Synovial fluid hyaluronan (HA) profoundly buffers pressure-driven fluid loss from joints. This may be due to flow-dependent HA accumulation at a partially reflecting interface (‘concentration polarisation’). We tested a discriminating prediction of concentration polarisation theory, namely that reflected fraction ( = 1-transmitted fraction) should fall at high fluid velocities (cf. rise to a plateau if no concentration polarisation).
The knee joint cavity of anaesthetised rabbits (30 mg kg-1 pentobarbitone, 500 mg kg-1 urethane, I.V.) was infused with 0.2 mg ml-1 rooster HA (2000 kDa, radius of gyration 145 nm) and fluorescein-dextran (FD, reference solute, 20 kDa) at a constant pressure and trans-synovial filtration rate (range 6-89 µl min-1) for 3 h. Joint-derived lymph was harvested from femoral lymphatics. Samples of mixed intra-articular fluid and subsynovial fluid were aspirated at the end. Samples were analysed by gel exclusion chromatography. HA transmitted fraction was calculated from the downstream (filtrate) to upstream (bulk infusate) concentration ratio, using [FD] to adjust for the dilution of joint lymph in femoral lymph. The animal was killed humanely at the end of the experiment.
Lymph flow, lymph [HA] and subsynovial [HA] increased with trans-synovial filtration rate. Lymph [HA] & subsynovial [HA] as fractions of infusate concentration were always lower than FD fractional concentrations in the same samples. The terminal intra-articular HA concentration, mean 0.47 ± 0.02 mg ml-1 (mean ± S.E.M., n = 31), always exceeded infusate concentration (0.20 mg ml-1), confirming ultrafiltration of HA.
The reflected fraction for the cavity-to-lymph barrier (synovium + lymphatic capillary endothelium), Rlymph, was a negative function of trans-synovial flow and fell from 0.93 at the lowest filtration rate (6 µl/min) to 0.14 at 89 µl/min (P < 0.0001, regression analysis, n = 33). Reflected fractions calculated from intra-articular HA accumulation (Rasp) or subsynovial/infusate [HA] ratio (Rsyn) were almost identical negative functions of filtration rate. Reflection by lymphatic capillary endothelium (Rendo), calculated from the HA lymph/subsynovial concentration ratio, was not significantly different from zero (mean -0.03 ± 0.18, n = 21). Rlymph, Rasp and Rsyn did not differ significantly and all exceeded Rendo (P < 0.001, ANOVA). Thus synovial interstitial matrix rather than lymphatic capillary endothelium is the chief HA-reflecting membrane.
The negative relation between reflection and filtration rate supports a concentration polarisation hypothesis. Concentration polarisation leads to osmotic buffering of fluid escape, which prevents the joint from wringing itself dry during flexion.
This work was funded by Wellcome Trust grant 056983/z/99