|Home | About | Journals | Submit | Contact Us | Français|
When ruptured, the anterior cruciate ligament (ACL) of the human knee has limited regenerative potential. However, the goal of this report was to show that the cells that migrate out of the human ACL constitute a rich population of progenitor cells and we hypothesize that they display mesenchymal stem cell (MSC) characteristics when compared with adherent cells derived from bone marrow or collagenase digests from ACL. We show that ACL outgrowth cells are adherent, fibroblastic cells with a surface immunophenotype strongly positive for cluster of differentiation (CD)29, CD44, CD49c, CD73, CD90, CD97, CD105, CD146, and CD166, weakly positive for CD106 and CD14, but negative for CD11c, CD31, CD34, CD40, CD45, CD53, CD74, CD133, CD144, and CD163. Staining for STRO-1 was seen by immunohistochemistry but not flow cytometry. Under suitable culture conditions, the ACL outgrowth-derived MSCs differentiated into chondrocytes, osteoblasts, and adipocytes and showed capacity to self-renew in an in vitro assay of ligamentogenesis. MSCs derived from collagenase digests of ACL tissue and human bone marrow were analyzed in parallel and displayed similar, but not identical, properties. In situ staining of the ACL suggests that the MSCs reside both aligned with the collagenous matrix of the ligament and adjacent to small blood vessels. We conclude that the cells that emigrate from damaged ACLs are MSCs and that they have the potential to provide the basis for a superior, biological repair of this ligament.
Much of the mechanical stability of the human knee is provided by the anterior cruciate ligament (ACL). With increased participation in sports, the frequency of ACL injuries is rapidly increasing, and over 100,000 patients rupture their ACL each year in the United States.1 Injury to the ACL presents enormous problems for both the patient and the orthopaedic surgeon.
Even with surgical repair, the ruptured ACL will not heal.2 However, if left unattended, it remains symptomatic and considerably increases the likelihood of developing premature osteoarthritis (OA) via exposure of pathologic loads to the cartilaginous joint surfaces in the unstable knee.3 Synthetic ACL substitutes have been evaluated, but these have had very limited clinical success due to mechanical failure and severe inflammatory reactions.4 For these reasons, it is common to surgically reconstruct the ACL using autograft hamstring or patellar tendon, and also allografts.5 Not only are these procedures highly invasive, with a protracted recovery period, but they are also very expensive, costing the U.S. healthcare system ~$100 million per annum.1 Further, they fail to obviate the development of secondary OA.3
Recent data from Murray and colleagues challenge the dogma that the ACL lacks any intrinsic ability to heal. When the human, ruptured ACL is placed into organ culture, there is a rapid egress of cells.6,7 Under suitable culture conditions, these cells divide and form a collagenous repair tissue that resembles neoligament; indeed, if provided with a suitable scaffold, the cells participate in the successful repair of damaged ACL in animal models.8–10 These findings offer the prospect of developing strategies for the biological repair of the ACL with the potential to be more effective, less invasive, quicker, and more economical than the existing practice of surgical reconstruction. Because the outgrowth cells are central to the development of regenerative approaches to healing ACL ruptures, we have examined their properties in detail and were surprised to find them almost indistinguishable from mesenchymal stem cells (MSCs) derived from human bone marrow. Although the term MSC is controversial, it is used neutrally throughout this article to conform to the abundant literature on the subject.
MSCs are multipotent, fibroblastic cells11,12 first identified in bone marrow.13 Similar cells have since been isolated from an expanding list of connective tissues, including fat,14 muscle,15 skin,16 bone,17 periosteum,18 synovium,19 meniscus,20 cartilage,21,22 intervertebral disc,23 tendon,24 and only recently ligaments.25 Their phenotypic plasticity has generated considerable enthusiasm for using them to repair and regenerate connective tissues either with ex vivo, tissue engineering strategies,26 or by in situ techniques.27,28
Lack of specific markers impedes the detailed study of MSC biology, and many investigators define them operationally on the basis of their ability to differentiate along multiple lineages, particularly those leading to chondrogenesis, osteogenesis, and adipogenesis. Kolf et al.29 recently collated the literature concerning the surface immunophenotype of human MSCs, noting that, according to most authors, cultures of these cells stain positively for CD13, CD29, CD44, CD90, and CD105. Staining for STRO-1 and CD106 is more variable, and CD11b, CD31, CD34, and CD45 are normally absent.
