The biophysical mechanisms of altered hyaluronan concentration in synovial fluid after anterior cruciate ligament transection




The residence time of hyaluronan (HA) in knee joint synovial fluid (SF) was investigated using a rabbit anterior cruciate ligament transection (ACLT) model. The aims of this study were to assess, at 7 and 28 days after surgery, the 1) HA concentration and molecular mass (Mr) distribution in the SF, 2) endogenous replenishment of HA after saline washout, 3) HA residence times in the SF, and 4) synovium and subsynovium cellularity of the knee joints of rabbits subjected to ACLT, compared to sham-operated and nonoperated control joints.


Adult NZW rabbits underwent ACLT or sham surgery on one hind limb, while each contralateral limb was the nonoperated control. On day 7 or 28 after surgery, the joints were aspirated for SF, lavaged with saline, and injected with saline or polydisperse HA, and samples were obtained for analysis at set time points up to 8 hours after injection. Joint fluid samples were analyzed for the concentration and Mr distribution of HA to calculate the HA residence time constant.


Analysis of HA concentrations and Mr distributions showed 1) loss of high-Mr HA in the SF on day 7 and a shift toward a lower-Mr distribution on day 28, 2) endogenous replenishment of high-Mr HA after washout, and 3) Mr-dependent loss of HA from the knee joints after ACLT, particularly on day 7 postsurgery. The HA residence time decreased with decreasing HA Mr (residence time ∼27 hours with an Mr load of 7,000–2,500 kd, to ∼7 hours with an Mr load of 250–50 kd). HA residence time also decreased (by ∼70%) in the knee joints on day 7 after ACLT. The subsynovium of the joints subjected to ACLT displayed increased cellularity and neovascularization on days 7 and 28 postsurgery.


The residence time of HA in the SF is transiently decreased after ACLT, suggesting that a biophysical transport mechanism is responsible for the altered composition of the SF after joint injury or during inflammation.

Synovial fluid (SF) is an ultrafiltrate of plasma with additional molecules secreted by local cell populations. Hyaluronan (HA) is a major component of SF that is secreted in high–molecular mass (Mr) form (high-Mr defined as ∼4–6 mDa, ∼2–4 mg/ml) (1–4), mostly by synoviocytes (5) in the synovium, the inner lining of the joint. HA is also found on the intraarticular surface of the synovium, providing substantial resistance to fluid and macromolecular outflow (6), thus helping to maintain the concentration of HA in the SF. The protein composition of SF reflects its origins as plasma, although the distribution of these proteins in SF differs from that in plasma (7–10) due to the size-selective nature of synovium (11, 12), which allows a higher flux of smaller molecules.

The composition of SF is normally in dynamic equilibrium, through the processes of convection from plasma to SF, secretion of molecules by local cells, diffusion out of the SF to the lymphatic vessels or via cellular uptake, accumulation at surfaces, and degradation within the joint cavity. Plasma, including its small proteins and metabolites, is filtered into the SF from capillaries. Larger molecules, such as HA, are secreted locally and are removed from the SF mainly by diffusion through the synovium to the lymphatic vessels (13, 14), but also by accumulation at tissue surfaces (15, 16) and by degradation in the SF (13). Normally, a balance exists between the secretion and loss processes that maintain a steady-state concentration of high-Mr HA in the SF.

Injury and pathologic conditions alter the composition of proteins and the levels of HA in the SF, due to shifts in the rates of diffusion and transport of molecules out of the SF. The permeability of the synovium to proteins is increased in arthritic joints (17, 18), with proportionately larger increases in permeability for larger proteins (19). Protein flux between the microvasculature and SF is measured by a variety of techniques, including determination of the ratio of SF to serum (19), injection of radioactive tracers and SF sampling (20, 21), and tracer injection with clearance measurements (17, 22, 23). These measurements consistently show that protein flux is increased in the presence of synovitis or other types of joint inflammation. For HA, the steady-state concentration in the SF is typically decreased and the Mr distribution shifts toward species with a lower Mr in the joints after injury (24) and during inflammation in patients with rheumatoid arthritis (RA) (25, 26).

