Hypoxia Inducible Factor 1a (Hif1a) and Hypoxia in Tendon Development and Disease

Tendons and ligaments are collagen-rich tissues required for the transmission of force from muscle to bone or bone to bone, respectively. During postnatal development, cells within the tendon/ligament and withing the tendon-bone interface (e.g., the enthesis or osteochondral interface) synthesize and organize the collagen-rich extracellular matrix (ECM) required for tissue maturation. Prior studies have shown that postnatal tendon development is dependent on mechanical cues; however, the extent to which local oxygen availability, and oxygen-sensing pathways such as Hif1a modulates tendon and enthesis cell behavior and the ECM deposition needed to establish mechanical functions remain poorly defined. Our studies aimed to investigate tendon development and disease by 1) interrogating how the loss of the transcription factor Hypoxia Inducible Factor 1a (Hif1a) in tendon progenitors drives changes in Achilles tendon and enthesis structure and function; 2) determining how Hif1a and hypoxia (e.g., low oxygen) regulate tendon fibroblast physiology and matrix deposition in vitro; and 3) evaluating how the reduction or increase in Hif1a expression affects tendon healing post-injury.

Tendinopathies and enthesopathies are among the most prevalent musculoskeletal diseases. These disorders are generally characterized by reduced mechanical function, largely due to the reduction in quality native ECM (e.g., organized Collagen-I). Often, tendinopathy or enthesopathy may be caused by an acute injury such as an Achilles tendon rupture, or they can be caused by repeat use over time where gradual injury is accumulated causing partial tears. Tendons and the interfacial tissue connecting tendons to bone have limited regenerative capabilities post-injury largely due to the reduced cellularity and thus the reduced capacity to generate native-type ECM. In our studies, we characterize the effect oxygen tension and Hif1a has on generating ECM post-injury using bulk-RNA sequencing, histology, and radiography. Further, we interrogate how hypoxia and a hypoxia-inducible gene regulate tendon and enthesis development, cell physiology, and tendon and enthesis healing post-injury. To do this, we utilize 2D and 3D in vitro approaches using genetically modified mice and/or pharmacological and environmental perturbation of Hif1a. Additionally, in our animals lacking Hif1a or exhibiting an overexpression of Hif1a in tendon and enthesis progenitor cells, we track tendon and enthesis development and healing over time. We use multiple imaging methods to look at tendon, enthesis, and bone health, in addition to biomechanical testing of the tendon and enthesis, and transcriptomic analyses paired with cellular data to better understand how cells regulate tendon- and enthesis-ECM deposition.

In our studies, we have found that the developing enthesis is hypoxic and Hif1a is required for tendon and enthesis postnatal development. Additionally, we have found that the overexpression of Hif1a post-injury improves tendon and enthesis healing by increasing collagen alignment in the tendon scar tissue. Further, RNA-sequencing in tandem with cell-based assays revealed Hif1a is required for cell survival and tissue delineation processes in tendon progenitor cells. This research has identified targetable pathways for tendon and enthesis regenerative approaches that may be used to treat tendon disease and injury.

After euthanasia, intact left limbs from Ctrl (Cre-negative, n= 15; Cre-positive, n=3 males) and cKO mice (n=15) at 8wks of age were frozen at -20°C. Achilles tendons and calcaneus were dissected and kept hydrated (with removal of the plantaris tendon and surrounding muscles). Tendon volume was imaged using optical coherence tomography with polarization (TEL221PSC1, OCT-LK4 lens, ThorLabs Inc., Newton, New Jersey). Intensity and polarization images were acquired in 3D (A-scan averaging = 4; spacing = 10μm in x and y, 2.5μm in z; sampling rate = 28kHz) and intensity and retardation images were exported (32-bit TIFF). Slice analysis was used to determine tendon cross-sectional area (CSA) and bifurcation distance (Dragonfly, Comet Technologies Canada Inc., Montreal, Canada). After imaging, the proximal Achilles tendon was clamped in a thin film grip (Imada, Northbrook, IL, USA) with the calcaneus secured in a custom 3D-printed fixture (FormLabs 3B, Somerville, MA, USA) prior to uniaxial tensile testing.