Here we report than human ACL outgrowth cells share very similar properties with MSCs derived from bone marrow and collagenase digests of ACLs, but subtle differences were found that will be discussed in detail further in the article. Moreover, MSCs are common within the body of the ACL, both aligned to its collagen fibers and around blood vessels. The unexpected discovery of a mobile population of MSCs within the ACL offers a new insight into the biology of this ligament and new possibilities for its biological repair.
This study was approved by the Institutional Review Boards of Children's Hospital and Brigham and Women's Hospital, Boston, MA, and the University of Würzburg, Würzburg, Germany. All patients gave informed consent. Human injured ACLs were harvested aseptically from a total of 20 patients (M:F ratio, 5:15), mean age 24.6 years (age range, 18–35 years), undergoing ACL reconstruction following complete ACL rupture after trauma during skiing (n=6), rollerblading (n=1), soccer playing (n=12), or biking (n=1). Surgery was performed within a mean time after trauma of 8 weeks (range, 4–16 weeks).
For comparison, we also retrieved intact ACL cells from 10 patients suffering knee OA (aged 46–68 years), whose intact ACLs were removed during total knee arthroplasty surgery.
Bone marrow cells were harvested from intramedullary reamings from the femoral shafts of 20 patients (M:F ratio, 15:5), mean age 56 years (age range 45–67 years), undergoing total hip replacement surgery after informed consent, as approved by the local Institutional Review Board.
For ACL explant cultures, ACLs were rinsed twice with serum-free Dulbecco's modified Eagle's medium (DMEM)/F-12 media (PAA Laboratories GmbH, Pasching, Austria) containing 2% antibiotic/antimycotic solution (Invitrogen, Carlsbad, CA). The synovial sheath and the ends of the ACL remnants were removed meticulously, and the ACL fascicles were dissected into pieces of about 3mm3. The ACL fragments were placed in 12-well plates for 2h at 37°C with a minimal amount of medium to promote adherence, followed by the addition of 1mL complete medium consisting of DMEM/F-12 supplemented with 10% fetal bovine serum (FBS) (Invitrogen), antibiotics (50IU penicillin/mL and 50μg streptomycin/mL; PAA Laboratories), and 50μg/mL ascorbate 2-phosphate (Sigma, St. Louis, MO) for 3–4 weeks. During this period, cells migrated from the tissue fragments forming a population of outgrowth cells (ACLOUT). Second passage cells were used for all experiments.
For ACL digest cell cultures, cells were isolated by direct digestion of ACL fragments in a buffer containing 0.1% (w/v) collagenase 1 and 3 (Invitrogen). The cells released from the ACL in this fashion (ACLDIG) were plated in monolayer culture in complete DMEM/F-12 medium. Second passage cells were used for all experiments.
Human bone-marrow-derived mesenchymal stromal cells (BMSCs) were isolated on marrow reamings (10–20mL) harvested from the femoral shaft, which were transferred to 50mL conical tubes containing 20mL DMEM/F-12 (PAA Laboratories). The tubes were vortexed to detach marrow cells from the bone plugs and centrifuged (1000rpm for 5min) to pellet cells and bone plugs. The supernatant was discarded and the pellets were reconstituted in 10mL complete medium. After repeated vortexing, the marrow cells were collected by aspiration with 10-mL syringes fitted with 20-gauge needles. The remaining cells in the bone plugs were extracted using the identical procedure for a total of four times until the bone plugs appeared yellowish-white. The collected cells were pelleted (1000rpm for 5min), resuspended in the complete medium, counted with a hemocytometer, and plated at a density of 8×107 nucleated cells per 150cm2 tissue culture flask (TPP, Trasadingen, Switzerland). Nonadherent cells were removed after 2 days, and attached BMSCs were washed with phosphate-buffered saline (PBS) and cultured in the complete medium for 10–14 days with medium changes every 3–4 days. Cells to be analyzed for chondrogenic differentiation were expanded in the presence of 10ng/mL fibroblast growth factor-2 (PeproTech, Rocky Hill, NJ), which has been shown to maintain the chondrogenic potential of MSCs in monolayer.30 All cells were observed in tissue culture using a light microscope (Model Axioscope 2; Carl Zeiss, Göttingen, Germany) and photographed with a digital camera (AxioCam MRc; Carl Zeiss).
