Spatial Variation in Sympathetic Influences on the Vasculature of the Synovium and Medial Collateral Ligament of the Rabbit Knee Joint


  • Jason J. McDougall,

    1. McCaig Centre for Joint Injury & Arthritis Research, Department of Surgery, University of Calgary, T2N 4N1 Canada
    Search for more papers by this author
  • William R. Ferrell,

    1. McCaig Centre for Joint Injury & Arthritis Research, Department of Surgery, University of Calgary, T2N 4N1 Canada
    2. Institute of Biomedical & Life Sciences, University of Glasgow, Glasgow G12 8QQ, UK
    Search for more papers by this author
  • Robert C. Bray

    1. McCaig Centre for Joint Injury & Arthritis Research, Department of Surgery, University of Calgary, T2N 4N1 Canada
    Search for more papers by this author


  • 1Laser Doppler perfusion imaging was used to assess the role of the sympathetic nervous system in the control of blood flow to the medial collateral ligament and capsule (synovium and overlying fibrous tissues) of the rabbit knee joint.
  • 2Electrical stimulation of the saphenous nerve (width 1 ms; amplitude 20V; 1–30 Hz) produced a frequency-dependent vasoconstriction of knee joint vasculature. The response was maximal at 30 Hz and gave the greatest fall in perfusion at the femoral insertion of the ligament (by 33.8 ± 7.4%, mean ±s.e.m.; n= 5–6) and the smallest decrease at the tibial insertion of the ligament (by 10.6 ± 2.9%).
  • 3Topical application of phentolamine (10−6 mol) had no significant effect on basal knee joint blood flow. However, it abolished the nerve-mediated constrictor responses in all regions of the medial collateral ligament and synovium at all frequencies.
  • 4Topical administration of adrenaline (10−14 to 10−7 mol) caused a dose-dependent decrease in knee joint blood flow with the highest dose producing > 75% reduction in perfusion at all areas.
  • 5There was no evidence of a reactive hyperaemia in the 5 min following a 5 min period of femoral artery occlusion. Artificial manipulation of arterial blood pressure by intravenous infusion or withdrawal of blood caused a proportional change in ligament and synovial blood flow. These observations may indicate a lack of autoregulation in the joint and its exclusion from baroreflex modulation.
  • 6These results suggest a potential role for the sympathetic nervous system in the control of knee joint blood flow. Neuromodulation of ligament perfusion appears to predominate at the femoral insertion and this could prove to have functional significance.

Ligaments of load-bearing joints are essential to maintain stability and provide dynamic control of joint movement. Following trauma, ligaments show poor healing responses with consequent increased joint laxity, and osteoarthritis invariably ensues (Jacobsen, 1977; Gollehon, Torzilli & Warren, 1987; Hirshman, Daniel & Miyasaka, 1990; Brandt, 1993). Effective healing responses are dependent on an adequate blood supply to provide the mediators necessary for tissue repair and to maintain joint homeostasis during injurious episodes. The medial collateral ligament has an extensive network of blood vessels which mainly occupy the medial epiligament, a layer which constitutes the superficial 180 μm of the tissue (Bray, Fisher & Frank, 1990; Chowdhury, Matyas & Frank, 1991). Surgical insult to the medial collateral ligament causes an increase in blood flow to the tissue (Bray, Butterwick, Doschak & Tyberg, 1996) which probably serves as a healing response to the injured ligament. Paradoxically, other studies have shown that injury-induced medial collateral ligament hyperaemia correlates with a degeneration in the mechanical properties of the ligament (Doschak, Bray & Tyberg, 1994) suggesting that poor adaptive responses to joint injury may be related to ligament perfusion. To date, most studies have centred on the regulation of synovial blood flow but at present very little is known of the factors which regulate ligament blood flow.