The residence time of HA in the affected joints might also be expected to be decreased, analogous to the effects on protein transport, although results from tracer studies, involving injection of the joints with radiolabeled high-Mr HA and analysis of the serum and SF, have been mixed. In a type II collagen injection model in sheep, the half-life of HA residence time in the SF decreased to 55% of normal (from ∼21 hours to ∼12 hours) after 2 weeks (27), while in a partial meniscectomy model of osteoarthritis (OA) in rabbits, the high-Mr HA half-life tended to increase to 135% of normal at 4 weeks, before returning to baseline by 12 weeks (28). HA residence times for the rabbit anterior cruciate ligament transection (ACLT) model of joint injury and OA (29–31) have not been reported.

The rate of diffusion of molecules out of the SF is determined by their Mr and shape, as well as by the cellularity and extracellular matrix of the synovium. Synovium is distinguished by densely populated synoviocytes 1–3 layers deep, although the cells are not laminar, do not share tight junctions, and lack a basement membrane (32, 33). Molecular size–dependent diffusive transport occurs between cells in the highly fibrillar interstitial matrix (34, 35), which has an effective pore size of ∼20–90 nm (12, 36). Direct manipulation of the matrix content by digestion of the synovial matrix (37) and by stretching of the matrix through increased intraarticular pressure (38) delineated the importance of matrix molecules in determining synovium resistance to fluid flow and molecular diffusion. Pathologic changes that can occur in the synovial matrix, due to cell infiltration or synovitis, may be responsible for the observed increases in the permeability of proteins and, possibly, the increased loss of HA.

Thus, the hypothesis of this study was that loss of HA from the SF is dependent on the Mr distribution of HA, and that this Mr-dependent loss of HA is increased in rabbit knee joints after ACLT surgery, when compared to sham-operated and nonoperated control joints. Using the rabbit ACLT model, the present study was undertaken to assess, at 7 and 28 days after surgery, the 1) HA concentration and Mr distribution in the SF, 2) endogenous replenishment of HA after saline washout, 3) HA retention over 8 hours in the SF, measured as the HA residence time, and 4) synovium and subsynovium cellularity of the knee joints of rabbits in the ACLT group compared to the nonoperated and sham-operated control groups.


Study design.

The residence time of HA in the knee joint SF was investigated using the rabbit ACLT model. All animal procedures were approved by the local Institutional Animal Care and Use Committee. Young adult (11–13-month-old) NZW rabbits (n = 26 rabbits, n = 52 knees) underwent ACLT (n = 20) or sham surgery (n = 6) on the right hind limb, while each contralateral limb was left unmanipulated as the nonoperated control. At 7 days after surgery (n = 12 ACLT, n = 6 sham-operated) or 28 days after surgery (n = 8 ACLT), the joints were aspirated for SF, lavaged with saline, injected with either saline or polydisperse HA, and flexed. Directly after flexing (0+ hours) and at 1, 3, or 8 hours after injection, 50 μl of knee joint fluid was withdrawn (information on sample numbers in each group are available from the corresponding author upon request). Rabbits were killed after the 8-hour time point, and joint fluid samples and intact joints were stored at −80°C. Joint fluid at 0+, 1, 3, and 8 hours postsurgery was analyzed for the concentration of HA (cHA) and the HA Mr distribution, which allowed us to calculate the HA residence time in the presence of various Mr loads. SF and serum samples were also obtained from the knee joints to analyze the cHA. In addition, samples of synovium were analyzed for cell density.

Surgeries and injections for HA transport studies.

ACLT or sham surgery was performed on the right knee joint. Rabbits were anesthetized and intubated, and the right hind limb was shaved and cleaned. The patella was displaced laterally and an ∼3-cm–long incision medial to the patellar ligament was made through the skin and joint capsule. The infrapatellar fat pad was displaced and the ACL was exposed. For ACLT surgery, the ACL was cut using curved, fine-tip scissors, and transection of the joint was verified by Lachman test. Thereafter, the knee joints from the ACLT and sham-operated groups were rinsed with saline and the patella was realigned. The capsule was closed with a 2-0 suture, and the skin was closed with a 4-0 suture.

