- Email: [email protected]
Determination of kinetic changes of aggrecan-hyaluronan interactions in solution from its rheological properties
J. Biomechmics, Vol. 27, No. 5, pp. 571-579,
Pergrmon
1994
Elsevier Science Ltd Printed in Great Britain. All rights -cd OOU-9290/W
Sa.CdI+.a,
DETERMINATION OF KINETIC CHANGES OF AGGRECAN-HYALURONAN INTERACTIONS IN SOLUTION FROM ITS RHEOLOGICAL PROPERTIES WENBO ZHU,* VAN C. Mow,~ LAWRENCEC. ROSENBERGS and LIH-HENG TANGS *Departmentof Mechanical Engineering, University of Maryland, Baltimore, Maryland, U.S.A.; tDepartment of Mechanical Engineering and Orthopedic Surgery, Columbia University, New York, U.S.A.; SOrthope& Research Laboratories, Montefiore Hospital Medical Center, Bronx, New York, U.S.A. Abstract-The kinetics of interactions between aggregating cartilage proteoglycan (aggrecan) and hyaluronan was examined through their rheological flow behavior using a cone on plate viscometer. The mixing of the two types of molecules was carried out directly on the plate of the viscometer, and aggregation process was monitored through the changes of the sample’s steady-shear viscosity and/or dynamic shear modulus as a function of time. The effect of flow conditions on the aggrecan-hyaluronan interaction rates was examined by subjecting samples to steady-shearing motions at specified shear rates, and to oscillatory shear motions of specified frequencies and amplitudes. The characteristics of the kinetics of interaction between aggmcan and hyaluronan molecules depended not only on the flow conditions under which proteoglycan aggregation took place, but also on the concentration of the components in the solution. At high shear rates (> 10 s- ‘), viscosity of the mixture solution increased monotonically, starting near the viscosity of the aggrecan solution, and reaching the viscosity of the aggregate solution in approximately 35 min. Surprisingly, under slow shearing motions (< 10 s-t), the viscosities of the mixture solutions exceeded those of control aggregate solutions at identical hyaluronan : aggrecan ratios and concentrations. In addition, the aggregation under oscillatory motions took place near physiologic Bequency (10 rads- r) although the rate of aggregation process was much slower than under steady-shearing motion (> 100 mm). However, the highfrequency oscillatory shearing (62.8 rads-t) tended to impede aggregation resulting in a reduction of dynamic modulus over time. The influence of loading conditions on the rate of aggregation and aggregam size observed in this study seems to suggest a close relationship between proteoglycan structure and content, and degree of physiological stress throughout the joint and frequency of the joint motion. INTRODUCMON
Proteoglycan aggregates, consisting of aggrecan, link proteins and hyaluronan, form a significant part of the cartilage extracellular matrix. Aggrecan consists of a protein core to which a large number of keratan sulfate and chondroitin sulfate chains are covalently attached. In cartilaginous tissues, most aggrecans exist as aggregates by binding to a linear hyaluronan chain. This binding is stabilized by link proteins. The significance of aggregate formation is to restrict the movement of proteoglycans within the collagen fiber network, and thereby to remain proteoglycans at a high concentration in the tissue (Muir, 1983; Pottenger et al., 1982). In cartilage matrix, the components of aggrecan molecules, hyaluronic acid and link proteins arc manufactured by chondrocytes. These components are secreted independently from the cells, and then assembled as aggregates within the extracellular matrix (Bay&s et aI., 1983; Kimura et al., 1979,198O; Lohmander and Kimura, 1986 Prehm, 1983). Aggregation is an important event in the development of a stable matrix, and if perturbed, could result in a less-
Received in jnal form 25 June 1993. Author to whom correspondence should be addDr Wenbo Zhu, Department of Mechanical Engineering, University of Maryland Baltimore County Campus, Baltimore, MD 21228, U.S.A.
ordered structure with inferior material properties (Hardingham et al., 1987; Mow et al., 1989; Zhu and Mow, 1990). The size and content of proteoglycan aggregates depend on the quality and distribution of the proteoglycan components in the tissue (Manicourt et al., 1991). Many researchers have demonstrated sign&ant changes of proteoglycan structure and content in articular cartilage with age and degeneration as well as the degree of physiological stress throughout the joint. For example, studies of the topographical proteoglycan distribution have shown that the proteoglycan content was higher in the articulating, high weightbearing regions than in the relatively low weightbrering areas of the joint (Sweet et al., 1977). Analysis of proteoglycan aggregates nondissociatively extracted from cartilage showed enhanced sire of proteoglycan aggregates in high weight-bearing area (Manicourt et al., 1991). A decrease in proteoglycan aggregation and an increase in proteoglycan turnover were found to be the two main biochemical characteristics of osteoarthritic and aging cartilages (Carney et al., 1984, McDevitt and Muir, 1976; Muller et al., 1989). A progressive and sign&ant decrease in hyaluranic acid content has been detected in the early stage of experimental osteoarthritis models (Manicourt and Pita, 1988; McDevitt and Muir, 1976). This decrease may limit the amount of proteoglycans to be involved in aggregation. 571
512
W. ZHu et al.
While the ultrastructure and content of proteoglyparable to those found in native cartilage can molecules throughout the joint and changes in (lo-50 mg ml- ‘), and the effect of varying concentraproteoglycan structure during aging and degeneration tions on the kinetic interactions were also examined. have been documented, little is known about the aggregation processes of matrix components MATERIALS AND METHODS occurring within the extracellular matrix and the mechanisms controlling the proteoglycan extracelluProteoglycan preparations lar assembly. It is also not known how external Aggrecan molecules were prepared from fresh loading or deformation may influence the rate of bovine articular cartilage as previously described assembly of aggrecan with hyaluronan. Recent studies (Rosenberg et al., 1973). Briefly, cartilages were exof kinetic aggregation and the stability of proteoglytracted in ice-cold 4 M guanidine HCl with 0.15 M can aggregates showed an age-related decrease in the sodium acetate at pH 6.3 containing proteinase inrate of extracellular assembly of proteoglycan aggreghibitors. After slow stirring at 5°C for 24 h, the extracts ates (Bayliss and Davidson, 1991). Most of the experiwere filtered, and dialyzed against 20 volumes of mental studies on aggregation have exclusively been 0.15 M sodium acetate, 0.05 M EDTA, pH 6.3, conperformed in vitro under rather dilute conditions taining protease inhibitors at 5°C for 16 h. The extract (Kimura et al., 1980), while the in situ concentration of was then fractionated by equilibrium density gradient the aggregate components in cartilage are very high. centrifugation under associative conditions in 3.5 M These high concentrations will undoubtedly influence CsCl at 5°C for 60 h at 40,000 rpm. The fraction the kinetics of the interactions involved in aggregate AlAl was taken up in 5.5 M guanidine HCl, 0.15 M formation. sodium acetate, pH 6.3, and was stirred overnight at Recent biorheological studies of concentrated pro5°C. Equilibrium density gradient centrifugation teoglycan solutions have provided quantitative inunder dissociative conditions was then carried out at formation on the effects of various proteoglycan struc5°C for 60 h in 3 M CsCl, 4 M guanidine HCl at tural changes on the mechanical properties of pro40,008 rpm. The dissociative gradient was divided into teoglycan solutions. There is considerable evidence to six equal fractions called Dl-D6. Proteoglycan fracshow that the proportion of proteoglycans aggregtion AlAlDlDl was dialyzed against 0.15 M sodium ated, the size of aggregates and their stabilization by acetate at pH 6.3 and precipitated with 3 volumes of link protein are important factors in determining the ethanol, washed three times with ethanol and ether elastic and viscous properties of the molecular netand dried in a vacuum oven. works they formed in solution (Hardingham et al., Three types of solutions were prepared: (I) aggrecan 1987; Mow et al., 1989; Zhu et al., 1991). For example, solutions, (2) aggregate solutions, and (3) hyaluronan the molecular networks formed by proteoglycan agsolutions. The aggrecan solutions were prepared from gregates were shown to have higher dynamic shear the AlAlDlDl fraction by re-equilibrium in PBS of stiffness and greater resistance to shear flow than those 0.15 M NaCl at pH 7. The proteoglycan aggregate formed by the proteoglycan monomers. For aggregsolutions were prepared by mixing aggrecan and ates formed without link protein, shear stiffness was hyaluronan components in PBS of 0.15 M NaCl at reduced by approximately 20% as compared to agpH 7. A weight ratio of hyaluronan:aggrecan at gregates with stabilizing link protein. The strength of 0.008: 1 was used for optimal proteoglycan aggregaintermolecular interactions for the link-free aggregtion (Hardingham and Muir, 1972). Thorough mixing ates was also significantly decreased (Mow et al., 1989; of aggrecan and hyaluronan molecules was carried out Zhu et al., 1991). by continuously rotating the sample on a mechanical The objectives of this study are (1) to examine the roller at 5°C for 48 h to assure a complete formation of effect of proteoglycan aggregation on the flow properaggregates. The solutions of aggrecan and aggregate ties of proteoglycan solutions, (2) to examine the were prepared at lo,30 and 50 mg ml - ‘, respectively. kinetics of aggrecan-hyaluronan aggregation, (3) to Both aggrecan and aggregate preparations served as determine the factors such as flow conditions and control solutions to provide reference values for flow solution concentrations that might influence the rate properties measurements. The hyaluronan preparof aggregation. These were achieved by examining the ation was obtained from rooster comb with an averdirect interactions of aggrecan and hyaluronan moleage molecular weight of 10’ D. The pure hyaluronan cules in vitro through their rheological flow properties. solutions were prepared at concentrations ranging The kinetic behavior of aggrecan-hyaluronan solufrom 2.4 to 20 mg ml- ‘. Additional aggrecan solutions tions was compared, respectively, to the properties of and hyaluronan solutions were prepared to be mixed corresponding control solutions of the aggrecan on the plate of the viscometer just prior to the flow alone and aggregates formed at identical measurements. hyaluronan : aggrecan ratios and concentrations. The effect of loading conditions on the rate of Measurements of control solutions aggrecan-hyaluronan interaction was observed under Rheological flow measurements of concentrated dynamic as well as steady-shear conditions at various proteoglycan solutions were made with a low-frefrequencies, amplitudes and shear rates. The studies quency mechanical spectrometer (Rheometrics model were carried out at proteoglycan concentrations com-
