In vivo analysis

All horses had moderate to marked ultrasonographic changes in the tendons (Figure 2). Most had generalised lesions with mixed central hypoechogenicity. Only three horses (2 stem cell and 1 saline control) had more focal hypoechoic core lesions. In all cases the contralateral limb had either no significant abnormalities on ultrasound (8/13 horses) or milder pathology (5/13 horses). No horses exhibited adverse effects after the intra-tendinous injections and all horses underwent the same post-injection exercise programme on the horse-walker (walking and trotting only) without lameness in 12 horses. One horse treated with stem cells was eliminated in further analysis due to sudden onset lameness during the rehabilitation phase of the study, which was associated with recurrent injury of a very severe tendon injury in the treated limb.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Ultrasonography of the superficial digital flexor tendon.

Representative ultrasonographic appearance of (A) stem cell injected horses and (B) control (saline-injected) horses at the time of injection and at the conclusion of the study (6 months). Both transverse and longitudinal images are shown to illustrate the injury which is visible as a hypo-echoic area (arrows) denoting the loss of extracellular matrix, which is not present at 6 months. The linear organization of the healed tendon is apparent in the longitudinal images at 6 months.

https://doi.org/10.1371/journal.pone.0075697.g002

Cross-sectional area of the affected SDFT at the time of treatment (cell or saline injection) was larger than the normal cross-sectional area, consistent with injury [6], but was not significantly different between the groups (Figure 3A). After 6 months, however, there was no significant difference in the cross-sectional area in the stem cell treated group compared to the contralateral tendons, however the cross-sectional area in the saline control group remained significantly larger compared to the contralateral (p=0.006) and to stem cell treated (p=0.045) tendons. This was supported by the analysis of the change in cross-sectional area as a percentage of the cross-sectional area at the time of implantation, which showed that the treated group had a reduction in area while the control group increased although this was not statistically significant (Figure 3B). The cross-sectional area of the SDFT in the contralateral limbs was smaller than the affected tendons and inside the normal range, confirming the mild nature of pathology in the contralateral limbs. There was no statistically significant change in the cross-sectional area of the untreated contralateral limbs over the period of the rehabilitation. As there was no significant difference in the contralateral limbs between groups, the data for the contralateral limbs were pooled for the analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Cross-sectional area of the superficial digital flexor tendon.

(A) Cross-sectional areas were measured ultrasonographically at the time of implantation (Stem Cell 1, Saline 1 and Contra 1) and at 6 months (Stem Cell 6, Saline 6 and Contra 6). (B) Relative change in cross-sectional area between groups. The Wilcoxon Signed-Rank test showed that there was no difference in cross-sectional area between 1 and 6 months for either the stem cell treated group or the saline group, or between the contralateral limbs in each group. Contralateral limbs were therefore combined from both groups for statistical analysis. The values for normal horses are shown as mean (thick dashed line) with two standard deviations (thin dashed lines). * p = 0.006, ** p= 0.003, ***p<0.045.

https://doi.org/10.1371/journal.pone.0075697.g003

In vitro analysis

The hydration of the SDFT at the injury site of the treated group (65.63±3.78%) was not significantly different to the contralateral SDFT (62.73±2.2%) but both were significantly different to the saline treated control group (69.55±3.6%), (p=0.04 compared to the stem cell group; p=0.01 compared to the contralateral SDFT).

A similar pattern was seen for tendon stiffness with the treated SDFT having significantly lower overall tendon structural stiffness than the control SDFT (by ~25%; p=0.015) and was similar to the stiffness of the contralateral tendon (2.2% different) (Figure 4). When a measure of the modulus of elasticity (tissue material property) was calculated using the cross-sectional area measured ultrasonographically at the maximum injury zone immediately prior to euthanasia, there was a higher modulus for the treated tendons compared to controls, although this was not statistically significant (Figure 4). In both cases, the modulus was not significantly different to the contralateral tendons.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Biomechanical properties of tendons.