Bolus injection of the knee joints and longitudinal sampling of the knee joint fluid were performed on operated and nonoperated hind limbs at 7 or 28 days after surgery. Rabbits were anesthetized and neat SF samples were withdrawn using a 22-gauge needle. Joints were lavaged 3 times with 0.5 ml saline, and then injected with either 0.5 ml of saline or 2.5 mg/ml polydisperse HA (Mr 2,500–50 kd; Lifecore Biomedical) and flexed 10 times. In addition, for tracking of high-Mr HA, 3 nonoperated joints were injected with a fluorescein isothiocyanate (FITC)–labeled HA preparation (39) that included HA in the 4,000-kd Mr range (Healon). Samples of the knee joint fluid (50 μl) were obtained with a 25-gauge needle at 0+, 1, 3, and 8 hours after injection. Blood was withdrawn for serum analysis after the 8-hour time point. Thereafter, the animals were killed, and the hind limbs were removed. The SF, joint fluid, serum, and limbs were stored at −80°C until processed.

Determination of HA concentration and Mr distribution, and analysis of HA transport.

Joint fluid was analyzed for both the cHA and the HA Mr distribution. Portions of the SF and joint fluid were digested overnight at 37°C with proteinase K, loaded at 1 μl per sample (0.3–3 μg HA) in a 1% agarose gel, and then electrophoresed at 150V in Tris–acetate–EDTA buffer. Gels were fixed in 25% isopropanol, stained overnight in Stainsall (Sigma-Aldrich), destained in water, and imaged on a light box. Hyalose monodisperse HA standards (200 ng each; Lifecore Biomedical) at Mr values of 4,000 kd, 2,400 kd, 1,156 kd, 450 kd, 262 kd, 160 kd, and 31 kd were loaded as a calibration ladder. The results of the gel distribution analysis were validated on test gels by loading 0.1 μg, 0.3 μg, 1 μg, and 3 μg of 51-kd HA, 1,680-kd HA, and 4,000-kd HA, separately and combined in equal parts. The gel distributions of HA varied by <5% per bin (mean deviation 2%) for each of the different Mr test loads, independent of the Mr. The HA distribution of the combined Mr standards was consistent with the sum of distributions in the individually loaded standards (mean deviation 6%; <15% deviation per bin).

The efficiency of proteinase K digestion was tested by staining the SF-loaded gels for the presence of proteins after digestion, using a Sypro Ruby stain (Life Technologies). The results indicated no visible protein bands (<0.1 ng/μl protein). Gels containing joint fluid samples with FITC-labeled HA were scanned on a fluorescence scanner (Storm; GE Healthcare). A custom MatLab program (Mathworks) was used to quantify the percentage of HA intensity (40) in each lane, at HA Mr loads of 7,000–2,500 kd, 2,500–1,000 kd, 1,000–500 kd, 500–250 kd, and 250–50 kd. The cHA in each joint fluid and serum sample was determined by digesting portions of the samples with proteinase K and inhibiting enzyme activity by boiling for 10 minutes, and then using an enzyme-linked immunosorbent–like assay (Corgenix or R&D Systems) to detect HA. The cHA was multiplied by the percentage of HA intensity, which yielded the total cHA per Mr load.

The residence time constants for HA by Mr load were then calculated. The endogenous secretion of HA was approximated by determining the mean cHA measured in the joint samples after saline injection, the majority of which was localized in the 7,000–2,500-kd Mr bin. This endogenous secretion concentration was subtracted from the total cHA determined after injection of polydisperse HA at each time point. A 2-parameter exponential decay was fit through the remaining cHA at 0+, 1, 3, and 8 hours postinjection for all bins, except the bin containing 7,000–2,500-kd HA (due to the high rate of HA secretion when compared to the injected cHA in this bin), which yielded the best-fit HA residence time constant for each group.

Histologic analysis of the synovium.

The matrix structure and cellularity of the synovium and subsynovium were determined. Limbs were thawed at 4°C, and samples of synovium were harvested from the medial and lateral distal patellar regions. Samples were fixed overnight in 4% paraformaldehyde, and were then either dehydrated and embedded in paraffin or embedded in OCT medium and sectioned at a thickness of either 10 μm or 100 μm, respectively. Thin sections were stained with hematoxylin and eosin or Alcian blue for histologic analysis of the tissue, or were probed for HA (with HA binding protein–horseradish peroxidase; Corgenix), CD4+ lymphocytes (with ab25804; Abcam), or CD11b+ macrophages (with ab8878; Abcam), and imaged at 20× or 40× magnification on an inverted microscope (Nikon Instruments). Thick sections were stained with 2 μg/ml propidium iodide in saline and imaged on a confocal microscope (Leica Microsystems) with a 20× objective, 0.75-μm–square voxel size, and 1,024 × 1,024 pixels.