Kinetic changes of aggrecan-hyaluronan
RMS-800) using a cone-on-plate viscometer (cone diameter = 50 mm and apex angle = 0.04 rad). The flat plate of the viscometer is driven by a precision d.c. motor. The stationary cone is attached to a sensitive biaxial transducer that measures the torque and normal force required to shear the solution. The viscometer is enclosed in an environmental chamber which maintains a constant temperature of 20°C and 100% humidity. The apparatus is controlled by a microcomputer for testing, data acquisition and analysis. In this study, the following flow properties of proteoglycan solutions were measured: (1) the magnitude of dynamic shear modulus lG*l and tangent of the phase shift angle, tans, from an oscillatory shear experiment; (2) the shear-rate dependent apparent viscosity qaPPand primary normal stress difference u1 from a steady-state shear experiment; and (3) transient stress growth function r(t) from a step shear rate experiment. Flow measurements were first performed on the control solutions of aggrecan, pre-formed aggregate and pure hyaluronan molecules. The magnitude of dynamic shear modulus of the control solutions was measured, respectively, at the frequencies of 1, 10,100 rad s-l with an applied strain amplitude of 50%, followed by a frequency sweep ranging from 1 to 100 rad s- ‘. Subsequently, the apparent viscosities of the solutions were measured at the shear rates of 1,10 and 100 s-l, respectively. A shear-rate sweep experiment was then performed for four decades of shear rate (0.25250 s- ‘). Finally, the transient stress growth experiments were performed at the step shear rates of 25 and 100 s- ‘. The values of apparent viscosity and dynamic shear modulus of the homogeneous aggrecan solutions define the initial flow properties, whereas those of preformed aggregate solutions define the flow properties of the equilibrium state at which the aggregation process was completed. The steady-state and transient properties of the pure hyaluronan solution were also determined for comparison. Two complete sets of measurements were ma& for each control proteoglycan solution. Since the test samples were obtained from the same preparation, the maximum difference between the two measurements was found to be less than 5%. Kinetics of aggregate formation
An experimental procedure was developed to study the kinetic interaction of aggrecan and hyaluronan molecules under a specific mechanical loading condition. In order to follow the initial kinetics and the rate of aggregation process the aggrecan and hyaluronan solutions were placed directly onto the plate of the viscometer using a pipette. Each solution volume was precisely measured ensuring a hyaluronan: aggrecan ratio of 0.8% upon mixing for optimal aggregation. Upon placing the samples, the upper cone was lowered by the stepper motor until the 50 cun gap between the cone and plate was reached. It took approximately 2 min to load the sample. The entire test section was
interactions
513
then enclosed in a temperature and humiditycontrolled chamber. Two types of tests were performed on separate but identical mixtures: (1) a steady-shear test at a constant shear rate, and (2) a dynamic-shear test at a constant frequency and amplitude. In the steady-shear test, a constant shear rate was initiated immediately after the loading of the solution. The aggregate formation as a function of time was observed through the viscosity measurements of the mixture solution. The viscosity measurement began immediately upon shearing, and the measurements were taken every 2 min during continuous shearing until an equilibrium value was reached. To examine the effect of shear rate on the proteoglycan aggregation process, separate kinetic experiments were performed at three different shear rates (1,lO and 100 s- ‘). A similar procedure was adapted for kinetic experiments under oscillatory shearing motion. The kinetics of aggregate formation was monitored by the change of the dynamic shear modulus of the sample. It was necessary to pre-mix the hyaluronan and aggrecan solutions before the dynamic shearing motion was applied to the sample. Upon loading, the sample was first subjected to a constant shearing for a pre-set amount of time followed by an oscillatory shearing motion at a specified frequency and amplitude. The effect of pre-mixing on the aggrecan-hyaluronan interaction was examined by applying a constant steady-shearing motion to the sample for 2 and 4 min, respectively. The effects of frequency and amplitude of oscillation were also examined. Two frequencies (10 and 62.8 rads-‘) and two amplitudes of shear strain (60 and 100%) were used. Two kinetic measurements were made for each loading condition and solution concentration. The differences between the two measurements were found to be approximately 1%. RIWJLTS
Flow properties of control solutions
Figure 1 shows the magnitude of dynamic-shear modulus as a function of frequency for solutions of aggrecan, pre-formed aggregate and pure hyaluronan molecules at various selected concentrations. As shown in the figure, for all three types of solutions, the magnitude of dynamic-shear modulus lG*l was dependent directly on the frequency. In the high-frequency domain (> 10 rad s-l), the values of lG+l of the proteoglycan aggregate solutions were approximately twice those of the aggrecan solutions at similar concentrations. This difference was generally larger in the low-frequency domain (l-10 rad s- ‘). Figure 2 shows the apparent viscosity qappas a function of shear rate for solutions of aggrecan, pre-formed aggregate and pure hyaluronan molecules. All three types of solutions exhibited a shear-rate dependent apparent viscosity where PI,~ decreased nonlinearly with shear rate. The dependence of qapp on shear rate was not significant for aggrecan solutions of low concentra-
W. ZHU et al.
574 10’
-
Hyalurunan
-
bB=Este Aggrenm
20 rng/rnl 50 me/ml
10 mg/ml
IO’
Frequency (rad/s) Fig. 1. Variations of the dynamic shear modulus JG*l with frequency and concentration for solutions of aggrecan, pre-formed aggregate and hyaluronan alone.
104
10”
. . . . ...r
.
. . ..
10’
..I
.
.
10’
.
.
. ..I
lof
.
.
.
.
.
1
10’
Shear Rate (se’)
Fig. 2. Variations of the apparent viscosity tl.PPwith shear rate and concentration for solutions of aggrecan, pre-formed aggregate and hyaluronan alone.
tions (10 mg ml- ‘), indicating a Newtonian behavior of the solution. The apparent viscosity values of proteoglycan aggregate solutions were substantially higher than those of aggrecan solutions for low shear rates and such differences tended to diminish as shear rate increased. The dependence of IG*l and qIPP on concentration is shown in Figs 3(a) and (b) for the three types of solutions. At comparable concentrations, hyaluronan solution had significantly higher IG*l and tl.PP values than aggregate and aggrecan solutions. The hyaluronan solution exhibited high nonlinear behavior with varying concentration. The variation of rapt, with concentration was similar to that of IG*l for all three types of solutions. The transient responses to the step change of shear rate were significantly different between hyaluronan solutions and proteoglycan aggregate solutions. As shown in Fig. 4, both hyaluronan and proteoglycan aggregate solutions exhibited stress-overshoot effects which was not observed for aggrecan solutions. The hyalur-
onan solution exhibited a larger, more distinct peak stress than the aggregate solutions. The stress overshoots became less pronounced at lower concentrations. Kinetic properties
of mixture solutions
Aggregation under steady-shearing
motion. Figure 5 shows a representative variation of viscosity as a function of time for a 30 mgml- ’ mixture solution under a constant shearing of 100 s-l. The aggregate formation was evident and took place immediately following the shearing motion. Starting near the value for an aggrecan solution, the viscosity of the mixture solution increased monotonically, and reached a steady equilibrium aggregate viscosity. The aggregation was completed in approximately 35 min under the constant-shearing motion of lC10s-~. It was observed from the steady-shear experiment that the viscosity-time characteristic depends on the concentration of the aggrecan-hyaluronan mixture solution
(a) 1w
=
Kinetic changes of aggrecan-hyaluronan interactions
Freqaoncy10rad s-’
00
-
1Control
Hyalaronm
-.%stmtmti
IR
.4gatqate
-4&P-n
575
8.’
Vtscwtty
\7----
““““““““““““““““““”
““”
/ C‘m*ral
hggresm
vtscdty
k_-_--_______----___
“““---“““““““---“”
es! -I
lb'
lb’
lb’
Time (main)
Fig. 5. Kinetic interaction of aggrecan and hyaluronan molecules under constant steady shear of lOOs-‘. The viscosity of mixture solution, qapp(t), varied with time. Starting near the value of aggrecan solution, q,,(t) increased monotonically, and reached a steady equilibrium pm-formed aggregate viscosity in approximately 35 minutes.
