Scleraxis is required for the development of a functional tendon enthesis

Abstract

The attachment of dissimilar materials is a major engineering challenge, yet this challenge is seemingly overcome in biology. This study aimed to determine how the transcription factor Scleraxis (Scx) influences the development and maturation of the tendon-to-bone attachment (enthesis). Mice with conditional knockout (cKO) for Scx (Scx^flx/−, Prx1Cre⁺) and wild-type [(WT) Scx^flx/+ or Scx^flx/flx] littermates were killed at postnatal days 7–56 (P7–P56). Enthesis morphometry, histology, and collagen alignment were investigated throughout postnatal growth. Enthesis tensile mechanical properties were also assessed. Laser microdissection of distinct musculoskeletal tissues was performed at P7 for WT, cKO, and muscle-unloaded (botulinum toxin A treated) attachments for quantitative PCR. cKO mice were smaller, with altered bone shape and impaired enthesis morphology, morphometry, and organization. Structural alterations led to altered mechanical properties; cKO entheses demonstrated reduced strength and stiffness. In P7 attachments, cKO mice had reduced expression of transforming growth factor (TGF) superfamily genes in fibrocartilage compared with WT mice. In conclusion, deletion of Scx led to impairments in enthesis structure, which translated into impaired functional (i.e., mechanical) outcomes. These changes may be driven by transient signaling cues from mechanical loading and growth factors.—Killian, M. L., Thomopoulos, S. Scleraxis is required for the development of a functional tendon enthesis.

Tendon attaches to bone across a specialized transitional tissue called the enthesis. The enthesis consists of tendon transitioning into fibrocartilage (FC) and then into mineralized FC and finally inserting into bone (1–3). This tissue solves the mechanical challenge of attaching dissimilar materials, tendon and bone, which would otherwise be prone to failure due to stress concentrations that would arise at their interface (4). This functionally graded attachment system is established during late fetal time points and matures during the early postnatal period (5). These attachments are localized to either protrusions near articulating joints, such as the greater tuberosity of the humeral head (HH), or protrusions along long bones, such as deltoid tuberosity on the diaphysis of the humerus (6). They can be fibrocartilaginous or fibrous in nature, depending on the anatomic requirement and attachment footprint (2, 3). For example, the supraspinatus tendon of the rotator cuff attaches to the greater tuberosity across a fibrocartilaginous enthesis, and this tissue is established and matures in the perinatal period (7). Aligned collagen fibers insert into bone across FC with a mineral gradient, resulting in a mechanically competent attachment (1, 2, 5, 8, 9). Although fetal and postnatal development leads to a strong and tough enthesis, the structural and compositional mechanisms necessary for effective attachment are not recreated during tendon-to-bone healing, and the repaired attachment often fails (10–14). A better understanding of the biochemical and biophysical cues necessary for the development of a functional enthesis could guide repair and regeneration strategies for tendon-to-bone repair.

The tendon enthesis is initiated embryonically through a number of molecular signals. Opposing expression gradients of Scleraxis [Scx; a transcription factor necessary for tenogenesis (6, 15–19)] and sex-determining region Y-box 9 (Sox9; a transcription factor necessary for chondrogenesis) establish a progenitor cell population during embryonic development for the tendon attachment unit (20, 21). Deletion of Scx leads to small, rudimentary force-transmitting tendons (15) and prevents the formation of bony tuberosities onto which tendons attach (6). Growth factors in the TGF-β and bone morphogenetic protein (BMP) families likely regulate early enthesis formation (6, 22, 23) and molecules such as Indian hedgehog and parathryroid hormone-related protein likely regulate late mineralization events (24–26). After establishment of the attachment unit, muscle forces can be transmitted across the interface without rupturing the connection, and regulation of growth and mineralization are coupled to increased muscle activity (6, 15, 22, 27). Muscle unloading leads to impaired enthesis mineralization (28), organization (28), and mechanical strength (27) during postnatal development.

Although a number of studies have shown that Scx is critical for establishing the progenitor cell population that forms the enthesis, it remains unclear what role Scx plays in enthesis maturation (i.e., cellular organization, molecular signaling, and extracellular matrix organization) and how mechanical loading regulates that process. Therefore, the purpose of this study was to determine the role of Scx in the formation of a functional tendon enthesis. Maturation and mineralization of the supraspinatus enthesis was examined during key postnatal time points for the murine rotator cuff. Morphologic, organizational, molecular, structural, and functional outcomes were assessed. We hypothesized that Scx is required for supraspinatus enthesis maturation, FC collagen organization, FC bone mineral density (BMD) and morphometry, and structural and mechanical properties and that the structure and function of the supraspinatus enthesis would be impaired in the absence of Scx.