Knee joints are known to be richly innervated with both myelinated and unmyelinated nerve fibres (Samuel, 1952; Skoglund, 1956), the density of which are highest in the ligaments compared with the capsule (Marinozzi, Ferrante, Gaudio, Ricci & Amenta, 1991). Cobbold & Lewis (1956a) showed that the sympathetic nerve supply in dog stifle joints exert a tonic constrictor tone on the joint vasculature. Electrical stimulation of the nerve supply to the knee is known to cause a sympathetically mediated decrease in capsular blood flow in dogs (Cobbold & Lewis, 1956a), cats (Khoshbaten & Ferrell, 1990), rats (Lam & Ferrell, 1993; Karimian, McDougall & Ferrell, 1995) and rabbits (Najafipour & Ferrell, 1993a, b). The nature of these constrictor responses has been examined further by exposure of the knee joint vasculature to adrenergic drugs. Initial experiments indicated that administration of noradrenaline produced a vasoconstriction which was greater in magnitude than the fall in blood flow associated with an equivalent dose of adrenaline (Cobbold & Lewis, 1956b). Although not commented upon at the time, this phenomenon was probably due to noradrenaline acting on only α-adrenoceptors producing a pure vasoconstriction, whereas adrenaline acted on both α- and β-adrenoceptors, the latter producing a dilator response which partially offsets the α-adrenoceptor vasoconstriction. This differential effect of adrenoceptor agonists was confirmed in human synovium where articular adrenergic responses were mediated by both α- and β-adrenoceptors (Dick, Jubb, Buchanan, Williamson, Whaley & Porter, 1971). More recently, Najafipour & Ferrell (1993a, b) have made an extensive study of the relative contributions of α- and β-adrenoceptors in the control of rabbit synovial perfusion. It was found that pretreatment of the knee joint with the α1-adrenoceptor antagonists prazosin or YM-12617 had no effect on the nerve-mediated constrictor response, whereas rauwloscine, an α2-adrenoceptor antagonist, converted the constrictor response into a dilator response. Furthermore, the dilator response produced by the β-agonist isoprenaline was substantially reduced by the β1antagonist atenolol; however, the selective β2-antagonist ICI118551 had no effect on the isoprenaline dilatation. The conclusion drawn from these data is that the nerve-mediated adrenergic control of rabbit synovial blood flow is carried out predominantly via α2- and β1-adrenoceptors with the constrictor response mediated by α-adrenoceptors being by far the dominant effect.

The present study set out to examine the role of the sympathetic nervous system in the control of knee joint blood flow by comparing its effects on the medial collateral ligament with those in the synovium. Previous studies on this subject area have used techniques which typically give perfusion values to either the tissue as a whole (e.g. microspheres) or to a very small, discrete area (e.g. laser Doppler flowmetry). The present study used laser Doppler perfusion imaging (LDI) to enable monitoring of joint tissue blood flow to a designated area and thereby assess differential flow responses in different regions within a similar time frame. Additional experiments investigated the ability of the joint to autoregulate in response to changes in systemic arterial blood pressure and following a period of articular ischaemia.


Seventeen adult female New Zealand White rabbits (4.2–5.2 kg) were premedicated with acepromazine maleate (25 mg ml−1, 0.2 ml i.v.) and then deeply anaesthetized by an intraperitoneal injection of urethane (1 g kg−1). Following abolition of a flexor withdrawal reflex, the right knee of the animal was shaved and the rabbit placed in a supine position. Rectal body temperature was maintained in the 35–37 °C range by a thermostatically controlled blanket (American Pharmaseal Company, California, USA) placed under the animal. All experimental protocols had prior approval by the University of Calgary Animal Care Committee and were in accordance with the Canadian Council for Animal Care guidelines.

Surgical procedures

The carotid artery was cannulated with a heparinized saline-filled cannula (PE-90, 0.86 mm internal diameter; Clay Adams, Sparks, MD, USA) which in turn was attached to a pressure transducer (Elcomatic EM752, Neilston, UK) to allow monitoring of arterial blood pressure using Codas software (Dataq Instruments Inc., Akron, Ohio, USA). The jugular vein was cannulated in a similar manner (Clay Adams PE-60, 0.76 mm internal diameter) to allow removal and infusion of heparinized blood as part of the auto-regulation study. A longitudinal incision was then made in the medial aspect of the knee joint and the overlying skin reflected to expose an area from the medial collateral ligament to the distal edge of the patellar tendon. All fascial and aponeurotic tissues covering the regions of interest were removed to expose the medial collateral ligament and the medial aspect of the joint capsule (synovium and overlying fibrous tissues). Warmed physiological saline (0.9% NaCl) was regularly applied to the joint to prevent desiccation of the uncovered tissues. A 1.5 cm section of the femoral artery was dissected free and a piece of moistened surgical thread hooked round it to facilitate placement of an arterial clip during the occlusion experiments. The saphenous nerve, which innervates the medial segment of the rabbit knee, was isolated at the midthigh and sectioned as proximally as possible. The distal end of the nerve was placed over a pair of silver bipolar stimulating electrodes and the exposed nerve stump covered with mineral oil to prevent tissue dehydration.

