Women are at greater risk of tearing their knee anterior cruciate ligament (ACL) than men participating in similar athletic activities. There is currently no conclusive explanation for this disparity; however, as ACL injuries in women have been linked with estrogen fluctuations during the menstrual cycle, one hypothesis is that estrogen has a direct detrimental effect on knee ligament mechanical properties. This study investigated the influence of estrogen and its receptors (ERα and ERβ) on knee ligament mechanical properties. This was achieved by testing the viscoelastic and tensile mechanical properties of knee medial collateral ligaments (MCL) and ACLs from: 1) male Sprague-Dawley rats treated with either estrogen (17α-ethynylestradiol; 0.03 mg/kg) or an ERα-specific agonist (propyl pyrazole triol; 2 mg/kg), and 2) female mice with a null mutation of the gene encoding for ERβ. Estrogen treatment had no significant effects on the viscoelastic or tensile mechanical properties of the rat MCL or ACL. Similarly, pharmacological stimulation of ERα using a selective agonist in rats and genetic modulation of ERβ by null mutation of its gene in mice did not influence MCL or ACL properties. These data indicate that estrogen does not have a major direct effect on ligament mechanical properties. Energies for the prevention of the disproportionately high rate of knee ligament injuries in women may be better spent focusing on more established and modifiable risk factors, such as abnormalities in neuromuscular control about the knee.
- anterior cruciate ligament
- sex differences
- medial collateral ligament
- neuromuscular control
women are at two to eight times greater risk of tearing their knee anterior cruciate ligament (ACL) than men participating in similar athletic activities (1, 2, 17). There is currently no conclusive explanation for this disparity; however, differences in sex-specific hormones have been shown to play a role (3, 37, 49, 55, 56). In particular, estrogen fluctuations during the menstrual cycle have been linked to ACL injuries in women (3, 37, 49, 55, 56), with preliminary data reporting that significantly more ACL injuries than expected occur during the midcycle surge of estrogen (ovulatory phase) than during phases where estrogen concentrations are low (follicular and luteal phases) (55). Although other investigators have found ACL injury risk to be heightened during alternative phases of the menstrual cycle (3, 37, 49), the link between the menstrual cycle and injury has raised questions regarding direct detrimental effects of estrogen on knee ligament mechanical properties and strength. These questions have been further stimulated by recent findings of the presence of estrogen receptors (ERs) in ligaments (30, 43–45) and the response of fibroblasts (28, 29, 31, 59, 60) and ligaments (48) to estrogen.
Estrogen effects are mediated by two ligand-activated transcription factors called ERs: ERα (14, 15) and ERβ (26). Both ER subtypes are widely expressed, where they have distinct biological functions. The actions of estrogen on cellular proliferation seem to require ERα, whereas the effect of estrogen on cellular differentiation may be mediated principally by ERβ (7). Given these differing roles, the function of each ER subtype is important to determine when investigating the biological effect of estrogen on a particular tissue. In terms of ligaments, transcripts for ERs have been identified in mRNA isolated from rabbit and human knee medial collateral ligaments (MCLs) and ACLs (44), and ERs have been identified via immunohistochemistry in human ACL fibroblasts (30) and ovine (45) and rodent (43) ACLs, although the relative expression of each ER subtype has not been confirmed in knee ligaments.
The identification of ERs in ligament fibroblasts suggests that estrogen may directly modulate fibroblast functioning. Indeed, studies have shown that fibroblasts derived from knee ligaments are responsive to estrogen (28, 29, 31, 59, 60), indicating that their ERs are functional. Although findings have been contradictory with reports of both increased (28, 29) and decreased (31, 59, 60) fibroblastic collagen production in response to physiological levels of estrogen, the data indicate that estrogen can directly influence knee ligament fibroblast activity. As fibroblasts maintain ligaments and are responsible for both collagen production and degradation, estrogen may directly influence knee ligament mechanical properties.