The tendon and grip were placed in a room temperature phosphate buffered solution (PBS) bath and 38 a pin was used to secure the tendon-grip unit. Uniaxial tensile tests were conducted with load measured using a multi-axis load cell (±70 N; Mach-1 VS500CST, Biomomentum, Laval, Quebec, Canada). Samples were preloaded to 0.2N and preconditioned for 10 cycles (±0.025mm at 0.1 Hz). Grip-to-grip gauge length (L0) was measured at the start of each test after preload and was consistent between tests (3.6 ± 0.5 mm). Load and torque data were collected while applying displacement to failure at 0.01mm/min to assess axial and off-axis loading. Mechanical properties (e.g., maximum load, maximum strain, linear stiffness; linear modulus and maximum stress calculated using load and CSA) of the Achilles tendon were calculated from force-displacement data using a piecewise linear segmentation by dynamic reprogramming recursion package (dpseg) in R (v4.2.2 or later, The R Project for Statistical Computing, Vienna, Austria). Data were presented as mean ± standard deviation and compared using a one-way ANOVA and corrected for multiple comparisons using Tukey’s multiple comparisons tests (Prism v9.0+; Graphpad, LaJolla, CA, USA).

We generated mice for stable (non-degradable) expression of HIF1a by breeding loxP-stop-loxP (LSL)-HIF1dPA (proline to alanine substitutions) mice (JAX stock 009673) in Scx-lineage cells (HIF1dPA; homozygous LSL, Cre+ mice) and wildtype (Cre-negative or Cre-positive heterozygous LSL mice). Mice were euthanized at 8-weeks of age (N>4 per genotype; at least 2 per sex), and one hindlimb was paraffin processed for sectioning and hematoxylin and eosin staining and imaging while the other was frozen at -20°C until preparation for dissection for biomechanics. 24 hours before the start of mechanical testing, limbs were thawed at 4°C and prepared for photogrammetry and biomechanical testing. Both the plantaris tendon and muscle were carefully removed and the Achilles tendons and calcaneus were left intact. Cross-sectional areas (CSA) of the Achilles tendon were measured using photogrammetry with a custom pinch clamp holder attached to a motor controller (Arduino, Ivrea, Italy). With the muscle and Achilles hanging free of the pinch clamp, at least 50 consecutive images of the pinch clamped tendon were acquired using a 12mm focal length lens (Basler Fujinon Lens, Ahrensburg, Germany). Using Metashape software (Agisoft, St. Petersburg, Russia), images were aligned and converted first to a sparse point then to a dense point cloud, to create an STL surface mesh of the tendon. CSA was defined as the smallest area from the STL surface mesh generated from the Achilles tendon and was measured using a slice analysis tool in Dragonfly (Comet Technologies Canada Inc., Montreal, Canada). A custom 3D printed fixture secured the calcaneus into place (FormLabs 3B, Somerville, MA, USA), and the proximal Achilles tendon was clamped in a textured grip and screwed into place (Imada, Northbrook, IL, USA). The assembled grip with the secured Achilles tendon was placed into a phosphate buffered solution (PBS) bath and secured with a pin to a tensile testing frame with a multi-axis load cell (±70 N; Mach-1 VS500CST, Biomomentum, Laval, Quebec, Canada). Samples were preloaded to 0.1 N, and gauge length was measured as the distance between the secured calcaneus and the textured grip. The Achilles tendons were then preconditioned for 10 cycles (±0.1 N at 0.1mm/sec) followed by the load to failure test at 0.1mm/sec. Off-axis load (forces in X and Y, torques in X, Y, Z) were collected to assess off-axis loading for the duration of the experiment. Instantaneous grip-to-grip strain was calculated as the displacement divided by the original gauge length, L0. Maximum load was calculated using a custom MATLAB script (R2017 or later, MathWorks, Natick, USA). Data between Cre+ Ctrl, Cre- Ctrl, and Hif1adPA were compared using a one-way Anova with Dunnett’s multiple comparisons test.

We used mice stably expressing HIF1A (JAX stock 009673) in constitutive Scx-lineage cells (Hif1adPA; homozygous LSL-HIF1dPA; Cre+ mice) and wildtype (Cre- or Cre+ heterozygous LSL mice). Mice received either unilateral or bilateral Achilles and plantaris tendon transection (Hif1adPA, n=6; Cre- control (Ctrl), n=3; 12-16wks of age for RNA-seq; Hif1adPA, n=7; Cre- control (Ctrl), n=7; 1yr of age for histology). Mice with bilateral injuries were euthanized at 14 days post injury (DPI) for bulk RNA-sequencing and mice with unilateral injuries were euthanized at 56DPI for histological outcomes.

