The Establishment of B Cell Tolerance in Humans

Eric Meffre: Thank you for that very kind introduction. This cartoon depicts the three mechanisms of central B-cell tolerance. Antibodies are generated during B cell differentiation by random joining of immunoglobulin gene segments. Immunoglobulin heavy chain genes rearrange first at the pro-B cell stage while light chain genes rearrange second at the small pre-B cell stage. After these series of rearrangements, immunoglobulins are first expressed at the immature B cell stage, but since the whole process of recombination is random, you can generate self-reactive antibodies. And it is at the immature B cell stage that non-reactive antibody-expressing clones are selected and there are three mechanisms that have been shown to ensure B cell tolerance. The first one is deletion, which corresponds to when a self-reactive antibody is expressed at that stage, signaling induced deletion and removal of the immature B cell at that stage.

The second mechanism is anergy, and this has been extensively studied by the group of Chris Goodnow and, instead of inducing deletion, the cell can remain still expressing a self-reactive antibody, but B cells are unable to respond to antigenic stimulation. Finally, there is a third mechanism, receptor editing, discovered by the group of David Nemazee and Martin Weigert, where in fact, in this case ongoing recombination replaces some chains and thereby change the specificity of antibodies, and can rescue the self-reactive clone by transforming a self-reactive antibody into an innocuous antibody. When the cell is fixed, it can migrate to the periphery as the new emigrant, and then be further selected into the mature naïve B cell pool. So these mechanisms have been shown to be very important to establishing B cell tolerance in the mouse -- all of these have been demonstrated by using transgenic models or animals and nothing has really been studied on physiological conditions. Even if we assume that random joining of all these immunoglobulin segments can produce self-reactive antibodies, nothing has been done in order to determine the percentage of the repertoire that has been generated and can be self-reactive, and, at what stage, especially in humans, the self-reactive B cell clones can be removed.

So this would be the first part of my talk. I am going to talk to you about how B cell tolerance is established in humans. In order to answer that question, we decided to do this analysis by doing single B cell analysis. For this we had different markers that allowed us to sort different B cell populations from the bone marrow, like pre-B cells or immature B cells, as well as the in the periphery, where we could sort new emigrant B cells and mature naïve B cells. By definition, pre-B cells expressed heavy chain transcripts but no light chains. We were surprised to find that in this compartment that we thought were pre-B cells, we could identify cells that expressed in frame heavy and light chain transcripts, so they seemed to be able to produce an immunoglobulin. We called these cells early immature cells and separated them from pre-B cells.

The way we analyzed them is that we sorted single B cells; we plated in each well one cell; and then did c-DNA synthesis that would allow us to amplify the genes by RT-PCR. The strategy for the RT-PCR is a PCR reaction, where the first reaction amplified the immunoglobulin VDJ genes or the kappa genes or the lambda genes, and, in fact, a second PCR introduced restriction sites that we used to clone these individual rearrangements into expression vectors that contained the constant region of all these immunoglobulins.

I just would like to mention that all the clones during the early B cell differentiation were all IgM clones. However we turned them into IgG clones in vitro for production purposes because IgGs are way easier to produce in vitro compared to IgMs. By doing this technique, we cloned more than 300 bone marrow and peripheral B cell antibodies. And out of these 300 antibodies, we transfected these in the fibroblast 293 A cells, and we could obtain 29 early immature antibodies, 72 immature, 54 new emigrant and 93 antibodies from mature naïve B cells.

In fact, we reproduced what was already described in the literature, which is that we start with longer IgH CDR3s. Heavy chains that have a long CDR3 decrease while the B cells mature and you can see those that have short CDR3 early on are rather amplified while the B cells mature. It seems that during B cell development you have a counter selection of immunoglobulins that have long IgH CDR3s. Similarly, we could identify that there was an increase in highly positively charged IgH CDR3s, and if this is of interest, it is because autoantibody and anti-DNA have been shown to have an increase in positively charged residues. We found that these were highly enriched in very early precursors. While B cell progresses, you lose these highly positively charged residues and this is statistically significant if you compare these fractions to the last fractions.