Cell proliferation was assessed by measuring adenosine 5′ triphosphate (ATP) using the sensitive luminometric Cell Titer-Glo® assay system (Promega, Mannheim, Germany) according to the manufacturer's instructions. Briefly, first passage ACL-outgrowth cells, ACL-digest cells, or BMSCs were seeded at 1000 cells per well in 96-well plates (Thermo Fischer Scientific Nunc, Langenselbold, Germany) and cultured in 100μL complete DMEM/F-12 mediu per well for 12 days, with medium changes every 2 days. At days 2, 4, 6, 8, 10, and 12, luminescence of n=10 wells per cell type and donor was measured for 0.1s in a luminometer after an equal volume of Cell Titer-Glo Reagent was added per well followed by 5min incubation at room temperature. A total of five donors for each cell type were included.
Cells were detached with 10mM ethylenediaminetetraacetic acid, and suspended in 4°C PBS with sodium azide (to minimize internalization or shedding of surface molecules) to a final concentration of 2×107 cells/mL. Samples were distributed into 96-well V-bottom plates (BD Biosciences, San Jose, CA) and spun at 400g for 3min. Supernatants were decanted and the cell pellets washed twice with PBS. After washing, samples were suspended in 100 (microL blocking buffer, and incubated for 30min at 4°C. Direct single or multicolour immunofluorescent staining was performed as follows: 100μL antigen-specific fluorescent mAb or an immunoglobulin isotype-matched control were added to samples and incubated for 30min at 4°C. After incubation, samples were pelleted at 400g for 3min and the supernatant decanted. Samples were resuspended in 4°C PBS with sodium azide, and this wash step repeated three times. Samples were analyzed immediately or stored in 2% paraformaldehyde for later analysis. Samples were analyzed using a Cryonics FC 500 flow cytometer (Beckman Coulter, Fullerton, CA). All antibodies (Serotec, Düsseldorf, Germany; BD Biosciences; R&D Systems, Minneapolis, MN; Acris Antibodies GmbH, Hiddenhausen, Germany; Santa Cruz Biotechnology, Santa Cruz, CA) were conjugated to the fluorochromes fluorescein isothiocyanate, phycoerythrin, or allophycocyanin for the following cluster of differentiation (CD) antigens: CD11c, CD14, CD29, CD31, CD34, CD40, CD44, CD45, CD49c, CD53, CD73, CD74, CD90, CD97, CD105, CD106, CD133, CD144, CD146, CD163, and CD166, as well as alkaline phosphatase (ALP), HLA A,B,C, and STRO-1. Nonspecific mAbs for all fluorochromes were used as comparative controls. Labeling, manufacturer, and marker specification for each antibody is listed in Table 1.
Monolayer cultures were washed three times with PBS and then fixed with 100% methanol cooled to 4°C. Staining of ACL tissue sections or constructs was performed after fixation in 10% buffered formalin for 5 days. The fixed tissues were embedded in paraffin and sectioned at 5μm. Sections were deparaffinized for 2min in xylene and rehydrated in graded alcohols. For histological analyses, ACL sections or constructs were stained using hematoxylin and eosin (H&E), Azan, and Masson/Goldner (M/G). For immunohistochemical analyses of ACL tissue sections or monolayer cultures, slides were washed for 20min in Tris buffered saline, and then incubated in 2% bovine serum albumin (Sigma) and 2.5% normal horse serum. After washing in Tris buffered saline, sections were trypsinized (1mg/mL) for 30min at 37°C, and then incubated with 5μg/mL primary antibody. Based on the fluorescence-activated cell sorting (FACS) data (Table 1), immunostaining for selected markers was performed to locate these cells within the bulk tissue, using primary diluted antibodies for the CD44 (1:200), CD90 (1:25), CD105 (1:50) (all DAKO, Hamburg, Germany), and STRO-1 (1:20) (R&D Systems) antigens. Antibody binding was detected using the Link-Label IHC Detection System (BioGenex, San Ramon, CA), as directed by the supplier, using Fast Red (BioGenex) as substrate chromogen. Sections were then analyzed for red color by light microscopy.