Statistical analysis.

Data are presented as the mean ± SEM. The fixed effects of surgery (for the nonoperated, ACLT, or sham-operated groups) and time (day 7 or day 28 postsurgery) and the repeated effects of time (1, 3, or 8 hours postinjection) and Mr (by bin) on the cHA after injection were assessed by four-way analysis of variance (ANOVA). The fixed effects of group (nonoperated, ACLT day 7, sham-operated day 7, or ACLT day 28) and repeated effects of Mr on the steady-state neat SF cHA were assessed by two-way ANOVA with Dunnett's post hoc tests, to determine differences in comparison with the nonoperated group within each Mr bin. The fixed effects of group-on-time constant and serum cHA were assessed by one-way ANOVA with Dunnett's post hoc tests, to determine differences in comparison with the nonoperated group. Significance was set at a P value less than 0.05, and statistical analyses were performed using Systat software, version 10.2.


Concentration and Mr distribution of HA in the SF after surgery.

The HA Mr distribution in the SF shifted toward a predominance of lower-Mr HA after surgery. The cHA in the SF varied with the shift in Mr (P < 0.0001) and also varied by surgical group (P < 0.0001), with an interaction of these variables (P < 0.0001) (Figure 1). The total cHA, which was ∼2.5 mg/ml in nonoperated joints, was decreased by 47% in the ACLT group and by 49% in the sham-operated group on day 7 after surgery. On day 28 after ACLT, the total cHA was decreased by 26%. Both the ACLT and sham-operated groups contained decreased levels of high-Mr (7,000–2,500 kd) HA on day 7 (P < 0.001 versus nonoperated joints) (Figures 1A and B). By day 28, the joints subjected to ACLT still displayed less high-Mr HA (P < 0.01), but also contained more lower-Mr HA including significantly more HA in the 1,000–50-kd Mr range (P < 0.01–0.05), than was observed in the nonoperated joints (Figures 1C–F).

Figure 1.

Distribution of hyaluronan (HA) molecular mass (Mr) in the synovial fluid (SF) of rabbit knee joints after anterior cruciate ligament transection (ACLT) surgery, as compared to sham-operated or nonoperated (NonOp) control knee joints. A, Gel distribution analysis shows an altered HA Mr distribution in the SF of ACLT- and sham-operated joints compared to nonoperated joints. B–F, Quantification of HA in the joint fluid samples shows that high-Mr (7,000–2,500-kd) HA (B) is decreased in the joints on day 7 after ACLT or sham surgery, when compared to nonoperated joints, and by day 28, there is also an increase in lower-Mr HA, in ranges of 2,500–1,000 kd (C), 1,000–500 kd (D), 500–250 kd (E), and 250–50 kd (F). Bars show the mean ± SEM. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001, versus nonoperated joints.

HA residence times in the SF.

After washout and injection of the joints with lower-Mr HA, evaluation of the cHA and Mr distribution of HA in the SF showed endogenous replenishment of high-Mr HA, while lower-Mr HA was lost from the SF, in particular on day 7 after ACLT (Figure 2, lane a). Rabbit SF normally contains mainly high-Mr HA (>4,000 kd).

After washout of the joints, high-Mr HA was indeed replenished over time in both the nonoperated and ACLT-operated joints (Figure 2, lanes b–g). After washout, the joints were injected with HA at Mr loads that were lower than what is typically found in the endogenous secretion profile, and the joint fluid was resampled at 0+ hours postinjection. The results immediately after injection showed a moderate dilution effect on the cHA in both the nonoperated and ACLT-operated joints, although the distribution of HA was similar to the injection profile (Figure 2, lanes h–j). Determination of the distributions of HA in the joint fluid at 1, 3, and 8 hours after injection showed an Mr-dependent loss of HA over time, with exacerbated loss occurring in the ACLT-operated joints by 7 days after surgery (Figure 2, lanes k–v).

Figure 2.