.““““--““““_______
“““”
Concentration (mglml)
Fig. 3. (a) Variations of the dynamic shear modulus IG*l with concentration at 10 rad s-l frequency;(b) variations of the apparent viscosity qePPwith concentration at lOOs-’ shear rate for solutions of aggrecan, pre-formed aggregate and hyaluronan alone.
“_
lb-’
I;*
12
lb
Time (min)
Fig. 6. Kinetic interaction of aggrecan and hyaluronan molecules under constant steady shear of lOOs-’ for three concentrations. The viscosities were normalized by the viscosities of pre-formed aggregate solution at corresponding shear rates.
$4 m&II 30 mu/ml
,&l
1b.l
Ii’
Ii’
Time (a)
Fig. 4. Transient shear stress overshoot effects following a step shear rate of 100 s- 1 for solutions of aggrecan, preformed aggregate and hyaluronan alone.
as well as the magnitude of shear rate. At the shear rate of 100 s- l, Fig. 6 shows the kinetics of aggregation for three different concentrations, 1430 and 50 mgml- ‘. The apparent viscosity of each sample was normalized by its corresponding viscosity of the pre-formed aggregate solution determined from the control experiments. Under the same shearing conditions, the proteoglycan aggtegation process was much slower for the high concentration solutions than for the low concentration solutions, For example, starting from the aggrecan viscosity value, the viscosities of the 10
and 30 mg ml- ’ mixture specimens increased monotonically with time, and reached their equilibrium viscosities of pre-formed aggregate solutions in about 26 min, while in the same time interval, the viscosity of 50 mg ml- 1 specimen reached only 73% of its equilibrium aggregated solution value. Figure 7 shows the variations of normalized viscosity with time for the applied shear rate of 10 s-‘. It was observed that the values of viscosity for the 10 mg ml- ’ sample became greater than those of the final equilibrium values by as much as 40%, followed by a tendency of slow decrease in the viscosity value. This overshoot phenomena was not observed in the case of higher shear rate (Fig. 6), but was more pronounced at lower shear rate (1 s-l) (Fig. 8). As shown in Fig. 8 for a 30 mg ml- ’ sample, the rate of aggregation depends strongly upon the rate of shear. The viscosity of the mixture solution approached steadily to the equilibrium value at 100 s- ‘, whereas at 1 s- 1 there was a much slower aggregating process in the earlier times (< 10 min), and a rapid rise
576
W. ZHU et al.
in viscosity occurred after approximately 20 min of shearing. In both 1 and 10 s-l cases, the viscosities overshoot their presumed equilibrium values. Greater overshoot occurred under slower shearing motion. After the shearing motion stopped for a period of time, the viscosity of the mixture solution was measured again and the magnitude was found to decline toward the equilibrium value (Fig. 8). Aggregation under oscillatory shear motion. The rate of aggregation was also examined under oscillatory shearing conditions by examining the change of the magnitude of shear modulus IG*I with time. We found that the aggregation did not take place for mixture solutions subjected to oscillatory motion immediately after being placed on the plate of the viscometer. The IG*l remained at the control aggrecan value for a long period of time. Thus, a pre-mixing of hyaluronan and aggrecan molecules was necessary for a possible aggregation to occur under oscillatory motions. This pre-mixing was done by subjecting the mixture solution to a continuous shearing motion.
h .* ,,*__________
____
Y
F E .;
E o.s-
10 mglml
L
30 mglml 50 na/rnl
0.04
IO"
IO'
10'
10'
Time (min)
Fig. 7. Kinetic interaction of aggrecan and hyaluronan molecules under constant steady shear of 10 s-l for three concentrations. The normalized viscosities for the 10 and 30 mgml- ’ samples overshoot the viscosity values of preformed aggregate solution.
1.5
g
‘I
1.0
I
Figure 9 shows two typical variations of IG*l normalized by the corresponding control aggregate values upon pre-mixing. In test 1, the mixture solution was subjected to the continuous shearing motion at 100 s- ’ for 2 min, followed by the oscillatory motion at the frequency of 10 rads-‘. Note that the premixing caused the value of IG*I for the mixture solution to reach about 60% of its equilibrium aggregate value before the initiation of the oscillatory motion. After 100 min of continuous oscillatory shearing at lOrads_ ’ frequency and 60% amplitude of shear, IG*I reached about 90% of its equilibrium value. In test 2, and a shear rate of 10 s- ’ was performed on the mixture for 4 min, followed by an oscillatory shear at the frequency of 62.8 rads-’ and 60% amplitude of shear. The value of IG*l tended to decrease with time. In addition, the amplitude of shear did not seem to change these trends significantly, since a similar trend was observed for amplitude of shear @lOO%. DISCUSSION
The integrity of the extracellular matrix is crucial for the biomechanical properties of articular cartilage and thereby for the function of the joint as a whole. This integrity depends on the structure and content of the various matrix components as well as the specific interactions among these matrix components. In this study, we first examined the effect of proteoglycan aggregation on the mechanical properties of the network formed by the aggrecan and aggregate molecules, respectively. Our results of control samples demonstrated that proteoglycan aggregation enhanced network stiffness and increased resistance of the network to shear flow (Figs 1 and 2). On the other hand, by comparison to intact cartilage, the dynamic shear modulus is much smaller for proteoglycan networks formed in solutions (Zhu et al., 1993). This result suggests that proteoglycan aggregation does not contribute directly to the overall stiffness of cartilage in
30mg/ml
___-_______-----____--
--
‘s
Measured afkr shearin motion
stoppe B
P
loo
10’
10’
Time (min) Fig. 8. E&t of shear rate on the kinetic interaction of aggrecan and hyaluronan molecules under steadyshear flow conditions. The overshoots occurred for shear rates of 1 and 10 s-l. The viscosity value decreased after the shearing motion stopped for a period of time.
Kinetic changes of aggrecan-hyaluronan
1.1
1
0
30 mglml
7.0
40
60
SO
I@0
110
Time (min)
Fig. 9. Kinetic interaction of aggrecan and hyaluronan molecules under oscillatory shearing conditions. Test 1: 2min shearing at constant shear rate of lOOs_’ followed by oscillation at frequency of 10 rad s- ‘; Test 2: 4 min shearing at 10 s-l followed by oscillation at 62.8 rads- ‘.
shear. Rather, the supra-molecular size of proteoglycan aggregates assures the immobilization of these molecules at high concentration within the collagen network and maintains the collagen network in its proper spatial configuration. As shown in our previous study (Zhu et nl., 1993), the disruption of proteoglycan structure, such as dissociation of proteoglycan aggregates, defeats one of the major functions of the proteoglycan aggregates, namely, to inflate and pre-stress the collagen network. As a result of such a disruption, the shear stiffness of cartilage tissue is greatly reduced. Furthermore, with the results of control samples as the base line, we examined the kinetic interactions between aggrecan and hyaluronan molecules and factors which might influence the rate of interactions. Hyaluronan is a high-molecular-weight polysaccharide and it has been visualized on the electron microscope as a linear unbranched chain. In solution it is highly hydrated in a random coil configuration and exhibits molecular exclusion of other macromolecules, and has marked physical properties (Laurent and Fraser, 1986). The molecular weight of hyaluronan is the predominant factor determining the size of proteoglycan aggregates (Buckwalter and Rosenberg, 1983). The results presented in this study showed that rheological flow properties of pure hyaluronan solutions differ significantly from the aggrecan and aggregate solutions especially as a function of solution concentration. At a comparable concentration, frictional resistance in hyaluronan solution due to constant shearing as measured by the solution viscosity was much greater (Fig. 2) and more energy was required to reorient hyaluronan molecules from random to aligned flow directions than aggrecan and aggregate solutions (Fig. 4) (Zhu et al., 1991). It is possible that these unbranched hyaluronan molecules tend to interact more easily and form more dense networks in solution than the proteoglycan molecules which have a more rigid structure.