Chondrogenic differentiation was assessed by the pellet culture method as modified recently.31 Cells were suspended to a concentration of 1×106 cell/mL in serum-free DMEM and 200-μL aliquots (2×105 cells) were distributed to a polypropylene, V-bottom 96-well plate (Corning Inc., Corning, NY). The plate was centrifuged at 400 g for 5min, and the supernatant aspirated and replaced with the chondrogenic medium consisting of DMEM-high glucose (Invitrogen) with 1% antibiotic/antimycotic cocktail, 1% ITS+Premix (BD Biosciences), 40μg/mL proline, 100nM dexamethasone, and 50μg/mL ascorbate-2-phosphate (all from Sigma). To certain aggregates, 10ng/mL recombinant human transforming growth factor beta1 (TGF-β1) (PeproTech) was added to enhance chondrogenesis. Media were changed every other day, and aggregates were collected at 3 or 4 weeks for analysis.
The DNA content of the aggregates was determined using the Hoechst Dye 33258 method.32 Samples were digested overnight at 65°C in 100μg/mL proteinase K (Sigma). Pellet digests were taken through three freeze–thaw cycles, and digest aliquots were diluted with 100ng/mL Hoechst Dye 33258 (Sigma) in 10mM Tris (pH 7.4), 1mM Na2 ethylenediaminetetraacetic acid, 100mM NaCl. The fluorescence intensity was measured with a DQ300 Fluorometer (Hoefer Scientific Instruments, San Francisco, CA), and DNA concentration was determined from a standard curve of calf thymus DNA (Sigma). Proteinase K digests were also analyzed for glycosaminoglycan (GAG) content using the dimethylmethylene blue (Sigma) dye binding assay.33 Aliquots of digest, diluted as necessary, were combined with dimethylmethylene blue solution, and sample absorbances were measured at 595nm. GAG concentrations were interpolated from a standard curve of shark chondroitin sulphate (Sigma), and results were normalized by DNA content.
Before histology, aggregates were fixed for 30min in 4% paraformaldehyde, encapsulated in 0.5% agarose gels, embedded in paraffin, and sectioned at 5μm thickness. Sections were mounted onto glass slides, de-paraffinized with three xylene washes (5min each), and re-hydrated in graded alcohol solutions. For detection of matrix proteoglycan, representative sections were stained with 1.0% (w/v) Toluidine blue (Sigma), pH 3.0 for 30min. Slides were rinsed in deionized water, dehydrated in graded alcohols, rinsed three times in xylene, and cover slipped with Cytoseal™ XYL mounting medium (Richard-Allan Scientific, Kalamazoo, MI).
For immunohistochemistry, aggregate sections were digested with 0.1U/mL chondroitinase ABC (Sigma) in PBS with 1% bovine serum albumin (Sigma) for 1h at 37°C. Endogenous peroxidases were quenched in 0.3% hydrogen peroxide solution for 30min. After blocking with normal horse serum for 1h, slides were incubated overnight (2°C–8°C) with polyclonal rabbit anti-human collagen type II antibody (Santa Cruz) in blocking buffer. Antigens were observed using a biotinylated horse anti-rabbit secondary antibody and VectaStain® Elite ABC reagent (Vector Laboratories, Burlingame, CA). Slides were rinsed, counterstained with hematoxylin, dehydrated, and cover slipped.