SDFTs were subjected to mechanical testing within 2 h of post euthanasia to determine (A) structural stiffness and (B) material stiffness (modulus), which was calculated by expressing the ratio of the structural stiffness to the ultrasonographically-determined cross-sectional area. Structural stiffness was significantly reduced in stem cell treated tendons and closer to that of the uninjured (contralateral) tendons. * p=0.015.

https://doi.org/10.1371/journal.pone.0075697.g004

Histologically, treated tendons had lower (indicating improved) organisational scores at the injured site compared to the control group, which were also significantly different to the contralateral tendons (Figure 5A; p=0.003 and p=0.002, respectively). The contralateral tendons had a score of 14.2±9.5 and so were not histologically normal reflecting the high frequency of bilateral (but milder) disease and there was no significant difference in scores between the contralateral limbs of both groups. This improved histological score in the treated tendons was supported by similar significant differences in the crimp scoring from polarised light microscopy (Figure 5B). It was also evident that the MSC-treated group had low cellularity scores (Figure 6; p=0.007), which again were comparable to remote, less injured sites. The lower cellularity was supported by lower DNA content (Figure 6; p=0.037). The treated group had significantly lower vascularity compared to the controls, which was similar to the contralateral limbs (Figure 7, p=0.026).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Morphological parameters of tendons.

(A) Organisation and (B) crimp as assessed by blinded scores were taken from two separate histological sections for each tendon. Representative micrographs of haematoxylin and eosin stained histological SDFT sections from the treated and remote (less injured) sites in both MSC-treated and saline-treated groups are shown in the panels on the right. Normal tendon has a score of zero and the higher the score the more disrupted the tendon matrix and less crimp pattern, respectively. (A) * p = 0.003; ** p < 0.002; (B) * p = 0.046; ** p < 0.0001 (scale bars 50 µm).

https://doi.org/10.1371/journal.pone.0075697.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Cellularity of tendons.

Both (A) cellularity scores and (B) DNA content demonstrated reduced cellularity in MSC-treated tendons. The values for normal horses are shown as mean (thick dashed line) with two standard deviations (thin dashed lines). (A) * p=0.007; ** p=0.001; (B) * p=0.037; ** p<0.001.

https://doi.org/10.1371/journal.pone.0075697.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Vascularity of tendons.

There was reduced vascularity in the MSC-treated tendons, similar to normal tendon, compared to saline-injected controls. * p = 0.026; ** p = 0.004.

https://doi.org/10.1371/journal.pone.0075697.g007

Total collagen (as hydroxyproline) content in MSC-treated SDFTs was unaltered between groups (Figure 8), while sGAG content was reduced and closer to the contralateral tendon, which was in the normal range published previously for normal tendons [38]. The sGAG content of the control group remained significantly elevated (p=0.005). There was no significant difference in the tissue-linked fluorescence between treatment and control groups although both were significantly lower than the contralateral limb (p=0.039 for treated and p=0.003 for controls) (Figure 9), which suggested similar amounts of new collagen in the injured limbs. A higher remodelling rate, exemplified by MMP-13 activity, was also demonstrated for the control group compared to the treated group (p=0.045) and contralateral limbs (p=0.015) (Figure 9).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 8. Compositional analysis of tendons.

Concentrations for (A) hydroxyproline and (B) sGAGs are shown. The values for normal horses are shown as mean (thick dashed line) with two standard deviations (thin dashed lines). Note that the sGAG content of MSC-treated tendons was closer to the normal range and not significantly different from the relatively uninjured contralateral limbs (p = 0.24). * p = 0.05.

https://doi.org/10.1371/journal.pone.0075697.g008

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 9. Collagen re-modelling in injured and stem cells treated tendons.

(A) Tissue-linked fluorescence and (B) MMP-13 activity. The increased MMP-13 activity seen in the saline-treated group matched the lower tissue-linked fluorescence suggesting greater remodelling of the collagen matrix in the saline-treated compared to the MSC-treated tendons. (A) * p = 0.039; ** p = 0.003; (B) * p = 0.045; ** p = 0.015.

https://doi.org/10.1371/journal.pone.0075697.g009

Conclusion

This study has demonstrated that the technique of MSC implantation with marrow supernatant is not only safe but also indicates significant benefits to the healing tendon in normalising biomechanical (reduced stiffness), histological (better organisation) and compositional parameters (lower GAG content) towards those levels in normal tendon, which could be considered surrogate indicators of regeneration. The findings provide evidence that the implantation of autologous MSCs has beneficial therapeutic effects on the healing of superficial digital flexor tendon over-strain injuries which supports the reduction in re-injury rate seen clinically with this treatment [36]. In addition, these data support the potential use of MSCs in bone marrow supernatant for treating human tendon injuries.