So what about the reactivity? We purified the antibodies that we expressed in vitro and tested them by ELISA for ANA activity, and we used ANA plates that you can buy. They are standard plates to test for ANA antibodies in patients. Here we depicted the situation for two different healthy donors. You can see that in early immature, the very earliest B cell that could express antibodies, almost 75% of the antibody expressed in this population in fact are ANA antibodies. Thus you start the repertoire with a tremendous amount of self-reactivity.

When the cell progressed to the immature B cell stage, you see that you still have a lot of self-reactive antibodies. But this percentage is now decreased to about 40%. So we showed a first step of selection at this very early immature B cell stage where you remove already about half of the self-reactive antibody you started with.

Then the cell migrates to the periphery in the blood and there are new emigrant B cells, but you see that it seems like there is no selection between these two steps. It just seems to be a release of these immature B cells in the periphery, but there is no further counter selection of self-reactive B cells.

Here there is no further selection, but when they mature and you get mature naïve B cells, what you see is another 50% of these removed and, again, this is statistically significant. So this first assay allowed us to show that there are two steps for counterselection of ANA autoreactive B cells, the first one very early on in the bone marrow between early immature and immature B cells, the second one in the periphery between new emigrant B cells and mature naïve B cells.

To confirm these results, we did stainings - immunofluorescent stainings - and that is what I was talking to you about. ANA, which are the real true antinuclear antibody, and you see these antibodies really stain the nucleus of the cells with different patterns or another kind of stain, which is subnuclear and cytoplasmic staining. And finally, the only exclusively cytoplasmic, where you see the nuclei are not stained.

If we look at the specificity of these antibodies to self-reactive, including all of these, and ANA including these two, you can see that early immature B cells show a really high percentage of true ANA antibodies, and very few antibodies are non-self-reactive. You can see the evolution of the non-self-reactive antibody. The proportion keeps increasing while B cells develop and mature, and again, here first a counter-selection of these anti-ANA antibodies and no further selection between the mature and new emigrant B cells, and then again the second removal of some ANA antibodies.

Finally, I would just like to point your attention to these cytoplasmic self-reactive anti-cytoplasmic antigens that seem to be very poorly removed and poorly counterselected. However, we may consider that if these proteins are cytoplasmic they may be soluble proteins and they may not have such power of BCR cross linking compared to nuclear antigens like DNA that would be very good at cross linking antibody and stimulating B cells.

We have now checked very specific antigens and we tested the reactivity of the antibodies by ELISAs, including single-stranded DNA, double-stranded DNA, as well as proteins like insulin. What you can see is that in the first fraction, the early immature B cell, you have again all these antibodies that in fact are reactive with all these antigens and which should be, by definition, polyreactive antibodies. However, when these (antibodies) are from mature B cells, it seems that all the polyreactive antibodies are removed from the repertoire. You can see that you start with about 50% of the antibodies that are reactive and drop to about 6.9%, 7.4 and then 4.3. So somehow, there is no further counterselection really of these antibodies. Thus, here there are early immature B cell antibodies that have auto-reactivity against a lot of antigens. How can we correlate some of these autoreactive antibodies with some of the features that I have previously described to you. First, what about the lengths of the IgH CDR3s? What you can see is if we segregate non-reactive and self-reactive antibodies, we found that you have a real increase in very long IgH CDR3s in the antibodies that are self-reactive, suggesting that this feature definitely is favoring self-reactivity.

In conclusion, at least half of the bone marrow B cell precursors that are first generated express self-reactive antibodies that usually show long CDR3 and high frequency of positively charged amino acids. So it seems that the price we pay for high diversity is really enormous, because we start by creating so many clones, but most of them in fact are not good clones and you have to remove them.

Here, by the way, when I say "positively charged", I mean positively charged CDR3s of the heavy chain. Of course, you can take into consideration the CDR3 of the light chain that also contains some positive charges and also contribute. Here we considered only the IgH CDR3 but the CDR2 and CDR1 may contain some charges too.

Thus, there is a tremendous price to pay for B cell tolerance and the creation of diversity to actually remove a lot of auto-reactive B cell clones that are generated very early on. Second, these self-reactive B cell clones are removed in a two-step process. As I showed you, the first one in the bone marrow between early immature and immature B cells and the second one in the periphery between the new emigrant and mature B cells.