We evaluated the differentially expressed genes (DEGs) in the Hif1adPA mouse tendons compared to Ctrl at 14DPI (proliferation phase of healing). Bulk RNA from the tendon stumps and scar were extracted using column-based isolation (Purelink RNA Mini kit, Invitrogen) with on-column DNase treatment (RIN±8.1). Libraries with Ply-A enrichment were prepared for 151bp paired-end sequencing (Illumina NovaSeqXPlus). BCL convert conversion software v4.0 (Illumina) was used to generate de-multiplexed Fastq files. Reads were trimmed and mapped (GRCm38) and analyzed in RStudio (fold change cutoff=1.5; p-value≤0.1) to identify DEGs. Biological processes differentially regulated in the Hif1dPA tendons compared to Ctrl (Cre-).

We evaluated the structure of tendon healing using quantitative polarized light imaging (qPLI) on PicroSirius Red stained sections, imaged within the healing scar region. Images were acquired at 20x objective on an inverted microscope (Motic) with a circular polarizer, polarization camera, and ThorCam software (ThorLabs Inc.). The standard deviation of the angle of polarization (AoP) and the degree of linear polarization (DoLP) were analyzed using the Math and SciPy Stats libraries in Python3. Differences in DoLP and AoLP of tendon were compared using t-tests (Graphpad v10.1, Prism). We also evaluated overall healing of the tendon, enthesis, and bone using Masson’s trichrome and Heamtoxylin and Eosin (H&E) stained sections with a blinded cumulative histology score adapted from the Bonar and OARSI scoring systems.177–179 We evaluated the differentially expressed genes (DEGs) in the Hif1adPA mouse tendons compared to Ctrl at 14DPI (proliferation phase of healing).

Randomization of samples prior to Achille’s tendon dissection was performed and tester was blinded to all experimental groups. After euthanasia, intact left limbs and injured right limbs from Ctrl (Cre-negative, n= 8; Cre-positive, n=6) and dPA mice (n=10) at 56 days post injury were frozen at -20°C. Achilles tendons and calcaneus were dissected and kept hydrated (keeping the plantaris tendon intact while removing surrounding muscles). Tendon volume was imaged using optical coherence tomography with polarization (TEL221PSC1, OCT-LK4 lens, ThorLabs Inc., Newton, New Jersey). Intensity and polarization images were acquired in 3D (A-scan averaging = 4; spacing = 10μm in x and y, 2.5μm in z; sampling rate = 28kHz) and intensity and retardation images were exported (32-bit TIFF). After imaging, the proximal Achilles tendon was clamped in a thin film grip (Imada, Northbrook, IL, USA) with the calcaneus secured in a custom 3D-printed fixture (FormLabs 3B, Somerville, MA, USA) prior to uniaxial tensile testing. The tendon and grip were placed in a room temperature phosphate buffered solution (PBS) bath and a pin was used to secure the tendon-grip unit. Uniaxial tensile tests were conducted with load measured using a multi-axis load cell (±70 N; Mach-1 VS500CST, Biomomentum, Laval, Quebec, Canada). Samples were preloaded to 0.2N and preconditioned for 10 cycles (±0.025mm at 0.1 Hz). Grip-to-grip gauge length (L0) was measured at the start of each test after preload and was consistent between tests (3.6 ± 0.5 mm). Load and torque data were collected while applying displacement to failure at 0.01mm/min to assess axial and off-axis loading. Mechanical properties (e.g., maximum load, maximum strain, linear stiffness; linear modulus and maximum stress calculated using load and CSA) of the Achilles tendon were calculated from force-displacement data using a piecewise linear segmentation by dynamic reprogramming recursion package (dpseg) in R (v4.2.2 or later, The R Project for Statistical Computing, Vienna, Austria). Data were presented as mean ± standard deviation and compared using a one-way ANOVA and corrected for multiple comparisons using Tukey’s multiple comparisons tests (Prism v9.0+; Graphpad, LaJolla, CA, USA).