1. Introduction

The mechanical properties of skeletal muscle, from cellular to organ scales, govern function and are determined by both the tissue constituents, e.g., muscle fibers (myofibers), adipose tissue, and extracellular matrix (ECM), and their complex structural arrangements [1]. Diseases such as muscular dystrophies modify both the constituents and structure of muscle tissue, resulting in scale-dependent variations in mechanical properties [2,3]. Characterizing mechanical properties on the cellular scale using techniques such as atomic force microscopy (AFM) [4–7], optical tweezers [8,9], and micropipette aspiration [10,11], has provided insights into the development of skeletal muscle disorders. On the macro-scale, the advent of ultrasound elastography [12] and magnetic resonance elastography [13] has enabled the characterization of variations in mechanical properties over entire organs. Importantly, skeletal muscle diseases often manifest on the intermediate scale, i.e., that between cells and organs. For example, in Duchenne muscular dystrophy, mutations in the dystrophin gene alter force transmission between the ECM and cytoskeleton, resulting in progressive muscle degeneration [14], with localized regions of muscle necrosis and increased fibrofatty tissue [15–17]. In another example, the muscular dystrophy dysferlinopathy, caused by mutations in the dysferlin gene, results in a unique muscle pathology characterized by loss of function and muscle wasting, inflammation, altered lipid metabolism, accumulation of lipid droplets in slow twitch myofibers and, in later stages, pronounced replacement of skeletal muscle with adipose tissue [18–22].

However, it remains challenging to accurately measure the mechanical properties of skeletal muscle on the intermediate scale in three-dimensions (3-D). For example, AFM is limited to surface measurements, and micropipette aspiration and optical trapping are limited to characterization of isolated cells. This makes it challenging to characterize micro-scale mechanical properties in 3-D environments that are more representative of the conditions found in vivo [23–25]. On the macro-scale, whilst ultrasound elastography and magnetic resonance elastography can provide 3-D characterization, they are typically limited to spatial resolution greater than several hundred micrometers [26], restricting quantification of the mechanical properties of muscle constituents [27].

Optical coherence elastography (OCE) can potentially bridge the gap in mechanical characterization of skeletal muscles by quantifying the 3-D micro-scale mechanical properties of skeletal muscles to a depth of 1–2 mm. OCE refers to a range of techniques that use optical coherence tomography (OCT) to map the mechanical properties of tissue into an image, known as a micro-elastogram [28]. OCE comprises three key steps: (1) deforming the tissue by applying a mechanical load, (2) measuring the resulting deformation using OCT, and (3) relating deformation to a mechanical property using a mechanical model [29–31]. OCE imaging of skeletal muscle has been demonstrated using both transverse wave OCE [32] and compression OCE [33]. An advantage of compression OCE is that applying compressive loading over the entire sample surface provides a practical means to rapidly image over relatively large, typically 10–45 mm, fields of view in the en face plane [34]. A previous study demonstrated that compression OCE can provide mechanical contrast in skeletal muscle tissue between control mice and the mdx mouse strain [33], a widely used pre-clinical animal model for Duchenne muscular dystrophy [35]. However, the results in that study were not quantitative, as only the local strain in the tissue was imaged. Quantifying an intrinsic mechanical property, such as elasticity, which describes how an object deforms due to an applied force (i.e., a measure of tissue stiffness), is critical to enabling comparisons between different muscles.

Quantitative micro-elastography (QME) is a variant of compression OCE that uses a pre-characterized compliant layer, placed between the sample and the imaging window, to map tangent modulus (equivalent to Young’s modulus in a linear elastic material) throughout a sample volume [36]. Importantly, many tissues exhibit a linear elastic response when they are subjected to strains up to 0.1, implying that careful control of the mechanical load allows elasticity to be determined using QME [37]. QME has been demonstrated in a range of applications, notably breast tumor margin assessment [38,39] and cell mechanobiology [40–42]. However, small tissues, including murine skeletal muscles, are difficult to manipulate and their uneven geometries make it challenging to achieve even contact between the muscle and imaging window. This leads to both uneven preload strains at different locations in the same sample and inconsistent preload strains between different samples. Large variations in preload strain can result in artificial mechanical heterogeneity, making quantification and subsequent longitudinal comparison difficult [43].

In this study, we use QME to characterize the 3-D micro-scale elasticity of freshly excised murine skeletal muscle. We address the challenges of muscle size and uneven surface by encapsulating the muscles in 3-D gelatin methacryloyl (GelMA) hydrogels to generate samples with improved compatibility with the QME imaging system. This is an extension of previous work that encapsulated silk fibroin tissue scaffolds in silicone [44]. Using this approach, we demonstrate the capability of QME to quantify the elasticity of excised quadriceps. In addition, we present a comparison of aging and disease in quadriceps between normal wild-type (WT) C57BL/6J mice and the dysferlin-deficient BLAJ mouse model for dysferlinopathy [35], at 3, 10, and 24 months of age (n = 3 per group). In OCT images of the 24-month-old dysferlin-deficient BLAJ quadriceps relative to the WT quadriceps at all ages, we observe a marked increase in disordered striations in the muscle, with substantial replacement of skeletal muscle with adipose tissue, as expected [18,19]. In the corresponding quantitative micro-elastograms of the 24-month-old BLAJ quadriceps relative to the WT quadriceps at all ages, we observe a decrease in elasticity of the individual myofibers, and a decrease in elasticity of the muscle over the field of view. We provide a quantitative comparison of each group and observe that BLAJ quadriceps become 77% softer (mean elasticity), with a 60% reduction in mechanical heterogeneity (standard deviation) compared to normal 24-month-old WT muscle. These results indicate that QME can quantify the micro-scale mechanical properties of skeletal muscle and evaluate variations in elasticity between normal and dysferlin-deficient quadriceps.

