Well, you are asking very important, but very difficult question. it has been documented that blood vessels in chronic inflammatory tissues are somehow resembled to high endothelial cells (HEV) in lymph node where lymphocytes enter into lymph nodes. Blood vessels within inflamed synovial membrane would express adhesion molecule , similar to that expressed by HEV, thus, circulating lymphocytes bind to vessels of synovial membrane, then transmigrate into synovial tissues. Some of them might further migrated into tissues by cleaving ECM and finally reach to epithelium of synovial membrane, then leak to synovial fluid. It was shown that lymphocytes within inflamed synovial membrane express MMP, thus cleaving ECM and epithelium of synovial membrane. What you can do is to check the type of adhesion molecules and MMP expressed by normal synovial, inflamed synovial membranes, and/or lymphocytes isolated from inflamed synovial membrane.
T and B cells traffic in synovium like any other tissue. There is no special barrier at the surface since there is no epithelium as such. We have some evidence that B cells autodestruct if they enter the surface layer and very few B cells are found in the fluid.
Perhaps I would ask why you want to know how these cells get into the fluid since as far as I know we have no reason to think they do anything interesting in the fluid - it is a sort of dustbin or septic tank of dead and dying cells that probably do not do much of importance. Cognate immune cell interactions are very unlikely to occur there.
I have a question regarding how neutrophils cross synovial membranes, I guess these cells can accumulate in the fluid during flares of RA?
About twenty years ago I was specifically interested in these issues and wrote various papers that you might find on PubMed. Cells in synovium was what I got asked to do chapters on. The synovial intima is actually rather fascinating in its relation to white blood cells because the fibroblasts of the intima express very high levels of VCAM-1, which is the ligand for VLA-4 (a4b1 integrin if I remember right). What that ought to mean is that VLA-4 + white cells should stick in the intima but VLA-4 - ones should go through. Of course we find lots of VLA-4 + white cells in the intima - they are the intimal macrophages. Neutrophils, on the other hand, which are VLA-4 - are not found in the intima but go through to the fluid. Lymphocytes are VLA-4 + but our immunofluorescent studies indicaed that when B cells get into the intima they fragment and die. My speculation was that the could not survive there because of a lack of stromal CD21 and maybe because the stromal cells are spewing out hyaluronan. T cells seem to wander everywhere and survive but B cells are much more dependent on stromal signals for survival than T cells.
RA joint fluid is often full of neutrophils. This raises the curious question of what tactic stimulus calls these cells into the fluid if the immunological events generating the inflammation all derive from the tissue. It is maybe even more surprising because the fluid and the intimal layer are very rich in complement inhibitors, especially DAF (CD55). Our conclusion was that there is something paradoxical about the effect of DAF in relation to the ability of small immune complexes to trigger cell activation through CD16 - maybe it actually enhances the effect. Intimal macrophages have much more CD16 then any other cells in the tissue so cytokine production might be chiefly from the tissue surface (intima). This would have the effect of drawing all immigrant neutrophils to the surface of the tissue, from which they would just drop off into the fluid! Once the fluid was full of neutrophils, further interactions with complexes might create a cycle of activation that would only be stopped if the intimal macrophages phagocytosed back all the neutrophils. As far as I know nobody really understands the kinetics but it ought to be something like this.
What I always thought was most fascinating was that many of the gene products upregulated in synovial intimal fibroblasts are the same as in bone marrow nurse cells, suggesting that bone marrow 'borrowed' these molecules from synovium when it evolved in teleosts (the basic embryological stock is the same - the skeletal anlagen). This seems to apply to VCAM-1, DAF and LGP-1 and probably some others.