Multiscale Poroviscoelastic Compressive Properties of Mouse Supraspinatus Tendons Are Altered in Young and Aged Mice

One of the most common causes of shoulder pain and disability in the aging population is disorders of the rotator cuff, which has been reported to affect greater than 50% of the aged population [1]. Unfortunately, one single cause for this degenerative process has been difficult to identify [2–4]. External compression of the rotator cuff tendons, particularly in the most commonly injured supraspinatus tendon, is a factor thought to contribute to tendon degeneration in the clinic [5–7]. Several factors associated with age, such as rotator cuff weakness and poor posture, can cause increased compression of the tendon and may ultimately lead to tendon degeneration, particularly when coupled with repetitive use and normal age-related changes. However, only a few studies have investigated the compressive response of tendons and none of these studies have focused on the rotator cuff tendons [8,9].

Tendons are composed primarily of tenocytes, water, and an extracellular matrix (ECM) comprised predominantly of collagen type I organized in a hierarchical manner, minor collagens, elastin, and proteoglycans with their associated glycosaminoglycan (GAG) chains [10–14]. GAGs have net negative charge, which leads to increased tissue hydration due to osmotic and electrostatic interactions [15–17]. This unique structure dictates tendon's mechanical function, specifically in the supraspinatus tendon, which is one of the most heterogeneous and complex tendons in the human body. Tendon is a compliant, anisotropic material, which exhibits nonlinear biomechanical behavior and several time-dependent properties including viscoelastic stress relaxation, hysteresis, and creep [18–20]. Tendons are also thought to exhibit time-dependent poroelastic behavior when localized compression causes fluid flow through the tendon's ECM to become the rate-limiting process that determines creep, stress relaxation, and dynamic biomechanical behavior [21,22]. In fact, we recently adapted our high-frequency AFM-based rheology technique to characterize the regional poroelastic behavior of tendon, and found that tendons exhibit a complex dynamic response to compression with both viscoelastic and poroelastic regimes over the large frequency range studied [23].

While it is well known that injury and disease can alter mechanical function, studies on age-related changes in macroscale tendon function have been inconclusive, warranting further investigation. Given that the prevalence of tendon injuries and tendon disease increases with age, it is likely that age-related changes in tendon function are more complicated than that revealed by traditional macroscale mechanical evaluation. Furthermore, dynamic loads are more often reported in vivo and may be more clinically relevant to tendon function, thus necessitating the need for more sophisticated multiscale mechanical evaluations. Several recent studies have shown that changes in dynamic responses can occur without changes in macroscale mechanical properties, suggesting the importance of changes at the molecular and microscale in early response to loading [9,24–26]. In addition, several studies have found that the dynamic viscoelastic mechanical responses in tension are altered in aged tendons, but it is still unclear how these properties are altered in compression [15,25,27,28].

Therefore, the objective of this study was to measure the multiscale dynamic response of young, mature, and aged mouse supraspinatus tendons to compression, and to utilize the poroelastic response in particular to better understand how fluid pressurization and flow is altered with age. We did this by characterizing the compressive mechanical, viscoelastic, and poroelastic properties of young, mature, and aged tendons at the tissue level using macroscopic compression testing and at the nanoscale using our high-frequency AFM-based rheology method. We hypothesized that aged tendons would have inferior macroscale and nanoscale poroelastic properties.

This study used C57BL/6J male mice at 1 (young, n = 12), 4 (mature, n = 14), or 18 months (aged, n = 14) of age obtained from the Aging Rodent Colony at the National Institute on Aging. The ages were chosen for consistency with previous structure–function studies of maturity and aging in mouse tendon [15,17,27,29,30] as well as due to their mouse-to-human equivalency relationship to clinically relevant ages for age-related rotator cuff injury [31]. While we did not observe any differences between groups, the level of physical activity was not quantified in any of the animal groups and the reduced activity of aged mice should be considered in the future. In addition, body weight was higher in the aged group than the other two groups (young: 22.9 ± 2.1 g, mature: 29.5 ± 1.8 g, aged: 35.7 ± 2.3 g).

Supraspinatus tendon and muscle along with the humerus were dissected free from the shoulder, as described previously [32,33]. Excess soft tissue was gently removed from the tendons, but the bony insertion and myotendinous junction were kept intact. Samples were then fixed to either a metal disk for macroscale compression testing (Fig. 1) or a custom stage for wide bandwidth rheological testing (Fig. 2) using cyanoacrylate glue. All samples were kept hydrated with phosphate-buffered saline throughout experiments and tested only on the articular surface of the tendon.