MATERIALS AND METHODS

Animals

All in vivo experiments were approved by the Washington University School of Medicine Animal Studies Committee. Mice were housed in pathogen-free barrier conditions and were provided food and water ad libitum. Scx^flx/flx and paired related homeobox 1-Cre (Prx1Cre) mice were used in this study to target Scx deletion in limb-bud progenitor cells. Mice used in this study were of a C57BL/6 background and were generously provided by Dr. Ronen Schweitzer (Scx^flx/flx, Shriner’s Research Center, Portland, OR, USA) or purchased from The Jackson Laboratory (Bar Harbor, ME, USA) (Prx1Cre). Heterozygous (Scx^flx/+), Prx1Cre-positive females (29) were crossed with male Scx^flx/flx homozygotes (15, 29), which created the Scx-null allele by exploiting the stochastic, ubiquitous Cre recombination in oocytes of Prx1Cre females (15). Additionally, male heterozygous, Cre-positive males were crossed with Scx^flx/flx homozygote females as additional breeding pairs. The genotypes of mice were determined by PCR-based analysis of DNA samples obtained from toe and tail biopsies and deletion of Scx was confirmed by genotyping for Cre, Scx-null, Scx-WT, and Scx-floxed alleles as previously described (15). Conditional knockout (cKO) mice were defined as mice that were positive for Prx1Cre, Scx-floxed, and Scx-null, as well as negative for Scx-WT allele (Scx^flx/−, Prx1Cre⁺), following PCR analysis of genomic DNA obtained from toe samples. Scx^flx/flx and Scx^flx/+ mice (Cre-negative, negative for Scx-null allele as detected via PCR) were used as wild-type (WT) littermate controls. Approximate Mendelian ratios for this breeding scheme were achieved (∼1:4 pups were Scx^flx/-, Prx1Cre⁺). To separate the effects of Scx deletion from the effects of muscle unloading, a set of unloading experiments was performed: Following palpation of the shoulder, 0.15 U (10 μl) of botulinum toxin A (BTX) in 0.1% saline were injected using a 27 gauge needle into supraspinatus muscles of WT mice at postnatal day 0 (P0) and P4 prior to killing at P7. Localized paralysis was observationally confirmed as the pups’ inability to raise their denervated arm, as previously described (30). Contralateral shoulders were injected with an equivalent volume of 0.1% saline at P0 and P4.

Histology and micro-computed tomography

To determine the effect of Scx on maturation and organization of the rotator cuff enthesis, supraspinatus muscle, tendon, and bone were carefully dissected as intact units, free of surrounding tendons and musculature, from WT and cKO mice aged to P14, P28, and P56 (n = 4–10 per group per time point). Supraspinatus muscle–tendon–bone units were placed in 4% paraformaldehyde immediately after dissection with the tendon positioned at 180° relative to the humerus (i.e., at 90° abduction) and fixed overnight. Samples were then rinsed in PBS and stored in ethanol until imaging. Samples were scanned using cone-beam micro-computed tomography (CT) at 16 μm resolution at 300 ms sample time, energy of 45 kV, and intensity of 177 μA. Parameters determined from micro-CT included (1): volume of the HH, including the compact bone (2), BMD (mg hydroxyapatite per cubic cm) of the HH including the compact bone (3), mineralized FC volume of the supraspinatus enthesis (4), tissue mineral density of the mineralized FC (5), normalized FC volume relative to HH volume, and (6) trabecular architecture of the HH, excluding the compact bone (trabecular thickness, number, and spacing, and connectivity density). Contours were drawn using Scanco software (Scanco Medical, Brüttisellen, Switzerland) by a single, blinded assessor.

Following micro-CT scanning, samples were decalcified using 14% ethylenediaminetetraacetic acid, rinsed in PBS, dehydrated in 70% ethanol, and embedded in paraffin. Paraffin sections were obtained in the coronal plane at 8 μm thickness and subsequently stained. General morphology of the supraspinatus enthesis was assessed using toluidine blue and hematoxylin and eosin staining. Cellular/nuclear shape (an indicator of cell phenotype) was assessed at P28 from histologic slides stained with toluidine blue (n = 3). Briefly, sections were imaged under bright field light at ×20 magnification. Histologic images were adjusted for threshold to isolate nuclear and cell membranes using ImageJ (U.S. National Institutes of Health, Bethesda, MD, USA), with exclusion criteria set to collect particles of size ranging between 2 and 500 pixels, including holes. Assignment of distinct regions (tendon, FC, and bone) was selected using the wand tool in a blinded fashion by a single assessor. The area selected for the tendon region was defined in the main belly of the tendon, within the field of view, and proximal to the FC and bone. The area selected for the FC region using an ellipsoidal shape where the tendon and bone connect. The bone region was selected as the area between the FC region and the distal edge of subchondral bone, defined by where the trabecular bone/marrow begins. Regional areas were comparable in size between WT and cKO groups (average pixel count: tendon = 4500–6000 pixels; FC = 6000–7000 pixels; bone = 7500–8000 pixels). Images were analyzed using the Particle Analysis plug-in (ImageJ), and the circularity (0–1; 1.0 = perfect circle, 0.0 = increasingly elongated shape) was calculated for cells in each distinct region.