Few studies have specifically studied the direct influence of estrogen on knee ligament mechanical properties with those that have provided conflicting results. Slauterbeck et al. (48) showed that postovariectomy estrogen treatment for 30 days reduced ACL ultimate force in rabbits. In contrast, Räsänen and Messner (42) found 5 mo of estrogen treatment in normal growing rabbits to have no effect on MCL ultimate force or elastic modulus; however, estrogen treatment did significantly reduce MCL cross-sectional area. Most recently, Strickland et al. (52) found estrogen to have no influence on knee MCL or ACL mechanical properties in sheep. Given these contrasting results, a need persists to establish the direct effect of estrogen on knee ligament mechanical properties.
The aim of this study was to further investigate the influence of estrogen on knee ligament mechanical properties and investigate the contribution of each ER. This was achieved using a multidirectional approach by 1) assessing ligaments from male rats treated with either estrogen (nonspecific ER agonist), an ERα-specific agonist, or a vehicle solution; and 2) assessing ligaments from female mice with and without a null mutation of the gene encoding for ERβ.
MATERIALS AND METHODS
Estrogen and ERα effects were investigated in 24 male Sprague-Dawley rats purchased at 6 wk of age from Harlan Sprague Dawley (Indianapolis, IN). Male rats were selected to minimize the effects of endogenous estrogen production. ERβ effects were investigated in female ERβ homozygous mutant (ERβ−/−) mice (BERKO) with a C57BL/6 background (N = 10; gift from P. Chambon, MD, Institut de Génétique et de Biologie Moléculaire et Cellulaire, France). These animals were generated by inbreeding of heterozygotes for ERβ disruption (ERβ+/−), as described by Dupont et al. (11). Control mice consisted of female wild-type (ERβ+/+) littermates (WT; N = 10). Progeny genotype was determined from tail-derived DNA by polymerase chain reaction amplification using established primers (11). Animals in both studies were maintained under standardized environmental conditions with ad libitum access to standard mouse or rat chow and water. All procedures were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee of Indiana University.
To investigate estrogen and ERα effects, rats were acclimatized until 8 wk of age and randomly divided into three intervention groups (N = 8/group): 1) estrogen treated (EE2); 2) propyl pyrazole triol treated (PPT); and 3) vehicle-treated controls (VEH). The EE2 group was treated via oral gavage with 17α-ethynylestradiol (0.03 mg/kg; Sigma-Aldrich, St. Louis, MO), which is a synthetic estrogen that is used in oral contraceptive formulations. The dose was chosen as it has been shown to correct the body weight and skeletal effects of estrogen deficiency in rats (12), whereas higher doses have been found to be detrimental to development in young male rats (53). The PPT group was treated via subcutaneous injection with propyl pyrazole triol [4-propyl-1,3,5-Tris(4-hydroxyphenyl)pyrazole, 2 mg/kg in 50% DMSO-50% Dulbecco's phosphate-buffered saline (PBS); Sigma-Aldrich]. PPT is an ERα-specific agonist that has a binding affinity that is 410-fold greater for ERα than ERβ (51), and the chosen dose of PPT corrects the bone loss caused by estrogen deficiency in rats (19). The VEH group received via an oral gavage vehicle solution (0.25 ml of 20% hydroxypropyl-β-cyclodextrin; Sigma-Aldrich). All rats were treated 7 days/wk for 5 consecutive weeks and killed at 13 wk of age. To investigate ERβ effects, mice were aged until 20 wk before being killed. There was no specific intervention in these animals.
Immediately after death, the right hindlimb was removed from each animal at the hip joint, placed intact in a plastic bag, and stored at −80°C. Postmortem storage by freezing does not change ligament properties during destructive mechanical testing (57). On the day of mechanical testing, the hindlimbs were allowed to thaw to room temperature in PBS. Dissection of extraneous tissue exposed the knee joint, and, with further careful dissection, femur-ACL-tibia and femur-MCL-tibia complexes were created. Because rats were concurrently being used to investigate estrogen effects on the skeleton, only one knee joint was available from these animals for ligament testing. To permit testing of both the MCL and ACL in the same knee in these animals, a Dremel drill (Robert Bosch Tool, Mount Prospect, IL) with attached diamond-embedded wafer blade (Super Flex Diamond Disc; Miltex, York, PA) was used to cut the femur and tibia sagittally. This was not performed in the mouse knees, as there were sufficient numbers to test the MCL and ACL in different animals. At all times, the ligament and insertion sites were kept hydrated with PBS.
