Dear Jean,

I have not studied this but J Hill et al have published this paper which you may be interested in::

Hill JA, Southwood S, Sette A, Jevnikar AM, Bell DA, Cairns E. Cutting edge:the inversion of arginine to citrulline allows for a high-affinity peptide interaction with the rheumatoid arthritis-associated HLA-DRB1*0401 MHC class II molecule. J Immunol. 2003 Jul 15;171(2):538-41

Best wishes

Frank Wollheim

I agree with Dr Wollheim that the work of Hill et al is the first definitive treatment of citrulline binding to pocket 4 of the shared epitope alleles. In direct support of their findings, the crystal structure of various HLA-DR proteins clearly shows that amino acids 70-74 are included in pocket 4 of the binding cleft. The basic residues of the 'shared epitope' version of these residues confers a positive charge to this binding pocket. It is not surprising therefore that a positively charged arginine residue is excluded from binding in that pocket. In contrast, when arginine is converted into a polar citrulline residue by PAD enzymes it anchors quite well there. Notably, PAD expression is increased due to inflammation and polymorphisms in certain isoforms of PAD enzymes have been reported to be genetically associated with RA. This creates a scenario where altered-self epitopes which have been post-translationally modified in the periphery can elicit enhanced immune responses and circumvent tolerance mechanisms.

I would be curious to know how your group has studied this, as our findings are quite clear. In our 2010 (James et al. Arthritis Rheum. 62:2909-18), we replicated and extended the work of Hill et al, showing with binding experiments that citrulline is preferred over arginine in multiple pockets for DR1001. In this and other work (including ongoing collaborative work with Vivi Malmstrom, Lars Klareskog, and Omri Snir of the Karolinsksa Institutet) we have shown that CD4+ T cells specific for citrullinated peptides are detectable in RA patients both by HLA-DR tetramer staining and CD154 up-regulation. As such there is good published evidence that citrullinated peptides have increased binding to HLA-DR that contain the shared epitope and are recognized by T cells in RA patients.

Of course the clinical importance and disease specificity of antibodies against citrullinated protein antigens is quite clear from the literature and this serology is now a routine clinical diagnostic. It seems only logical that an antigen specific T helper response would accompany or precede these antibody responses.

Dear Frank, Dear Eddie,

The main strength of the Hill paper you are referring to is its trendy title. However, it demonstrates shared epitope binding for ONE citrullinated vimentin peptide.

At the same time, we have done a thorough study of the direct binding of 200 peptides encompassing the entire A and B chains of fibrinogen, to HLA-DRB1*0401, *0404, *0101, *0402, *0701. Whenever we met an arginine residue, we synthesized both the citrullinated and native forms of the peptide. In the end, for every allele, we observed binding of as many arginine as citrulline peptides. Indeed, it would have been easy to pick ten citrullinated shared epitope binders and ignore another ten non citrullinated shared epitope binders. It would have made easier the publication of this serious paper which finally made it into Arthritis and Rheumatism in 2005 (Auger et al, Arthritis and Rheumatism 52,3424-3432).

A recent paper from Alex Sette and Paola Migliorini (J Rheumatol 2012;39;1490-1493) also fails to demonstrate preferential binding of citrullinated viral peptide to shared epitope positive HLA-DR molecules.

At this point, there is no serious point to the flashy hypothesis that shared epitope binds citrullinated peptides. Demonstrating this point would require showing preferential binding of most citrullinated peptides from a given protein as compared to their uncitrullinated versions, to shared epitope positive alleles and not to shared epitope negative alleles.

We did not observe this with fibrinogen, and Sette and Migliorini did not observe it either with their peptides from EBV.

The "shared epitope binds citrullinated peptides" is a flashy, trendy, UNdemonstrated idea.

Fashion does not replace good science!