Laser-capture microdissection and quantitative RT-PCR

Supraspinatus muscle–tendon–HH specimens were dissected immediately after killing via decapitation of P7 WT, cKO, and WT BTX-unloaded shoulders (n = 3–6 per group). This premineralization time point was chosen to characterize genes that may play a role in the subsequent mineralization events at P10–P14 (5). Samples were placed in optimal cutting temperature compound (Sakura FineTek, Torrance, CA, USA) and frozen at −80°C for 24 h prior to sectioning. Sectioning of 20–25 μm thick laser-capture microdissection (LCM) samples was performed in RNase-free conditions on a cryostat by a histology technician using polyvinylidene chloride transfer tape (Section-lab, Hiroshima, Japan) and acrylic LCM slides (Section-lab). Sections were deidentified using numerical cataloging by the histology technician for blinding purposes and kept frozen on dry ice until microdissection. Microdissection of muscle, tendon, and FC was performed (Fig. 1C) using gravity-drop isolation (Leica Microsystems, LMD6500, Buffalo Grove, IL, USA) into β-mercaptoethanol-treated lysis buffer (Norgen Biotek, Thorold, Ontario, Canada) within 4 h of sectioning. No less than 16 sections were pooled per region per mouse, and dissected tissue was stored on dry ice for <6 h until RNA extraction. Samples were reidentified and total RNA was isolated following LCM using a commercially available kit for small yield samples (Total RNA Purification Micro kit, Norgen Biotek). First, microdissected tissue was homogenized using micropipette and vortex in lysis buffer and ethanol. The tissue lysate was then treated with proteinase K followed by DNase I treatment. RNA was eluted using elution buffer provided in the commercially available kit, and RNA quality and quantity was measured using RNA 6000 Pico Kit and 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). RNA isolations were repeated from LCM harvests if the concentration of RNA were < 250 pg/μl. Total RNA (2 ng) was then reverse-transcribed using Superscript VILO (Life Technologies, Grand Island, NY, USA) in 20 μl reaction volumes. Subsequent quantitative PCR was performed using a 48 × 48 Dynamic Array integrated fluidics circuit (Fluidigm Corporation, South San Francisco, CA, USA) and Taqman assay probes for genes listed in Table 1 (Applied Biosystems, Grand Island, NY, USA). All samples isolated using LCM were partially or strongly degraded, with RNA integrity number (RIN) ranging from 3 to 6. Specifically, tendon and FC regions were strongly degraded (RIN= 2–4), and muscle was partially to strongly degraded (RIN = 3–6). Therefore, Taq-based assays were used to reduce nonspecific amplification and improve detection sensitivity. All cDNA samples were subject to specific target amplification (Fluidigm Gene Expression Specific Target Amplification; 95°C hold for 10 min followed by cycles of 15 s at 95°C, 4 min at 60°C for 14 cycles) with pooled Taqman assay mix (×0.2). Samples were then amplified in triplicate using quantitative PCR and ΔC_T values were calculated individually using ipo8 as the reference gene. ΔC_T values were averaged between triplicates and converted to fold change values (2^−ΔCT). Quantitative PCR data were log base 2-transformed and presented as mean ± 95% confidence intervals, on a linear scale, for each region.

Figure 1.

A) Supraspinatus tendon-to-bone attachment of WT (left column) and cKO (right column) mice at P14, P28, and P56. Black arrowhead indicates enthesis and dotted line in WT P56 panel indicates the tidemark. Scale bar, 200 μm. B) Quantification of cell circularity (height/width), indicative of cell shape, from tendon, FC, and bone of WT (gray bars) and cKO (black bars) mice. C) P7 supraspinatus tendon-to-bone attachment after laser capture microdissection of the FC region. Similar cuts were made to isolate tendon (dotted circle) cells and supraspinatus muscle from 20–25 μm thick frozen, unfixed sections. The secondary ossification center of the bone appeared black in unstained sections and the enthesis and tendon are morphologically distinct from each other and from the bone. D–G) Gene expression (2^−ΔCT) of bmp-4 (D), tgf-β3 (E), smo (F), and ptch1 (G), normalized to ipo8, are shown for muscle, tendon, and FC of WT (gray bars) and cKO (black bars) groups. B, bone template; E, enthesis; T, tendon. Gene expression data are log base 2-transformed. All data are presented as the mean ± 95% confidence interval. Solid line indicates significant difference between regions within a group or between groups for a specific region (P < 0.05); dashed line indicates significant difference between regions or groups (P = 0.0525).

TABLE 1.

List of Taqman assay probes used for microfluidics quantitative PCR plate

Gene name	Gene symbol	Entrez Gene ID	Taqman Assay ID	RefSeq	Amplicon length
18s rRNA	rn18s	19781	Mm00432359_m1	NR_003278.3	61
Importin 8	ipo8	320727	Mm01255158_m1	NM_001081113.1	79
Ubiquitin C	ubc	22190	Mm02525934_g1	NM_019639.4	176
Bone morphogenetic protein 2	bmp-2	12156	Mm01340178_m1	NM_007553.3	58
Bone morphogenetic protein 4	bmp-4	12159	Mm00432087_m1	NM_007554.2	61
Cytochrome c oxidase subunit II	cox-2	17709	Mm03294838_g1	AF378830	79
β-Catenin	ctnnb1	12387	Mm00483039_m1	NM_001165902.1	77
Early growth response 2	egr2	13654	Mm00456650_m1	NM_010118.3	73
Histone cluster 4, H4	hist4h4	320332	Mm03031915_gH	NM_175652.2	140
Indian hedgehog	ihh	16147	Mm00439613_m1	NM_010544.2	64
Patched	ptch1	19206	Mm00436026_m1	NM_008957.2	69
Runt-related transcription factor 2	runx2	12393	Mm00501584_m1	NM_001145920.2	91
Scleraxis	scx	20289	Mm01205675_m1	NM_198885.3	59
Smoothened	smo	319757	Mm01162710_m1	NM_176996.4	58
Sp7 transcription factor 7	sp7	170574	Mm04209856_m1	NM_130458.3	88
TGF-β3	tgfβ-3	21809	Mm00436960_m1	NM_009368.3	60

Collagen organization

The organization of the tendon and FC collagen fibers was assessed at P28 for both WT and cKO mice using Picrosirius red-stained sections and quantitative polarized light microscopy (n = 5 per group) (31). Histologic sections were imaged in a blinded fashion at ×40 magnification. The distribution of collagen fiber angles was evaluated for the enthesis exclusive of the tendon and up to the subchondral bone using custom-written software (MATLAB, MathWorks, Natick, MA, USA). Angular deviation was calculated for each region within each sample using circular statistics.

