Online Resource Appendix 1. Supplementary protocols for animal surgery, vivaCT imaging,
biomechanical test, histology and immunohistochemistry
Animal surgery
To perform ACL reconstruction, animals were sedated by intramuscular injection of 10%
ketamine/2% xylazine (Kethalar, 0.3mL: 0.2mL) and maintained sedation with 10% ketamine injected
intravenously (Sigma Chemical Co, St Louis, MO, USA). The ipsilateral flexor digitorum longus
tendon (20mm in length and 1mm in diameter) was harvested through a longitudinal medial incision.
The ACL was excised and completely removed. A tibial tunnel and a femoral tunnel of 1.1mm in
diameter and about 7mm in length were created from the footprint of the original ACL to the medial
side of the tibia and lateral-anterioral femoral condyle, respectively, with an angle of 55 o to the articular
surface. The graft was inserted and tight fitted into the bone tunnels, fixed on the femoral and tibial
tunnel exits with suture tied over the neighboring periosteum. The graft was sutured to the femoral
tunnel exit first. Then a tension of 0.7N was applied by hanging an object of constant weight of 72g at
30° of knee flexion while suturing the graft to the tibial tunnel exit. No suture was found within the
bone tunnel. Soft tissue was closed in layers. Lachman test was performed both after ACL excision and
after ACL reconstruction. Only one randomly-selected limb was operated. 0.1 ml of Temgesic was
injected intra-muscularly when the rat regained consciousness. The animals were allowed to have free
cage movement immediately after operation. Except redness and swelling at the reconstructed site, no
other postoperative complications were observed after the surgery.
vivaCT imaging and data analysis
After thawing the femur-graft-tibia complex and removing the surrounding muscles, the region
covering the entry and exit of bone tunnel was scanned with vertical displacement of 30μm for about
220 and 350 consecutive sections, respectively, for the femoral and tibial tunnels at the resolution of
35m. The whole process took 20 minutes and the sample was kept moist during scanning. The sample
was then subjected to biomechanical test immediately after scanning. The consecutive sections were
3-dimensionally (3D) reconstructed and rotated to align the bone tunnel vertically using the built-in
software. To measure the mass and density of the mineralized tissue inside the bone tunnel, a circular
region of interest (ROI) of 1.2mm in diameter, covering the bone tunnel and closest to the initial tunnel
diameter that was allowed by the vivaCT built-in software and confirmed to be appropriate by the low
bone volume/total volume fraction (BV/TV) of the ROI of the samples harvested at day 0, was used to
represent the perimeter of the bone tunnel. The same circular ROI was selected for different sections
along the length of the bone tunnel and these ROIs were 3D reconstructed after thresholding to remove
the signals from the soft tissue (e.g. tendon graft) (Figure 1B). To measure the peri-tunnel bone mass
and density, a 0.1cm x 0.1cm square ROI was chosen next to the bone tunnel at the trabecular bone
region. The selected square ROIs in consecutive sections were 3D reconstructed after thresholding
(Figure 1C). Bone mineral density (BMD) as defined by vivaCT (unit: mg HA/ccm (milligram of
hydroxyapatite per cubic centimeter)) and BV/TV were calculated for the selected volume covering the
whole femoral tunnel, the epiphyseal and the metaphyseal segments of tibial tunnel (Figure 1A). Due to
its shorter length and located only at the epiphyseal region, the femoral tunnel was not further divided
into 2 segments in vivaCT analysis. About 50 sections were analyzed in each segment and were very
consistent among different animals.