Depicted on this cartoon you can see that when the small pre-B-cell express light chains, these early immature B cells are so enriched in self-reactive antibodies that they seem somehow to be stuck there and they do not express the antibody because probably they already recognize right away something and it (antibody) isn't expressed on the cell or is rapidly internalized. Since we lose a lot of the characteristics of this population of long CDR3s and positively charged residues, we believe that some of the clones probably would undergo deletion and are removed right away, and some others may be rescued by receptor editing, by finding another light chain that may silence these clones and migrate to the immature B cell stage. then in the new emigrant B cell stage again potentially some other clones may be silenced. Deletion potentially, may be involved in that stage to remove these clones before they migrate into the long-lived B cell compartment.

Now in the second part of my talk, I would like to talk to you about a very specific and small B cell population in the periphery of healthy donors that express this molecule called V-pre-B. In fact V-pre-B is normally expressed here in the pre-B cell stage and usually is associated with IgM to produce a pre-B cell receptor. This complex is down-regulated during the transition of large to small pre-B cell and is not expressed any further during B cell development. However we have found a very small population of B cells in humans that express still this very immature early marker. V-preB cells represent only 0.5 to 1% of the total peripheral B cells and they express RAG genes at low levels. They are the genes that produce the protein that are responsible for VDJ recombination.

In fact, these cells seem to have a mature phenotype. It is not depicted here on the slide, but they are neither new emigrant B cells nor bone marrow precursors and, when we analyze the antibody repertoire of these B cells, we find that they have very unique features, and I am going to detail a few of these features. This is the study of three different healthy donors and these results were obtained by RTP-CR. When we look for instance at the JH usage in heavy chains, we find that it is very similar among the three donors in conventional B cells that do not express V-preB, while here in V-preB positive cells, there is a real significant increase in JH6 usage.

In the three healthy donors, we also found that the IgH CDR3 here in black was on average way longer than in conventional B cells. You see these two populations, and these results were reproducible in all three independent donors. Somehow JH6 can contribute to this increased length because it is the JH segment that has the longest amino acid encoding sequence in the human genome. However, there is another contributor to these long IgH CDR3s which is the very usually infrequent D-D fusions in VDJ rearrangements. Conventional B cell show VDJ rearrangements instead of in V-preB-positive B cells, in which you can find some cells that can display VDDJ rearrangements, which is completely unusual and it can represent up to 10% depending on the VH family that you study. They are very easy to recognize because there is absolutely no doubt that there are two Ds in tandem.

Finally, the reading frame of these Ds that are using these cells is also very unusual because it favors hydrophobic residues that are encoded by reading frame 3, while in conventional B cell it is normally hydrophilic residues that are encoded by the reading frame 2. So you see that for all these major D segments that are used very commonly, V-preB positive B cells really do favor these hydrophobic residues instead of hydrophilic, and this is unusual because the IgH CR3 loop is in contact with the solvent and potentially with the antigen. All of these hydrophobic residues should confer a very specific configuration to these antibodies. Thus, this is summarized here in that these very specific B cells have a very biased repertoire to JH6 and have long CDR3s including D-D fusions and unusual D reading frame.

This was very interesting because in fact long CDR3s with D-D fusion were typical for autoimmune repertoire and the autoimmune mouse model MRL/lpr mice. Second, there is this gene, VH1-69, in this population that really resemble those that are found in B-CLL because VH1-69 is specially associated with JH6 and that is what we find in these cells and B-CLL are known to express self-reactive antibodies. V-preB positive B cells may be cells that have escaped deletion in the bone marrow and could have migrated to the periphery.

I would say that we have no evidence whatsoever that these cells can be dangerous, especially since we find them in all healthy donors. I don't have any explanation why they keep this marker, and I will show you later on during the talk that they may have some other B cell functions.

I try to depict V-pre positive B cells in the cartoon, it seems that they have escaped deletion because we found V-pre B cells in the periphery. But are they anergic? or have they suffered receptor editing? So what about receptor editing? Here is depicted the human kappa locus and with a prearranged V-kappa, J-kappa 1 for instance. In fact, the J-kappas that are downstream of this J-kappa 1 all have recombining sequences that are compatible for recombining sequences of upstream V-kappas. Secondary rearrangements would result in the deletion of the first rearranged genes and thereby skew the repertoire to downstream J-kappas and upstream V-kappas if this phenomenon occurred.