Reverse transcriptase (RT)-polymerase chain reaction (PCR) analyses were used to evaluate expression of cartilage-specific genes. Total RNA was isolated from six aggregates, harvested after 21 days. To facilitate lysis of the cells in aggregate culture, they were first frozen in liquid nitrogen and ground to a powder with a custom-made pellet pestle. Total RNA was subsequently extracted using 1mL of Trizol reagent with an additional purification step using RNeasy separation columns (RNeasy kit; Qiagen, Hilden, Germany) according to the manufacturer's instructions. For cDNA synthesis, 1μg of total RNA from each group was reverse transcribed using random hexamer primers and MoMLV-H reverse transcriptase (Promega), according to the manufacturer's instructions. Equal amounts of each cDNA synthesized (100ng) were used as templates for PCR amplification in a 50μL reaction volume using Taq DNA polymerase (Amersham, Braunschweig, Germany) and 50pmol of specific primers for the following human messenger (m)RNAs: collagen type II alpha 1 (COL II), aggrecan core protein (AGC), cartilage oligomeric matrix protein (COMP), and elongation factor 1α as an internal control. Primer sequences are provided in Table 2. The RT-PCR products were electrophoretically separated on 1.5% agarose gels containing 0.1μg/mL ethidium bromide.
Cells were seeded at a density of 1×105 cells/cm2 in four-well chamber slides and 25cm2 flasks (Thermo Fischer Scientific Nunc). At confluence, osteogenesis was induced by supplementing the media with 100nM dexamethasone, 50μg/mL ascorbate, 10mM β-glycerophosphate, and 25ng/mL recombinant human bone morphogenetic protein 2 (BMP2) (R&D Systems); control cultures lacked osteogenic supplements. Media were changed every 3 days for 3 weeks. Cultures were stained histochemically for ALP, using a commercial kit according to the manufacturer's protocol (Sigma), and for matrix mineralization using Alizarin red as described previously.17 Additionally, markers of osteogenesis were analyzed by RT-PCR, using the protocol described above. Expression of the following human mRNAs was examined: ALP, collagen type I alpha 2 (COL I), osteocalcin, and core binding factor alpha 1 (Runx2). Primer sequences are provided in Table 2.
Cells were grown to 50%–70% confluence in four-well chamber slides, or 25cm2 flasks (Thermo Fischer Scientific Nunc) in DMEM supplemented with 10% FBS. Adipogenesis was induced by supplementing the media with 1μM dexamethasone, 1μg/mL insulin, 0.5mM 3-isobutyl-1-methylxanthine, and 100μM indomethacin. Control cultures without adipogenic supplements were also maintained. After 3 weeks, cultures were examined for evidence of adipogenesis by fixing in 10% formalin and staining with freshly prepared Oil red-O solution for the detection of lipid droplets as previously reported.17 As additional markers of adipogenesis, RT-PCR, using the protocol described above, was used to detect expression of the following human genes: lipoprotein lipase and proliferator-activator receptor γ2. Primer sequences are provided in Table 2.
Ligamentogenic induction of cell cultures was performed using an in vitro assay that we described previously.34 Briefly, ACLOUT, ACLDIG, and BMSCs were seeded at 3.6×106 cells/175cm2 flask and transduced at 100 multiplicities of infection of Ad.BMP12, in 5mL of the serum-free medium. Marker gene (green fluorescent protein [GFP]) or untransduced cells were used as comparative controls. At 24h after transduction, the cells were recovered and placed in collagen hydrogel constructs, with 3×105 cells being suspended in 100μL gel neutralization solution, followed by the addition of 100μL collagen type I stock solution (Arthro Kinetics AG, Esslingen, Germany). After polymerization, the constructs were cultured in 48-well plates with 500μL complete culture medium with medium changes every 2–3 days throughout the 21-day culture period. All constructs were evaluated histologically (H&E, Azan, and M/G), immunohistochemically (collagen type III [COL III], elastin, vimentin, and fibronectin), and by RT-PCR (COL I, tenomodulin, fibronectin, and vimentin) after 21 days as described above. Primer sequences are provided in Table 2.
The numerical data from the ATP and GAG analyses were expressed as mean values±standard deviation. Each experiment was performed at least in triplicate per group and patient and repeated on at least 3 and up to 10 individual ACL/marrow preparations from different patients (n=3–10), as indicated in the respective experiments. Stainings for ALP, Alizarin red, and Oil red-O were evaluated histomorphometrically employing four high power fields of four representative slides per group and donor using the Zeiss Axio Vision Software Rel. 4.6.3 (Zeiss, Ulm, Germany). All numerical data were subjected to variance analysis (one- or two-factor analysis of variance) and statistical significance was determined by Student's t-test, with <0.05 considered significant.