Blood flow measurement

Relative changes in articular perfusion were measured using a standardized protocol (Karimian et al. 1995) involving a laser Doppler perfusion imager (Moor Instruments Ltd, Axminster, UK). This technique has been validated for use in ligament blood flow determination studies (Bray, Forrester, McDougall, Damji & Ferrell, 1996) and provides a two-dimensional representation of tissue perfusion. The imager scanner head was placed 19 cm above the exposed joint and a scan region chosen which included the medial collateral ligament, medial capsule and medial patellar tendon. Since more than 90 % of capillaries in the joint capsule are located in the synovium (Knight & Levick, 1983), perfusion values generated in this region were considered a good representation of synovial blood flow. Scans typically took 30 s to complete and were taken immediately before (control) and during experimental manipulation (test). The various manipulations produced haemodynamic effects which were typically of much longer duration than the scan time, thus the possibility of the response being only partially captured was avoided. At the end of the experiment, the rabbit was killed by an overdose of sodium pentobarbitone (360 mg intracardiac) and a further scan was obtained. The resultant ‘biological zero’ for that animal was typically 5–10% of the control scan and this value was subtracted from each image before any calculations were carried out. In previous work it was demonstrated that the imager can penetrate through the capsule to detect changes in synovial perfusion as it is possible to detect a rapid and localized reduction in flux over the capsule following ultra-articular administration of adrenaline (Karimian et al. 1995). Furthermore, the composition of the fibrous components of the capsule has some similarity to dermal fibrous tissues which present little by way of an optical barrier to light at the wavelength (635 nm) of the laser used (Anderson & Parrish, 1981).

Experimental protocol

At the beginning of each experiment, warmed (37 °C) 0.9% saline was administered topically to the exposed joint for 20 min to confirm that perfusion was stable. Experiments would then follow one of three different protocols: (1) pressure autoregulation; (2) test for reactive hyperaemia followed by nerve stimulation under normal conditions and in the presence of phentolamine; and (3) the effect of phentolamine on basal blood flow and then adrenaline dose-response experiments. Pressure autoregulation and reactive hyperaemia experiments are described below as autoregulation studies, whilst the other experiments are described in the sympathetic vasomodulation studies section.

Autoregulation studies

The ability of the rabbit knee joint to autoregulate was examined by two different experiments. The first experiment, performed to test for reactive hyperaemia, involved occlusion of the femoral artery as it emerges from the abdomen into the upper thigh by clipping the vessel with an arterial clip. The clip remained in situ for 5 min and was then removed. LDI scans were performed immediately after and at the fourth minute of the period of occlusion. Scans were then taken immediately and 5 min after removal of the clip to test for any signs of reactive hyperaemia. Pressure autoregulation was examined in a different series of four animals by either removal or infusion of blood from the animal via the jugular cannula. This procedure was repeated a few times in each animal to give a range of different blood pressures. Removal of 20 ml of blood from a normotensive animal (101 ± 5 mmHg; n= 4) rendered it hypotensive (75 ± 6 mmHg; n= 9), whilst infusion of 20 ml of heparinized blood caused systemic hypertension (118 ± 4 mmHg; n= 9). Only when the blood pressure remained steady at its new level was a scan taken. Simultaneous tissue perfusion and blood pressure measurements were obtained and regression analysis was performed on the resultant data.

Sympathetic vasomodulation studies

In a group of six animals, the saphenous nerve was electrically stimulated by a Harvard Stimulator (Model 6012; Harvard Apparatus, Canada), with the delay set at 1 ms, pulse width at 1 ms and voltage at 20 V, parameters known to be supramaximal for eliciting vasoconstrictor responses (Ferrell & Najafipour, 1992). The nerve was stimulated at various frequencies ranging from 1–30 Hz so that a frequency–response profile could be generated. The α-adrenoceptor antagonist phentolamine (10−6 mol) was then applied topically to the joint and the nerve stimulation protocol repeated.