Biomechanics

The supraspinatus tendon was chosen for biomechanical analysis based on its clinical relevance in rotator cuff disease (1, 7, 28, 32). To evaluate the mechanical properties of the supraspinatus tendon, humerus–supraspinatus tendon–supraspinatus muscle specimens were dissected intact from shoulders of P28 and P56 WT and cKO mice (n = 8–10 per group/time point). Before mechanical testing, specimens were scanned using micro-CT to determine the tissue cross-sectional near the enthesis. To grip specimens for testing, the humerus was embedded in marine epoxy, the muscle was carefully removed without disturbing the tendon, and the tendon was clamped using thin-film grips (Imada, Northbrook, IL, USA). Specimens were tested in filtered 37°C PBS using a materials testing frame (Instron ElectroPuls e1000, Norwood, MA, USA). The uniaxial tensile loading protocol consisted of 1) preconditioning (5 cycles of 0.1%/s rate between 0 and 5% strain); 2) stress-relaxation cycle and recovery; and 3) uniaxial tension to failure (0.2%/s). Engineering stress was calculated as load divided by the original cross-sectional area (as derived from micro-CT scans). Young’s modulus was determined from the slope of the linear portion of the stress-strain curve. Ultimate load and ultimate stress were defined as the maximal values on load-displacement and stress-strain curves, respectively. Ultimate strain was defined as the strain corresponding to ultimate stress. Resilience was calculated by determining the area under the stress-strain curve prior to yield. Failure location was determined visually.

Activity monitoring

WT and cKO mice were housed in individual cages at P42 (n = 5 WT, n = 3 cKO) to establish baseline metabolism and activity. Gas exchange [CO₂, O₂, Vco₂ (ml/h/kg), Vo₂ (ml/h/kg)], food (g) and water (mL) consumption, and individual mouse weights (g) were measured at 13 min increments for 120 h (Phenomaster, TSE Systems International, Chesterfield, MO, USA). Cage activity was also measured for the first 48 h in each cage using counts of infrared beam breaks (Inframot, TSE Systems International), including time in the center of the cage, time at the perimeter, and rearing. The first 24 h of cage activity was used as an equilibration period, and the activities during 24–48 h were compared. Respiratory exchange rate and calorimetry (kcal/h/kg) were calculated from metabolic measurements using Daco Inhalation software (TSE Systems International, Chesterfield, MO, USA).

Statistical analysis

Statistical analyses and data presentation were performed using Prism software (GraphPad Software, La Jolla, CA, USA, version 6.0d). Results are presented as mean ± 95% confidence interval for biomechanical testing, micro-CT, total activity, rearing, metabolic measurements, and cell shape analysis. Results from LCM gene expression are presented as the log base 2-transformed data as the mean ± 95% confidence intervals. For polarized light microscopy, angular deviation of the distribution of collagen fibril angles for P28 WT and cKO entheses were compared using a one-tailed unpaired t test (α = 0.05). For LCM (log base 2-transformed 2^−ΔCT values), micro-CT (parameters), biomechanical testing, total activity, rearing, metabolic measurements, and cell shape analysis, differences between WT and cKO, WT and BTX-unloaded (LCM only), and regions (LCM only) were tested using planned comparisons via 2-way ANOVA with Holm-Sidak’s multiple comparisons tests (α = 0.05), with total activity, rearing, metabolic measurements, and cell shape comparisons made using repeated measures.

RESULTS

Scx is necessary for the formation of a cell phenotype gradient along the enthesis

Morphologic differences were observed in the supraspinatus entheses of cKO mice compared with WT mice prior to and upon entheseal maturation (Fig. 1A). The entheses of WT mice developed normally, with distinct fibrocartilaginous regions observed at all three time points investigated (Fig. 1A, left column). Specifically, the unmineralized FC at P14 was populated with fibrochondrocytes, with a gradient in phenotype from round hypertrophic cells in the cartilage template, to smaller chondrocytic cells in the enthesis, to elongated fibroblasts in the tendon. By P28, the enthesis was established and distinct between the tendon and subchondral bone, consisting primarily of aligned fibrochondrocytes and increased proteoglycan staining, with the surrounding articular cartilage of the HH not yet fully mature. By P56, the subchondral bone of the enthesis as well as the articular cartilage of the HH were mature, a tidemark was present at the FC-mineralized FC interface, and the enthesis appeared larger than at P28. Enthesis maturation (i.e., the formation of a gradient in cellular and matrix morphology) was impaired in the cKO mice. Specifically, at P14, cKO entheses appeared smaller, with an abrupt transition between the tendon and the cartilage template (Fig. 1A, right column). However, cellular morphology of the cartilage template did not appear altered. By P28, the cKO enthesis was amorphous and smaller compared with that of the WT littermate. At P56, the mature cKO enthesis was populated with rounded chondrocyte-like cells and no noticeable tidemark of the FC was present, in contrast to the WT FC. Cell nuclear shape was generally more round, from tendon to bone, in the cKO mice compared with WT mice at P28. Specifically, the WT cell nuclei and stained membranes had lower circularity in the tendon (i.e., spindle-shaped) compared with the circularity of the enthesis and bone (i.e., round), indicative of a cell morphology (and presumably phenotypic) gradient from tendon to bone (Fig. 1B). This change in cell shape from tendon to bone was not as striking in the P28 cKO enthesis, as indicated by a relatively comparable cell shape from analyzed regions of tendon, FC, and bone.