Throughout this study, cells that migrated from the ACL during explant culture (ACLOUT) were compared to ACL cells released by collagenase digest (ACLDIG) and BMSCs. It was not possible to obtain ACL and bone marrow from the same individuals, so the samples are not patient- or age-matched. To prevent contamination of ligament cells by synoviocytes, the synovial sheaths were meticulously dissected from sections of human ACL before use.
After human ACL samples were recovered, ligaments were dissected into explants (~3mm2) and placed into organ culture in 12-well plates. After a lag of ~5 days, fibroblastic cells emigrated from the cut ends of the ACL. Although the rate of outgrowth and cell morphology varied between samples (Fig. 1a), all cultures formed confluent monolayers in ~3 weeks.
Cells recovered from collagenase-digested ACL (ACLDIG) were seeded directly into monolayer culture, where their growth pattern and morphology appeared almost indistinguishable from ACLOUT and BMSCs, which are shown for comparison (Fig. 1a).
We next compared cell proliferation rates in all three cell types, which were estimated by quantitation of ATP using the Cell Titer-Glo luminescent assay and revealed increased cell proliferation rates of the BMSCs at all time points measured compared to the ACLDIG and ACLOUT, which showed no significant differences (Fig. 1b).
FACS was used to measure the presence of key surface markers that are displayed in Figure 2a and Table 1. ACLOUT and ACLDIG populations were both strongly positive for CD29, CD44, CD49c, CD73, CD90, CD97, CD105, CD146, CD166, HLA A, B, C, and ALP, weakly positive for CD14, but negative for CD11c, CD31, CD34, CD40, CD45, CD53, CD74, CD106 and CD133, CD144, CD163, and STRO-1 (Table 1). The ACLOUT populations appeared to contain a higher percentage of cells that were CD146 positive compared to the ACLDIG, although this observation did not reach levels of significance (p=0.33; Fig. 2a).
BMSCs displayed a similar, although not identical, pattern (Table 1); in particular, they stained more strongly for CD106 and CD146, and were weakly positive for CD11c and STRO-1. However, they stained less strongly for CD97 (Table 1) compared to the ACL cell populations. FACS data for selected surface markers are shown in Figure 2a.
The expression of certain markers was confirmed by immunocytochemical staining of ACLOUT, ACLDIG, and BMSCs (Fig. 2b). In agreement with the FACS data, these cells showed strong immunoreactivity toward CD44, CD90, and CD105 (Fig. 2b). Staining for STRO-1 was stronger than expected from the FACS data (Fig. 2a). This may reflect the properties of the anti-STRO-1 immunoglobulin M antibody used in these experiments, or the fact that cells used for FACS had gone through more passages.35
When placed into pellet culture, cells derived from both ACL sources and bone marrow underwent chondrogenic differentiation in the presence, but not absence, of TGF-β1 as judged by histology and immunohistochemistry (Fig. 3a), the accumulation of GAG (Fig. 3b), and the expression of transcripts encoding COMP, AGC, and COL II (Fig. 3c). Nevertheless, there were subtle differences between the chondrogenic behavior of ACLOUT, ACLDIG, and BMSCs. Pellets formed by the latter were larger and stained more uniformly for proteoglycans with toluidine blue and for cartilage-specific anti-type II collagen antibodies (Fig. 3a). Cartilaginous pellets formed by ACLOUT, and ACLDIG in contrast, had a core that stained only weakly with these reagents (Fig. 3a). Chondrogenic pellets from all cell types produced significantly more GAG compared to nonchondrogenic controls (Fig. 3b). However, chondrogenic pellets formed from ACLOUT and ACLDIG contained significantly less GAG than those formed from BMSCs, both in total and when normalized for DNA content (Fig. 3b). Consistent with this, chondrogenic pellets from all cell types produced significantly more mRNA transcripts encoding COMP and induced expression of AGC and COL II compared to the respective nonchondrogenic controls (Fig. 3c).