What about the J-kappa usage in V-preB-positive B cells. Here in conventional B cells you can see that J-kappa 1 and J-kappa 2 are the major contributors to the repertoire, while here in V-preB-positive cells you can see that J-kappa 4 is significantly increased in all three donors. This is suggestive that these B cells have suffered receptor editing since they show a bias to more downstream J-kappa usage.

For all of these donors, a range between 30 to 40 sequences has been analyzed for all individual donors. Here on this cartoon, we presented the V-kappa locus, at least the proximal locus, with the more downstream genes and the more upstream V-kappa genes, and you can see that, yes, it seems that there is some increase in upstream V-kappa genes. However, the one that was really statistically significant was the most downstream of all in the V-preB-positive cells, and in fact all of this is the result of the expression of a single gene in this compartment, which is the most downstream of all, V-kappa 4-1. We were intrigued by these results, but then we started to look closer the literature. We found that the V-kappa 4-1 and the usage of this gene was increased in patients with lupus. So it seems that there may be some kind of auto-reactivity that is conferred to this gene and, on the top of all these, we found some very unusual 11 amino acid-long this time in the Ig-kappa CDR3s. This is here in the V-preB-positive cells; this is found in 11% of the cells that have these very unusual, very long 11 amino acid-long CDR3s, while in conventional B cells the proportion is usually about 2 to 3% maximum. This was interesting because these very long Ig-kappa CDR3s were found in patients with rheumatoid arthritis, again linking these features to potentially some autoreactivity.

It is summarized here -- that we found that there was downstream J-kappa usage and potentially upstream V-kappa usage suggest that receptor editing may have occurred in these cells; that V-kappa 4.1, the more downstream of V-kappas is over expressed and is thought to incur potentially self-reactive antibodies and potentially anti-DNA and is frequently found in lupus; that we have these increased very rare 11 amino acid-long CDR3s in Ig-kappas, that are found in the joint of patients with rheumatoid arthritis. Thus, when you combine all the features from the heavy chain or the light chains, everything would seem to suggest that there is some self-reactivity going on in these cells.

Do V-pre B-positive B cells really express self-reactive antibodies? We used exactly the same method that I described to you previously: Single cell analysis. As I told you, V-preB-positive B cells represent about 0.5 to 1% of the peripheral B cells. Here they are already pre-enriched with some magnetic beads up to 4% and then by a first sort at 64% enrichment and then by a second sort finally to have the population as pure as possible, and here, as a control, we have conventional B cells that do not express V-pre-B.

We were first very happy to find, in fact, all the features that I have talked to you about in the batch study with two new individual donors, with again biased to JH6 usage compared to conventional B cell; way longer IgH CDR3s for both the kappa clones and the lambda clones; and you have unusual D reading frame. We found previously that some V-preB-positive clones could be mutated clones, but when we did the single cell analysis and by doing two sorts, of course, we get the cells with a degree of purity that was way better than previously, which was a single sort. And we realized in fact that they had all their Ig sequences almost in germline configuration, so unmutated antibodies.

Finally for the Ig-kappa again increased downstream J-kappa usage more combined to upstream V-kappa usage, and still this very unique V-kappa 4.1 usage once more and these 11 amino acid-long Ig-kappa CDR3s while these features are very rare in conventional B cells.

Q: It is the most downstream of all.

Dr. Meffre: Yes. It doesn't fit; it seems more like a special selection. On the other hand, you cannot completely exclude receptor editing because you know that the kappa locus is duplicated and you have the proximal and the distal locus. All the distal genes rearrange by inversion, which means that the V-kappa 4.1 will remain in the genome and will be flipped upstream. Which means that if you have a secondary rearrangement and you can bring back downstream V-kappa 4-1 and that is what may happen actually. So that is why it is difficult to tell if it is really only from proximal locus, but there is definitely an increase of this gene in V-preB-positive B cells compared to conventional B cells.

What we found also is that there is an increase in these positively charged residues in V-preB-positive B cell CDR3s compared to the conventional B cells which I've just showed you before that again it was a feature that is suggestive of self-reactivity but no increase in negatively charged residues.