Verhoeff's stain lines were applied to the tissue at 1 mm increments (Fig. 1(a)), to define the insertion site (0–1 mm) and the midsubstance (1–2 mm). Tendons were then subjected to a dynamic compression protocol using a custom-built rounded 0.5 mm diameter indenter. The protocol began with an initial compressive preload to 0.025 g with a 300 s hold (Fig. 1(b)). The tissue was then subjected to consecutive ramp-and-hold indentations to 2%, 4%, 8%, 12%, 16%, 20%, and 24% cumulative compressive strain at 0.1% strain/second followed by 1000 s of relaxation. Our maximum cumulative compression value was chosen based on previous studies in mouse tendon [8], as well as unpublished pilot studies indicating that this maximum provided multiple points within the linear region of the equilibrium stress–strain response in order to calculate equilibrium modulus. Following the 4% step and subsequent relaxation, a 2% dynamic strain-amplitude sinusoidal frequency sweep spanning from 0.001–2.0 Hz was applied. We then calculated the compressive equilibrium modulus (the slope of a linear fit to the equilibrium stress–strain points) as well as the magnitude |E*| (stress amplitude divided by the strain amplitude) and phase angle δ (between the measured stress and the applied dynamic strain) of the dynamic modulus (Fig. 1(b)). A higher phase angle would represent more energy dissipation, while a lower phase angle would represent more elastic storage. Two-way analyses of variance with effects for age, location, and the interaction between age and location were then used to analyze the data statistically, followed by Bonferroni-corrected post hoc comparisons. Significance was defined as p < 0.025 (solid bar) and a trend defined as p < 0.05 (dashed bar or hash (#) symbol).

For regional nanoscale evaluation, indentations were performed at fourteen regions along the length of the tendon from bone to muscle (Fig. 2(a)). The first region was defined as the closest indentation site to the humeral head, typically less than 100 μm from the bone, and this was done consistently for all samples. Samples were then indented every 50 μm for the first 200 μm and then every 100 μm until the junction with muscle. Using this procedure, there was variation in the number of regions between samples, but there was no relationship between the number of regions and any particular age group. Given the variance in the data, we estimated that approximately seven samples were necessary for sufficient power for statistical comparisons and therefore, we only considered regions where there were at least seven samples in each group, which were the first 14 regions in the present study. Indentations at each region were performed in the middle of the tissue (away from the glued edges) at nine locations within each region. Built-in top-view optics and translational stage were used to visualize positions of locations.

We indented supraspinatus samples on the articular surface of the tendon with ∼25 μm diameter polystyrene colloidal probe tips attached to tipless cantilevers having nominal spring constant K ∼ 7.4 N/m (Fig. 2(b)). Data were obtained over a wide frequency range (1 Hz to 10 kHz) using our custom rheology system coupled to the MFP-3D AFM (Asylum Res, CA), as described [34–36]. This system allows us to measure how the response of the tissue varies with loading rate, which is critical toward understanding injury mechanisms including those involving very high impact rate. In addition, the poroelastic (but not the viscoelastic) phase response of soft tissues to compressive deformation scales with the square of the characteristic length associated with the fluid-flow response to deformation (see below) and thus, AFM-based indentation at high frequencies can reveal biomechanical behavior very relevant to normal tissue loading of larger tendons by larger indenters (such as the subacromial bursa in the human rotator cuff).

The indentation loading profile (Fig. 2(c)) consisted of an initial load-controlled ramp-and-hold pre-indentation of approximately ∼3.5 μm followed by superimposed random binary sequence displacements of 8–12 nm, applied using previously established methods [23,34,35]. The contact between the indenter probe and the tissue was maintained for the duration of experiments. While we are indenting with microscale-sized probe tips, our study utilizes applied indentation displacements that are truly nanoscale (only ∼8–12 nm). As a result, our measurements interrogate tissue compressive strains that are nanoscale regarding depth dependence, and the corresponding nanoscale fluid flow profiles associated with these compressive strains; hence our use of “nanoscale” [34–37]. This is an important distinction and should be not confused with studies using nanoscale-sized probes. A discrete Fourier transform was used to obtain the fundamental frequency component of the force and displacement signals and we then obtained the magnitude and phase of the dynamic complex modulus from the measured force and the applied displacement as we have described extensively previously [34,35].