Quantification of the HA concentration over time in samples of SF and whole joint fluid in the presence of various HA Mr loads. On day 7 after ACLT, rabbit knee joints were injected with HA at different Mr loads, and SF samples were obtained for analysis at 0+ hours postinjection (lane a). Joint fluid samples were obtained from nonoperated and ACLT-operated rabbit knees on day 7 or day 28 after surgery, and the HA concentration was analyzed at 0+, 1, 3, and 8 hours postinjection (time 0− represents baseline [time of injection]). The joints were subjected to saline lavage (lanes b–g) or HA injection at the activated Mr loads (lanes h–v). See Figure 1 for definitions.

The breakdown of cHA in each sample according to Mr load was determined to allow quantitative comparisons between surgical groups and time after surgery. After washout and injection of the joints with different Mr loads (Figure 3, panels A.i–A.v), the distributions of HA directly after injection (at 0+ hours) were similar between the nonoperated and ACLT-operated joints (P = 0.82) (Figure 3, panels B.i–B.v), with the cHA being diluted, on average, to ∼60% of the injection concentration. The total cHA in each Mr bin was analyzed separately as the endogenous concentration of HA (white-shaded bars in Figures 3C and D) and the injected concentration (black-shaded bars in Figures 3C and D). The endogenous concentrations of HA were mainly found in the 7,000–2,500-kd bin (Figures 3C and D, panels C.i and D.i). The injected concentrations of HA varied by surgical group (P < 0.0001), day postsurgery (P < 0.05), Mr load (P < 0.0001), and time postinjection (P < 0.0001), with interactions of Mr–surgery (P < 0.01), time–day (P < 0.0001), and Mr–time (P < 0.01).

Figure 3.

A–D, HA concentrations in whole joint fluid samples over time after injection of different Mr loads of HA (panels i–v), at the time of injection (A), at 0+ hours postinjection on day 7 after surgery (B), and at 1, 3, and 8 hours postinjection on day 7 (C) or day 28 (D) after surgery in nonoperated (N), ACLT (A)–operated, or sham-operated knees. White-shaded bars indicate endogenous replenished levels of HA, and black-shaded bars indicate the concentration of injected HA. Dotted horizontal lines indicate the levels of HA normally found in rabbit synovial fluid. Bars show the mean ± SEM. Significant effects on the HA concentration were seen according to surgical group (P < 0.0001), day postsurgery (P < 0.05), Mr (P < 0.0001), and time postinjection (P < 0.0001). See Figure 1 for other definitions.

On day 7 after surgery, the cHA in the 2,500–1,000-kd bin had dropped to 75% of the injection concentration in the nonoperated group, compared to drops of 37% in the ACLT group and 57% in the sham-operated group, by 8 hours postinjection (Figure 3, panel C.ii), while in the 250–50-kd bin, the decline in cHA by 8 hours was 48% in the nonoperated group compared to 24% in the ACLT group and 45% in the sham-operated group (Figure 3, panel C.v). However, on day 28, the differences between the nonoperated group and the ACLT group were smaller, in that the cHA at 8 hours postinjection in the 2,500–1,000-kd bin had dropped to 85% of the injection concentration in the nonoperated joints compared to a drop of 68% in the ACLT-operated joints (Figure 3, panel D.ii), and in the 250–50-kd bin, this had dropped to 55% in the nonoperated group and 43% in the ACLT group (Figure 3, panel D.v).

The HA residence time not only decreased with lower Mr, but also decreased after surgery. The increased loss of HA, plotted as a percentage of the injection concentration over sampling times, was apparent in the ACLT group when compared to the sham-operated and nonoperated groups on day 7 after surgery (Figure 4, panels A.i–A.iv), whereas by day 28, the loss of HA was similar between the ACLT and nonoperated groups (Figure 4, panels B.i–B.iv). In the nonoperated joints, the residence time constants by Mr bin, which describe the slope of the HA loss, ranged from ∼27 hours in the 2,500–1,000-kd bin to ∼7 hours in the 250–50-kd bin (Figure 4, panels C.i–C.iv). After ACLT, the residence time constants decreased (by 64–75%) in all Mr bins on day 7. In the sham-operated joints, the residence time constants in the 2,500–1,000-kd bin also decreased (by 41%) compared to that in the nonoperated joints (Figure 4, panel C.i). These results indicate that ACLT has a major effect on the loss of HA, while the surgical procedure itself has only a minor effect. By day 28, the HA residence time constants in all bins were similar between the ACLT group and nonoperated controls (P = 0.39–0.50) (Figure 4).

Figure 4.