interactions
571
The addition of small amounts of hyaluronan (0.8% w/w) to aggrecan molecules produced large proteoglycan aggregate molecules. We observed that the kinetic interaction of these two types of molecules took place immediately following the mixing, and resulted in a large increase in solution viscosity. Constant shearing of the two solution samples produced a homogeneous mixture and assisted the interaction of these two types of molecules (Fig. 6). The viscosity of the mixture solution increased steadily until an equilibrium state of aggregation was reached. This increase in viscosity with added hyaluronan molecules appears to result mostly from the combined effects of proteoglycan size, molecular entanglement, exclusion volume or other forms of physical interactions of macromolecules, while hyaluronan itself would contribute little to the total viscosity due to its relatively small amount present (0.8% w/w) in solution. It is not surprising that the higher concentration solution took longer time to produce a homogeneous mixture (Figs 6 and 7). Possibly, at high concentrations (50 mgml- ‘), it is harder for the hyahn-onan backbone to be stretched limiting the percentage of aggrecan molecules on each hyaluronan chain and resulting in a magnitude of viscosity much lower than the corresponding controls. It was evident that the loading conditions affected the kinetic properties of proteoglycan-hyaluronan interactions as measured by the changes in viscosity and dynamic modulus with time. The magnitude of shear rate appeared to be the dominant factor controlling the kinetics of these interactions. Under steady shear conditions, the slow shear motion (c 10 s-l) caused the mixture solutions to achieve super-viscous values prior to settling to their equilibrium values (Fig. 7). The explanation for this overshoot phenomenon may be as follows: at low shear rates the hyaluronan molecules were being unfolded from their preferred coiled position to accommodate more aggrecans to be attached, thus producing super-large aggregates. This supra-optimal packing of proteoglycans caused the viscosity of the solution to increase beyond the values of control aggregate solutions. After the shearing motion stopped for a period of time, the viscosity of mixture solutions was measured again and a drop in the viscosity was observed (Fig. 8). It seemed that the hyaluronan molecules tended to recoil when the shearing motion stopped, and some aggrecans dissociated from the hyaluronan backbone to arrive at the final nonsaturated equilibrium state. This behavior suggested that the hyaluronan chains can be unfolded beyond their nonsaturated aggregation length by the shearing force. If the shearing motion was sufficiently fast (2 100 s-l), there would not be sufficient time for forming of super-large proteoglycan aggregates; thus, no overshoot in viscosity occurred. The rate of aggregation under oscillatory shearing motions depended upon the pre-mixing of aggrecan and hyaluronan molecules as well as the oscillatory frequencies. After pre-mixing, the aggrecan and hyal-
578
W. ZHu et al.
uronan molecules tended to interact and form aggregates under low frequency oscillation (10 rad s-l) (Fig. 9). However, the aggregating process was much slower under oscillatory motion than under steady shearing motion. It appears that frequent change in shear flow directions reduces the rate of association of hyaluronan and aggrecan molecules. On the other hand, the high frequency oscillatory shearing (62.8 rads- ‘) did not seem to assist the molecular association but rather impeded the aggregation as reflected by the trend of decrease in IG*I with time (Fig. 9). This frequency-dependent aggregation is of great importance in understanding the in uiuo assembly process of proteoglycans. These results indicate that either continuous shearing or oscillatory shearing in physiologic range of frequencies will assist proteoglycan aggregation process. The influence of loading conditions on the rate of aggregation and aggregate size observed in this study seems to suggest a close relationship between proteoglycan structure and content, and degree of physiological stress throughout the joint and frequency ofthe joint motion. It appears that the topographical differences in aggregate size and content between the regions which experience high stresses and low stresses (Manicourt et al., 1991) may be a result of this loadingdependent aggregate assembly process. The degenerative or atrophic changes in articular cartilage caused by the lack of joint motion and/or load-bearing (e.g. joint disuse) will result in changes in the structure and content of proteoglycans in the tissue. It is known that link protein plays an important role in aggregate formation. It provides the binding stability between aggrecan and hyaluronan and induces an increase in the basic structural dimensions of proteoglycan molecules in solution (Hardingham, 1979). The large preparation of link protein which would be required to conduct this in vitro study prevented us from examining the aggregation process with the presence of link protein. Nevertheless, the current study provided important information on the kinetics of proteoglycan-hyaluronan interactions and the factors which might influence these interactions. The results from this study also provided better understanding of the nature of the hyaluronanproteoglycan interactions and strongly suggested that mechanical loading may influence the aggregation process which occurs in the tissue. However, a more complex in vi00 aggregation process may involve additional factors such as the metabolic activity of the chondrocytes which may affect the hyaluronan-affinity of aggrecan. Other factors such as the availability of hyaluronan molecules and link proteins, and the density of the collagen network which controls the rate of proteoglycan diffusion, probably also have an influenci: on the rate of aggregation.
Acknowledgements-This research was supported by grants from the National Institutes of Health (AR38742)and the Whitaker Foundation.
REFERENCES
Bayliss, M. T. and Davidson, C. (1991) Human articular cartilage: an investigation of in situ aggregation of proteoglycan. Trans. orthop. Res. Sot. 16, 5. Bayliss, M. T., Ridgway, G. D. and Ali, S. Y. (1983) Differences in the rates of aggregation of proteoglycans from human articular cartilage and chondrosarcoma. Biochem. J. 215,705-708. Buckwalter, J. A. and Rosenberg, L. C. (1983) Structural changes during development in bovine fetal epiphyseal cartilage: electron microscopic studies of proteoglycan monomers and aggregates. Collagen. Rel. Res. 3,489-W. Camey, S. L., Billingham, M. E. J., Muir, H. and Sandy, J. D. (1984) Demonstration of increased proteoglycan turnover in cartilage explants from dogs with experimental osteoarthritis. J. orthop. Res. 2,201-206. Hardingham, T. E. (1979) The role of link protein in the structure of cartilage proteoglycan aggregates. Biochem. J. 177,231-241.
Hardingham, T. E and Muir, H. (1972) The specific interaction of hyaluronan acid with cartilage proteoglycans. Biochem. Biophys. Acta 279,401-405.
Hardingham, T. E., Muir, H., Kwan, M. K., Lai, W. M. and Mow, V. C. (1987) Viscoelastic properties of proteoglycan solutions with varying proportions present as aggregates. J. orthop. Res. 5, 36-46.
Kimura, J. H., Hardingham, T. E. and Hascall, V. C. (1980) Assembly of newly synthesized proteoglycan and link protein into aggregates in cultures of chondrosarcoma chondrocytes. J. Biol. Chem. 255, 7134-7143. Kimura, J. H., Hardingham, T. E. and Hascall, V. C., Solursh, M. (1979) Biosynthesis of proteoglycans and their assembly into aggregates in culture of chondrocytes from the swarm rat chondrosarcoma. J. Biol. Gem. 254, 2600-2609.
Laurent, T. C. and Fraser, J. R. E. (1986) The properties and turnover of hyaluronan. In Function of Proteoglycans (Edited by Evered, D. and Whelan, J.), Ciba Foundation Symposium 124, pp. 9-24, Wiley, Chichester. Lohmander, L. S. and Kimura, J. H. (1986) Biosynthesis of cartilage proteoglycan. In Articular Cartilage Biochemistry (Edited by Kuettner, K., Schleyerbach, R. and Hascall, V. C.), pp. 93-l 11. Raven Press, New York. Manicourt, D. H. and Pita, J. C. (1988) Progressive depletion of hyaluronic acid in early experimental osteoarthritis in dogs. Arth. Rheum 31, 538-544. Manicourt, D. H., Pita, J. C., Thonar, E. and Howell, D. S. (1991) Proteoglycans nondissociatively extracted from different zones ofcanine normal articular cartilage: variations in the sedimentation profile of aggregates with degree of physiological stress. Connective Tissue Res. 26, 231-246. McDevitt, C. A. and Muir, H. (1976) Biochemical changes in the cartilage of the knee in experimental and natural osteoarthritis in the dog. J. Bone Jt Surg. A59, 94-101. Mow, V. C., Zhu, W. B., Lai, W. M., Hardingham, T. E., Hughes, C. and Muir, H. (1989) The influence of link protein stabilization on the viscometric properties of proteoglycan aggregate solutions. Biochim. Biophys. Acta 992, 201-208.
Muir, H. (1983) Proteoglycans as organizers of the extracellular matrix. Biochem. Sot. Trans. 11, 613-622. Muller, F. J., Pita, J. C., Manicourt, D. H., Malinin, T. I., Schoonbeck, I. M. and Mow, V. C. (1989) Centrifugal characteristics of proteoglycans from various depth layers and weight-bearing areas of normal and abnormal human articular cartilage. J. orthop. Res. 7, 326-334. Pottenger, L. A., Lyon, N. B., Hecht, J. D., Neustadt, P. M. and Robinson, R. A. (1982) Influence of cartilage particle size and proteoglycan a&gation on immobi-&&ion of proteoalycans. J. Biol. Chem. 257. 11.479-11.485. Prehm, f i1983) Synthesis of hyaliroiate in differentiated teratocarcinoma cells: mechanism of chain growth. Eiothem. J. 211. 191-198.
Kinetic changes of aggrecan-hyaluronan Ratcliffe, A., Fryer, P. R. and Hardingham, T. E. (1984) The distribution of aggregating proteoglycans in articular cartilage. J. Histochem. Cytochem. 32.193-201. Rosenberg, L. C., Pal, S. and Beale, R. (1973) Proteoglycans from bovine proximal humeral articular cartilage. J. Biol. Ckm. 24lI, 3681-3690. Sweet, M. B. E., Thonar, E. J. M. A. and Immehnan, A. R. (1977) Regional distribution of water and glycosaminoglycan in immature articular cartilage. Bioch-Biophys Acta. 500, 173-186. Zhu, W. B., Lai, W. M. and Mow, V. C. (1991) the density and
interactions
579
strength of proteoglycan-proteoglycan interaction sites in concentrated solutions. J. Biontech. 24, 1007-1018. Zhu, W. B. and Mow, V. C. (1990) Viscometric properties of proteoglycan solutions at physiological concentrations. In Biomechanics ofDiarthro&al Joints (Edited by Mow, V. C., Ratcliffe, A. and Woo, S. L. Y.), pp. 313-344. Springer, New York. Zhy W. B., Mow V. C., Koob, T. J. and Eyre, D. R. (1993) Viscoelastic shear properties of articular cartilage and the effects of glycosidase treatment. J. orthop. Rex 11, 771-781.