Again these are the same conclusions that I have shown you before that we can recapitulate. We found for single V-preB-positive B cells compared to the batch study with again all the features, long CDR3s and all the specific kappa usage. All the antibody sequences were germline and all the antibody expressed are IgMs which would fit maybe potentially with a "natural antibody" profile, and then, IgH CDR3s revealed a high frequency of positively charged residues in these clones. What about their self-reactivity? Do they (V-preB-positive B cells) express antinuclear antibodies? In these sorted conventional B cell, we found again that the percentage of self-reactive antibodies was around 10% which was consistent to what we have described in the previous study, but then when we looked at the V-preB-positive B cells, in fact almost 70% of the cells expressed self-reactive antibodies again tested for ANA using ELISA available plates.

To confirm this result we did immunostainings on HEp-2 cells, and this was a negative control antibody that was isolated from conventional B cells, and you can see that all the nuclei are not stained. But when we compare this profile to the Vpre-B positive antibodies you can see that the nuclei are very well stained with some different patterns including some ribonucleic protein pattern or more homogenous pattern and some of the clones in fact were against the cytoskeleton. These look like actin stress fibers and these more like vimentin and these like vinculin.

What about their reactivity with specific antigens? We looked at specific antigens, here single stranded DNA, double stranded DNA, IgM rheumatoid factor activity, insulin or LPS and again in conventional B cells we find very few clones that have some self reactivity to these antigens and with a very low affinity. This is a positive control polyreactive antibody that was obtained from Paolo Casali at Cornell, and when we did the analysis of the V-preB positive clones, you can see that all these antibodies, a lot of them, have very good reactivity especially the rheumatoid factor IgM activity is very good in these clones. Thus, about a third of V-preB-positive B cells really express polyreactive antibodies.

Then when we aligned the sequences of these antibodies that are polyreactive, what we found was very interesting. Most of them were JH6 which I showed you was increased in this population, using a very unusual D-reading frame and have these very long CDR3s and these JH6 in fact combined their stretch of tyrosines that is in the genome to positively charged residues. And this seems to be kind of the consensus to all these antibodies that have this stretch of tyrosines combined to positively charged residues that seems to segregate in this population. And actually that was exactly the case with the positive control polyreactive antibody that was obtained from Paolo Casali. This is a JH6 clone with hydrophobic reading frame, it has a very very long CDR3 with this stretch of tyrosines combined to multiple positively charged residues.

In conclusion of this part, about 70% of the V-preB positive cells produce anti-ANA antibodies in healthy donors. Most of them produce anti-ANAs but some do recognize the cytoskeleton and are not true ANAs. In fact, both clones that we found with 11 amino acid-long kappas; both clones were found self-reactive. So about a 1/3 of the V-preB positive cells expressed polyreactive antibodies that share this common sequence in the IgH CDR3 and this, in fact, is a result of a combined stretch of aromatic residues or hydrophobic residues combined to positively charged amino acids.

But what is really the physiological relevance of V-preB-positive B cells in humans? Are they just cells that have escaped deletion from the bone marrow, are they anergic B cells? What about these B cells? It is very difficult to assess this question in humans so we decided to produce a mouse model. We did a transgenic mouse with one of the heavy chains that expressed the clone ED-45 which is a polyreactive clone that has the consensus IgH CDR3 sequence that I have just shown you. You can see that the B cells that expressed the transgene were stuck very early in B cell differentiation and seem to be deleted. The clones don't seem to progress in B cell differentiation. We could show that we have the expression of the transgene that is a human IgM transgene compared to wild type mice which don't express the transgene.

Most of the V-preB-positive H transgenic B cells are in fact deleted from the bone marrow but few cells can make it and migrate to the periphery and you can see that these cells express the transgene compared to the wild type B cells that don't express the transgene. So you do have potentially V-preB positive H transgenic B cells in these mice but what was very interesting is that that these B cells migrate in a very specific compartment which is the marginal zone.

There is an enrichment in the transitional B cells probably because of maturation problems I guess and then most of the cells that accumulate in these mice migrate to the marginal zone and do not mature very well in classic B-2 mature B cells. This is really interesting because it has been shown that marginal zone B cells are usually enriched with clones that are self-reactive or at least polyreactive and that would confirm that transgenic B cells are not only potentially deleted but maybe really have a physiological relevance and migrate to a geographic area of the spleen that are just in the front line of blood born pathogens which could be the case for the V-pre positive cells.