Loss of HA over time in the rabbit knee joint fluid after injection of different Mr loads of HA (panels i–iv) on day 7 (A) and day 28 (B) after ACLT surgery, and calculated HA residence time constants in the nonoperated, ACLT-operated, and sham-operated joints (C). The HA residence time increased in bins with higher-Mr HA, and decreased on day 7 post-ACLT, although recovery from the loss of HA was seen by day 28. The results in sham-operated joints were similar to those in nonoperated joints after injection of lower-Mr HA, and between ACLT-operated and nonoperated joints after injection of higher-Mr HA. Bars show the mean ± SEM. ∗ = P < 0.05; ∗∗ = P < 0.01. See Figure 1 for definitions.

To confirm that the rate of HA loss from the joint fluid was dependent on the Mr, we performed tracking experiments using FITC-labeled HA. The results showed that FITC-labeled high-Mr HA was cleared from the joint fluid slower than midrange-Mr HA, while low-Mr HA was quickly lost (results available from the corresponding author upon request).

Synovium and subsynovium matrix structure and cell density.

Microscopy images showed that the structure of the synovium was altered and the cell density of lymphocytes and macrophages was increased in the subsynovium of the knee joints after surgery. Compared to synovial tissue from nonoperated joints on day 7 (Figure 5, panels A.i–, thickening of the synovium and loss of matrix content (decreased matrix staining) occurred in the joint tissue on day 7 after ACLT (Figure 5, panels B.i– By day 28 after ACLT (Figure 5, panels C.i–, the intensity of matrix staining had returned, although the synovium lining was still thickened.

Figure 5.

Microscopy images of the synovium and subsynovium from nonoperated rabbit knee joints (A) and ACLT-operated rabbit knee joints on day 7 (B) and day 28 (C) postsurgery. Joint tissue was assessed histologically by hematoxylin and eosin (H&E) staining (panel i) or Alcian blue staining (panel ii) and by probing for HA (panel iii), CD4+ lymphocytes (panel v; higher-magnification [mag.] views in panels v.a and v.b), or CD11b+ macrophages (panel vi; higher-magnification views in panels vi.a and vi.b). Thick sections were stained with propidium iodide (panel iv). The images show increased synovium thickness and decreased matrix content and cellular infiltration by lymphocytes and macrophages after ACLT surgery, as well as increased cell density and neovascularization postsurgery. See Figure 1 for other definitions.

The subsynovium of ACLT-operated joints displayed substantial cellular density and evidence of neovascularization, as well as positive staining for lymphocytes and macrophages, on days 7 and 28 postsurgery (Figure 5). Although the increased rates of HA transport out of the joints after ACLT appeared to be resolved by day 28, the increased cell density of the subsynovium was still apparent. However, staining for CD4+ and CD11b+ cell expression markers (Figure 5, panels C.v and was no longer evident by day 28 after ACLT.

Changes in serum HA concentration.

The cHA in the blood serum of rabbits was also transiently increased after surgery. Treatment (nonoperated, ACLT day 7, sham surgery day 7, or ACLT day 28) significantly affected the serum HA content (P < 0.05) (results available from the corresponding author upon request). The serum HA concentration almost doubled, from ∼21 ng/ml to ∼41 ng/ml, on day 7 in both the ACLT and sham-operated groups, before returning to ∼22 ng/ml on day 28.


These results demonstrate that the residence time of HA in the SF is transiently decreased after ACLT, suggesting that there is a biophysical mechanism for the alterations in the HA Mr distribution in the SF, and in the SF composition in general, after injury or during inflammation. The HA Mr distribution shifted toward a predominance of lower-Mr HA after surgery (Figure 1). Direct sampling of the SF over time, after washout and injection, indicated an accumulation of high-Mr species, while lower-Mr HA diffused out of the SF, especially on day 7 after ACLT (Figure 2). Quantification of HA in these gels confirmed the effects of HA Mr, surgery, day after surgery, and time after injection on loss of HA from the SF (Figure 3).