Pergrmon
1994
Elsevier Science Ltd Printed in Great Britain. All rights -cd OOU-9290/W
Sa.CdI+.a,
DETERMINATION OF KINETIC CHANGES OF AGGRECAN-HYALURONAN INTERACTIONS IN SOLUTION FROM ITS RHEOLOGICAL PROPERTIES WENBO ZHU,* VAN C. Mow,~ LAWRENCEC. ROSENBERGS and LIH-HENG TANGS *Departmentof Mechanical Engineering, University of Maryland, Baltimore, Maryland, U.S.A.; tDepartment of Mechanical Engineering and Orthopedic Surgery, Columbia University, New York, U.S.A.; SOrthope& Research Laboratories, Montefiore Hospital Medical Center, Bronx, New York, U.S.A. Abstract-The kinetics of interactions between aggregating cartilage proteoglycan (aggrecan) and hyaluronan was examined through their rheological flow behavior using a cone on plate viscometer. The mixing of the two types of molecules was carried out directly on the plate of the viscometer, and aggregation process was monitored through the changes of the sample’s steady-shear viscosity and/or dynamic shear modulus as a function of time. The effect of flow conditions on the aggrecan-hyaluronan interaction rates was examined by subjecting samples to steady-shearing motions at specified shear rates, and to oscillatory shear motions of specified frequencies and amplitudes. The characteristics of the kinetics of interaction between aggmcan and hyaluronan molecules depended not only on the flow conditions under which proteoglycan aggregation took place, but also on the concentration of the components in the solution. At high shear rates (> 10 s- ‘), viscosity of the mixture solution increased monotonically, starting near the viscosity of the aggrecan solution, and reaching the viscosity of the aggregate solution in approximately 35 min. Surprisingly, under slow shearing motions (< 10 s-t), the viscosities of the mixture solutions exceeded those of control aggregate solutions at identical hyaluronan : aggrecan ratios and concentrations. In addition, the aggregation under oscillatory motions took place near physiologic Bequency (10 rads- r) although the rate of aggregation process was much slower than under steady-shearing motion (> 100 mm). However, the highfrequency oscillatory shearing (62.8 rads-t) tended to impede aggregation resulting in a reduction of dynamic modulus over time. The influence of loading conditions on the rate of aggregation and aggregam size observed in this study seems to suggest a close relationship between proteoglycan structure and content, and degree of physiological stress throughout the joint and frequency of the joint motion. INTRODUCMON
Proteoglycan aggregates, consisting of aggrecan, link proteins and hyaluronan, form a significant part of the cartilage extracellular matrix. Aggrecan consists of a protein core to which a large number of keratan sulfate and chondroitin sulfate chains are covalently attached. In cartilaginous tissues, most aggrecans exist as aggregates by binding to a linear hyaluronan chain. This binding is stabilized by link proteins. The significance of aggregate formation is to restrict the movement of proteoglycans within the collagen fiber network, and thereby to remain proteoglycans at a high concentration in the tissue (Muir, 1983; Pottenger et al., 1982). In cartilage matrix, the components of aggrecan molecules, hyaluronic acid and link proteins arc manufactured by chondrocytes. These components are secreted independently from the cells, and then assembled as aggregates within the extracellular matrix (Bay&s et aI., 1983; Kimura et al., 1979,198O; Lohmander and Kimura, 1986 Prehm, 1983). Aggregation is an important event in the development of a stable matrix, and if perturbed, could result in a less-
Received in jnal form 25 June 1993. Author to whom correspondence should be addDr Wenbo Zhu, Department of Mechanical Engineering, University of Maryland Baltimore County Campus, Baltimore, MD 21228, U.S.A.
ordered structure with inferior material properties (Hardingham et al., 1987; Mow et al., 1989; Zhu and Mow, 1990). The size and content of proteoglycan aggregates depend on the quality and distribution of the proteoglycan components in the tissue (Manicourt et al., 1991). Many researchers have demonstrated sign&ant changes of proteoglycan structure and content in articular cartilage with age and degeneration as well as the degree of physiological stress throughout the joint. For example, studies of the topographical proteoglycan distribution have shown that the proteoglycan content was higher in the articulating, high weightbearing regions than in the relatively low weightbrering areas of the joint (Sweet et al., 1977). Analysis of proteoglycan aggregates nondissociatively extracted from cartilage showed enhanced sire of proteoglycan aggregates in high weight-bearing area (Manicourt et al., 1991). A decrease in proteoglycan aggregation and an increase in proteoglycan turnover were found to be the two main biochemical characteristics of osteoarthritic and aging cartilages (Carney et al., 1984, McDevitt and Muir, 1976; Muller et al., 1989). A progressive and sign&ant decrease in hyaluranic acid content has been detected in the early stage of experimental osteoarthritis models (Manicourt and Pita, 1988; McDevitt and Muir, 1976). This decrease may limit the amount of proteoglycans to be involved in aggregation. 571
512
W. ZHu et al.
While the ultrastructure and content of proteoglyparable to those found in native cartilage can molecules throughout the joint and changes in (lo-50 mg ml- ‘), and the effect of varying concentraproteoglycan structure during aging and degeneration tions on the kinetic interactions were also examined. have been documented, little is known about the aggregation processes of matrix components MATERIALS AND METHODS occurring within the extracellular matrix and the mechanisms controlling the proteoglycan extracelluProteoglycan preparations lar assembly. It is also not known how external Aggrecan molecules were prepared from fresh loading or deformation may influence the rate of bovine articular cartilage as previously described assembly of aggrecan with hyaluronan. Recent studies (Rosenberg et al., 1973). Briefly, cartilages were exof kinetic aggregation and the stability of proteoglytracted in ice-cold 4 M guanidine HCl with 0.15 M can aggregates showed an age-related decrease in the sodium acetate at pH 6.3 containing proteinase inrate of extracellular assembly of proteoglycan aggreghibitors. After slow stirring at 5°C for 24 h, the extracts ates (Bayliss and Davidson, 1991). Most of the experiwere filtered, and dialyzed against 20 volumes of mental studies on aggregation have exclusively been 0.15 M sodium acetate, 0.05 M EDTA, pH 6.3, conperformed in vitro under rather dilute conditions taining protease inhibitors at 5°C for 16 h. The extract (Kimura et al., 1980), while the in situ concentration of was then fractionated by equilibrium density gradient the aggregate components in cartilage are very high. centrifugation under associative conditions in 3.5 M These high concentrations will undoubtedly influence CsCl at 5°C for 60 h at 40,000 rpm. The fraction the kinetics of the interactions involved in aggregate AlAl was taken up in 5.5 M guanidine HCl, 0.15 M formation. sodium acetate, pH 6.3, and was stirred overnight at Recent biorheological studies of concentrated pro5°C. Equilibrium density gradient centrifugation teoglycan solutions have provided quantitative inunder dissociative conditions was then carried out at formation on the effects of various proteoglycan struc5°C for 60 h in 3 M CsCl, 4 M guanidine HCl at tural changes on the mechanical properties of pro40,008 rpm. The dissociative gradient was divided into teoglycan solutions. There is considerable evidence to six equal fractions called Dl-D6. Proteoglycan fracshow that the proportion of proteoglycans aggregtion AlAlDlDl was dialyzed against 0.15 M sodium ated, the size of aggregates and their stabilization by acetate at pH 6.3 and precipitated with 3 volumes of link protein are important factors in determining the ethanol, washed three times with ethanol and ether elastic and viscous properties of the molecular netand dried in a vacuum oven. works they formed in solution (Hardingham et al., Three types of solutions were prepared: (I) aggrecan 1987; Mow et al., 1989; Zhu et al., 1991). For example, solutions, (2) aggregate solutions, and (3) hyaluronan the molecular networks formed by proteoglycan agsolutions. The aggrecan solutions were prepared from gregates were shown to have higher dynamic shear the AlAlDlDl fraction by re-equilibrium in PBS of stiffness and greater resistance to shear flow than those 0.15 M NaCl at pH 7. The proteoglycan aggregate formed by the proteoglycan monomers. For aggregsolutions were prepared by mixing aggrecan and ates formed without link protein, shear stiffness was hyaluronan components in PBS of 0.15 M NaCl at reduced by approximately 20% as compared to agpH 7. A weight ratio of hyaluronan:aggrecan at gregates with stabilizing link protein. The strength of 0.008: 1 was used for optimal proteoglycan aggregaintermolecular interactions for the link-free aggregtion (Hardingham and Muir, 1972). Thorough mixing ates was also significantly decreased (Mow et al., 1989; of aggrecan and hyaluronan molecules was carried out Zhu et al., 1991). by continuously rotating the sample on a mechanical The objectives of this study are (1) to examine the roller at 5°C for 48 h to assure a complete formation of effect of proteoglycan aggregation on the flow properaggregates. The solutions of aggrecan and aggregate ties of proteoglycan solutions, (2) to examine the were prepared at lo,30 and 50 mg ml - ‘, respectively. kinetics of aggrecan-hyaluronan aggregation, (3) to Both aggrecan and aggregate preparations served as determine the factors such as flow conditions and control solutions to provide reference values for flow solution concentrations that might influence the rate properties measurements. The hyaluronan preparof aggregation. These were achieved by examining the ation was obtained from rooster comb with an averdirect interactions of aggrecan and hyaluronan moleage molecular weight of 10’ D. The pure hyaluronan cules in vitro through their rheological flow properties. solutions were prepared at concentrations ranging The kinetic behavior of aggrecan-hyaluronan solufrom 2.4 to 20 mg ml- ‘. Additional aggrecan solutions tions was compared, respectively, to the properties of and hyaluronan solutions were prepared to be mixed corresponding control solutions of the aggrecan on the plate of the viscometer just prior to the flow alone and aggregates formed at identical measurements. hyaluronan : aggrecan ratios and concentrations. The effect of loading conditions on the rate of Measurements of control solutions aggrecan-hyaluronan interaction was observed under Rheological flow measurements of concentrated dynamic as well as steady-shear conditions at various proteoglycan solutions were made with a low-frefrequencies, amplitudes and shear rates. The studies quency mechanical spectrometer (Rheometrics model were carried out at proteoglycan concentrations com-