That is what is summarized here: most of the time these chains (expressed by V-preB-positive B cells) are removed during B cell differentiation. However, these cells escaped deletion and could migrate to the periphery. They could migrate to a very specific compartment: the marginal zone. This was also very interesting because Martin Weigert has described a population of cells that are in fact dual expression B cells, which means that they in fact express two B cell receptors with the same heavy chain with two different light chains. One is a kappa chain and the other is a lambda chain, and when these have dual expression, they all migrate to the marginal zone also. I would like to postulate that our cells in fact are the equivalent of dual expressing B cells because they express one potential receptor with a surrogate light chain (V-preB) and the other one with a conventional light chain (kappa or lambda). In these dual expressions in Martin Weigert's model, one is self-reactive and other is not self-reactive, and the one that is not self-reactive, the chain has a low isoelectric point and that is exactly the case in surrogate light chain in humans -- the isoelectric point is very low. In contrast, the surrogate light chain in the mouse has a very high isoelectric point therefore they would be very bad in trying to silence the self-reactive B cells, according to Martin Weigert's hypothesis.

How to fit these cells into the diagram that I have shown you before? In fact, these cells have exactly all the characteristics that I've shown you for the early immature B cells. A lot of them express polyreactive antibodies with very long IgH CDR3s, and lots of positively charged residues. So they (V-preB-positive B cells) seem to derive from this population and somehow be able to escape deletion and migrate to the periphery. What I can tell you is that these cells in the early immature compartment do not have the consensus sequence that you find for most of V-preB-positive cells. Therefore only a fraction of these seems to be able to escape and who knows maybe are selected to do so.

These cells (V-preB-positive B cells) express really true ANA antibodies in all healthy donors. Could these cells be involved in the development of autoimmune disease? We assessed this question by looking in the joint of a patient with rheumatoid arthritis and had a very interesting finding that V-preB positive cells were highly enriched in the joint in 30% of the patients that had joint replacement. You can't find that in the blood of these patients. The level of these cells is similar to normal healthy high donors. In fact, these cells when you look for CD10 expression which is an activation marker that is expressed in germinal centers, you can see this time a fraction of the cells expressing CD10, while in normal donors in the blood they do not express CD10. It really seems that some of V-preB-positive B cells get activated in the joint of RA patients.

[Indistinct Question]

We are actually planning a collaboration with Dr. Bob Hotchkiss to go back now to the synovium and try to really single sort these cells and really try to identify and isolate the antibodies and identify what are the antigens that they recognize in the joint.

What I think is in that scheme here. I think that there are two possibilities in RA. Like I said these cells (V-preB-positive B cells) express self-reactive antibodies. We know that natural antibodies are self-reactive antibodies that are beneficiary for the body. So potentially B cells maybe important in, for instance, producing antibodies to help clear out apoptotic cells, dead bodies. So you can see the scenario two ways for the joint. Either there is first some destruction of the joint, obviously there is a lot of dying cells. So maybe these cells maybe recruited to try to clear the mess or maybe to try to help the resorbing problem with the inflammation. Maybe you can think that down the road they may have help for instance from dendritic cells or T cells that may activate V-pre-B cells and this time turn the IgMs for instance into IgGs and the IgM is anti-inflammatory and the IgG will be pro-inflammatory and may help to fire the inflammation in the joint.

On the other hand, you can imagine the other scenario that since they (V-preB-positive B cells) produce and displayed self-reactive antibodies, maybe they first were recruited into the joint because they recognize some self antigens there and then, because they get activated there, maybe they can start, I would say almost initiate it. I don't really believe that they initiate it but we cannot completely exclude it either.

What I would say is first, I don't know if these cells in humans really go to the marginal zone; we haven't done it in humans, only in mice. Yes, usually what is found is that these clones that express self-reactive antibodies that seem to be beneficiary because as I said, natural antibodies are good to clear viral infections or bacteria. In fact you seem to segregate them (polyreactive and self-reactive B cells) in compartments to remove them from follicles and germinal centers where they can be activated. That is why you segregate them in the marginal zone, in the peritoneal cavity for B1 cells, you try to put them in geographic situations.