The time constants for HA residence in the SF by Mr are consistent with previously reported values, and demonstrate both an effect of the surgical injury and a more intense effect of ACLT itself in decreasing HA residence times (Figures 3 and 4). The decreased density and loss of matrix associated with synovial thickening and subsynovial cell infiltration after surgery (Figure 5) likely decreases the resistance to HA loss through the synovium to the lymphatic vessels, while the systemic turnover of HA, as measured via plasma concentrations, an additional measure of inflammation, was also increased on day 7 (results not shown). Taken together, these data suggest that increased HA transport due to an inflammatory state may be the biophysical mechanism underlying the loss of high-Mr HA from the SF after injury and during inflammatory conditions, although other inflammatory changes, such as increased lymph node drainage, local degradation, and biosynthetic alterations, may also contribute.

The in vivo data presented herein provide a description of the transport of HA from the SF that was logically consistent among the multiple mechanisms investigated, despite inherent limitations in this type of animal study. The numbers of animals were limited, and neat SF samples were obtained repeatedly from individual knees. Although 26 rabbits and 52 knees were utilized, they were divided among 13 groups in order to investigate the effects of surgery and time after surgery on HA transport. Despite limited sample numbers, sufficient power was achieved to show significant effects and confirm the hypothesis. In addition, endogenous noise in the high-Mr bins confounded the transport measurements, although FITC labeling of HA confirmed that transport trends were similar, as measured quantitatively, in bins with lower-Mr loads (results not shown). The rate of FITC-labeled HA loss from the joint is difficult to quantify and compare to the rate of concentration-based HA loss, because of the confounding effects of loss of label, quenching of the fluorescence in vivo, and unknown conversion between scanned fluorescence intensity and concentration. Nevertheless, qualitative trends confirming an Mr-dependent efflux of HA from the SF were observed.

Repeated sampling of the SF required multiple needle sticks through the joint capsule, which could have increased the loss of HA from the SF. However, the needle sampling did not appear to markedly affect HA loss, since large holes in the synovium would have negated the Mr dependence, which was not the case.

Finally, homeostasis reflects loss through degradation, transport, and cellular phagocytosis. Although degradation and cellular uptake were not measured directly, the uniformity and concentration of high-Mr HA that accumulated after saline washout suggest that the time scale for degradation or uptake was longer than that for diffusion, implying that our results were dominated by the effects of diffusion.

By day 28 after ACLT, the rates of HA loss were similar to those in nonoperated joints, although the steady-state cHA in aspirated SF showed increasing concentrations of HA in the low-Mr ranges. Taken together, these findings suggest that there is an additional mechanism operating to increase the low-Mr cHA. Two possibilities for this mechanism are the degradation of high-Mr HA due to increased activity of hyaluronidase or reactive oxygen species after surgery, or increased synthesis of smaller HA molecules by synoviocytes. After aspiration and saline injection, replenishment of HA did not result in higher concentrations of low-Mr HA (Figure 3, white-shaded bars), suggesting that degradation is primarily responsible for the shift.

Although endogenous replenishment appeared similar between saline- and HA-injected joints within the same experimental group, analysis at later time points after injection would be needed to determine the effect of saline injection as compared with HA injection on replenishment. As indicated by the dotted lines in Figure 3, panels C.i–C.v, after 8 hours, only ∼26–40% of the steady-state concentration had returned, as the time constant for replenishment is ∼14 hours. Sampling at later points in time, after 2–3 time constants, would be necessary to determine whether the rates were truly different.

The results are consistent with the concept of a biophysical mechanism being responsible for the shift to lower-Mr HA in the SF after injury and during inflammatory states (Figure 6). Normally, the Mr-dependent residence time of HA in the SF, accompanied by the secretion of high-Mr HA, leads to a steady state, with SF containing mostly high-Mr HA (Figure 1); low-Mr HA, created by HA turnover, diffuses relatively quickly out of the SF. After injury or during inflammation, the synovium becomes more permeable to HA (Figures 2–4), likely due to the fact that cellular infiltration displaces and digests the extracellular matrix barrier to molecular efflux through the synovium, thus decreasing the residence time of high-Mr HA. Faster diffusive efflux of high-Mr HA, coupled with decreased secretion rates and possibly increased degradation rates, would shift the steady-state Mr distribution to lower-Mr HA, as was observed in our experiments. Although the cell infiltration was still present on day 28 after surgery, the synovial matrix production between days 7 and 28 appeared to have replenished the extracellular matrix barrier (Figure 5), with a coincident restoration of residence times. In addition to major inflammatory events such as injury or during RA, this mechanism may also be responsible for altered SF composition in OA, where an inflammatory component is increasingly recognized (41).