Consistent with the cell morphology gradient in the WT FC, localized gene expression for a number of genes varied between muscle, tendon, and FC in WT mice (Fig. 1D, G). At P7, expression of the growth factors bmp-4 and tgf-β3 was increased in the FC region compared with muscle or tendon for WT mice (Fig. 1D, E). A gradient in expression was also seen for the hedgehog signaling-related gene patched (ptch1), with up-regulation of ptch1 in FC compared with tendon/muscle for WT mice (Fig. 1G). In contrast, cKO FC demonstrated reduced tgf-β3 and ptch1 expression compared with the WT group and regional expression levels of tgf-β3 were comparable across all 3 regions for cKO mice at P7 (Fig. 1E, G). The expression of ptch1 showed a trend toward lower expression in the FC region of cKO mice compared with WT mice, but these differences were not statistically significant (Fig. 2G, P = 0.11). No difference in expression of smo (Fig. 1F), ctnnb1, egr2, or cox2 (Supplemental Fig. 1A–C) was seen when comparing across tissue regions or between WT and cKO.

Figure 2.

A, B) Two-dimensional cut-plane of micro-CT reconstruction of proximal humerus for WT (A) and cKO (B) mice at P28 and P56, with black pixels highlighting higher density bone. Black arrow highlights supraspinatus FC. C, D) Normalized FC volume (FC volume/HH volume) (C) and FC BMD (D) were measured at P28 and P56. E, F) HH volume (E) and HH BMD (F) were measured at P14, P28, and P56. WT (gray bars) and cKO (black bars) data are represented for each time point; P14 FC volume and BMD were not measured, as the mineralized FC is not established at that time point. An over-bar indicates a significant difference (P < 0.05).

Loss of Scx leads to defects in deposition and mineralization of collagen fibers at the enthesis

Bone morphology and the shape of the HH differed between WT and cKO mice throughout postnatal growth (Fig. 2). In WT mice, the HH was approximately spherical, with tubercles at tendon attachment sites such as the supraspinatus enthesis (Fig. 2A, arrow). However, HHs of cKO mice appeared flattened (Fig. 2B). Mineralized FC was established and visible by micro-CT in WT and cKO mice by P28. The normalized FC volume (Fig. 2C) and the FC BMD (Fig. 2D) increased, for both WT and cKO, with age; however, at P28, the normalized FC BMD was reduced in cKO mice compared with WT mice (Fig. 2D). These differences were not apparent at P56. Additionally, HH volume proximal to the growth plate was significantly smaller in cKO mice compared with WT mice at all postnatal time points investigated (Fig. 3E). Trabecular BMD of the HH increased with age for both WT and cKO, but no differences in HH BMD were observed between groups (Fig. 3F). No significant differences were observed at either P28 or P56 for trabecular properties (Table 2).

Figure 3.

A, B) Polarized light micrographs of WT (A) and cKO (B) mice at P28. The regions of interest used for analysis of collagen fibril orientation are identified in dashed boxes, with tendon, enthesis, and bone and cross-polarizer orientation (45°/135°) annotated. C, D) Average distributions of extinction angles for combined WT (C) and cKO (D) analyses, represented from 45 to 135°, graphically illustrate the angular deviation of collagen fibrils in WT and cKO entheses. B, bone; E, enthesis; T, tendon. Scale bars, 200 μm.

TABLE 2.

Trabecular structure and morphometry of the HH for WT and CKO mice at P28 and P56

Group	Tb.N. (n/mm)	Tb.Th. (μm)	Tb.Sp. (μm)	Connectivity density
P28
WT	5.8 ± 0.7	32.9 ± 3.4	181.7 ± 27.9	294.5 ± 147.5
CKO	6.1 ± 1.3	31.0 ± 4.4	188.6 ± 74.7	193.9 ± 136.7
P56
WT	5.8 ± 0.8	64.7 ± 10.0	196.8 ± 37.9	322.6 ± 100.5
CKO	5.9 ± 0.8	60.5 ± 14.9	187.1 ± 30.7	340.1 ± 119.6

Results are average ± 95% CI. Tb.N., trabecular number; Tb.Sp., trabecular spacing; Tb.Th., trabecular thickness.