[Indistinct Question]

Yes, it seems that they are. They are just like mature naive B cells though they have self-reactive features. It is only in the joint that I could find some activation marker of B cells that suggest to me that, yes, they may be able to. So even if they are normally potentially anergic, which I don't believe either because the level of immunoglobulin that they express is similar to normal B cell, while usually anergic B cell have low level of B cell receptor on the surface. It definitely can be activated in the joint.

Yes, that is what I've showed, for instance, in the diagram of polyreactivity. If you check the self-reactivity of these antibodies, they definitely recognize human IgM and human IgG. Thus, they definitely also have the rheumatic factor activity. That is why, if they're activated in the joint, they can produce rheumatoid factors and they can be even more selected to produce the rheumatoid factor reactivity after activation.

I would like to finish by thanking Michel Nussenzweig, who was my mentor and still is, and helped develop all this system, all the work has been done in his laboratory. Hedda Wardemann who worked on the first part, all the bone marrow precursors and who worked a bit on the V-preB positive B cell population. Patrick Wilson and Anne Schaefer who developed the single cell method and produced recombinant antibodies from V-preB positive B cells. Serguey Yurasov who joined lately to help Hedda with the bone marrow precursors. Eric Davis and Louis Staudt from the NIH who provide us the B cell leukaphoresis that make this work possible for the V-preB cells because we need a tremendous amount of B cells to be able to select and purify this small population of B cells. Jim Young at Sloan Kettering who provides us with all the bone marrow samples and Lionel Ivashkiv who provided us with the rheumatoid arthritis synovium and samples.

Thank you very much for your attention.

Dr. Paget: In patients with rheumatoid arthritis, where are they in this spectrum and when does it occur? Do they have this all the time, and when they flare with a clinically obvious disease through some antigen stimulation, it becomes obvious?

Dr. Meffre: So the question is: what happens in rheumatoid disease? What are the different scenarios that you can imagine? As I said, the first one for rheumatoid arthritis, where I believe V-preB positive B cells may be involved obviously. You can also imagine that in autoimmune disease, for instance, all these counter-selection processes in the bone marrow may not occur properly. You may have the threshold of the B cells signaling to delete, for instance, the self-reactive clones may be affected and you may release not only just the V-preB positive cells but new emigrant B cells that have a higher percent of self-reactivity. So you would have an accumulation of B cells with self-reactivity in these patients and down the road they may be activated and finally get to produce really self-reactive antibodies.

To really answer your question, then what happens during the flare when you have really provided lots of cytokines to activate and secrete the antibodies and then I would say in the non-active stage, I would believe that the bone marrow would still keep producing self-reacting antibodies. By deleting all the B cells, you would resolve the flare, but you would not resolve the cause because you would keep producing from the bone marrow self-reactive clones at a higher percentage than healthy individuals do and at some point then you would end up again flaring later on.

[Indistinct Question]

I believe they do. That is now what we are going to study in the laboratory. We are going to start with scleroderma and rheumatoid arthritis. We are going to try to do exactly what was done with single cell analysis at different fractions in a patient with autoimmune disease and see if there is a central defect in establishing B cell tolerance that would release constantly more self-reactive B cells in the periphery of the patient with autoimmune disease. If this is the case then it means that all the treatments are not available to treat and remove all the B cells and that seems to work very well in autoimmune disease, there would not be a cure I would say because you don't go to the source.

That is what we are trying now to see if there is really a problem in removing self-reactive clones, and then if that is the case at what stage is it, is it in the bone marrow, or is it further down as I showed you in the periphery.

I believe that the female bias for all these diseases, I would potentially agree with Polly Matzinger, is because of their cycle of menstruation because there is obviously a lot of apoptotic cells, a lot of cells dying on a regular basis. You may expose more of the B cells to self-antigen. There may be something related to that that you don't have in males. That was her idea of the danger model, which of course is controversial, but maybe there is something to do with that.

The bone marrow donors were healthy donors. Of course, we are not expecting to get bone marrow samples from an autoimmune patient because it would be really difficult. But since we can really identify new emigrants, we really have a reflection of what is going on in the bone marrow because we know that new emigrants are exactly similar to the immature B cells in the bone marrow, we have already this reflection and we know that the most ANA enriched population is in fact in the early immature B cell fraction that produces all these polyreactive antibodies that you don't find in the new emigrants. So if we found them there, we would say that the defect occurs in the bone marrow.