Monolayer cultures of ACLOUT, ACLDIG, and BMSC responded equivalently to the osteogenic medium in terms of staining for ALP by cytochemistry (Fig. 4a) and for mineralization with alizarin red (Fig. 4b). Histomorphometric analyses revealed significantly increased areas of positive staining for ALP (Fig. 4a) and Alizarin red (Fig. 4b) in the respective osteogenic cultures of all three cell types compared to controls. Among the osteogenic cultures the BMSCs showed the largest mean areas of positive staining for ALP (64.7%) and Alizarin red (70.2%), followed by the ACLDIG (56.8%/55.1%) and ACLOUT (52.9%/51.2%) cultures, respectively. However, significant differences between the cell types could not be resolved (p>0.07/p>0.09 for all comparisons). This corresponds to the induction of transcripts encoding osteocalcin and core binding factor alpha 1/Runx2, and the increased expression of ALP and the α-1 chain of COL I in all osteogenic cultures compared to controls (Fig. 4c).
Monolayer cultures of ACLOUT, ACLDIG, and BMSCs responded equivalently to adipogenic conditions, as assessed by staining with Oil red-O (Fig. 5a). Again, histomorphometry showed significantly increased areas of positive staining for Oil red-O in the respective adipogenic cultures of all three cell types compared to controls (Fig. 5a). Again the adipogenic BMSCs showed the largest mean areas of positive staining for Oli red-O (39.6%), followed by the ACLOUT (33.8%) and the ACLDIG (30.6%) cultures; however, differences were not significant. Equivalently, adipogenic culture conditions induced production of transcripts encoding lipoprotein lipase and proliferator-activator receptor γ2 compared to controls within all groups (Fig. 5b).
This trilineage differentiation capacity of ACLDIG and ACLOUT was also confirmed, when intact ACLs from osteoarthritic donors were used and were also confirmed on single colony level for each donor (data not shown).
To show that ACLOUT have the capacity for self-renewal, ligamentogenesis was induced using an in vitro assay, which was employed previously to reveal the ligamentogenic capacities of ACLDIG and BMSC populations.34
After 21 days, staining with H&E showed a homogenous cell distribution of ACLOUT, ACLDIG, and BMSC fibroblasts within the hydrogel constructs (Fig. 6a). Moreover, the BMP12-modified constructs revealed that the cells were embedded in a dense collagenous matrix compared to GFP-modified controls where less matrix formation was seen, as evidenced by matrix staining with Azan and M/G (Fig. 6a).
Immunohistochemistry for the ligament matrix proteins fibronectin, vimentin, COL III, and elastin revealed more intense stainings in the Ad.BMP12 transduced hydrogel constructs compared to corresponding controls with ACLOUT, ACLDIG, and BMSC fibroblasts, where no red staining of the matrix was detectable (Fig. 6a). For each immunostaining, controls were performed without primary antibody, which were negative in all cases (not shown).
This is in correspondence with increased expression of mRNAs encoding COL I, tenomodulin, fibronectin, and vimentin in all ligamentogenic cultures compared to marker gene controls (Fig. 6b).
Longitudinal sections of ACLs stained with H&E, Azan, or M/G confirmed the typical pattern of undulating collagenous fibers along which align extended fibroblastic cells (Fig. 7a). Vasculature was represented by small, rare blood vessels.
Surprisingly, most of the fibroblastic cells, as well as cells associated with the blood vessels, stained positively for CD44, CD90, and CD105. Staining for STRO-1 was also positive, but less widespread (Fig. 7b). Examples of positively stained ACL fibroblasts within the matrix were labeled using white arrows, and near vasculature using black arrows, as shown in the 500×panels (Fig. 7b). We did not observe marked differences between torn and intact osteoarthritic ACLs, and representative sections of a 56-year-old male OA patient are shown (Fig. 7).
Thus, multipotent ACL cells appear to be common residents within the ligament, both as components of the ligament itself and associated with blood vessels. This indicates that they apparently do not arise from a separate, small, discrete pool of progenitors that are selected and amplified by the explant culture system.