MCL thickness and width were measured optically, and the MCL area was estimated using an elliptical geometry (40). Geometry of the ACL was not determined due to its dual bundles and intimate position within the knee joint. The femoral and tibial portions of the femur-ACL-tibia and femur-MCL-tibia complexes were placed in a custom-built testing jig, and the knee was positioned at 90° and 0° flexion for ACL and MCL testing, respectively (Fig. 1, A and B). The femoral and tibial portions were embedded in Wood's low-melting-point metal (Bismuth alloy LMA-117; Small Parts, Miami Lakes, FL) for fixation, and the jig was coupled to an electromagnetic material-testing device (TestBench 200 N ELF LM-1; EnduraTEC Systems Group, Bose, Minnetonka, MN). A 50-N load cell was used to measure forces during testing of rat ligaments, which resulted in a force and displacement resolution of 0.01 N and 0.001 mm, respectively. A 5-lb (22 N) load cell was used to measure forces during testing of mouse ligaments, which resulted in a force and displacement resolution of 0.001 N and 0.001 mm, respectively. A preload of 0.05 N was applied to each ligament, and the ligaments were preconditioned in displacement control by cyclically loading at 1 Hz for 10 cycles to ∼1% strain to rectify any alteration in low-load properties of the ligament due to deep freezing (57). The ligaments were subsequently unloaded and allowed to recover for 10 min before testing of viscoelastic properties.
Ligament viscoelastic properties.
Viscoelastic properties were determined via stress relaxation tests. After reestablishment of preload (0.05 N), ligaments were stretched to ∼5% strain and held for 100 s. To achieve this strain, rat MCLs and ACLs were lengthened 0.32 and 0.12 mm, respectively, whereas mouse MCLs and ACLs were lengthened 0.05 and 0.03 mm, respectively. A strain magnitude of 5% is below the threshold for structural damage in ligaments (39), and testing for 100 s enables accurate estimation of the viscoelastic behavior of ligaments (34). Loading was ramped at a strain rate of ∼10%/s, and data were collected at 20 Hz during testing. Force-time data were plotted on both linear-linear and log-log scales. The percent force reduction from the peak force was calculated from linear-linear plots [(peak force − final force)/peak force × 100] (Fig. 2A). Log-log scales were curve fitted with a power law, tn (Fig. 2B). The dimensionless parameter n, which is the slope on a log-log plot, was considered as the “rate” of relaxation (34, 38).
Ligament tensile mechanical properties.
After viscoelastic testing, ligaments were unloaded and allowed to rest for 1,000 s (16–17 min) while being kept hydrated with PBS. Experiments subjecting ligaments to displacements <5% tissue strain have shown that a recovery time equal to 10 times the test time allows the return to the initial preload gauge length (38). Following tissue recovery, preload (0.05 N) was reestablished, and the ligaments were pulled to tensile failure in displacement control at a strain rate of ∼10%/s (Fig. 1, C and D). To achieve this strain rate, rat MCLs and ACLs were lengthened at 0.8 and 0.5 mm/s, respectively, whereas mouse MCLs and ACLs were lengthened at 0.5 and 0.3 mm/s, respectively. During testing, force and displacement data were collected at 100 Hz. From the force-displacement curves, the mechanical properties of ultimate force and stiffness were obtained (Fig. 2C). Ultimate force data from the MCLs were divided by MCL area to obtain ultimate stress.
Statistical comparisons were performed with the Statistical Package for Social Sciences (SPSS 6.1.1; Norusis/SPSS, Chicago, IL), with a level of significance set at 0.05. Differences between VEH, PPT, and EE2 rat groups for body weight, viscoelastic properties, and ultimate stress were assessed by using one-way ANOVAs followed by Fisher's protected least significant difference test for pairwise comparisons. Analyses of covariance were used to assess for rat group differences in ligament ultimate force and stiffness with body weight as the covariate. Differences between WT and BERKO mice were assessed using unpaired t-tests. All data are presented as means ± SD.