In another group of six rabbits, the presence of a tonic sympathetic vasoconstrictor influence on the knee joint vasculature was assessed by topical application of phentolamine (10−6 mol). Scans were taken immediately after drug administration and 10 min later before the knee was washed with warmed 0.9% saline for 40 min. This time period was found to be sufficient to allow recovery of the joint vasculature from the effects of phentolamine, as evidenced by nerve-mediated constrictor responses after saline washing being of the same order of magnitude as pre-phentolamine treatment (see Results). After confirming that the antagonistic effects of phentolamine had worn off, adrenaline (10−14 to 10−7 mol) was topically applied to the knee with each dose of adrenaline being administered consecutively to produce a cumulative dose–response curve.


Phentolamine hydrochloride and urethane were supplied by Sigma; Atrovet (acepromazine maleate) by Ayerst Laboratories (Montreal, Canada); adrenaline hydrochloride by Epiclor (Calgary, Canada); sodium pentobarbitone by MTC Pharmaceuticals (Ontario, Canada). Where appropriate, drugs were dissolved in 0.9 % saline to give the necessary concentrations.

Perfusion and statistical analyses

LDI images were analysed using MoorLDI software (Moor Instruments Ltd) using a standardized protocol (Karimian et al. 1995). For each image, four different analysis areas were chosen which corresponded to the femoral insertion, midsubstance and tibial insertion points of the medial collateral ligament, as well as the synovium (Fig. 1). The orientation of the knee and the imager meant that scans proceeded in a rostrocaudal direction, and thus the femoral insertion was reached early on during each scan. It could be argued that subsynovial structures, e.g. the infrapatellar fat pad, could be contributing to the synovial perfusion signal. For this reason, the synovial region of interest was restricted to the medial gutter of the synovium adjacent to the medial collateral ligament. This region of synovium only covers bone which is too dense to transmit a perfusion signal and avascular articular cartilage. The dose–response and frequency–response curves were expressed as a percentage change in perfusion units compared with control, whereas raw perfusion values were used to illustrate the occlusion and pressure autoregulation data. Individual data points are presented as means ±s.e.m. Data were tested for normality by a modified Shapiro–Wilk test and data conforming to a Gaussian distribution were analysed using parametric statistics (i.e. one-way ANOVA, a general linear model of variance (unbalanced two-way ANOVA) or Student's t test), whilst non-Gaussian data were tested using non-parametric statistical tools (i.e. Mood's median test). Data sets were considered significantly different if P < 0.05.

Figure 1.

A diagrammatic representation of the different sites in the knee joint from which perfusion measurements were taken

The measurement sites were the femoral insertion (FI), midsubstance (MID) and tibial insertion (TI) of the medial collateral ligament, in addition to a medial portion of gutter synovium (SYN). For clarity, the gap between the femur and tibia has been exaggerated to illustrate the delineation of the bones. In vivo the condyles are in close proximity to each other and the tissues underlying the gutter synovium are bone and articular cartilage.


The medial collateral ligament of the rabbit knee joint was found to be well perfused with greatest flow occurring at the tibial insertion. Comparison of each of the sites showed that synovial blood flow (91 ± 8.1 perfusion units) was significantly less (P < 0.015; Student's paired t test; n= 16) than all areas of the medial collateral ligament (120 ± 17.3, 119 ± 12.6 and 138 ± 20.5 perfusion units for the femoral insertion, midsubstance and tibial insertion, respectively).

Autoregulation of the rabbit knee

As shown in Fig. 2, occlusion of the femoral artery caused blood flow to fall by 95 ± 1.3% (femoral insertion), 95 ± 2.8% (midsubstance), 89 ± 2.8% (tibial insertion) and 93 ± 2.5% (synovium) compared with pre-occlusion values. In the medial collateral ligament this fall in blood flow was still apparent 4 min into the occlusion, however, synovial blood flow was starting to recover such that blood flow at 4 min of occlusion was significantly higher (P < 0.05) than synovial flux values obtained immediately after the artery was clamped. This suggests that the synovium derives its blood supply from other sources in addition to the femoral artery. Removal of the arterial clip caused ligament and synovial perfusion to rise again, however, blood flow was unable to rise above basal levels even 5 min after reestablishing femoral blood flow. Mean arterial blood pressure remained stable throughout the occlusion experiments (Table 1).