Biomechanical test
After vivaCT scanning, the two bone ends of the femur-graft-tibia complex were placed in
polytethylene tubes and potted in methyl-methacrylate. The samples were then stripped of all soft
tissues, including the capsule and other knee ligaments. Cross pins were passed through the potting
tube and the bone shaft to provide additional fixation. The whole construct was mounted onto a
material testing machine (Hounsfield H25K-S, Hounsfield Test Equipment Ltd., Salfords, Redhill, UK)
with the two tunnels oriented along the direction of tensile load. The sutures fixing the tendon to the
bone tunnel were removed. To set the “zero” load state and ensure that the graft was now under tension,
a displacement was applied with a 50N or 10N load cell slowly until the loading force increased (about
0.02N), and then the load cell was returned towards the initial position slightly. The construct was then
loaded at a displacement rate of 20mm/minute using a 50N or 10N load cell until failure after 5 cycles
of preconditioning at a maximum displacement of 0.5mm, which was the smallest value possible for
the machine, in moist condition. Failure of the construct was confirmed by the complete dissociation of
the femoral and tibial halves of the complex which normally occurred at the displacement of
0.6-1.5mm and we stopped pulling at the displacement of about 20mm. The sample was kept moist
during the whole experiment. The ultimate load to failure (N) and the failure mode were recorded. The
slopes of the linear portion of the load (Y) - displacement (X) curve were calculated at the fixed
displacement interval of 67.5µm using the formula (Y2-Y1) / (X2-X1) and the maximum slope was
taken as the stiffness (N/mm) of the healing complex. Absolute values were compared as all the
specimens had the same tunnel diameter and similar graft diameter. The ultimate load and stiffness of
normal ACL of another group of rats (n=9) equivalent to post-operative week 6 were used for
comparison.
Histology and immunohistochemistry
The femur-graft-tibia complexes were fixed in 4% neutral buffered formalin, decalcified in 9%
formic acid for about 4-6 weeks with the exact time determined by X-ray and embedded in paraffin.
The femoral and tibial tunnel complexes were divided into two, with the intra-articular graft
mid-substance preserved and attached to the tibial tunnel. 5μm-thick longitudinal sections parallel to
the direction of the bone tunnel were cut and stained with Hematoxylin and Eosin (H&E) for the
examination of healing at the tendon graft to bone tunnel interface and graft mid-substance under light
microscopy. Sharpey’s fiber was examined by polarized light microscopy (Leica DMRXA2, Leica
Microsystems Wetzlar GmbH, Germany). Only those sections that the entry and exit of bone tunnel
could be identified were selected for histological assessment. One section was selected for histological
assessment of each tunnel segment of each femur-graft-tibia complex.
The immunohistochemical staining of MMP1, MMP13 and CD68 (a monocyte-macrophage
lineage marker that labels the potential osteoclast precursors and osteoclasts) was done using the Lab
Vision kit (Thermo Fisher Scientific Inc., MI, USA). Briefly, after removal of paraffin and rehydration,
endogenous peroxidase activity was quenched with 3% hydrogen peroxide for 20 minutes at room
temperature. Antigen retrieval was then performed with citrate buffer at 65oC for 20 minutes. After
blocking with Ultra V block of the Lab Vision kit, the sections were stained with polyclonal rabbit
antibody against rat/human/mouse MMP1 (NBP1-72209; 1:100; Novus Biologicals, CO, USA), mouse
monoclonal antibody against rat/human MMP13 (ab3208; 1:100; Abcam, Cambridge, UK) or mouse
monoclonal antibody against rat/mouse CD68 (ab31630; 1;100; Abcam, Cambridge, UK) in a humid
chamber at 4oC overnight. The sections were then treated with primary antibody amplifier of the Lab
Vision kit for 10 minutes at room temperature, followed by horse radish peroxidase (HRP) polymer
Quanto from the same kit (which is a micropolymer conjugated with HRP) for another 10 minutes at
room temperature. The spatial and temporal expression of the target proteins was then visualized by
adding DAB Quanto Chromogen for 3 minutes at room temperature. Afterwards, the sections were
rinsed, counterstained in hematoxylin, dehydrated with graded ethanol and xylene, and mounted with
p-xylene-bis-pyridinium bromide (DPX) permount (Sigma Aldrich, St Louis, MO, USA). Primary
antibody was replaced with blocking solution in the controls. For good reproducibility and
comparability, all incubation times and conditions were strictly controlled. The sections were examined
under light microscopy (Leica DMRXA2, Leica Microsystems Wetzlar GmbH, Germany).