2. Methods

2.1 Animals

This study used male C57BL/6J (the WT parental control strain) and dysferlin-deficient BLAJ mice, aged 3, 10, and 24 months (sample size, N = 18; n = 3 in each group). Mice were maintained at the Preclinical Animal Facility at The University of Western Australia, housed individually in cages with food and water ad libitum, maintained in a 12-hour light/dark regime at 20–22°C. All experiments were approved by the Animal Ethics and Experimentation Committee of The University of Western Australia (RA/3/100/1436), in accordance with guidelines of the National Health and Medical Research Council of Australia. Mice were euthanized via cervical dislocation under anesthetic (2% volume-to-volume (v/v) Attane isoflurane). Quadriceps muscles were excised and placed into oxygenated Krebs on ice for transport and subsequent encapsulation and imaging, which occurred within four hours of excision.

2.2 Encapsulating skeletal muscles in hydrogels

Accurate QME measurements require uniform compression throughout the sample, which is challenging to achieve in small skeletal muscles with uneven surfaces. QME uses a preload compressive strain to ensure uniform contact between the imaging window, compliant layer, sample, and motorized translation stage, described further in Section 2.3. This is a problem as the muscle is subjected to spatially varying compression that, when coupled with the non-linear stress-strain behavior of the muscle, can introduce artificial mechanical heterogeneity, making quantification and longitudinal comparison difficult. To address this, we encapsulated the muscles in GelMA hydrogels to create samples with flat surfaces and controlled thicknesses. Encapsulation enables uniform contact to be achieved at low preload strains and allows a consistent preload strain to be applied to each sample, described further in Section 2.4. Encapsulating muscles in a hydrogel also ensures that the muscles remain hydrated throughout the scan acquisition.

GelMA, a gelatin-based hydrogel material [45–47], was used to encapsulate the muscles. GelMA, like other hydrogels, is advantageous due to its permeability to oxygen and nutrients, which help to maintain tissue viability [48–53]. GelMA has highly tunable elasticity, controlled by exposure to ultraviolet (UV) light and the concentration of GelMA and photoinitiator solution, and thus can be fabricated to have similar elasticity to soft biological tissues (∼1–200 kPa) [46]. Furthermore, GelMA is well-suited to muscle encapsulation as it does not damage the muscle tissue, due to cytotoxicity or high temperatures, which is the case for other hydrogels [54,55].

GelMA was synthesized by methacrylation of gelatin as described previously [45]. Briefly, 10 g of gelatin powder was added to 100 mL of phosphate buffered saline (PBS; Gibco, USA) in a 500 mL round bottom flask and heated to 60°C in an oil bath until fully dissolved. Methacrylic anhydride (8 mL) was then added dropwise to the solution under vigorous stirring, and the reaction mixture was maintained at 60°C for two hours. PBS (100 mL warmed to 60°C) was added to the solution and kept at 60°C for a further 30 minutes. The resulting solution was dialyzed using a dialysis tube (#132676, Spectrum Laboratories, New Zealand) in deionized water at 40°C for seven days, after which, the solution was lyophilized and stored at -20°C until further use. Freeze-dried GelMA was processed through a desiccator vacuum and then dissolved in PBS (Gibco, USA) at 37°C to achieve a 7.5% weight-to-volume (w/v) GelMA solution. Light-sensitive photoinitiator Irgacure-2959 (Sigma Aldrich, USA), dissolved in ethanol, was added at a concentration of 0.1% (w/v). Polystyrene bead (0.2 µm) stock solution (2.6% w/v; #07304-15, Polysciences, USA) was added to the GelMA solution at a concentration of 3% (v/v) to provide optical scattering within the hydrogel. The resulting prepolymer solution was then enclosed in aluminum foil to reduce exposure to light and stored at 4°C overnight. On the day of use, GelMA was heated to and maintained at 37°C. The multi-layer muscle encapsulation method is illustrated in Fig. 1. Molds for encapsulation, illustrated in Figs. 1(f) and 1(g), comprised acrylic walls with no base or top, with internal dimensions of 15 × 15 × 7 mm (xyz). Molds were 3-D printed (Form 2, FormLabs, USA) using clear resin (FLGPCL04; FormLabs, USA). After printing, uncured resin was removed using a Form Wash (FormLabs, USA) containing isopropyl alcohol (Sigma-Aldrich, USA), and exposed to 405 nm UV light for 30 minutes at 60°C using a Form Cure (FormLabs, USA). The mold was then removed from the support struts and sanded to create a smooth surface.

 

Fig. 1. Multi-layer muscle encapsulation method. (a) GelMA solution (65 µL) was pipetted on a DCDMS-coated glass slide, (b) a 15 × 15 mm (xy) coverslip (DCDMS-coated on surface in contact with GelMA) was placed on top and (c) the solution was polymerized with ultraviolet (UV) light to form a flat 15 × 15 × 0.29 mm (xyz) layer of hydrogel. (d) The resulting GelMA layer, attached to the coverslip, was inverted, and (e) the muscle was placed on top. (f) A mold was placed around the muscle and GelMA layer, and a small amount of GelMA was added and polymerized to hold the muscle in place. The mold was filled with GelMA, covered with a DCDMS-coated glass slide, and (g) polymerized. (h) The encapsulated muscle was removed from the mold, glass slides, and coverslip for QME imaging.

Download Full Size | PDF

Prior to muscle encapsulation, glass coverslips (15 × 15 × 0.15 mm (xyz); BBAD01500150#A*, Menzel-Gläser, Germany) and glass slides were coated with dimethyldichlorosilane (DCDMS; Sigma Aldrich, USA) to stop GelMA from sticking to them after polymerization. GelMA solution (65 µL) was pipetted on a glass slide and a coverslip was placed on top (Fig. 1(a) and 1(b)). GelMA was polymerized by exposure to 365 nm UV light for 60 seconds using a transilluminator (MLB-16, Maestrogen, Taiwan), forming a 15 × 15 × 0.29 mm (xyz) layer of hydrogel (Fig. 1(c)). The coverslip, with GelMA attached, was inverted and a muscle was placed on the GelMA layer at a neutral length, with a mold placed around the gel (Figs. 1(d)–1(f)). A small amount of GelMA was added and polymerized for 45 seconds to hold the muscle in place. The mold was then filled with GelMA, covered with a glass slide, and polymerized for six minutes (Fig. 1(g)). Finally, the hydrogel containing the muscle was removed from the mold, glass slides, and coverslip for QME imaging (Fig. 1(h)).