Ethical Statement and recruitment of horses

The study was carried out following informed consent from the owners for horses to be donated to this study and under approval from the Ethics and Welfare Committee of the Royal Veterinary College (URN 2013 1230R 2005) and under UK Home Office Licences. Thirteen Thoroughbred and Thoroughbred cross horses (all male castrates), aged between 5 and 15 years of age (7.8±3.0 years), and suffering career-ending severe superficial digital flexor tendonitis within the metacarpal region were recruited for this study (Table 1). The selection criteria were that each horse had a recent injury to the SDFT in the mid-metacarpal region of less than 2 months (average 30 days between injury and entry into the study). Additional requirements were that the paratenon of the SDFT had to be intact and no horse had received previous intra-tendinous injections, thereby reducing potential loss of cells after implantation. Horses could have received anti-inflammatory agents but all systemically administered drugs had been cleared by the time of implantation because of the interval between entry into the study and treatment. It was not possible to determine with confidence if a previous injury had been present, although, as the horses had suffered career-ending injury which rendered them uneconomic to attempt to return them to racing, the injuries were severe, in many cases completely disrupting the entire tendon matrix (Figure 2). Naturally-occurring tendinopathy frequently manifests with bilateral involvement (a comprehensive epidemiological study documented a 35% frequency of bilateral injuries [6]) although the contralateral limb usually has milder pathology or else was injured previously (e.g. chronic disease). For the horses recruited for this study, pathology in the contralateral limbs was recorded but the limbs were not treated and were recovered for comparative evaluation.

Ultrasonography of forelimbs

The SDFTs were examined by ultrasonography of the palmer aspect of the forelimbs, performed with the horse in weight-bearing stance as described previously [6,58]. Briefly, the limbs were prepared by clipping the hair, cleaned with a surgical scrub and then spirit and acoustic coupling gel applied. Images were obtained with a 13 MHz linear transducer to generate serial on-incidence transverse and longitudinal images of the soft tissues throughout the palmar metacarpal region (sequentially labelled levels 1-7 [6]) and stored digitally. Scans were performed at admission into the trial (time 0), then at 1 month (time point for cell or saline injection), 3 and 6 months. Transverse images of the SDFT were used to obtain the cross-sectional area (CSA) at the maximal injury zone (MIZ) of the injured tendon and at the same level of the contralateral limb with the aid of image analysis software (Scion Image, Scion Corporation, USA). All scans were obtained by a single operator (R.K.W.S).

MSC isolation, in vitro culture expansion and characterisation

Autologous BM was recovered from the sternum under standing sedation using a standardised protocol [36] and shipped to the laboratory (VetCell BioScience Ltd, Kingham, Oxfordshire, UK) using validated shipping containers for MSC recovery and expansion.

In addition to cell expansion for in vivo implantation, the MSCs were characterised for their ability to undergo tri-lineage differentiation, population doubling time (PDT) and colony forming unit-fibroblast (CFU-F) assays.

Osteogenic differentiation

MSCs (between passage 2-4) were seeded at a density of 1 ×10⁴ cells/cm² in 35 mm² dishes and allowed to expand until about 60-70% confluent in D10 medium (Dulbecco’s modified Eagle’s medium (DMEM), supplemented with foetal bovine serum (10% v/v), 100 U/ml penicillin, and 100 U/ml streptomycin) (Sigma-Aldrich, UK). The D10 was replaced with osteogenic media [59] which consisted of D10 medium supplemented with 10^-6 M dexamethasone, 10 mM β-glycerophosphate, 0.02 mM L-ascorbic acid-2-phosphate (all from Sigma-Aldrich, UK). Control cultures were cultured in D10 medium alone. Media were replaced every 3-4 days for a total of 24 days. To assess mineralization, monolayers were fixed in 4% formaldehyde then stained with the von Kossa technique [60] using silver nitrate (Sigma-Aldrich, UK).