Tendons typically exhibit a dynamic phase response having two characteristic peaks in the frequency range measured here, but only the 2nd peak, which occurs at higher frequencies, is associated with fluid–solid poroelastic energy dissipation [23]. The poroelastic peak in the phase response (Fig. 2(d)) occurs at the frequency, $fp=(kEL/d2)$⁠, where k is the hydraulic permeability, E_L is the low frequency modulus (Fig. 2(d)), and γ is a constant associated with probe tip geometry. The probe size used in this study (25 μm) insures that the poroelastic phase peak is higher than the viscoelastic peak (e.g., see Fig. 2(d), [23], for the properties of these tendons), but lower than the resonance frequency of the piezo testing system. From the measured dynamic modulus at low and high frequency limits, we quantified the equilibrium (low-frequency, E_L) and instantaneous (high-frequency, E_H) moduli, the self-stiffening ratio (ratio of high to low frequency moduli [38]), and the angle and frequency of the peak viscoelastic (δ_v, f_v) and poroelastic (δ_p, f_p) phase angles (Fig. 2(d)). For pooled comparisons, the insertion site was defined as the first 300 μm from the first indentation and the midsubstance was defined as the next 400–1100 μm. Pooled comparisons were made using two-way analyses of variance, and with Bonferroni corrected post hoc t-tests where applicable (p < 0.025 significant (solid bar); p < 0.05 trend (dashed bar)).

In macroscale compressive testing, the equilibrium modulus was increased in the aged tendons compared to the young and mature tendons at both the insertion site and the midsubstance (Fig. 3(a)). The magnitude of the dynamic modulus was increased in the aged tendons compared to the mature tendons at all frequencies at the insertion site and at the lowest frequency at the midsubstance (Fig. 3(b)). The young tendons also had increased dynamic modulus magnitude when compared to the mature tendons at the insertion site, which reached significance at the highest frequency. The phase angle was increased in the young tendons compared to the mature and aged tendons at both low and high frequencies at the midsubstance, as well as compared to the mature tendons at the lowest frequency in the insertion site. No other significant differences were found in either region (Fig. 3(c)).

At the nanoscale, the magnitude of the low frequency modulus (E_L, Fig. 4(a)) was higher in the young and aged tendons compared to the mature tendons at the insertion site and in the aged and mature tendons compared to the young tendons at the midsubstance. There were no significant differences at either region in the high frequency modulus (E_H, Fig. 4(a)). Similarly, self-stiffening ratio (E_H/E_L) was no different between the three ages at either the insertion site or the midsubstance (Fig. 4(b)). Interestingly, poroelastic properties were significantly altered in the young and aged tendons compared to the mature tendons. The frequency of the poroelastic peak phase angle (f_p, depicted in Fig. 2(d)) was slightly lower in the aged tendons than the mature tendons at the insertion site (Fig. 5). At the midsubstance, the frequency of this poroelastic peak was higher in both young and aged tendons when compared to mature tendons. The phase angle at the peak was also higher in the young and aged tendons when compared to the mature tendons at both regions. The frequency of the viscoelastic peak phase angle (f_v, depicted in Fig. 2(d)) was slightly but significantly higher in the aged and young tendons when compared to the mature tendons at the midsubstance as well (Fig. 6). However, there were no significant differences in the phase angle of the viscoelastic peak between groups in either region.

Direct comparisons of parameter values between the midsubstance and the insertion site were made to detect regional differences within each age group (Fig. 7). The midsubstance of the mature tendons, when compared to the insertion site, had higher indentation modulus, higher low and high frequency moduli, higher viscoelastic and poroelastic peak phase angles, higher viscoelastic and lower poroelastic peak frequencies. Regional variation was lost in aged tendons, with no significant differences between insertion site and midsubstance in almost all parameters. In the young tendons, regional variation was lost in the poroelastic peak phase angle and frequency. Furthermore, the indentation modulus, as well as the low and high frequency moduli, was greater at the insertion site than the midsubstance, opposite to the mature tendons. However, there were no differences in regional variation in either of the viscoelastic peak parameters between groups.

