
which we started looking at because we had become very frustrated. With anti-TNF agents, we are making most of our arthritis patients so much better; however, there was this group of children who we were not making better. We decided to look into the pathogenesis of this disease.
This form of juvenile arthritis can be one of the most severe.
It has no age predilection, no gender predilection, and is completely different from all the other types of juvenile arthritis. Patients never have uveitis or autoantibodies. To me, this was a clue that this was a disease of a different category, and it looked from the beginning as if this could be an autoinflammatory disease with recurrent symptoms — fever being such an important part of it that we have to at least look at it and try to understand it. We did not have a good treatment, nor did we have a good specific diagnostic test. The average time from initiation of symptoms to diagnosis in our clinic — and we have a lot experience with this disease — is 3 months. These children go from pediatrician to pediatrician; they are hospitalized, they get all their blood tests, they get marrow aspiration — and it takes us 3 months to make a diagnosis.
In terms of pathogenesis, I think that we were very lucky because we chose to do a simple experiment that had already been successful in lupus. We took healthy serum, or serum from four patients in the systemic phase of the disease, and we cultured healthy blood cell with this serum.
Six hours later, we extracted the RNA and performed a micro-array analysis. The results were very straightforward. The first thing we saw was significant upregulation of different members of the interleukin-1 family.
Of course, interleukin-6 and interleukin-18 have been implicated in the pathogenesis of this disease. We could not find any significant up-regulation in these short-term cultures; later on, of course, we did find interleukin-6 and interleukin-18. Nevertheless, in the first 6 hours, what we found was interleukin-1. Then we found other molecules, such as chemokines, chemokine receptors, and innate immunity receptors such as pentraxin-3, which is directly induced by interleukin-1. One of the most significantly up-regulated was a potassium rectifier channel encoding gene. We know that these potassium fluxes are fundamental for the secretion of IL-1. All of these genes, in a way, were telling us that maybe IL-1 was an important pathogenic clue in the disease.
From time to time we saw this in 2003 until we published a paper on it last year, we did many other experiments to prove that this was not only at the gene level, that we could also find protein, that the PBMNCs of these patients upon certain stimuli will make more IL-1. After I saw these results, we started treating children with anakinra before we did any other experiments. The results were spectacular.
We started in a subset of patients who were refractory to everything else — children who had had this disease for years, were refractory to very high steroid doses, anti-TNF agents (at least half of them), methotrexate, and every other possible drug that you can think of. We started anakinra, and the fever went away immediately.
Arthritis went away in about eight of nine patients, and we’ve had many more since. Laboratory parameters also improved tremendously: white blood cell count, hemoglobin, platelet count, sedimentation rate, etc.
I believe systemic arthritis — like lupus — is a heterogeneous disease. Not everyone responds to blocking IL-1, but in our experience, most children do. Additionally, most children respond with a sustained response, although some kids do not. But at the least, we are able to induce very nice partial responses.
Finally, I want to describe where we are in trying to develop better diagnostic tests for the disease. We have a good drug that treats most patients. If we can use it early, I think we will be able to prevent even the development of arthritis. That is our next goal.