Figure 6.

Schematic representation of the biophysical processes, including secretion, degradation, and loss, contributing to the altered Mr distribution of HA in pathologic synovial fluid. CAP = capillaries; sSYN = subsynovium; LYM = lymphatic vessels; CART = cartilage (see Figure 1 for other definitions). Color figure can be viewed in the online issue, which is available at journal/10.1002/(ISSN)1529-0131.

The cHA variable includes the effects of joint fluid volume, which is typically altered after injury and during inflammation. Since it is difficult to accurately measure total SF volume, the aspirated volume of neat SF is a reasonable approximation. Larger volumes of neat fluid were aspirated from the knee after surgery, increasing from a mean ± SD ∼20 ± 13 μl in the nonoperated joints, to 290 ± 120 μl on day 7 and 540 ± 335 μl on day 28 after ACLT. However, the transport study was conducted after fluid aspiration and injection of the joints with 500 μl in each group, which normalized the volume at time zero. Dilution effects are unlikely to have confounded the interpretation of the transport data, as the volumes were highest on day 28, when the transport rates and concentrations were similar to those in the nonoperated group. Although the cHA and Mr distribution shifted to lower-Mr species by day 28 after ACLT, the total mass of HA of all sizes was increased, due to the increase in fluid volume.

Understanding the mechanisms leading to altered HA composition in the SF, such as altered transport rates due to inflammation, is important, since the function of SF is determined by composition. Changes to the HA concentration and Mr in the SF alter the biophysical properties of the fluid, including viscosity and viscoelastic properties (42, 43), as well as the lubrication function (44). SF effusion is characteristic of inflammatory conditions, and increased plasma filtration rates into the SF, which depend on SF hydrostatic pressure (45) and joint capsule strains (46) that vary with flexion, likely contribute to decreases in the concentration of HA. In addition to HA, lubricant molecules secreted locally into the SF, such as lubricin, likely also have decreased residence times in the SF. A decreased barrier to diffusion out of the SF also indicates decreased resistance to convective and diffusive molecular flux from plasma into the SF. This may explain the increased concentration of higher-Mr plasma proteins, including from the inter-α trypsin inhibitor family, that are found in pathologic SF (47, 48).

HA in the SF is modified with various HA binding proteins (HABPs), especially during inflammation, which likely affects the residence time of HA in vivo. To characterize HA transport independent of the effects of HABPs, the samples in the present study were incubated with proteinase K to digest protein. The calculated residence time rates may be an underestimate of the residence time in vivo, due to the decrease in Mr after protein removal. The HA distributions were similar overall, with or without proteinase K digestion, although an additional light, high-Mr signal in the area between the well and main hyperintense band at ∼4,000 kd was apparent before digestion (results not shown). These high-Mr species are probably serum-derived HA-associated protein–HA aggregates (47, 48), with modification by inter-α trypsin inhibitor heavy chains, which likely develop over time on exogenously injected HA as well.

The transport mechanisms suggested by these results indicate potential avenues, and pitfalls, for therapeutic interventions. The Mr-dependent HA residence times in the SF in nonoperated joints and joints after ACLT provide an estimate of the residence time for other proteoglycans and proteins in the SF, and highlight the difficulty in attaining localized retention of injected pharmaceutical agents, especially during the inflammatory response phases following injury. The inflammatory state of the synovial lining may be a potential target for intervention to restore the normal homeostasis of the SF. Early clinical intervention that decreases the synovial inflammatory response and restores lubrication following injury (49, 50) may be able to limit or retard the progression from injury to posttraumatic OA.

The present study investigated the state of transport of HA in the joint at various times after ACLT, demonstrating the Mr-dependent rate of HA loss, which could be attributed to inflammatory cell infiltration into the synovial and subsynovial joint capsule lining. Such altered transport rates help explain the altered HA content and general compositional changes in the SF after injury and during inflammatory states. In addition, our findings characterize the molecular conditions under which any pharmaceutical intervention must function, as well as suggest potential targets for clinical intervention.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Sah had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. McCarty, Firestein, Masuda, Sah.

Acquisition of data. McCarty, Cheng, Hansen, Yamaguchi, Masuda, Sah.

Analysis and interpretation of data. McCarty, Cheng, Firestein, Masuda, Sah.