The organization of collagen fibers at the enthesis differed between WT and cKO mice at P28 (Fig. 3). At P28, collagen fibers of WT entheses showed a preferred orientation distribution centered around the longitudinal direction of the tendon (i.e., the direction of muscle loading) (∼90°, Fig. 3C). In contrast, collagen fibers in cKO entheses had a much broader orientation distribution (Fig. 3D). Consistent with the qualitative observation of reduced collagen organization in cKO entheses, the angular deviation was significantly greater in cKO entheses (26.3 ± 5.6°) compared with WT entheses (19.2 ± 3.5°).

Loss of Scx leads to impaired attachment between tendon and bone

Representative stress-strain curves for uniaxial tensile tests to failure for WT and cKO supraspinatus entheses at P56 are shown in Fig. 4A. All mechanically tested samples failed at the enthesis, as characterized by visual inspection and video analysis. The ultimate strength (σ_max) of supraspinatus tendon-to-bone attachments increased significantly with age in WT mice (Fig. 4B). cKO attachments were significantly less strong than WT mice at both P28 and P56. Likewise, Young’s modulus increased with age in WT mice from P28 to P56, but not in cKO mice (Fig. 4C). Supraspinatus entheses of cKO mice had significantly reduced Young’s modulus compared with WT mice (Fig. 4C).

Figure 4.

Biomechanical outcomes of the supraspinatus tendon-to-bone uniaxial tensile tests. A) Representative stress-strain curves of P56 WT and cKO uniaxial tensile tests; dotted circle indicates highest measured stress for calculating ultimate stress (σ_max), and the slope of the linear curve was used to calculate Young’s modulus. B, C) Ultimate strength (B) and Young's modulus (C) for WT (gray bars) and cKO (black bars) at P28 and P56. Data are presented as means ± 95% confidence interval. An overbar indicates a significant difference (P < 0.03). E, Young’s modulus.

The role of Scx in enthesis development is likely distinct from the role of muscle loading

A lack of Scx led to smaller animal sizes throughout most of postnatal growth (Supplemental Fig. 2). Cage activity of cKO mice was significantly reduced compared with WT mice at P42 (Fig. 5A). Specifically, total activity in both horizontal (X and Y) directions (XT + YT counts) was significantly lower in cKO mice compared with WT mice (Fig. 5Aa), and cKO mice also exhibited less rearing compared with WT mice over 48 h (Fig. 5Ab). Although oxygen consumption and carbon dioxide respiration overall tended to be reduced in cKO mice compared with WT mice (P = 0.1081 and P = 0.1022, respectively), there were no differences in respiratory exchange rate (Fig. 5Ac), Vo₂, or Vco₂.

Figure 5.

A) Activity of WT and cKO mice at P42 measure for 24–120 h. a) Total activity counts in both horizontal (X and Y) directions (XT + YT; counts) for the first 24 h following a 24 h period of acclimation. b) Rearing at 0–24 and 24–48 h of activity monitor for WT and cKO mice. c) Respiratory exchange ratio for WT and cKO mice in 24 h increments for 5 d of monitoring. Significant differences between groups are indicated by solid black lines (P < 0.05). B) Relative expression of scx (a), bmp-4 (b), tgf-β3 (c), smo (d), and ptch1 (e) for WT and BTX-unloaded shoulders from LCM-isolated regions (muscle, tendon, and FC). Data are log base 2-transformed and presented as means ± 95% confidence interval. A solid overbar indicates a significant difference between regions (P < 0.05); a dashed overbar indicates a significant difference between WT and BTX groups (P = 0.0715).

Due to the reduced activities and animal sizes in cKO mice compared with WT mice, cKO entheses likely developed under less mechanical load that WT entheses. Therefore, developmental defects could have been due to absence of Scx and/or reduced loading. To separate these 2 effects and examine the effect of unloading in isolation, WT entheses were unloaded via local muscle paralysis. Regional and group differences in gene expression were statistically compared for 8 of the 13 candidate genes, as 6 candidate genes were either 1) not expressed in any tissue (e.g., runx-2; ihh; bmp-2 only expressed in WT FC) or 2) not expressed in all 3 groups or regions (e.g., sp7, hist4, and scx). Given the nature of this study, regional expression of scx was compared between WT and BTX groups (Fig. 5Ba). Muscle unloading via BTX led to increased expression of scx in tendon compared with WT tendon at P7 (Fig. 5Ba, P = 0.072). Additionally, for BTX groups, expression of scx, bmp-4, tgf-β3, smo, and ptch1 was increased in the FC region compared with BTX muscle (Fig. 5Bb–e), with similar expression levels in WT compared with BTX groups within all 3 regions. No differences in expression by region or between groups (WT, BTX) were observed for ctnnb1, egr2, or cox2 (Supplemental Fig. 2A–C). Descriptive comparisons of sp7 and hist4 genes are presented in the supplement documentation (Supplemental Fig. 2D, E).

DISCUSSION

Entheses are critical for vertebrate mobility, as these tissues anchor tendons to the skeleton for transmission of muscle forces and subsequent joint motion (2, 3, 23). The structure and function of the enthesis is adaptable during development and growth and unique to each attachment, allowing for minimization of stress concentrations at the tendon–bone interface for a variety of different muscle sizes and types (4, 33–36). Scx plays a crucial role in the fetal development of the bone eminences onto which tendons attach (6, 15–17). The current study demonstrated that Scx is also required for the formation of a functional fibrocartilaginous tendon-to-bone attachment during postnatal growth. Organizational and structural defects developed in the absence of Scx, resulting in mechanical (i.e., functional) impairments in the supraspinatus enthesis and suggesting a role for this transcription factor in postnatal maturation of the enthesis.