2.3 Quantitative micro-elastography

QME measurements were performed using a fiber-based spectral-domain OCT system (TEL220, Thorlabs Inc., USA). A schematic of the experiment setup is shown in Fig. 2. The light source is a superluminescent diode with a mean wavelength of 1300 nm and a spectral bandwidth of 170 nm (full-width at half maximum (FWHM)). The measured OCT axial resolution in air is 4.8 µm (FWHM). In this study, QME was performed in a common-path configuration where the interface between the imaging window and the compliant silicone layer acts as the OCT reference reflection. The compliant layer is used to measure the stress applied at the sample surface, as described in the next paragraph. The scan lens (LSM03, Thorlabs) has a numerical aperture of 0.063, a measured lateral resolution of 7.2 µm (FWHM) and a working distance of 25.1 mm. A 0.1 preload strain was applied to each sample using a motorized translation stage (MLJ050, Thorlabs Inc., USA) to ensure uniform contact between the sample, compliant layer, and the imaging window (Edmund Optics, USA). Micro-scale compression was applied to the sample using an annular piezoelectric actuator (Piezomechanik GmbH, Germany) fixed to the imaging window through which the optical beam illuminates the sample. The annular piezoelectric actuator has an aperture of 65 mm and a maximum stroke of 10 µm. The 75 mm diameter imaging window, fixed to the actuator, transfers the compressive load to the sample. QME volumes comprised 1,000 A-scans per B-scan, and 1,000 B-scans over a 10 × 10 mm (xy) lateral region, resulting in a lateral sampling density of 10 µm per voxel. In all cases, the actuator was driven in a quasi-static regime by a 10 Hz square wave, collinearly with the imaging beam and synchronized with the acquisition of OCT B-scans. Two B-scans were acquired for each y-location such that alternate B-scans are acquired at different compression levels, referred to as loaded and unloaded levels. Local axial displacement, u_z, is calculated from the phase difference between B-scans acquired at the same y-location [56]. To extend the displacement dynamic range, axial displacement is computed from the unwrapped phase difference using a 3-D phase unwrapping algorithm described previously [57]. Two B-scan pairs were acquired at each y-location and averaged together to alleviate the effect of system noise, resulting in a total volume acquisition time of approximately 200 seconds. Local axial strain, ε_zz, is calculated from the gradient of axial displacement using weighted least squares linear regression over a sliding window (Δz of 100 µm), equivalent to 29 pixels in air [58]. In this approach, the weights are determined by the OCT signal-to-noise ratio (SNR) corresponding to each displacement measurement [59]. The axial resolution using this approach is approximately Δz/√2 [60]. The resulting elasticity system resolution is approximately 10 × 10 × 72 µm (xyz). Both the displacement sensitivity and elasticity sensitivity of the QME system used in this study have been rigorously characterized in a previous study by our group. The measured displacement sensitivity is approximately 3.6 nm at an OCT SNR of 35 dB [61]. For the system parameters used in this study, the elasticity system sensitivity is approximately 1 kPa at an OCT SNR of 35 dB [61].

To estimate local stress at the sample surface, we use a compliant silicone layer with a total thickness of approximately 500 µm (P7676 2:1:0.3 (crosslinker:catalyst:silicone oil ratio)), (Wacker, Germany) placed between the sample surface and imaging window [36]. AK50 silicone oil (Wacker, Germany) is applied to lubricate the stress layer-imaging window interface to reduce friction. The compliant layer comprises two layers: a non-scattering layer bonded to a scattering layer, each with a thickness of approximately 250 µm. Measuring layer strain by detecting the interface between the two layers reduces artefacts in the preload strain from poor contact between the silicone layer and GelMA hydrogel, improving QME accuracy and repeatability [62]. The initial thickness of the non-scattering layer, L_I, is measured using OCT prior to performing QME. After the muscle and compliant layer are subjected to a preload strain, the final thickness of the non-scattering layer, L_F, is measured by identifying the interface between the non-scattering and scattering layers in each B-scan using an automatic algorithm based on the Canny edge detector in post-processing [63]. The preload strain, ε, at each location is computed as the change in layer thickness of the non-scattering layer divided by the initial thickness of the non-scattering layer,

The preload strain is calculated at each lateral (xy) location across the sample surface. Local axial strain imparted to the layer by the actuator is used to calculate the local stress at the interface between the two layers in the compliant layer using the stress-strain relationship of the compliant layer. The stress-strain relationship of the compliant layer is characterized using a uniaxial compression testing apparatus prior to the QME experiment, as described previously [36,64]. Local axial strain in the compliant layer is measured from the gradient of axial displacement with depth in the scattering region. The local axial stress in the compliant layer resulting from the micro-scale actuation is estimated by multiplying the local axial strain in the compliant layer by the tangent modulus (i.e., the gradient of the tangent of the stress-strain curve) at the preload strain. Assuming uniaxial compression, axial stress is constant with depth and tangent modulus in the sample is calculated by dividing the local axial strain in the sample by the axial stress at the sample surface [36]. In the linear regime at low preload strains, tangent modulus is assumed to be equal to Young’s modulus. Quantitative micro-elastograms are presented in false color, overlaid on grayscale OCT images.

 

Fig. 2. QME experiment setup using a phase-sensitive OCT system and compression applied from an annular actuator. GelMA: gelatin methacryloyl; SLD: superluminescent diode.