Kinetic changes of aggrecan-hyaluronan
RMS-800) using a cone-on-plate viscometer (cone diameter = 50 mm and apex angle = 0.04 rad). The flat plate of the viscometer is driven by a precision d.c. motor. The stationary cone is attached to a sensitive biaxial transducer that measures the torque and normal force required to shear the solution. The viscometer is enclosed in an environmental chamber which maintains a constant temperature of 20°C and 100% humidity. The apparatus is controlled by a microcomputer for testing, data acquisition and analysis. In this study, the following flow properties of proteoglycan solutions were measured: (1) the magnitude of dynamic shear modulus lG*l and tangent of the phase shift angle, tans, from an oscillatory shear experiment; (2) the shear-rate dependent apparent viscosity qaPPand primary normal stress difference u1 from a steady-state shear experiment; and (3) transient stress growth function r(t) from a step shear rate experiment. Flow measurements were first performed on the control solutions of aggrecan, pre-formed aggregate and pure hyaluronan molecules. The magnitude of dynamic shear modulus of the control solutions was measured, respectively, at the frequencies of 1, 10,100 rad s-l with an applied strain amplitude of 50%, followed by a frequency sweep ranging from 1 to 100 rad s- ‘. Subsequently, the apparent viscosities of the solutions were measured at the shear rates of 1,10 and 100 s-l, respectively. A shear-rate sweep experiment was then performed for four decades of shear rate (0.25250 s- ‘). Finally, the transient stress growth experiments were performed at the step shear rates of 25 and 100 s- ‘. The values of apparent viscosity and dynamic shear modulus of the homogeneous aggrecan solutions define the initial flow properties, whereas those of preformed aggregate solutions define the flow properties of the equilibrium state at which the aggregation process was completed. The steady-state and transient properties of the pure hyaluronan solution were also determined for comparison. Two complete sets of measurements were ma& for each control proteoglycan solution. Since the test samples were obtained from the same preparation, the maximum difference between the two measurements was found to be less than 5%. Kinetics of aggregate formation
An experimental procedure was developed to study the kinetic interaction of aggrecan and hyaluronan molecules under a specific mechanical loading condition. In order to follow the initial kinetics and the rate of aggregation process the aggrecan and hyaluronan solutions were placed directly onto the plate of the viscometer using a pipette. Each solution volume was precisely measured ensuring a hyaluronan: aggrecan ratio of 0.8% upon mixing for optimal aggregation. Upon placing the samples, the upper cone was lowered by the stepper motor until the 50 cun gap between the cone and plate was reached. It took approximately 2 min to load the sample. The entire test section was
interactions
513
then enclosed in a temperature and humiditycontrolled chamber. Two types of tests were performed on separate but identical mixtures: (1) a steady-shear test at a constant shear rate, and (2) a dynamic-shear test at a constant frequency and amplitude. In the steady-shear test, a constant shear rate was initiated immediately after the loading of the solution. The aggregate formation as a function of time was observed through the viscosity measurements of the mixture solution. The viscosity measurement began immediately upon shearing, and the measurements were taken every 2 min during continuous shearing until an equilibrium value was reached. To examine the effect of shear rate on the proteoglycan aggregation process, separate kinetic experiments were performed at three different shear rates (1,lO and 100 s- ‘). A similar procedure was adapted for kinetic experiments under oscillatory shearing motion. The kinetics of aggregate formation was monitored by the change of the dynamic shear modulus of the sample. It was necessary to pre-mix the hyaluronan and aggrecan solutions before the dynamic shearing motion was applied to the sample. Upon loading, the sample was first subjected to a constant shearing for a pre-set amount of time followed by an oscillatory shearing motion at a specified frequency and amplitude. The effect of pre-mixing on the aggrecan-hyaluronan interaction was examined by applying a constant steady-shearing motion to the sample for 2 and 4 min, respectively. The effects of frequency and amplitude of oscillation were also examined. Two frequencies (10 and 62.8 rads-‘) and two amplitudes of shear strain (60 and 100%) were used. Two kinetic measurements were made for each loading condition and solution concentration. The differences between the two measurements were found to be approximately 1%. RIWJLTS
Flow properties of control solutions
Figure 1 shows the magnitude of dynamic-shear modulus as a function of frequency for solutions of aggrecan, pre-formed aggregate and pure hyaluronan molecules at various selected concentrations. As shown in the figure, for all three types of solutions, the magnitude of dynamic-shear modulus lG*l was dependent directly on the frequency. In the high-frequency domain (> 10 rad s-l), the values of lG+l of the proteoglycan aggregate solutions were approximately twice those of the aggrecan solutions at similar concentrations. This difference was generally larger in the low-frequency domain (l-10 rad s- ‘). Figure 2 shows the apparent viscosity qappas a function of shear rate for solutions of aggrecan, pre-formed aggregate and pure hyaluronan molecules. All three types of solutions exhibited a shear-rate dependent apparent viscosity where PI,~ decreased nonlinearly with shear rate. The dependence of qapp on shear rate was not significant for aggrecan solutions of low concentra-
W. ZHU et al.
574 10’
-
Hyalurunan
-
bB=Este Aggrenm
20 rng/rnl 50 me/ml
10 mg/ml
IO’
Frequency (rad/s) Fig. 1. Variations of the dynamic shear modulus JG*l with frequency and concentration for solutions of aggrecan, pre-formed aggregate and hyaluronan alone.
104
10”
. . . . ...r
.
. . ..
10’
..I
.
.
10’
.
.
. ..I
lof
.
.
.
.
.
1
10’
Shear Rate (se’)
Fig. 2. Variations of the apparent viscosity tl.PPwith shear rate and concentration for solutions of aggrecan, pre-formed aggregate and hyaluronan alone.
tions (10 mg ml- ‘), indicating a Newtonian behavior of the solution. The apparent viscosity values of proteoglycan aggregate solutions were substantially higher than those of aggrecan solutions for low shear rates and such differences tended to diminish as shear rate increased. The dependence of IG*l and qIPP on concentration is shown in Figs 3(a) and (b) for the three types of solutions. At comparable concentrations, hyaluronan solution had significantly higher IG*l and tl.PP values than aggregate and aggrecan solutions. The hyaluronan solution exhibited high nonlinear behavior with varying concentration. The variation of rapt, with concentration was similar to that of IG*l for all three types of solutions. The transient responses to the step change of shear rate were significantly different between hyaluronan solutions and proteoglycan aggregate solutions. As shown in Fig. 4, both hyaluronan and proteoglycan aggregate solutions exhibited stress-overshoot effects which was not observed for aggrecan solutions. The hyalur-
onan solution exhibited a larger, more distinct peak stress than the aggregate solutions. The stress overshoots became less pronounced at lower concentrations. Kinetic properties
of mixture solutions
Aggregation under steady-shearing
motion. Figure 5 shows a representative variation of viscosity as a function of time for a 30 mgml- ’ mixture solution under a constant shearing of 100 s-l. The aggregate formation was evident and took place immediately following the shearing motion. Starting near the value for an aggrecan solution, the viscosity of the mixture solution increased monotonically, and reached a steady equilibrium aggregate viscosity. The aggregation was completed in approximately 35 min under the constant-shearing motion of lC10s-~. It was observed from the steady-shear experiment that the viscosity-time characteristic depends on the concentration of the aggrecan-hyaluronan mixture solution
(a) 1w
=
Kinetic changes of aggrecan-hyaluronan interactions
Freqaoncy10rad s-’
00
-
1Control
Hyalaronm
-.%stmtmti
IR
.4gatqate
-4&P-n
575
8.’
Vtscwtty
\7----
““““““““““““““““““”
““”
/ C‘m*ral
hggresm
vtscdty
k_-_--_______----___
“““---“““““““---“”
es! -I
lb'
lb’
lb’
Time (main)
Fig. 5. Kinetic interaction of aggrecan and hyaluronan molecules under constant steady shear of lOOs-‘. The viscosity of mixture solution, qapp(t), varied with time. Starting near the value of aggrecan solution, q,,(t) increased monotonically, and reached a steady equilibrium pm-formed aggregate viscosity in approximately 35 minutes.