Adipogenic differentiation

Cells seeded and expanded as above were induced to adipogenic differentiation in D10 supplemented with 1% ITSS media supplement (insulin, transferrin, sodium selenite), 1 µM dexamethasone, 100 µM indomethacin, 500 µM 3-isobutyl-1-methyl xanthine (all from Sigma-Aldrich, UK). After 72 h the cells were cultured in adipogenic maintenance media which consisted of D10 medium supplemented with 1% ITSS for a further 3 days. Four cycles (16 days) of this regime were performed, following which the cells were maintained in adipogenic maintenance media for a further 7 days [61]. Control cells were grown in maintenance medium for the entire period of differentiation. Cells were then fixed with 4% formaldehyde prior to staining with fresh Oil-Red-O solution to assess lipid droplet accumulation [61].

Chondrogenic differentiation

2.5 x 10⁵ cells were centrifuged at 500 ×g for 5 minutes in 15 ml polypropylene conical tubes (FACS tubes; BD Biosciences, UK) and the pellet cultured in 0.5 ml of serum-free chondrogenic medium supplemented with 10 ng/ml transforming growth factor-β3, 0.1 µM dexamethasone, 0.17 mM ascorbic acid-2-phosphate, 1 mM sodium pyruvate, 0.35 mM L-proline, 1% ITSS, 1.25 mg/ml BSA, 5.33 µg/ml linoleic acid (Sigma-Aldrich, UK). Control cell pellets were grown in growth media. Media was replaced every three days up to 24 days after which the diameters of cell pellets were measured before fixing in 10% buffered formalin for 24 h and embedding in paraffin and sectioning. Sections (5–8 µm thick) were stained with 1% Safranin O and 0.02% Fast Green and counterstained with haematoxylin.

Assessment of differentiation

The quality of the differentiation ability of the MSCs was visually assessed on the amount of coverage of respective stains for osteogenic and lipogenic differentiation. Thus, this was noted as negative if stained cells covered less than 10% of the dish; moderate if 10–50% and positive if greater. Chondrogenic differentiation was assessed as positive if pellet diameters grew to greater than 2 mm; moderate if 1-2 mm and poor if less than 1 mm. Proteoglycan deposition in cell pellets was confirmed by Safranin O staining of histological sections.

CFU-F assays

Cells at low density (5 and 10 cells per cm²) were seeded in 35 mm dishes and cultured in D10 medium for 14 days at which time cell colonies were fixed in 4% formaldehyde for 10 mins. Colonies were visualised by staining with 0.5% crystal violet for 10 mins.

Population doubling assays

Cells were seeded in triplicate at a density of 800 cells per cm² and cultured for 3 days in D10 medium. Cells were then lifted by trypsin treatment and suspended in D10 for counting. Cells were re-seeded at the same density and cell counting repeated another 5 times. The population doubling time was calculated from the log₁₀ of the ratio of the cell numbers at 36 h to the starting cell number.

MSC implantation

10 x 10⁶ MSCs were suspended in 2 ml of citrated BM supernatant which was collected at the same time as the other BM samples in sodium citrate anticoagulant, centrifuged (1,500 x g) to remove cells and particulate material, and stored at -20 °C until required. The control group received 2 ml of isotonic saline (Dechra Veterinary Products, UK). For all horses, injections were performed on the most severely affected limb 3-4 weeks after bone marrow aspiration. A 2 inch 19G needle was guided under ultrasound guidance into the middle of the lesion (maximum injury zone) at 2-4 sites to ensure adequate spread throughout the lesion (as assessed ultrasonographically). The limb was then immediately bandaged and a single peri-operative dose of intra-muscular antibiotics (Norocillin, Norbrook Laboratories (GB) Ltd, Corby, UK) administered.