Download Full Size | PDF

2.4 Effect of encapsulating quadriceps in GelMA hydrogels

Encapsulating quadriceps in GelMA enables uniform contact to be achieved at low preload strains, as illustrated in Fig. 3 that presents QME images of three-month-old WT C57BL/6J quadriceps without (Figs. 3(a), 3(c), 3(e), and 3(g)) and with encapsulation (Figs. 3(b), 3(d), 3(f), and 3(h)). In both cases, the OCT B-scans in Figs. 3(a) and 3(b) show the compliant layer placed on top of the sample after the preload strain is imparted. In a volume, the variation in preload strain calculated at each lateral (xy) location across the sample surface is shown for the unencapsulated and encapsulated muscles in Figs. 3(c) and 3(d), respectively. In Fig. 3(c), due to the uneven surface of the quadriceps, obtaining contact over a sufficiently large area of the muscle results in a highly uneven preload strain. For example, in Fig. 3(c), in the center of the muscle, corresponding to the thickest region of the unencapsulated sample, the muscle is subject to preload strains greater than -0.65. In addition, despite the center being under very high preload strain, regions toward the edge of the muscle are not in contact and the preload strain at these locations is undefined, where the preload is only defined for approximately 19.1 mm² in the OCT field of view. In contrast, in Fig. 3(d), the flatter surface of the encapsulated quadriceps enables uniform contact to be achieved at much lower preload strains, where most regions are at approximately -0.2 strain or lower, and the preload strain is defined over almost the entire 100 mm² OCT field of view. Figures 3(e) and 3(g) show en face (xy) and B-scan (xz) quantitative micro-elastograms of the unencapsulated muscle, respectively, and Figs. 3(f) and 3(h) show en face (xy) and B-scan (xz) quantitative micro-elastograms of the encapsulated muscle, respectively. The elasticity in the surrounding GelMA is removed using manual segmentation to help improve interpretations of the quantitative micro-elastograms. The elasticity measured in the unencapsulated muscle, particularly in the center, is higher than the encapsulated muscle. Importantly, the preload strain imparts an offset along the non-linear stress-strain curve of the tissue, which, at higher strains, will typically result in QME overestimating elasticity, reducing accuracy. Encapsulation helps to extend the QME field of view and improves QME accuracy by reducing the variation and magnitude of the preload strain.

In addition, as the compression is applied to the entire sample surface, due to mechanical heterogeneity in the muscle and surface unevenness, when the micro-scale actuation is applied, there can be regions of positive strain both in the tissue and the compliant layer [65]. By convention, compression is represented by negative values of strain. The mechanical model used in QME assumes that there is only compressive uniaxial stress and strain, and regions of positive strain violate this mechanical model and cannot be easily interpreted in the subsequent analysis. Therefore, elasticity measurements corresponding to regions of positive (or undefined) preload strain, denoted by green dashed lines in Figs. 3(e) and 3(f), are filtered out, described further in Section 2.5. Importantly, improving the surface flatness by encapsulating the muscle reduces the prevalence of positive strain in the tissue, where the elasticity appears to be more continuously defined in the encapsulated muscle relative to the unencapsulated muscle. Note that in Fig. 3(f), there are regions, particularly towards the center of the muscle, denoted by dashed red lines, where there are no elasticity values. These correspond to regions of positive local strain in the compliant layer from the micro-scale actuation and are also filtered out, described further in Section 2.5. Approaches to mitigate this effect are discussed further in Section 4.

 

Fig. 3. The effect of encapsulating quadriceps in GelMA hydrogels on preload strain and measured elasticity. OCT B-scans (xz) showing the compliant layer and quadriceps (a) without and (b) with encapsulation in GelMA. The cyan and red lines indicate the initial and final thickness of the non-scattering region of the compliant layer, respectively, following the preload strain. En face (xy) images of the preload strain applied to the quadriceps (c) without and (d) with the quadriceps encapsulated in GelMA. Corresponding quantitative micro-elastograms in the xy plane (e) without and (f) with encapsulation, and in the xz plane (g) without and (h) with encapsulation. Red arrows indicate the locations of the respective cross-sections in the xy and xz planes. Green dashed lines in (e) and (f) denote regions corresponding to positive (or undefined) preload strain. Red dashed lines in (f) indicate regions corresponding to positive local strain in the compliant layer. Yellow arrow in (f) denotes an air bubble. Scale bars represent 500 µm.

Download Full Size | PDF

2.5 Filtering out elasticity corresponding to positive strain

The process of removing elasticity values corresponding to regions of positive strain for a 3-month-old WT C57BL/6J quadriceps encapsulated in GelMA is illustrated in Fig. 4, which shows an OCT en face (xy) image (Fig. 4(a)), with the corresponding strain elastogram (Fig. 4(b)) and quantitative micro-elastogram with the elasticity corresponding to positive strain filtered out (Fig. 4(c)). Example regions of positive local strain in the quadriceps (indicated by blue arrows in Fig. 4(b)) and the surrounding GelMA are removed using a combination of thresholding and manual segmentation. Elasticity values corresponding to regions of positive strain in the compliant layer, indicated by the dashed red lines in Fig. 4(c), are removed using thresholding. The remaining elasticity values presented in the quantitative micro-elastogram (Fig. 4(c)) correspond to regions of compressive local stress and local strain and are used in the results described in Sections 3.2, 3.3, and 3.4.

 

Fig. 4. Filtering out regions corresponding to positive strain. (a) OCT en face (xy) image, with corresponding (b) strain micro-elastogram and (c) quantitative micro-elastogram overlaid on the OCT image in (a), for a 3-month-old WT C57BL/6J mouse quadriceps muscle. Blue arrows in (b) indicate example regions of positive local strain in the muscle. Green dashed lines in (c) denote regions corresponding to positive (or undefined) preload strain. Red dashed lines in (c) indicate regions corresponding to positive local strain in the compliant layer. Yellow arrows denote an air bubble. Scale bars represent 1 mm.