.““““--““““_______
“““”
Concentration (mglml)
Fig. 3. (a) Variations of the dynamic shear modulus IG*l with concentration at 10 rad s-l frequency;(b) variations of the apparent viscosity qePPwith concentration at lOOs-’ shear rate for solutions of aggrecan, pre-formed aggregate and hyaluronan alone.
“_
lb-’
I;*
12
lb
Time (min)
Fig. 6. Kinetic interaction of aggrecan and hyaluronan molecules under constant steady shear of lOOs-’ for three concentrations. The viscosities were normalized by the viscosities of pre-formed aggregate solution at corresponding shear rates.
$4 m&II 30 mu/ml
,&l
1b.l
Ii’
Ii’
Time (a)
Fig. 4. Transient shear stress overshoot effects following a step shear rate of 100 s- 1 for solutions of aggrecan, preformed aggregate and hyaluronan alone.
as well as the magnitude of shear rate. At the shear rate of 100 s- l, Fig. 6 shows the kinetics of aggregation for three different concentrations, 1430 and 50 mgml- ‘. The apparent viscosity of each sample was normalized by its corresponding viscosity of the pre-formed aggregate solution determined from the control experiments. Under the same shearing conditions, the proteoglycan aggtegation process was much slower for the high concentration solutions than for the low concentration solutions, For example, starting from the aggrecan viscosity value, the viscosities of the 10
and 30 mg ml- ’ mixture specimens increased monotonically with time, and reached their equilibrium viscosities of pre-formed aggregate solutions in about 26 min, while in the same time interval, the viscosity of 50 mg ml- 1 specimen reached only 73% of its equilibrium aggregated solution value. Figure 7 shows the variations of normalized viscosity with time for the applied shear rate of 10 s-‘. It was observed that the values of viscosity for the 10 mg ml- ’ sample became greater than those of the final equilibrium values by as much as 40%, followed by a tendency of slow decrease in the viscosity value. This overshoot phenomena was not observed in the case of higher shear rate (Fig. 6), but was more pronounced at lower shear rate (1 s-l) (Fig. 8). As shown in Fig. 8 for a 30 mg ml- ’ sample, the rate of aggregation depends strongly upon the rate of shear. The viscosity of the mixture solution approached steadily to the equilibrium value at 100 s- ‘, whereas at 1 s- 1 there was a much slower aggregating process in the earlier times (< 10 min), and a rapid rise
576
W. ZHU et al.
in viscosity occurred after approximately 20 min of shearing. In both 1 and 10 s-l cases, the viscosities overshoot their presumed equilibrium values. Greater overshoot occurred under slower shearing motion. After the shearing motion stopped for a period of time, the viscosity of the mixture solution was measured again and the magnitude was found to decline toward the equilibrium value (Fig. 8). Aggregation under oscillatory shear motion. The rate of aggregation was also examined under oscillatory shearing conditions by examining the change of the magnitude of shear modulus IG*I with time. We found that the aggregation did not take place for mixture solutions subjected to oscillatory motion immediately after being placed on the plate of the viscometer. The IG*l remained at the control aggrecan value for a long period of time. Thus, a pre-mixing of hyaluronan and aggrecan molecules was necessary for a possible aggregation to occur under oscillatory motions. This pre-mixing was done by subjecting the mixture solution to a continuous shearing motion.
h .* ,,*__________
____
Y
F E .;
E o.s-
10 mglml
L
30 mglml 50 na/rnl
0.04
IO"
IO'
10'
10'
Time (min)
Fig. 7. Kinetic interaction of aggrecan and hyaluronan molecules under constant steady shear of 10 s-l for three concentrations. The normalized viscosities for the 10 and 30 mgml- ’ samples overshoot the viscosity values of preformed aggregate solution.
1.5
g
‘I
1.0
I
Figure 9 shows two typical variations of IG*l normalized by the corresponding control aggregate values upon pre-mixing. In test 1, the mixture solution was subjected to the continuous shearing motion at 100 s- ’ for 2 min, followed by the oscillatory motion at the frequency of 10 rads-‘. Note that the premixing caused the value of IG*I for the mixture solution to reach about 60% of its equilibrium aggregate value before the initiation of the oscillatory motion. After 100 min of continuous oscillatory shearing at lOrads_ ’ frequency and 60% amplitude of shear, IG*I reached about 90% of its equilibrium value. In test 2, and a shear rate of 10 s- ’ was performed on the mixture for 4 min, followed by an oscillatory shear at the frequency of 62.8 rads-’ and 60% amplitude of shear. The value of IG*l tended to decrease with time. In addition, the amplitude of shear did not seem to change these trends significantly, since a similar trend was observed for amplitude of shear @lOO%. DISCUSSION
The integrity of the extracellular matrix is crucial for the biomechanical properties of articular cartilage and thereby for the function of the joint as a whole. This integrity depends on the structure and content of the various matrix components as well as the specific interactions among these matrix components. In this study, we first examined the effect of proteoglycan aggregation on the mechanical properties of the network formed by the aggrecan and aggregate molecules, respectively. Our results of control samples demonstrated that proteoglycan aggregation enhanced network stiffness and increased resistance of the network to shear flow (Figs 1 and 2). On the other hand, by comparison to intact cartilage, the dynamic shear modulus is much smaller for proteoglycan networks formed in solutions (Zhu et al., 1993). This result suggests that proteoglycan aggregation does not contribute directly to the overall stiffness of cartilage in
30mg/ml
___-_______-----____--
--
‘s
Measured afkr shearin motion
stoppe B
P
loo
10’
10’
Time (min) Fig. 8. E&t of shear rate on the kinetic interaction of aggrecan and hyaluronan molecules under steadyshear flow conditions. The overshoots occurred for shear rates of 1 and 10 s-l. The viscosity value decreased after the shearing motion stopped for a period of time.
Kinetic changes of aggrecan-hyaluronan
1.1
1
0
30 mglml
7.0
40
60
SO
I@0
110
Time (min)
Fig. 9. Kinetic interaction of aggrecan and hyaluronan molecules under oscillatory shearing conditions. Test 1: 2min shearing at constant shear rate of lOOs_’ followed by oscillation at frequency of 10 rad s- ‘; Test 2: 4 min shearing at 10 s-l followed by oscillation at 62.8 rads- ‘.
shear. Rather, the supra-molecular size of proteoglycan aggregates assures the immobilization of these molecules at high concentration within the collagen network and maintains the collagen network in its proper spatial configuration. As shown in our previous study (Zhu et nl., 1993), the disruption of proteoglycan structure, such as dissociation of proteoglycan aggregates, defeats one of the major functions of the proteoglycan aggregates, namely, to inflate and pre-stress the collagen network. As a result of such a disruption, the shear stiffness of cartilage tissue is greatly reduced. Furthermore, with the results of control samples as the base line, we examined the kinetic interactions between aggrecan and hyaluronan molecules and factors which might influence the rate of interactions. Hyaluronan is a high-molecular-weight polysaccharide and it has been visualized on the electron microscope as a linear unbranched chain. In solution it is highly hydrated in a random coil configuration and exhibits molecular exclusion of other macromolecules, and has marked physical properties (Laurent and Fraser, 1986). The molecular weight of hyaluronan is the predominant factor determining the size of proteoglycan aggregates (Buckwalter and Rosenberg, 1983). The results presented in this study showed that rheological flow properties of pure hyaluronan solutions differ significantly from the aggrecan and aggregate solutions especially as a function of solution concentration. At a comparable concentration, frictional resistance in hyaluronan solution due to constant shearing as measured by the solution viscosity was much greater (Fig. 2) and more energy was required to reorient hyaluronan molecules from random to aligned flow directions than aggrecan and aggregate solutions (Fig. 4) (Zhu et al., 1991). It is possible that these unbranched hyaluronan molecules tend to interact more easily and form more dense networks in solution than the proteoglycan molecules which have a more rigid structure.