Body weight did not differ between rat groups at baseline (P > 0.05). However, after 5 wk of intervention, the groups differed significantly (P = 0.02), with the PPT group (304.5 ± 18.2 g) being lighter than both the EE2 (333.4 ± 33.4 g; P = 0.04) and VEH (342.3 ± 23.0 g; P = 0.01) groups. The reduction in weight gain in the PPG group is consistent with feminization of male rats, similar to that produced by high-dose estrogen (53). Genotype in the mice did not influence body weight (P = 0.92), with WT and BERKO mice weighing 19.2 ± 2.6 and 19.1 ± 1.7 g, respectively.
There were no significant group differences in the peak force obtained or percent force reduction that occurred during the stress relaxation tests in either rat (Table 1) or mouse (Table 2) ligaments. The rate of stress relaxation in rat MCLs and ACLs was not influenced by intervention group (P = 0.27 and 0.12, respectively; Fig. 3A). Similarly, mouse genotype did not influence the rate of stress relaxation in MCLs and ACLs (P = 0.10 and 0.13, respectively; Fig. 3B).
Tensile mechanical properties.
Estrogen and ERα-specific agonist treatment did not influence knee ligament strength in rats, with no differences being found between the VEH, EE2, and PPT groups for MCL (P = 0.62) or ACL (P = 0.30) ultimate force (Fig. 4A). Similarly, no differences were found between groups for knee ligament stiffness (all P = 0.29–0.41; Fig. 4B) or MCL ultimate stress (all P = 0.44; Fig. 4C).
Null mutation of the gene encoding for ERβ did not influence knee ligament strength in mice, with no differences being found between WT and BERKO groups for MCL (P = 0.49) or ACL (P = 0.29) ultimate force (Fig. 5A). Similarly, no genotype differences were for knee ligament stiffness (all P = 0.29–0.52; Fig. 5B) or MCL ultimate stress (all P = 0.39; Fig. 5C).
A possible cause for the disproportionately high rate of knee ligament injuries in women is a direct detrimental effect of estrogen on knee ligament mechanical properties and strength. This study used controlled animal models to investigate the influence of estrogen and its receptors on knee ligament mechanical properties. Estrogen treatment had no significant effects on the viscoelastic or tensile mechanical properties of the rat MCL or ACL. Similarly, pharmacological stimulation of ERα using a selective agonist in rats and genetic modulation of ERβ by null mutation of its gene in mice did not influence MCL or ACL viscoelastic or tensile mechanical properties. These data indicate that estrogen and its receptors do not directly influence the mechanical properties of these knee ligaments.
The current findings indicate that direct ligament effects of estrogen may not be responsible for the disproportionately high rate of knee ligament injuries in female athletes. However, statistical power and the potential of committing a Type II statistical error needs to be explored in any study that observes a nonsignificant finding. Power calculations for ultimate force indicated that there was 80% probability of detecting a mouse genotype effect of 10 and 21% for the MCL and ACL, respectively. Similarly, there was 80% probability of detecting a difference of 15–24% in ligament ultimate force between VEH, PPT, and EE2 rats. Although post hoc power calculations need to be interpreted cautiously (24), these minimal detectable differences are considered to be clinically relevant and indicate that we addressed our research question with sufficient statistical power.
It is possible that estrogen and its receptors had effects smaller than could be statistically detected. Previous studies have shown estrogen to negatively impact the collagen content of a variety of connective tissues (16, 47). As collagen fulfills the major load-bearing role in ligaments, these studies suggest that estrogen may also influence the mechanical properties of ligaments. However, this has been confirmed by only one study, which showed that estrogen treatment following ovariectomy reduced ACL ultimate force in rabbits by 11.3% (48). This small effect was generated in animals exposed to high levels of estrogen, with the serum estrogen level being two times higher than normal and equivalent to that experienced by rabbits during pregnancy. Such an effect was not observed in the present study using more physiological levels of estrogen, nor has it been confirmed by previous studies (42, 52). Also, the removal of estrogen does not result in the opposite effect of an increase in ligament mechanical strength and subsequent decrease in injury risk (50, 54). These data indicate that any direct detrimental effect of physiological levels of estrogen on knee ligament mechanical properties is, at most, small.