Download Full Size | PDF

2.6 Statistical analysis

Normality and variance equality were tested using Shapiro-Wilk tests and Levene's Test for Equality of Variances, respectively. Data were examined using separate two-way Analyses of Variance. Post-hoc comparisons with Tukey HSD corrections were conducted for each statistical test where appropriate, with statistical significance taken as p < 0.05. Statistical analyses were performed using Jamovi (Version 1.6). Data are presented as individual values with median and interquartile range for each tissue sample, unless otherwise specified.

3. Results

In this section, we demonstrate the impact of age and dysferlin-deficiency on the elasticity of excised quadriceps muscles, from normal WT and dysferlin-deficient BLAJ mice aged 3, 10, and 24 months (n = 3 for each group). In Section 3.1, we compare the body and muscle mass between the experimental groups of age and disease. In Sections 3.2 and 3.3, we present quantitative micro-elastograms with comparison to co-located OCT images of tissue structure. The quantitative micro-elastograms (presented on a kPa scale) are representative en face (xy) and B-scan (xz) slices taken from the 3-D data sets. In the en face images, the en face OCT images are taken from the beginning of the fitting range of the corresponding quantitative micro-elastograms. In the xy plane, the elasticity system resolution is isotropic and matched to the OCT lateral resolution (10 × 10 µm (xy)). In the B-scans, the OCT B-scans are taken at the same y-location as the corresponding quantitative micro-elastograms. In the xz plane, the elasticity system resolution is 10 × 72 µm (xz). For each muscle, red arrows denote the location of the corresponding en face and B-scan slices. Finally, in Section 3.4, we present a detailed analysis of the contrast provided by QME in skeletal muscle and compare the elasticity of quadriceps between the experimental groups. For this, we show the mean and standard deviation of all lateral elasticity values in the quadriceps in a volume with a depth range of 50 µm. Despite observing local variations in elasticity with depth, we found no significant difference between the elasticity mean and standard deviation as a function of depth for this volume in each quadriceps. Therefore, we only show one representative en face plane corresponding to a region of high OCT SNR sufficiently far from the interface between the top of the quadriceps and GelMA in each case in Figs. 6 and 7. Quadriceps surface topology and optical scattering resulted in this region varying from approximately 10–100 µm from the muscle surface between each quadriceps.

3.1 Body and muscle mass of age and disease groups

Body and muscle mass were measured at the time of sampling to provide further insight into the impact of age and disease on the mice used in this study. Shown in Fig. 5, body mass was significantly higher at 10 and 24 months of age, compared with 3 months, (+24% and +17%, respectively, p < 0.01; Fig. 5(a)), and did not differ between normal WT and dysferlin-deficient BLAJ mice. The mass of the BLAJ quadriceps muscles, normalized to body mass, at 24 months was significantly lower than WT at 24 months (-56%), and also lower compared to BLAJ quadriceps at 3 and 10 months (-58% and -54% respectively, p < 0.001; Fig. 5(b)), while the WT quadriceps mass was unchanged with age.

 

Fig. 5. Body and muscle masses of normal wild-type (WT) and dysferlin-deficient BLAJ mice aged 3, 10, and 24 months (n = 3). (a) Body mass measured at sampling. (b) Quadriceps muscle mass normalized to body mass. BLAJ significantly different to WT (*** p < 0.001). Significant differences between age groups (# p < 0.05, ## p < 0.01, ### p < 0.001). Data are presented as individual values with median and interquartile range.

Download Full Size | PDF

3.2 Normal WT C57BL/6J mice

In this section, we compare representative, co-located OCT images and quantitative micro-elastograms of quadriceps from normal WT mice at 3, 10, and 24 months of age. We show OCT en face images (Figs. 6(a)–6(c)) and B-scans (Figs. 6(d)–6(f)) along with their corresponding quantitative micro-elastograms (Figs. 6(g)–6(l)), with additional corresponding magnified views of the regions denoted by red squares for both OCT en face images (Figs. 6(m), 6(o), and 6(q)) and quantitative micro-elastograms (Figs. 6(n), 6(p), and 6(r)).

 

Fig. 6. Normal WT C57BL/6J quadriceps at different ages. (a)-(c) OCT en face (xy) images and (d)-(f) OCT B-scans (xz) with corresponding quantitative micro-elastograms in the (g)-(i) xy and (j)-(l) xz planes. Red squares highlight the ordered striation patterns shown in these images. Magnification of these red squares is shown for both the en face (m), (o), (q) OCT images and (n), (p), (r) quantitative micro-elastograms. Red arrows indicate the locations of the respective cross-sections in the xy and xz planes. Yellow arrow in (a) denotes an air bubble. Scale bars represent 500 µm in (a)-(l) and 250 µm in (m)-(r).

Download Full Size | PDF

In the OCT en face images across all three age groups, the quadriceps feature ordered striation patterns (highlighted by the red squares), as was also observed in a previous study that characterized excised quadriceps using compression OCE and are indicative of the direction of the myofibers [33]. In that study, the striation patterns in the corresponding quantitative micro-elastogram of local strain followed the direction of the myofibers closely. Similarly, in the corresponding en face quantitative micro-elastograms in this study, the variations in elasticity follow the structure of the muscle in the OCT en face images. At 24 months, the OCT en face image (Fig. 6(c)) shows slightly disordered striations in comparison to the 3- and 10-month-old quadriceps, suggesting a deterioration of both myofiber composition and structure with age [66]. In the quantitative micro-elastograms, there appears to be a gradual stiffening of the quadriceps with age where the 24-month-old quadriceps (Figs. 6(i) and 6(l)) appear stiffer compared to the 3- and 10-month-old quadriceps. Also, there is a distinct lack of adipose tissue in the WT quadriceps; adipose tissue is typically denoted by a characteristic honeycomb structure in OCT images [33,67].