Figure 2.

Effect of femoral artery occlusion on knee joint blood flow compared with control

Data shown for: A, femoral insertion; B, midsubstance; C, tibial insertion; and D, synovium. The recovery period is also shown following removal of the arterial clip. Means ±s.e.m.; n= 7 at each site.

Table 1. Mean arterial blood pressure (MABP) in response to femoral occlusion, nerve stimulation and administration of phentolamine or adrenaline
 MAPB (mmHg)
  1. Values are expressed as means ±s.e.m.

  Control98 ± 5
  Occlusion98 ± 5
  4 min of occlusion98 ± 6
  Recovery93 ± 6
  5 min of recovery93 ± 5
Phentolamine (mol)
  Control88 ± 6
  10−687 ± 6
Stimulation frequency (Hz)
  Control99 ± 5
  1100 ± 5
  299 ± 5
  598 ± 4
  10103 ± 4
  30102 ± 5
Adrenaline (mol)
  Control71 ± 6
  10−1360 ± 6
  10−1168 ± 7
  10967 ± 8
  10−763 ± 7

Alteration of mean arterial pressure caused knee joint blood flow to change proportionately. A plot of LDI flux against mean arterial pressure (Fig. 3) shows a significant, linear relationship between the two variables at the femoral insertion (correlation coefficient r= 0.85), medial collateral ligament midsubstance (r= 0.90), tibial insertion (r= 0.75) and synovium (r= 0.91); P < 0.05 and n= 22 at each site.

Figure 3.

Relationship between mean arterial blood pressure (MABP) and LDI-determined blood flow to the knee joint

Data shown for: A, femoral insertion; B, midsubstance; C, tibial insertion; and D, synovium. A significant linear relationship exists between the two variables at the femoral insertion (r= 0.85), ligament midsubstance (r= 0.90), tibial insertion (r= 0.75) and synovium (r= 0.91). P < 0.05 and n= 22 at each site.

Effect of nerve stimulation and phentolamine

Blood vessels in the femoral insertion, midsubstance and tibial insertion of the medial collateral ligament, and the synovium showed a frequency-dependent vasoconstriction (P < 0.05 repeated measures one-way ANOVA or Mood's median test; n= 5–6) following electrical stimulation of the saphenous nerve (Fig. 4). Maximal vasoconstriction occurred during 30 Hz stimulation with the greatest fall in blood flow occurring in the femoral insertion (by 33.8 ± 7.4%) and the smallest fall in blood flow appearing in the tibial insertion (by 10.6 ± 2.9%). The constrictor responses occurring in the synovium and medial collateral ligament midsubstance at this frequency were found to be similar in magnitude (18.5 ± 6.0 and 18.2 ± 5.6%, respectively).

Figure 4.

The response of the femoral insertion (A), ligament midsubstance (B), tibial insertion (C) and synovium (D) to electrical stimulation of the saphenous nerve

Data show control (□) and phentolamine-treated (▴) knee joints. In control joints, the vasoconstriction was frequency dependent (P < 0.05; one-way ANOVA or Mood's median test) at all sites. The constrictor response was abolished following phentolamine administration. Means ±s.e.m.; n= 5–6 for each site.

Application of phentolamine to the rabbit knee had no significant effect on systemic blood pressure (Table 1) or on basal blood flow compared with control (P > 0.6, Student's paired t test – from 10 min data). Although phentolamine is an α1α2-adrenoceptor antagonist and would therefore be expected to lower peripheral resistance with consequential lowering of arterial blood pressure, topical application permits a high local concentration to be achieved with limited systemic effects. Using values generated at the femoral insertion (which was the first area scanned), a Student's unpaired t test showed that the change in perfusion obtained immediately after phentolamine administration was not significantly different (P= 0.27) from a saline control performed at the same time point. However, phentolamine did alter the constrictor effect of saphenous nerve stimulation (Fig. 4). One-way ANOVA showed that prior treatment of the joint with phentolamine abolished the nerve-mediated constrictor response, even at 30 Hz (P= 0.984, 0.743, 0.478 and 0.831 for femoral insertion, medial collateral ligament midsubstance, tibial insertion and synovium, respectively).