The present study was unable to find any effect of estrogen or its receptors on knee ligament mechanical properties with continuous exposure to intervention. Rats were treated daily for 5 wk with either estrogen or an ERα-selective agonist, and mice had life-long disruption of ERβ. These exposures are somewhat unphysiological, as they deviate from the clinical scenario where ligaments and their ERs are exposed to phasic levels of estrogen during the menstrual cycle. It is possible that estrogen may produce a different effect when introduced intermittently as opposed to continuously (41) and that intermittent doses may significantly alter ligament mechanical properties within a matter of days, as collagen turnover can occur rapidly (3–5%/day in adult rat skin) (27). However, normal physiological fluctuations in estrogen during the estrous cycle have not been found to alter the failure properties of the rat ACL (43), and clinically assessed knee ligament mechanical properties (laxity and stiffness) do not differ across the different phases of the menstrual cycle (5, 6, 8, 25). As such, we believe that intermittent intervention would not have altered the findings in the present study and that the absence of an effect, despite constant exposure to intervention, actually strengthens the conclusions. The latter theory is based on the finding that ligament fibroblasts exhibit a dose-dependent response to estrogen, with increasing estrogen concentrations resulting in progressive declines in fibroblast proliferation (31, 59, 60). This indicates that continuous intervention may magnify estrogen effects, resulting in a larger and more detectable effect than when estrogen levels vary phasically with intermittent intervention.
A possible further limitation of our study is the multidirectional approach used to investigate estrogen and ER effects. The effect of combined ERα and ERβ stimulation, isolated ERα stimulation, and no ER stimulation was investigated by treating rats with EE2, PPT, and VEH, respectively, whereas isolated ERβ effects were investigated by comparing mice with and without a null mutation in the gene encoding ERβ. We did not investigate isolated ERβ effects in rats, as no selective agonist of this receptor was available at the time of investigation. It is possible to use diarylproprionitrile for the stimulation of ERβ, as this reagent is more potent for ERβ than for ERα; however, it is not selective (18). Consequently, it was elected to investigate isolated ERβ effects in genetically modified mice. This multidirectional approach to address the research question may be viewed as a limitation; however, it is actually a strength. Irrespective of the model used, no effect of estrogen or the ERs was found. Further investigation of ERβ effects in rats and ERα effects in genetically modified mice may more completely answer the research question; however, we do not believe that data from these studies would alter our conclusions.
As direct ligament effects of estrogen may not explain the disproportionately high rate of knee ligament injuries in female athletes, other mechanisms need to be considered. The risk of knee ligament injury is multifactorial and results from a combination of extrinsic and intrinsic factors (4). Women have been shown to exhibit a number of intrinsic risk factors that have been linked with ACL injuries. These have been adequately reviewed elsewhere (10) and include anatomic features, such as a narrower intercondylar notch (46), smaller ACL size (9), and altered lower extremity alignment (32), as well as neuromotor features. The latter are of particular interest as they are potentially modifiable and may be influenced by the differing levels of estrogen during each phase of the menstrual cycle. Women have been found to utilize different muscular activation patterns compared with men during activity (23). This results in significant sex differences in mechanics, with women displaying increased hip internal rotation and a larger valgus at the knee (13, 21, 35, 58). This “dynamic valgus” knee motion has prospectively been shown to be a significant risk factor for ACL injury in female athletes (22), suggesting that clinical approaches targeting neuromuscular control issues may be useful in preventing ACL injuries in women. Supporting this, training programs for lower extremity neuromuscular control have been found to significantly decrease knee injury rates (20, 33, 36).
In summary, this study found in two separate animal models that estrogen and its receptors did not significantly influence the mechanical properties of the knee ligaments. While an effect smaller than could be statistically detected may exist, the clinical relevance of such an effect is questionable. Combined with previous studies, these data indicate that estrogen does not have a major direct effect on ligament mechanical properties. Energies for the prevention of the disproportionately high rate of knee ligament injuries in women may be better spent focusing on more established and modifiable risk factors, such as abnormalities in neuromuscular control about the knee.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 by American Physiological Society