3.3 Dysferlin-deficient BLAJ mice

Here, we compare representative, co-located OCT images and quantitative micro-elastograms of quadriceps from dysferlin-deficient BLAJ mice at 3, 10, and 24 months of age. We show OCT en face images (Figs. 7(a)–7(c)) and B-scans (Figs. 7(d)–7(f)) along with their corresponding quantitative micro-elastograms (Figs. 7(g)–7(l)). In addition, red squares highlight regions of interest including ordered striation patterns in 3- and 10-month-old muscle and cyan squares indicate regions of adipose tissue in the 24-month-old muscle. Magnification of these regions of interest is shown for both OCT en face images (Figs. 7(m), 7(o), and 7(q)) and en face quantitative micro-elastograms (Figs. 7(n), 7(p), and 7(r)).

 

Fig. 7. Dysferlin-deficient BLAJ quadriceps at different ages. (a)-(c) OCT en face (xy) images and (d)-(f) OCT B-scans (xz) with corresponding quantitative micro-elastograms in the (g)-(i) xy and (j)-(l) xz planes. Red squares highlight regions of interest including ordered striation patterns in 3- and 10-month-old muscle. Cyan squares indicate regions of adipose tissue in the 24-month-old muscle. Magnification of these regions of interest is shown for both the en face (m), (o), (q) OCT images and (n), (p), (r) quantitative micro-elastograms. Red arrows indicate the locations of the respective cross-sections in the xy and xz planes. Yellow arrow in (a) denotes an air bubble. Scale bars represent 500 µm in (a)-(l) and 250 µm in (m)-(r).

Download Full Size | PDF

Overall, in comparison to the normal WT quadriceps, the 3- and 10-month quadriceps do not appear to be considerably different. However, at 24 months, the BLAJ quadriceps is markedly changed. This is in contrast with the WT quadriceps that show only subtle changes with age. Both the 3- and 10-month-old BLAJ quadriceps appear similar to the WT (Fig. 6), with ordered striations in the OCT en face images (highlighted by the red squares), indicating a typical structural arrangement of myofibers and corresponding variations in elasticity in the respective quantitative micro-elastograms. Furthermore, in the 3- and 10-month WT and BLAJ quadriceps, there are no identifiable regions of adipose tissue. The marked histopathological effects of dysferlin-deficiency typically only begin to manifest from around eight months of age, becoming more pronounced over time [35,68,69]. As anticipated, for the 24-month BLAJ quadriceps, both the structure and elasticity appear altered by substantial muscle replacement with adipose tissue [18]: this appearance is markedly different to that of the 3- and 10-month WT and BLAJ quadriceps. In the OCT en face images of the 24-month BLAJ quadriceps, there are disordered striations, indicating disrupted myofiber arrangement, as well as much of the muscle being replaced by adipose tissue, denoted by the characteristic honeycomb structure in the OCT en face image (Figs. 7(c) and 7(q)). In the corresponding quantitative micro-elastogram in Fig. 7(f), there is both a decrease in elasticity of the individual myofibers, and a decrease in elasticity of the entire muscle over the field of view; in marked contrast to both the 3- and 10-month BLAJ mice, and WT mice at all ages. The replacement of myofibers with adipose tissue at 24 months is highlighted in the magnified views indicated by the cyan squares, highlighting the difference between the ordered myofiber arrangement and elasticity patterns at 3 and 10 months (Figs. 7(n) and 7(p)) and presence of adipose tissue and reduced elasticity at 24 months (Fig. 7(r)).

3.4 Micro-scale elasticity contrast in skeletal muscle

To further examine the contrast provided by QME, we analyzed all elasticity values acquired 25 µm above and below the en face planes shown, for each of the 6 quadriceps presented in Sections 3.2 and 3.3, and similar regions for all remaining 12 quadriceps. To demonstrate the effects of aging and dysferlin-deficiency, we firstly present histograms of these elasticity data, pooled for each age sub-group for WT (Fig. 8(a)) and BLAJ muscles (Fig. 8(b)). The overall mean elasticity and variability of elasticity, expressed as mean ± standard deviation, for the 3-, 10-, and 24-month-old WT quadriceps are 24.1 ± 32.5 kPa, 33.2 ± 42.1 kPa, and 45.6 ± 49.7 kPa, respectively, and for 3-, 10-, and 24-month-old BLAJ quadriceps are 25.8 ± 31.1 kPa, 35 ± 41.1 kPa, and 10.9 ± 20.3 kPa, respectively. These results indicate both a higher elasticity and increase in the variation of micro-scale elasticity of normal WT muscle tissue at 24 months of age, which is consistent with previous literature that suggest a deterioration of both myofiber composition and structure with age [66,70]. Furthermore, these results indicate both a lower elasticity and reduction in mechanical heterogeneity of the 24-month dysferlin-deficient BLAJ quadriceps as the myofibers are replaced by adipose tissue. The 24-month BLAJ quadriceps are significantly softer (-77%, p < 0.01; Fig. 8(c)) with less mechanical heterogeneity (-60%, p < 0.01; Fig. 8(d)), compared to WT, when individual muscles’ mean, and standard deviation of elasticity are calculated, and groups are compared. However, there is no statistically significant increase in elasticity in the WT at 24 months as the quantitative micro-elastograms and histograms suggest (p > 0.05).

 

Fig. 8. Quantitative comparison of age and disease in the quadriceps presented in Figs. 6 and 7. Histograms of the elasticity values pooled for 3-, 10-, and 24-month-old (a) normal WT and (b) dysferlin-deficient BLAJ quadriceps. (c) Mean elasticity and (d) variability of the elasticity, expressed as standard deviation, of WT and BLAJ muscles at 3, 10, and 24 months. Summary data in (c) and (d) are presented as individual muscle values, with median and interquartile range.