In this study, the first to measure the regional compressive properties of mouse supraspinatus tendons, the equilibrium compressive modulus was found to be lower than the tensile modulus as expected and similar to other studies of tendon in compression [8]. Aged tendons were found to be stiffer than mature tendons in both equilibrium and dynamic loading conditions at the tissue level and at the nanoscale, contrary to our original hypothesis. As the supraspinatus is a major joint stabilizer, this stiffening could be associated with general joint stiffening or could suggest an accumulation of GAG; ongoing studies are investigating this possibility. Interestingly, self-stiffening with increased frequency of loading over the frequency range studied was not different between the three age groups, which would be expected with alterations in GAG content or with aging, as in cartilage [34,35]. However, there currently is not a consensus on alterations in GAG or proteoglycan content with aging [15,39–41]. We do know that GAGs and proteoglycans are important for the dynamic tissue response in tendon, and thus they likely contribute in compression as well. Differences between groups in the magnitude of the dynamic modulus (Figs. 3(b) and 4(a)), however, were present throughout the frequency range studied, suggesting that the mode by which mature tendon is able to self-stiffen in compression is likely similar in young and aged tendons even if the aged tendons were overall stiffer.

A number of regional differences in the nanoscale compressive properties were present between the insertion site and midsubstance of the mature tendons. We originally hypothesized that the insertion site of the mature tendon would be stiffer than the midsubstance at both hierarchical scales, as the insertion site of the supraspinatus is thought to experience higher compressive loads due to its anatomical positioning around the humeral head and because of the increased glycosaminoglycans in the fibrocartilage of the insertion site. Interestingly, while we did not see a dramatic difference in either equilibrium or dynamic properties at the macroscale, our nanoscale experiments demonstrated that midsubstance was significantly stiffer than the insertion site (Figs. 3, 4, and 7). It has been well established that bulk tissue properties and nanoscale properties can vary dramatically because of the dynamic responses occurring at multiple hierarchical scales, and we hypothesize that this may be the case in our study. Furthermore, we posit that specialization for the regional response to tensile loading is related to regional response to compressive loading [24,33]. Collagen fibrils at the insertion site must be flexible to both compress when wrapped around the humeral head during loading and re-organize during tensile loading via collagen re-alignment and crimp. Similarly, fibrils in the midsubstance are primarily required to slide against one another during tensile loading and provide strength and durability against levels of stress past the toe region of the stress–strain response.

In addition, the insertion site of the mature tendons also had higher poroelastic peak frequency and lower energy dissipation than that of the midsubstance, confirming again our previous results [23]. Based on our previous studies using finite element and analytical modeling [34–36], there is a direct relationship between the frequency of the poroelastic peak in the phase response, $fp$⁠, and the hydraulic permeability of the tissue, $fp=(kEL/d2)$⁠, where k is the hydraulic permeability, E_L is the low frequency modulus (Fig. 2(d)), and γ is a constant associated with probe tip geometry. Therefore, we infer that hydraulic permeability is also increased at the insertion site compared to the midsubstance. This combination may play specific roles in tendon homeostasis, allowing for nutrient influx to the cells of the three transitional tissues present while also maintaining fluid pressurization when this region is subjected to compressive and shear loading during shoulder motion. Interestingly, this regional heterogeneity, which we also suspect may be responsible in part for directed fluid flow along the length of the tendon, was lost in the young and aged tendons. This is not only likely due to regional alterations in structure and composition, as we will discuss below, but also alludes to the possibility of different maturation and aging processes in the two regions of the tendon [25,27].

Furthermore, young and aged tendons both had higher energy dissipation and hydraulic permeability than the mature tendons, as demonstrated by the changes in poroelastic peak phase angle and frequency, respectively, (e.g., see Fig. 5). Unlike cartilage, in which the many GAGs of the large proteoglycan aggrecan are known to restrict fluid flow and thereby regulate matrix poroelasticity [35,38], we hypothesize that collagen organization is a more likely candidate to govern the poroelastic response of normal mature tendons. In tendons, the resident small leucine-rich proteoglycans are less abundant overall and each have only one or a few GAG chains; therefore, compared to the packed collagen architecture, this much lower GAG density would offer far less resistance to fluid flow associated with local compression-induced hydrostatic pressure gradients [23]. Although mice are considered to be skeletally and sexually mature at 28 days following birth, fibril diameter size, density, and organization are well known to fluctuate dramatically throughout life. Following fibrillogenesis, which is completed around 1 month of age in the mouse, tendon fibril diameters shift modestly toward a population with higher fibril diameter as the mouse ages [15,17,42], and therefore fibril diameters are smallest at the young tendons and largest at the aged tendons. However, organization and density of collagen fibrils have been shown to decreased in both young and aged tendons [15,17,43], which could be responsible for the increased hydraulic permeability in both groups as found here (Fig. 5). We hypothesize that fibril spacing mat also be a major determinant of intrinsic hydraulic permeability, as several modeling studies have shown that fluid preferentially flows in the direction of the collagen fibrils rather than across them toward the outside of the tendon [21,44,45]. However, there are many other likely candidates for involvement in the poroelastic response of tendon, including cross-linking of the collagen fibrils and the presence of elastic fibers, which both appear to impact multiscale tendon mechanics [46–50].