Why raise this question if you have already made up your mind about it? I am happy to politely disagree but to me your reasoning is faulty. It is not neccesary for all or even most citrullinated peptides to bind better to the shared epitope alleles than their unmodified counterparts (for one thing, not all peptides that bind are epitopes - in fact we find that many are not). Our results clearly show that it will certainly not be the case that all citrullinated peptides bind better than their unmodified counterparts. Did you read our paper and look at the binding data? One can easily find citrullinated pepitdes that bind the same as their un-modified counterparts and even some that bind worse (see our results for pocket 6). It all depends on the binding register of the peptide. However, it is a simple matter of physics that arginine cannot bind well in pocket 4 of alleles such as DR0401. As we have already said residues 70-74 (the shared epitope) make a significant contribution to the binding specificity of pocket 4 and hinder arginine from anchoring there. Likewise citrulline is preferred over arginine in position 7 probably because it lays down better on that shelf-like surface. Conceptually a single epitope that is preferentially bound in its modified form could be sufficient to drive disease progression by breaking tolerance. Likewise, modification at a T cell contact will have no impact at all on binding but can fundamentally alter TCR recignition and break tolerance. Please note that in our published work we have generated T cell clones that respond only to the citrullinated peptide and not the unmodified counterpart. These include clones that recognize three different epitopes. So we have established that T cells do exist that tolerate an unmodified peptide while responding to its citrullinated form. We are currently expanding these results so please stay tuned to the literature as we hope to report soon on our progress.

I have not made my mind. I just do not trust the "classical" Hill paper. I 'll read yours and get back to you.

Jean

Fair enough. I look forward to discussing this further.

I was interested in seeing the above discussion. I have two questions, which may be covered in the literature but do not seem to be resolved by the above. Firstly, I see no reason to require that citrullinated peptides should bind better than their arginine counterparts to SE to explain a link between SE and ACPA. As long as citrullinated peptides bind in a different way to SE than to e.g. DR2 or DR3 then we have a story, regardless of what the arginine versions do. The arginine versions presumably do not induce antibodies because tolerance will be more watertight for native epitope-candidates.

The other point is that if the citrulline is buried in a pocket in SE then presumably a TCR will not see it, so the epitope will not include a charge pattern for citrulline? One could imagine a 'sideways binding' option with citrulline showing to both ligands but not for a pocket?It therefore seems likely that the ACPA producing B cells are getting some sort of promiscuous of bypass help from T cells recognising citrullinated foreign material. Unless of course the citrulline makes the peptide kink in some way that means that what is shown to the TCR is a new charge pattern made up of other amino acids sticking up in a way they would not do in the native (arginine) form. Maybe one could even speculate that we do not need any preferential binding to SE, but that maybe bainding to SE is particularly likely to make the other residues stick up in an odd way. I guess that may need extensive crystal studies to be nailed?

I find the above debate very informative and relevant. However, the large amount of work of Holoshitz and co-workers seems not to be mentioned. What is the general concensus on the citrullinated calreticulin-SE binding hypothesis and activation of ROS/NOS in bystander cells?

Ling S, Liu Y, Jiaqi Fu J, Colletta A, Gilon C, Holoshitz J. Shared epitope antagonistic ligands - a new therapeutic strategy in mice with erosive arthritis. Arthritis Rheumatol. 67:2061-70, 2015.

An article from  John Sidney and Alessandro Sette  just came out in PLOS one. It settles the dispute by showing that for most peptides the citrullinated or native versions have identical affinities for most HLA-DR alleles.

So the shared epitope citrullin binding hypothesis may return to oblivion....

We have shown definitively that the effect of citrulline on HLA binding depends completely on the positioning of the residue.  Depending on the placement in pockets 1-9 of the minimal binding motif, conversion of arginine to citrulline can either inhibit or enhance binding.  Independent of HLA binging we have also shown that conversion  of arginine to citrulline can dramatically alter T cell recognition even for peptides that have reduced binding in their citrullinated form.

Is the shared epitope itself citrullinated? 

This graph is a nice analysis of the influence of citrullination of vimentin peptides on binding to HLA-DRB1*0401 and HLA-DRB1*0101

It is not really appropriate to look at the peptides in aggregate. As I have already mentioned the influence of citrullination on HLA-DR binding is completely dependent on the positioning of the residue within the binding groove. The simple minded approach of ignoring the placement of the cit/arg residue completely ignores crucial structural immunology. I would suggest that you read the study published by Scally et al in JEM

doi: 10.1084/jem.20131241