Post-implantation exercise regime

The horses were box-rested with the limb bandaged for 7 days after which the horses began a standardised ascending exercise rehabilitation regimen of walking (increasing from 10-45 mins over the first 3 months) and then combined with trotting (increasing from 5-20 mins) in the second 3 months using a horse-walker. Horses were trotted in a straight line at the time of BM aspiration, pre-implantation and 1, 3 and 6 months after implantation to assess for obvious lameness (tendinopathy of the SDFT characteristically only causes significant lameness during the acute inflammatory stages (typically 1-2 weeks). At the same time, the palmar aspect of the metacarpus was evaluated ultrasonographically using standard transverse and longitudinal images of the SDFT as detailed above.

Tendon analysis

Horses were euthanized after 6 months post-implantation and tendons were harvested from both forelimbs from each horse and analysed as follows

Mechanical properties

The SDFT was removed in its entirety from both forelimbs and mounted in freezer clamps within 2 h in a 50 kN electro-hydraulic material testing machine (Dartec HA50; Zwick Roell Ltd, UK; retrofitted with an Instron 8800 control system; Instron, High Wycombe, UK) to give an inter-clamp (gauge) length of between 120 and 250 mm consisting of the damaged area of the SDFT. Non-destructive testing was performed on the tendon with loading up to 4 kN (one quarter of the average failure load found in similar equine tendons) for 20 cycles (to allow for pre-conditioning) at 0.5 Hz. The stiffness was calculated from between 3 and 4kN on the last loading cycle where the points lay on a linear part of the force-deformation curve. The gradient of the best-fit line (R²>0.99 in all cases) was used as the stiffness (N/%).

After mechanical testing the SDFT was divided into seven 4 cm sections matching the extent of the ultrasonographic examination from proximal to the distal extent of the SDFT. Each 4 cm section was sub-divided into four 1 cm sections. Peripheral tissue was discarded before the samples were prepared for histology and compositional analysis.

Histological examination

Tendon sections were prepared for histology as previously described [58]. Briefly, tendon sections were embedded in optimal cutting temperature compound (OCT, Sakura Tissue-Tek®, The Netherlands), and snap frozen in pre-chilled (-80 °C) n-hexane and stored at -80 °C. Serial sections (8–10 µm thickness) in the longitudinal plane were cut on a cryostat (Bright Instrument, UK) and mounted onto poly-L-lysine coated slides (VWR, UK) and stained with Harris’s haematoxylin and 1% alcoholic eosin (Sigma-Aldrich, UK). Sections were cleared in Histoclear before mounting in DPX mounting medium (BDH, Poole, UK). Stained sections were imaged at the same magnification (x 20 objective lens; Olympus BX60 microscope) to obtain five images each for the peripheral and central (core) regions of the tendon. Two sections were imaged for each sample and evaluated in a blinded fashion for cellularity, vascularity and extracellular matrix organisation based on the system previously described [62,63] and summarised in Table 2. Scores from each image were combined for each variable to give a total score for the tendon for each horse.

For crimp analysis, cryosections were fixed in ice cold methanol (100%) for 5 minutes then washed twice in PBS. Sections were stained with 0.1% Direct Red 80 (Sigma Aldrich, UK) in saturated picric acid for 1 hour. After a 5 minute wash in 0.5% acetic acid, sections were dehydrated and mounted. Sections were imaged under polarised light (Olympus CX31-P microscope) and were scored in a blinded fashion for banding patterns indicative of crimp. A score of 0 indicated a normal banding pattern, a score of 1 indicated a reduced banding pattern, and a score of 2 indicated no banding (Table 2).

Compositional analysis

One tendon section (1 cm section from each level) was freeze-dried and ground to a powder prior to analysis of tissue content. Water content was measured from the weight of the tissue section before and after freeze-drying.

Collagen content was determined by a measure of hydroxyproline content. Aliquots (10 mg) of ground tissue were suspended in papain digestion buffer (PBS containing 5 mM EDTA, 5 mM L-cysteine HCl and 125 µg/ml papain (all from Sigma-Aldrich, UK) and digested at 60 °C for 24 h and the solubilised collagen hydrolysed with 6 N HCl at 105 °C for 24 hours in sealed tubes. The aqueous phase was then removed by freeze-drying and the solid residue reconstituted in distilled water. Aliquots in triplicate were then assayed for hydroxyproline in a microtitre plate assay. Briefly, the sample was made to 35% isopropanol, 12 mM Chloramine T, 0.06 M sodium acetate and 0.05 M trisodium citrate and 4 mM citric acid. This was then incubated at 70 °C in 0.3 M 4-dimethylaminobenzaldehyde, 7% Perchloric acid and 40% isopropanol. The absorbance of the samples at 540 nm was measured after 20 mins. Absorbance values were quantified using a standard curve consisting of hydroxyproline (Sigma-Aldrich, UK) dilutions (50 to 0.78125 µg/ml) treated as above.