The overall purpose of and the downstream biological consequences of increased hydraulic permeability unfortunately are unknown. We hypothesize that increased hydraulic permeability in young tendons may allow for increased communication between tenocytes along the length of the tendon via growth factors and signaling proteins, thus contributing to the maturation and appropriate cellular programming for assembly of the growing tendon structure. Alternatively, increased permeability in aged tendons could mean a decreased ability to resist compressive loading via fluid pressurization. This could result in direct loading of the collagen fibers and cells, particularly during high impact loading, promoting unknown downstream biological consequences. Tendinopathy samples have been reported to a contain cartilage-like phenotype, with rounded cells and GAG accumulation, that could be an effect of increased compressive damage [5–7,50,51] but this is also currently unknown. We must also consider the altered signaling response of aging tenocytes [51–54], which could further exacerbate tissue damage, leading to the development and/or progression of tendon damage. Finally, an increase in fluid flow or permeability could also indicate an increased ability of cytokines and/or growth factors to reach cells in the interior of the tendon, thereby potentially spreading any metabolic response from one tendon to the rest of the rotator cuff (cartilage, joint capsule, tendons, etc.) more easily. In the case of inflammation or matrix degradation, this would quickly devolve from single tendon injury to a multitissue joint degeneration as has been reported in several studies. However, this is beyond the scope of this study and future studies should investigate these potential mechanisms.

While this study was the first to directly investigate the multiscale compressive properties of mouse supraspinatus tendons, there are a number of limitations and considerations that should be addressed in future studies. First, indenting and measuring regions very close to the bone proved to be difficult due to the geometry of the supraspinatus tendon wrapping around the humeral head. Future studies should focus on determining a better method to probe closer to and possibly within the transition from tendon through fibrocartilage into bone as this region remains a major interest in tendon research. In addition, these studies only investigate the articular surface of the tissue. Since the bursal side is in direct contact with the subacromial bursa and therefore may be subjected to additional shear and compressive forces, future studies should investigate properties on the bursal surface of the tendon. This study is particularly focused on maturation and aging and as such used only postnatal mouse tendons. However, given the role of proteoglycans and GAGs in the developmental regulation of fibril assembly as well as the growing fibril structure, investigating poroelasticity during postnatal development could provide insight into the relative contributions of various structural elements to the overall tissue response. Finally, our study draws conclusions on tendon fluid flow, yet there is no determination yet of flow direction. This should be confirmed in follow-up studies using computational modeling and, potentially, the incorporation of strain-generated potential measurements that directly confirm flow direction [55], to understand directionality and mechanics of fluid flow.

Overall, we found that many of the properties governing fluid flow and energy dissipation are altered during maturation and aging. We hypothesize that this could be a contributing factor to the development of age-related tendinopathy. However, several questions still remain and should be answered in follow-up studies. While we hypothesize and provide evidence that fibril organization and morphology are likely the cause of age-related alteration in tendon hydraulic permeability, direct evidence is still needed. Altering the composition and structure of tendon in mouse models is challenging due to importance of collagen type I in almost every major organ system, but we believe we may be able to gain insight in this respect through the use of targeted transgenics. Nevertheless, these studies provide insight into the regional compressive response of and indicate that altered dynamic compressive properties and energy dissipation in aging tendons may be a major contributor to overall tendon degeneration.

The authors thank Dr. Eliot Frank and Han-Hwa Hung for their expertise and assistance.

• Division of Civil, Mechanical and Manufacturing Innovation (Grant No. CMMI-1536233).

• National Institute on Aging (Grant No. F32-AG052284).