Blood pressure remained stable throughout the nerve simulation experiments (Table 1), which was as expected since the nerve was centrally transected and therefore could not induce centrally mediated reflex hypertension.

Effects of adrenaline

Before adrenaline experiments could be carried out it was necessary to check whether the antagonistic effects of phentolamine had worn off. After 40 min of saline washing, nerve stimulation (at 30 Hz) caused a 36.8 ± 5.7 % reduction in blood flow to the femoral insertion of the medial collateral ligament. This was found to be not significantly different from the 33.8 ± 7.4% reduction in perfusion observed in non-phentolamine treated knees (P= 0.85, Student's two-tailed unpaired t test; n= 4–6).

A highly significant (P < 0.001 one-way ANOVA; n=6) dose-dependent constrictor effect was observed in response to topical application of adrenaline onto the joint surface (Fig. 5), which was significantly different from a saline control performed over a similar time period (P < 0.001; general linear model of variance). Maximal vasoconstriction occurred with the 10−7 mol dose of adrenaline, causing blood flow to decrease by 73.4 ± 11.4% (femoral insertion), 80.0 ± 7.1 % (ligament midsubstance), 77.0 ± 14.7% (tibial insertion) and 76.6 ± 6.5% (synovium). A general linear model of variance showed that the dose–response curves to adrenaline at the different sites were not significantly different from each other (P > 0.21).

Figure 5.

Cumulative dose–response curves to topical adrenaline administration in the femoral insertion (A), ligament midsubstance (B), tibial insertion (C) and synovium (D)

Data show administraton of adrenaline (□) and corresponding saline time controls (•). In each case the adrenaline responses were significantly different from the saline time control (P < 0.001; general linear model of variance). Means ±s.e.m.; n= 6 for each site.

At the doses used, topical application of adrenaline had no effect on mean arterial blood pressure (Table 1).


Ligaments have long been thought of as inert structures whose primary function is to provide stability to a joint. The present study has demonstrated that the medial collateral ligament is relatively well perfused and, for the first time, that its vasculature may be modulated by the sympathetic nervous system. By using LDI it was possible to monitor blood flow in three distinct regions of the medial collateral ligament almost concurrently, viz. the femoral insertion, ligament midsubstance and tibial insertion, as well as a medial segment of the joint capsule. In addition to revealing a heterogeneity in perfusion to different areas of the knee, LDI was also able to show that these regions exhibit differences in responsiveness to various physiological and pharmacological interventions. It could be argued that the spatial variation in ligament blood flow may be due to a variation in depth distribution of vessels in the tissue generating a differential LDI signal. However, LDI measures the dominant flow in a tissue irrespective of the location of the predominant blood vessels provided they are found within the working range of the laser. In the medial collateral ligament, the majority of vessels occur in the superficial 180 μm, which is well within the penetrative range of the laser source used in the imager. Furthermore, there is currently no histological evidence of a difference in density or depth distribution of vessels in different regions of the medial collateral ligament.

Knee joint autoregulation

The arterial occlusion experiments demonstrated that, under the prevailing experimental conditions, there was an absence of reactive hyperaemia following interruption of blood flow to the knee joint. Although the scan took 30 s to complete, the region of the femoral insertion was scanned within 10 s and yet there was still no evidence of reactive hyperaemia, suggesting that a modest response was not missed in these experiments. Furthermore, artificial alteration of systemic blood pressure caused a proportional change in knee joint blood flow suggesting that articular blood vessels may not possess myogenic activity and are therefore unable to change their resistance in response to changing transmural stretch. This lack of pressure autoregulation in the rabbit knee means it must rely heavily on extrinsic neural and hormonal influences to control its blood supply. This finding has profound repercussions in the diseased joint where the effectiveness of these neurohumoral mediators may be altered. In this instance, if the joint is unable to vasoregulate effectively tissue degeneration may ensue thereby exacerbating the disease state (McDougall, Karimian & Ferrell, 1994, 1995). The close correlation between mean arterial pressure and articular blood flow also suggests, in the anaesthetized animal at least, that the rabbit knee joint does not take part in the cardiopulmonary or baroreceptor reflex alteration of peripheral resistance. Changes in mean arterial pressure typically result in a reflex alteration in total peripheral resistance via altered firing patterns of sympathetic nerves so as to counteract the potentially detrimental effects of the original blood pressure change.