interactions
571
The addition of small amounts of hyaluronan (0.8% w/w) to aggrecan molecules produced large proteoglycan aggregate molecules. We observed that the kinetic interaction of these two types of molecules took place immediately following the mixing, and resulted in a large increase in solution viscosity. Constant shearing of the two solution samples produced a homogeneous mixture and assisted the interaction of these two types of molecules (Fig. 6). The viscosity of the mixture solution increased steadily until an equilibrium state of aggregation was reached. This increase in viscosity with added hyaluronan molecules appears to result mostly from the combined effects of proteoglycan size, molecular entanglement, exclusion volume or other forms of physical interactions of macromolecules, while hyaluronan itself would contribute little to the total viscosity due to its relatively small amount present (0.8% w/w) in solution. It is not surprising that the higher concentration solution took longer time to produce a homogeneous mixture (Figs 6 and 7). Possibly, at high concentrations (50 mgml- ‘), it is harder for the hyahn-onan backbone to be stretched limiting the percentage of aggrecan molecules on each hyaluronan chain and resulting in a magnitude of viscosity much lower than the corresponding controls. It was evident that the loading conditions affected the kinetic properties of proteoglycan-hyaluronan interactions as measured by the changes in viscosity and dynamic modulus with time. The magnitude of shear rate appeared to be the dominant factor controlling the kinetics of these interactions. Under steady shear conditions, the slow shear motion (c 10 s-l) caused the mixture solutions to achieve super-viscous values prior to settling to their equilibrium values (Fig. 7). The explanation for this overshoot phenomenon may be as follows: at low shear rates the hyaluronan molecules were being unfolded from their preferred coiled position to accommodate more aggrecans to be attached, thus producing super-large aggregates. This supra-optimal packing of proteoglycans caused the viscosity of the solution to increase beyond the values of control aggregate solutions. After the shearing motion stopped for a period of time, the viscosity of mixture solutions was measured again and a drop in the viscosity was observed (Fig. 8). It seemed that the hyaluronan molecules tended to recoil when the shearing motion stopped, and some aggrecans dissociated from the hyaluronan backbone to arrive at the final nonsaturated equilibrium state. This behavior suggested that the hyaluronan chains can be unfolded beyond their nonsaturated aggregation length by the shearing force. If the shearing motion was sufficiently fast (2 100 s-l), there would not be sufficient time for forming of super-large proteoglycan aggregates; thus, no overshoot in viscosity occurred. The rate of aggregation under oscillatory shearing motions depended upon the pre-mixing of aggrecan and hyaluronan molecules as well as the oscillatory frequencies. After pre-mixing, the aggrecan and hyal-
578
W. ZHu et al.
uronan molecules tended to interact and form aggregates under low frequency oscillation (10 rad s-l) (Fig. 9). However, the aggregating process was much slower under oscillatory motion than under steady shearing motion. It appears that frequent change in shear flow directions reduces the rate of association of hyaluronan and aggrecan molecules. On the other hand, the high frequency oscillatory shearing (62.8 rads- ‘) did not seem to assist the molecular association but rather impeded the aggregation as reflected by the trend of decrease in IG*I with time (Fig. 9). This frequency-dependent aggregation is of great importance in understanding the in uiuo assembly process of proteoglycans. These results indicate that either continuous shearing or oscillatory shearing in physiologic range of frequencies will assist proteoglycan aggregation process. The influence of loading conditions on the rate of aggregation and aggregate size observed in this study seems to suggest a close relationship between proteoglycan structure and content, and degree of physiological stress throughout the joint and frequency ofthe joint motion. It appears that the topographical differences in aggregate size and content between the regions which experience high stresses and low stresses (Manicourt et al., 1991) may be a result of this loadingdependent aggregate assembly process. The degenerative or atrophic changes in articular cartilage caused by the lack of joint motion and/or load-bearing (e.g. joint disuse) will result in changes in the structure and content of proteoglycans in the tissue. It is known that link protein plays an important role in aggregate formation. It provides the binding stability between aggrecan and hyaluronan and induces an increase in the basic structural dimensions of proteoglycan molecules in solution (Hardingham, 1979). The large preparation of link protein which would be required to conduct this in vitro study prevented us from examining the aggregation process with the presence of link protein. Nevertheless, the current study provided important information on the kinetics of proteoglycan-hyaluronan interactions and the factors which might influence these interactions. The results from this study also provided better understanding of the nature of the hyaluronanproteoglycan interactions and strongly suggested that mechanical loading may influence the aggregation process which occurs in the tissue. However, a more complex in vi00 aggregation process may involve additional factors such as the metabolic activity of the chondrocytes which may affect the hyaluronan-affinity of aggrecan. Other factors such as the availability of hyaluronan molecules and link proteins, and the density of the collagen network which controls the rate of proteoglycan diffusion, probably also have an influenci: on the rate of aggregation.
Acknowledgements-This research was supported by grants from the National Institutes of Health (AR38742)and the Whitaker Foundation.
REFERENCES
Bayliss, M. T. and Davidson, C. (1991) Human articular cartilage: an investigation of in situ aggregation of proteoglycan. Trans. orthop. Res. Sot. 16, 5. Bayliss, M. T., Ridgway, G. D. and Ali, S. Y. (1983) Differences in the rates of aggregation of proteoglycans from human articular cartilage and chondrosarcoma. Biochem. J. 215,705-708. Buckwalter, J. A. and Rosenberg, L. C. (1983) Structural changes during development in bovine fetal epiphyseal cartilage: electron microscopic studies of proteoglycan monomers and aggregates. Collagen. Rel. Res. 3,489-W. Camey, S. L., Billingham, M. E. J., Muir, H. and Sandy, J. D. (1984) Demonstration of increased proteoglycan turnover in cartilage explants from dogs with experimental osteoarthritis. J. orthop. Res. 2,201-206. Hardingham, T. E. (1979) The role of link protein in the structure of cartilage proteoglycan aggregates. Biochem. J. 177,231-241.
Hardingham, T. E and Muir, H. (1972) The specific interaction of hyaluronan acid with cartilage proteoglycans. Biochem. Biophys. Acta 279,401-405.
Hardingham, T. E., Muir, H., Kwan, M. K., Lai, W. M. and Mow, V. C. (1987) Viscoelastic properties of proteoglycan solutions with varying proportions present as aggregates. J. orthop. Res. 5, 36-46.
Kimura, J. H., Hardingham, T. E. and Hascall, V. C. (1980) Assembly of newly synthesized proteoglycan and link protein into aggregates in cultures of chondrosarcoma chondrocytes. J. Biol. Chem. 255, 7134-7143. Kimura, J. H., Hardingham, T. E. and Hascall, V. C., Solursh, M. (1979) Biosynthesis of proteoglycans and their assembly into aggregates in culture of chondrocytes from the swarm rat chondrosarcoma. J. Biol. Gem. 254, 2600-2609.
Laurent, T. C. and Fraser, J. R. E. (1986) The properties and turnover of hyaluronan. In Function of Proteoglycans (Edited by Evered, D. and Whelan, J.), Ciba Foundation Symposium 124, pp. 9-24, Wiley, Chichester. Lohmander, L. S. and Kimura, J. H. (1986) Biosynthesis of cartilage proteoglycan. In Articular Cartilage Biochemistry (Edited by Kuettner, K., Schleyerbach, R. and Hascall, V. C.), pp. 93-l 11. Raven Press, New York. Manicourt, D. H. and Pita, J. C. (1988) Progressive depletion of hyaluronic acid in early experimental osteoarthritis in dogs. Arth. Rheum 31, 538-544. Manicourt, D. H., Pita, J. C., Thonar, E. and Howell, D. S. (1991) Proteoglycans nondissociatively extracted from different zones ofcanine normal articular cartilage: variations in the sedimentation profile of aggregates with degree of physiological stress. Connective Tissue Res. 26, 231-246. McDevitt, C. A. and Muir, H. (1976) Biochemical changes in the cartilage of the knee in experimental and natural osteoarthritis in the dog. J. Bone Jt Surg. A59, 94-101. Mow, V. C., Zhu, W. B., Lai, W. M., Hardingham, T. E., Hughes, C. and Muir, H. (1989) The influence of link protein stabilization on the viscometric properties of proteoglycan aggregate solutions. Biochim. Biophys. Acta 992, 201-208.
Muir, H. (1983) Proteoglycans as organizers of the extracellular matrix. Biochem. Sot. Trans. 11, 613-622. Muller, F. J., Pita, J. C., Manicourt, D. H., Malinin, T. I., Schoonbeck, I. M. and Mow, V. C. (1989) Centrifugal characteristics of proteoglycans from various depth layers and weight-bearing areas of normal and abnormal human articular cartilage. J. orthop. Res. 7, 326-334. Pottenger, L. A., Lyon, N. B., Hecht, J. D., Neustadt, P. M. and Robinson, R. A. (1982) Influence of cartilage particle size and proteoglycan a&gation on immobi-&&ion of proteoalycans. J. Biol. Chem. 257. 11.479-11.485. Prehm, f i1983) Synthesis of hyaliroiate in differentiated teratocarcinoma cells: mechanism of chain growth. Eiothem. J. 211. 191-198.
Kinetic changes of aggrecan-hyaluronan Ratcliffe, A., Fryer, P. R. and Hardingham, T. E. (1984) The distribution of aggregating proteoglycans in articular cartilage. J. Histochem. Cytochem. 32.193-201. Rosenberg, L. C., Pal, S. and Beale, R. (1973) Proteoglycans from bovine proximal humeral articular cartilage. J. Biol. Ckm. 24lI, 3681-3690. Sweet, M. B. E., Thonar, E. J. M. A. and Immehnan, A. R. (1977) Regional distribution of water and glycosaminoglycan in immature articular cartilage. Bioch-Biophys Acta. 500, 173-186. Zhu, W. B., Lai, W. M. and Mow, V. C. (1991) the density and
interactions
579
strength of proteoglycan-proteoglycan interaction sites in concentrated solutions. J. Biontech. 24, 1007-1018. Zhu, W. B. and Mow, V. C. (1990) Viscometric properties of proteoglycan solutions at physiological concentrations. In Biomechanics ofDiarthro&al Joints (Edited by Mow, V. C., Ratcliffe, A. and Woo, S. L. Y.), pp. 313-344. Springer, New York. Zhy W. B., Mow V. C., Koob, T. J. and Eyre, D. R. (1993) Viscoelastic shear properties of articular cartilage and the effects of glycosidase treatment. J. orthop. Rex 11, 771-781.