Download Full Size | PDF

4. Discussion

The results presented in this study demonstrate the ability of QME to provide quantitative micro-scale elasticity of skeletal muscle. Skeletal muscle is a complex hierarchical tissue composed of myofibers, which are long, cylindrical, multinucleated cells organized into bundles surrounded by layers of ECM (connective tissue) and attached to the skeleton by tendons [71]. Individual myofibers are covered in the specialised ECM basement membrane and endomysium, bundles of which are encased by the perimysium to form fascicles. Finally, the whole muscle, comprised of many fascicles, is enveloped by the epimysium, and is connected to the skeleton by tendons [72]. Collagen fibers, which comprise the majority of ECM and run in the plane parallel to the surface of myofibers [73], are typically stiff under tension but relatively soft when compressed [1,74]. Therefore, the striation patterns in elasticity observed in the quantitative micro-elastograms likely represent the relatively soft ECM surrounding individual myofibers. The results demonstrate that QME can image the micro-scale variation in elasticity for skeletal muscle and can observe both differences in local elasticity within a muscle and between normal muscles and those impacted by a muscle disease.

Dysferlinopathy results in the replacement of skeletal muscle by adipose tissue at later stages [18,19]. The main result is that the 24-month dysferlin-deficient BLAJ quadriceps appeared markedly different to that of the 3- and 10-month-old BLAJ muscles, which closely resembled the muscle structure and elasticity of WT muscles. In contrast, the 24-month BLAJ quadriceps showed disordered striations in the OCT en face images, as well as much of the muscle replaced by adipose tissue. Adipocytes, of which adipose tissue is comprised, are relatively large lipid-filled cells, with previous studies showing that adipose tissue, at low strain levels (< 0.3), has low elasticity under compression (∼1 kPa) as reviewed in [75]. In the case of the 24-month BLAJ quadriceps, where there is a large, relatively homogeneous region of adipose tissue replacing skeletal myofibers, it is therefore unsurprising that we observed a significant reduction in elasticity and mechanical heterogeneity. While the substantial replacement of dysferlin-deficient muscle tissues with adipocytes is well established [18,19], this is the first time the effects of dysferlinopathy on the micro-scale variation of tissue elasticity have been characterized in 3-D with QME.

In the normal WT quadriceps at 24 months of age, the myofiber organization appeared to be disordered, which is consistent with literature that suggests a deterioration of both myofiber composition and structure with age [66,70]. Specifically, aging muscle is characterized by myofiber branching [76] and increased ECM with fibrosis [77,78], typically resulting in stiffer muscles. However, this disordered structure observed in the OCT en face images, likely indicative of myofiber branching, did not translate to a statistically significant increase in elasticity or mechanical heterogeneity in this study. Polarization-sensitive OCE [79] could be combined with QME to enable simultaneous measurement of both optical birefringence and elasticity. This could be used to provide additional information, such as myofiber orientation, to improve the correlation between tissue structure and elasticity, likely providing further insight into the relationship between muscle structure, function, and disease.

The quantitative micro-elastograms of all quadriceps, except for those from the 24-month-old BLAJ mice, show a region of elevated elasticity that corresponds to the distal region of the quadriceps where the muscle connects to the patellar tendon (i.e., myotendinous junction). The patellar tendon is considered a relatively stiff tendon that provides efficient transfer of mechanical power and precise control over the extension of the knee joint [80,81]. The myotendinous junction is a specialized and complex structure that enables the transmission of force from the myofibers to the collagen fibers in the tendon [82,83]. Typically, this region is characterized by invaginations within the muscle containing collagen fibers that are parallel to the longitudinal axis of the myofibers [84–86], and this junction region contains a higher proportion of collagen, particularly collagen type I [87,88]. Additional non-collagenous ECM within tendon, including elastin, proteoglycans, and glycosaminoglycans, are reported to play a large role in tendon elasticity under compression [88]. Elastin, for example, which, while comprising only approximately 4% on tendon dry weight, plays a disproportionately large role in transverse stiffness of tendon [89–91]. Therefore, the increased elasticity in the region corresponding to the myotendinous junction shown in the current study is unsurprising given the complex structure and increased proportion of tendon in this region and is consistent with previous studies [92].

Statistical comparisons in this study are limited by the high variability in elasticity within and between samples, and/or the low sample sizes used (n = 3). Notably, we showed that skeletal muscle elasticity is heterogeneous at the intermediate scale, with large variation in elasticity within a muscle. This result is unsurprising given the nature of skeletal muscle at this scale, with relatively soft myofibres surrounded by a much stiffer scaffold of ECM [72], where the quadriceps is a highly complex muscle comprised of four individual muscles held together by tendons and other connective tissue [80,93]. Moreover, the results indicate variable bulk elasticity and mechanical heterogeneity between muscles in the same experimental group, which is not surprising since variations between the same muscles in individual mice are widely reported [94] and the same muscles within one mouse can also vary in severity of pathology: as demonstrated for ‘leakiness’ of limb muscles in mdx mice [95]. While great care was taken to excise, encapsulate, and scan muscles consistently, there are undoubtedly differences in muscle structure and orientation that would impact the resulting bulk elasticity and mechanical heterogeneity measured for individual muscles. Finally, a small sample size was used due to the complex and time-consuming nature of the method presented in this study and the large number of experimental groups examined. However, despite these limitations, we demonstrated the impact of dysferlin-deficiency on skeletal muscle mechanical properties, showing the potential for this elastography technique to identify important biomechanical alterations in muscular dystrophies.