The time course of enthesis development has been elucidated in several fetal and postnatal developmental biology studies (6, 21, 25, 27, 20, 37, 38). At the time of limb bud development [approximately embryonic day (E)9.5 in mice], Scx expression is induced and sweeps from the proximal to distal regions of the limb bud (22). By E12.5, Scx-expressing tendon progenitors undergo alignment and organization, driven by TGF-β (19). By E13.5, progenitor cells condense and undergo differentiation into distinct, force-transmitting tendons (15). In the absence of Scx, condensation and differentiation of tendon progenitor cells into distinct tendons is obstructed, suggesting that Scx is required during progenitor condensation for tendon development (15). At the same time as tendon progenitor condensation (i.e., E13.5), bone ridge formation of the proximal humerus is initiated, eventually forming the deltoid tuberosity (6). The initiation of the deltoid tuberosity attachment unit is derived from a colocalized population of cells that are established earlier, at E11.5, and express both Sox9 and Scx (20). The initiation of the tuberosity is muscle independent; however, growth of the tuberosity is controlled by muscle loads (6). Mineralization and maturation of the enthesis occurs in the early postnatal period. A distinct population of hedgehog-responsive cells populates the enthesis and mediates FC mineralization (25). For the supraspinatus enthesis, mineralization begins at ∼P14 and reaches morphologic maturity in cell and extracellular matrix organization by P28 (5, 7, 39). In the current study, Scx was absent throughout the entire time course of enthesis development. Therefore, the dramatic defects in cKO mice for formation of a cell phenotype gradient, enthesis mineralization, collagen organization, and attachment mechanical properties may have been due to the role of Scx in formation of bone ridges during fetal development or due to an unknown role of Scx for enthesis mineralization or collagen modeling. Although it is still unknown what cell population establishes the enthesis and is responsible for collagen deposition and organization, recent work has shown that the resident population of cells at the attachment site express hedgehog-related markers such as gli1 (25, 40). These gli1-expressing cells regulate to the mineralization of the mature enthesis (25, 40); however, gli1 is not expressed until after the attachment is established.

Establishment of the attachment during embryonic growth is an orchestrated event cued by concomitant signaling via Sox-9 and Scx, whereby a pool of progenitors is established that eventually form the tendon–bone attachment units (20). The temporal regulation of Scx at attachment sites such as the deltoid tuberosity, and its coexpression with Sox9, has until recently been unknown (20). In situ hybridization for scx has identified this factor to be primarily expressed in differentiated tendon cells by E13.5 in mice (20). We chose to use the Prx1Cre promoter for limb-bud progenitor deletion of Scx given that the enthesis has a heterogenous population of cells. However, the secondary effects of the loss of Scx during postnatal tendon elongation were not identified in this study due to the loss of Scx throughout limb development. Of note, it is possible that some or all of the cKO mice used in this study had partial knockdown of Scx in all tissues, as the breeding schema chosen led to ubiquitous recombination of Scx in oocytes (i.e., female Prx1Cre females). However, the controls used for this study did not have Cre and these controls were negative for the Scx-null gene. Conversely, unpublished observations in our laboratory and by others (15) have shown that there are no gross phenotypic differences or differences in postnatal growth or viability observed in mice that have global deletion of Scx (Scx-null mice that are Cre-negative) compared with mice that have deletion of Scx via Prx1Cre (mice used in this study). Future studies using an inducible Cre or a more targeted deletion [e.g., Sox9-CreER (20, 41)], are needed to determine if Scx is dispensable in postnatal development. Previous work has demonstrated that Scx mediates BMP-4 signaling during limb formation and growth, driving bone ridge formation, and these molecular cues induce the coordinated attachment of tendon to bone (6). Conversely, during late embryonic and postnatal maturation, hedgehog-responsive cells take residence at the attachment unit site, and hedgehog signaling is required for mineralization and robustness of the mature FC (25, 26). In the current study, we showed that a phenotypic gradient formed during normal postnatal development, with increasing expression of BMP, TGF-β, and hedgehog-related genes from muscle to tendon to FC. Furthermore, this gradient was lost, and these genes were down-regulated, in the developing FC in the absence of Scx signaling. This implies that Scx (presumably expressed in a gradient, as shown at fetal time points) (21) is necessary for the formation of this gradient in cell phenotype across the tendon-to-bone attachment.

Muscle denervation via BTX-unloading did not lead to an altered expression gradient of the genes of interest in this study; however, BTX-unloading did lead to increased levels of scx expression. These changes may have been an injury response to the altered mechanical environment of the tendon imposed by loss of muscle contraction. We previously showed that, in adult rats, injury to the supraspinatus tendon leads to an increase in expression of scx, suggesting its role in the remodeling phase of healing (42). Conversely, scx expression decreases with age in mouse Achilles tendon, with high levels expressed at 2 mo of age and very little expression by 4 mo of age (43). During embryonic development in mice, tgf-β3 expression increases from E11.5 to E14.5 (38). However, there is a paucity of data on the temporal expression of scx, tgf-β3, and other tenogenic factors during the linear growth phase of postnatal development. In the present study, we did not investigate the normal expression of scx or other tenogenic factors throughout the postnatal linear growth period (P0–P56). Therefore, we cannot say with certainty that scx expression is altered or maintained during growth. However, as expression of both scx and tenomodulin decreases with increasing age in adult mouse tendons (43), it is possible that expression of scx peaks at the time of tendon maturation, when the structure of the tendon is established. It is also possible that, during the postnatal growth phase of tendon, scx expression is up-regulated because the cells in the growing tendon and peritenon are subjected to stretching, are proliferating, and/or are differentiating. Additional study is necessary to determine the temporal expression and necessity of scx and tenogenic genes during postnatal maturation with and without altered muscle loading.