Glycosaminoglycan (sGAG) content of papain digested tissue was determined by the dimethylmethylene blue (DMMB) assay [64]. Standards were made from bovine trachea chondroitin sulphate (Sigma-Aldrich, UK). 10 µL aliquots of standards or media samples were added to a 96 well flat-bottomed plate in triplicate. 100 µL of DMMB (Sigma-Aldrich, UK) solution consisting of 46 µM DMMB, 0.2% methanol, 40 µM sodium chloride, 40 µM glycine and 10 mM HCl (pH 3.5) was then added to each well and the absorbance measured at 530 nm within 15 minutes on a spectrophotometer (Applied Biosystems UK). sGAG concentration of samples was derived from the standard curve expressed as µg/ml and subsequently adjusted to represent sGAG per mg of tissue.

DNA and tissue linked fluorescence assays

DNA as a measure of tissue cellularity and tissue-linked fluorescence as a measure of the age of tissue components were determined on aliquots of papain digested tissue extracts prepared as described above. DNA was measured using bisBenzimide (Hoechst 33258, Sigma Aldrich, UK) and tissue-linked fluorescence was measured in the absence of bisBenzimide [62]. Samples were corrected for background fluorescence and the DNA concentration calculated from a standard curve prepared with calf thymus DNA (Sigma Aldrich, UK). DNA content of tendon samples was expressed as micrograms of DNA per mg of dry weight tissue and tissue linked fluorescence as arbitrary units per milligram of collagen (measured as hydroxyproline content).

MMP-13 assay

A sample of freeze-dried powdered tendon was mixed with 100 volumes (w/v) of extraction buffer (50 mM Tris pH 7.5, 150 mM NaCl, 5 mM CaCl₂, 1 µM ZnCl₂, 0.01% Brij 35) at room temperature for 90 mins. Supernatants were recovered by centrifugation (20,000 x g for 10 mins at room temperature and frozen at -20 °C until further analysis. MMP-13 in the supernatants was measured by a fluorogenic substrate assay (Calbiochem, UK) according to manufacturer’s instructions. Briefly, 1 µl of fluorogenic substrate was added to 25 µl of tissue extract and 74 µl of assay buffer (50 mM HEPES, 200 mM NaCl, 1 mM CaCl₂, 0.01% Brij 35) in a 96-well plate. Plates were incubated at 37 °C in the dark for 2 h and plates read at excitation wavelength 325 nm and emission wavelength 393 nm. Plates were incubated a further 6 h at 37 °C in the dark and measured every hour. A standard curve was constructed using human recombinant active MMP-13 (Calbiochem, UK). Standard and samples were assayed in duplicate and the results expressed as µg active MMP-13 per mg dry weight tissue.

Statistical analysis

Histology scores and biochemical data were analysed by a non-parametric Mann-Whitney U-test as described for Movin scores of tendon by Maffulli et al. [65] Tendon cross-sectional areas were subjected by Kruskal-Wallis non-parametric one way ANOVA to identify differences between groups (stem cell treated, saline treated and contralateral) followed by a post-hoc Mann-Whitney test with Bonferroni correction. A detailed comparison of the contralateral limbs within and between the groups (stem cell or saline) using an independent samples or paired T-test (or their non-parametric equivalent) showed that there was no difference between the groups or with time (implantation time compared to 6 months). Therefore the results are shown with the contralateral limbs as a single group pooled from both treatment groups. The n values therefore are n = 6 each for stem cell and saline injection groups and n = 12 for the contralateral group. Significance level was set at 5% (p≤0.05). Data are expressed as mean ± standard deviation (SD).