Encapsulating tissue in a hydrogel has several important advantages in QME. This approach has previously been used in OCE to keep the tissue hydrated during scan acquisition [96,97]. The main motivation of encapsulation in this study was to improve image quality by addressing issues related to mechanical deformation including the size and uneven surface of the muscle. For example, in QME, it is necessary to apply a preload strain to the sample to ensure uniform contact between the flat imaging window, compliant layer, and tissue. Importantly, many tissues exhibit non-linear elastic behavior [98]. Applying a preload strain will impart an offset along the non-linear stress-strain curve of the tissue which, at higher strains (i.e., typically >0.1), will typically result in QME overestimating elasticity. A primary advantage of encapsulation is that it enables good contact to be achieved at lower preload strains and ensures that the entire muscle remains in the linear-elastic region, improving QME accuracy and enabling longitudinal comparisons. In addition, making specimens with a known height provides a practical means to achieve an accurate preload strain using the axial translation stage. Furthermore, strong reflections at the interface between the bottom of the muscle and rigid plate can generate an additional common-path OCT signal that reduces QME image quality. In this study, optical scattering was added to the GelMA to attenuate the light below the muscle, further improving QME image quality.

Despite providing several key improvements in QME image quality, encapsulation has several disadvantages. For example, encapsulation mechanically couples the muscle to the surrounding material. As tissue is typically incompressible [99], it needs to expand laterally to deform axially. If the surrounding material restricts the lateral expansion of the muscle, it will cause the muscle to exhibit an artificially high elasticity and reduce QME accuracy. The elasticity of the surrounding material needs to be carefully considered to minimize this effect. UV exposure time and GelMA solution composition were chosen such that the GelMA elasticity was closely matched to that of the muscle, where muscle elasticity has been reported to be on the scale of 1–100 kPa [100,101]. Two other important considerations of encapsulation are that the material above the tissue will attenuate the OCT signal and reduce OCT SNR, and that encapsulation also increases sample preparation time, cost, and complexity. In this study, the OCT SNR in the muscle was sufficiently high, such that the attenuation of the OCT signal from the top GelMA layer above the muscle was minimal. Importantly, each muscle was encapsulated and scanned using the same protocol where the effects of the GelMA elasticity on QME accuracy are likely to be minimal and are consistent between muscles, enabling an accurate longitudinal comparison. Whilst it was outside the scope of this study to characterize the effects of GelMA elasticity on QME accuracy in encapsulation, this represents an important avenue of future research to further optimize QME in imaging skeletal muscle.

If the sample and compliant layer are incompressible and have different elasticities, friction can restrict lateral expansion of the sample and change its apparent elasticity, reducing QME accuracy. Furthermore, the compliant layer reduces imaging depth. If the sample is made of a hydrogel, rather than requiring a separate compliant layer, the hydrogel itself could be used to measure axial stress. This has potential to improve both QME accuracy and imaging depth. In such an approach, the stress-strain relationship of the hydrogel could be characterized prior to performing QME and a homogenous region of the hydrogel used to estimate axial stress. However, a challenge with this approach is that it requires a region of the hydrogel to be sufficiently far away from any mechanical heterogeneity such that the stress in that region is both uniform and uniaxial. One application where this is likely to effectively improve image quality is in cell mechanobiology where mechanical heterogeneity is typically localized within approximately 10–100 µm of the cell [40].

QME relies on several assumptions that can, in some instances, limit image quality. For example, the unloaded thickness of the compliant layer is assumed to be a constant value when placed between the sample and the imaging window. However, due to the uneven surface of the quadriceps, even when encapsulated, the incompressible compliant layer can become confined and forced to expand, resulting in a positive measured preload strain in some areas. Positive preload strain violates the assumption of uniaxial compression in the compliant layer and sample, and it is necessary to filter out these regions in post-processing to not confound the comparison of muscle elasticity. However, this reduces the ability to characterize the entire muscle. For example, in Figs. 6 and 7, there are regions, particularly towards the center of the muscle, where there are no elasticity values. One approach to address this in future studies is to instead bring the compliant layer/sample ensemble just into contact with the imaging window and acquire an OCT scan of the stress layer in this unloaded state. An edge detection algorithm, similar to that used to determine the preloaded layer thickness, could then be used to create a two-dimensional map of the unloaded layer thickness. This will help to improve the accuracy of estimating the true preload strain in the compliant layer and achieve elasticity measurements over the entire muscle. Additionally, there are several factors that are likely to contribute to positive local strain in the compliant layer, including layer incompressibility, the geometry of the sample surface, sample mechanical heterogeneity, and the strain SNR. In this study, consistent with previous studies of QME, the compliant layer is made of an incompressible silicone. Importantly, for an incompressible material to deform axially, it needs to be free to expand laterally. Since the compression is applied over the entire surface, friction at the interfaces between the imaging window, compliant layer, and sample, can restrict the lateral expansion of the compliant layer. Local variations in friction at these interfaces, due to uneven lubrication, can result in corresponding local variations in the deformation of the compliant layer, where it is possible for there to be non-axial and positive local strain. The effect of positive local strain is more prominent when the sample surface is uneven and when there is mechanical heterogeneity in the sample. In regions of non-axial strain, if the magnitude of the negative axial strain in this region is low (i.e., low strain SNR), noise can also result in the measured strain being positive. The elasticity in these regions also needs to be filtered out to avoid misinterpretation of muscle elasticity. Future work could reduce positive compliant layer strain by measuring all three spatial components of deformation and solving more complex mechanical models that make fewer assumptions regarding sample deformation [65,102,103].

5. Conclusion

We have demonstrated QME to characterize the micro-scale elasticity of excised mouse quadriceps on the intermediate scale in 3-D. In addition, we compared the effects of aging and disease on quadriceps elasticity, using normal WT and dysferlin-deficient BLAJ mice at 3, 10, and 24 months of age. This was achieved using a sample preparation method that encapsulates the muscles in GelMA hydrogels to create specimens with flat surfaces that are more suitable for QME imaging. Our results showed a significant decrease in elasticity at 24 months in dysferlin-deficient quadriceps compared to normal quadriceps. This represents the first time the effects of dysferlinopathy on the micro-scale variation of tissue elasticity have been characterized in 3-D with QME. We believe characterizing muscle elasticity on the intermediate length scale in 3-D with QME will help to improve the understanding of the development and progression of many skeletal muscle pathologies.