Sympathetically mediated responses

In addition to the vasoconstrictor effect on rabbit synovial blood vessels seen in other studies (Najafipour & Perrell, 1993a,b), electrical stimulation of the joint nerve supply causes a vasoconstriction of medial collateral ligament blood vessels. This is not surprising since the majority of nerve fibres in the medial collateral ligament occur in the epiligament and are often associated with blood vessels (McDougall, Bray & Sharkey, 1997). The articular nerves do not appear to exert a tonic constrictor effect on knee joint vessels, at least in the anaesthetized animal, as evidenced by the lack of an increase in basal blood flow following topical application of the α-adrenoceptor antagonist phentolamine. However, this finding does not preclude the possibility of sympathetic vasoconstrictor fibres being tonically active in the alert animal. Phentolamine did abolish nerve-mediated vasoconstriction, suggesting that this response was purely adrenergic in nature.

Spatial mapping of blood flow across the ligament showed that nerve stimulation caused the femoral insertion to produce the greatest fall in perfusion. This phenomenon was not due to a higher density of postjunctional α-adrenoceptors at the femoral insertion since cumulative administration of adrenaline caused blood flow to decrease by similar amounts at all of the knee sites tested. It could be argued that antidromic nerve stimulation is causing the release of vasodilator mediators from sensory nerve fibres which may be offsetting the constrictor responses in different areas. A study in the rat knee joint, however, has shown that depletion of articular sensory nerves by capsaicin treatment did not augment sympathetic vasoconstriction (Karimian et al. 1995). This suggests that the form of stimulation employed in these experiments does not lead to the release of sensory neuropeptides which could in theory attenuate the sympathetic vasoconstrictor response. One reason for the larger constriction at the femoral insertion may be due to a higher number of sympathetic fibres innervating this region. Alternatively, there could be a lower density of presynaptic α-adrenoceptors on the nerves innervating the femoral insertion compared with the fibres innervating other regions of the joint. Noradrenaline is known to act on sympathetic nerve prejunctional α-adrenoceptors by an ultrashort inhibitory feedback mechanism so as to modulate its own release (Gillespie, 1980). Hence, the potential for a negative feedback system regulating noradrenaline release at the femoral insertion would be less compared with the rest of the knee. Another possibility is that sympathetic fibres differentially release other vasoconstrictors in addition to noradrenaline at the various sites.

The observation that there is greater sympathetic vasomodulation at the femoral insertion may lie in its uniqueness in both chemical composition and mechanical behaviour of the ligament. It has been shown that the highest levels of strain typically encountered in the rabbit medial collateral ligament occur at the femoral insertion (Lam et al. 1995) and biochemical studies have shown this to be the region with highest water content (Frank, McDonald, Lieber & Sabitson, 1988). Since high water content makes ligaments more compliant (Chimich, Shrive, Frank, Marchuk & Bray, 1992), the femoral insertion has adapted biochemically to be able to withstand the relatively large forces placed upon it. Elevated water levels could be generated and maintained at the femoral insertion by high hydrostatic forces acting in the region. However, if the water content was to exceed a critical level then the ligament would become unduly viscoelastic with resultant joint instability. A regulatory mechanism must therefore exist to prevent excessive water build-up in the medial collateral ligament which could be mediated in part by the sympathetic nervous system. Sympathetic vasoconstriction would reduce capillary hydrostatic pressure preferentially at the femoral insertion, the net effect of which would be increased absorption of ligament interstitial fluid.

In conclusion, the present results indicate that the sympathetic nervous system can exert a profound effect on the vasculature of the rabbit knee joint and the magnitude of these responses is heterogeneous. This variation in medial collateral ligament vasoregulation could be related to differences in the functional properties of ligaments and may suggest new directions for the investigation of ligament healing.


This work was funded by the Arthritis Society of Canada and the Alberta Heritage Foundation for Medical Research (AHFMR). J. J. McDougall is an AHFMR and Canadian MRC postdoctoral fellow and R. C. Bray is an AHFMR clinical investigator. W. R. Ferrell was supported by The Wellcome Trust/Canadian MRC and AHFMR travel grants.