The entheseal defects observed in mutant mice in the current study support the hypothesis presented previously by others that tendon growth, in this case driven by Scx expression, influences the growth of periosseous tissues such as the tendon enthesis (6, 15, 20). Adaptations to the periosteal surfaces may be independent of changes to the medullary cavity, which may explain the finding in the current study of a lack of trabecular disruption of the proximal humerus in the absence of Scx. In contrast to a lack of changes in the trabecular bone, changes in mineralization were apparent in the enthesis FC when Scx was deleted. Reduced bone ridge formation may be a result of a single or concomitant factors associated with musculoskeletal growth in the absence of Scx, such as altered patterning of the bone during embryonic development, reduced loading throughout growth due to lack of robust force-transmitting tendon development, and/or altered establishment of the attachment unit progenitor cells. The current study, as well as previous work by others (6, 15, 20), demonstrated that Scx is required for establishing and maintaining the attachment unit.

The effects of Scx in the current study may have been due to a direct effect of Scx on postnatal enthesis development or an indirect effect of the absence of Scx during tendon development that led to impaired mobility and reduced loading across the developing enthesis. Lineage tracing studies using Scx-Cre transgenic mice identified Scx as being expressed at one time by all tendon and entheseal cells, but expression levels are thought to decrease significantly postnatally (25, 43). Previous studies utilizing the Scx-Cre transgene to delete downstream modulators of musculoskeletal assembly, such as bmp-4 and smoothened, have demonstrated periosteal and articular abnormalities independent of trabecular alterations, similar to the current study (6, 25). Mice lacking Scx have movement disorders, such as hyperdorsiflexion of the forepaw (15), which may be associated with decreased mobility and access to food during postnatal growth. Our results quantified these prior observations and demonstrated reduced mobility and activity. However, despite these changes, defects to trabecular bone morphometry (e.g., reduced trabecular number and increased trabecular spacing) were not seen. This is contrary to changes previously associated with movement disorders such as muscular dystrophy (44) and disuse osteopenia (45). Further evidence from the current study points to different mechanisms for unloading-induced effects and Scx-induced effects on enthesis development: muscle unloading via BTX-induced paralysis did not lead to apparent changes in local gene expression patterns, unlike the loss of the cell phenotype gradient seen when Scx was deleted. Therefore, abnormalities associated with bone morphometric outcomes in cKO mice are likely not entirely dependent on impaired mobility and growth.

There were a few limitations to our study. Although the work identified a critical role for Scx in enthesis growth, maturation, and mineralization, it did not identify the mechanism by which Scx expression influences enthesis organization and mineralization. Despite the lack of a clearly defined mechanism, the results of the current study are valuable for generating mechanistic hypotheses for future studies. Gene expression data implicated 3 major signaling pathways, BMP, TGF-β, and hedgehog, in enthesis postnatal development and related to the loss of Scx. Recently, it was shown that Scx-lineage cells control mineralization of the supraspinatus attachment unit, independent of the secondary ossification center of the proximal humerus, via the hedgehog pathway (25, 26). Targeted deletion of smoothened from Scx-expressing cells led to reduced FC volume, impaired mineralization, and reduced mechanical properties (25). The formation of skeletal superstructures, such as tuberosities and ridges, are also regulated by Scx/BMP signaling (6). Conversely, the mineralization and maturation of these structures is likely controlled by both the hedgehog pathway and mechanical loading (25, 27). Nonetheless, in the absence of either Scx or hedgehog signaling, an enthesis structure still forms to connect tendon to bone, albeit defective in organization and mechanical properties. Future work is needed to better identify the mechanisms of assembly of musculoskeletal structures such as the enthesis, as well as patterning of these structures and superstructures on the skeleton.

In summary, the current work demonstrated an essential role of Scx in the maturation of a fibrocartilaginous tendon enthesis. Loss of Scx during embryonic development prevented the formation of a cellular and molecular gradient that is typically seen in the normal enthesis. This was likely due to both regulation of factors associated with Scx (i.e., bmp-4, tgf-β) as well as biophysical changes due to the loss of Scx (i.e., reduced mobility).

Glossary

BMD

bone mineral density

BMP

bone morphogenetic protein

BTX

botulinum toxin A-treated

cKO

conditional knockout

CT

computed tomography

E

embryonic day

FC

fibrocartilage

HH

humeral head

LCM

laser capture microdissection

P

postnatal day

Prx1Cre

Scx^flx/flx and paired related homeobox 1-Cre

RIN

RNA integrity number

Scx

Scleraxis

Sox9

sex-determining region Y-box 9

WT

wild-type