These data demonstrate unequivocally a population of multipotent mesenchymal progenitor cells within the human ACL. The recent identification of such progenitor cells by Cheng et al.25 that were isolated by collagenase digestion of anterior and posterior cruciate ligaments supports this conclusion. The present study, however, focuses on the repair relevant outgrowth ACL cell populations (ACLOUT) and offers a unique perspective on ACL progenitors by examining their origins within the tissue and by comparing them directly to ACL populations released by collagenase digestion (ACLDIG) and BMSCs from human bone marrow. Both ACL populations are very similar to human BMSCs, but differ in certain details. All cell types examined show the ability of colony formation (Fig. 1a) and revealed almost similar osteogenic (Fig. 4) and adipogenic (Fig. 5) capacities. However, the ACL populations show less cell proliferation (Fig. 1b) as well as chondrogenic capacity (Fig. 3), and do not express CD106 or CD146 as highly or STRO-1 as consistently, but they express higher levels of CD97 (Fig. 2 and Table 1). CD146 in particular was less expressed in the ACLOUT compared to the ACLDIG, although levels of significance were not reached (Fig. 2a). Expression of STRO-1 by MSCs declines with cell passage,35 which may explain the marginal staining for this epitope on those cells examined by FACS (Table 1). This would be consistent with the considerable staining for STRO-1 noted by immunocytochemistry, and immunohistochemistry, where less subculture is involved (Fig. 2). However, this study is limited to the examination of only a finite number of characteristic cell surface markers, and analyses of, for example, whole transcriptomes might be a suitable approach, to further resolve differences between the cell types.
Although cells with the general properties of MSCs have been recovered from a number of connective tissues, there are small phenotypic variations depending on the tissue source.14,16–24,26,36–39 The MSCs that are released from ACL tissues fit this pattern, but differ in subtle ways from the BMSCs used as a reference population. However, the three cell types could not be age- and patient-matched, so these variables could also account for the differences.
Progenitor cells were isolated from ligaments recovered 4–16 weeks after injury, at the time of surgical reconstruction. Given the evidence that MSCs migrate to sites of injury,40 there is the possibility that they are not normally present in the ACL. Indeed, a recent article by Morito et al.37 suggests that MSCs enter the synovial fluid after rupture of the ACL. The surface immunophenotype of the synovial fluid cells is similar to that reported here for ACL cells, except for a much higher expression of STRO-1.29 Despite the presence of these cells in the synovial fluid, our immunohistochemical findings argue against them being the source of the MSCs within the ACL. Cells positive for CD44, CD90, CD105, and STRO-1 were noted deep within the body of the ACL, closely aligned with the collagen fibers, as well as surrounding small blood vessels. The latter may be pericytes, long suspected of being a source of mesenchymal progenitors within vascularized tissues.39 The former, however, are morphologically indistinguishable from what are normally considered to be ligament fibroblasts, lying adjacent to the collagen bundles of this tissue. Further, we confirmed our results by using ACL outgrowth and digest populations from intact ACLs, derived from OA patients undergoing knee arthroplasty surgery, and noted an almost identical surface antigen expression and trilineage differentiation potential (not shown), compared to the data from cells recovered from torn ACLs (Figs. 1–5), supporting the concept of a tissue-inherent MSC population and confirming findings in the tendon.24
The capacity of self-renewal is also an important criterion of a stem cell that is met by the ACLOUT population as evidenced by an in vitro assay using BMP12 adenovirus and three-dimensional culture in a collagen hydrogel for ligamentogenic induction (Fig. 6). Using a similar approach, we were recently able to delineate the ligamentogenic potential of the ACLDIG and BMSC populations,34 which were used as comparative controls here (not shown). Our study, however, is limited to its in vitro nature, and only in vivo experiments can provide the necessary evidence for final validation of ACL regenerative approaches using inherent stem cells. The strong regenerative response of the BMP12 adenovirus, a strong ligamentogenic inductor of stem cells,41 in a rat model of Achilles tendon healing supports this concept.42
The results of our study differ in some ways to the data of Cheng and colleagues,43 who recently compared the cell proliferation and differentiation capacities of BMSCs to that of collagenase-digested ACL cells. In contrast to their study, where ACL cells revealed higher proliferation rates compared to BMSC,43 we found the BMSCs to elicit higher rates of cell proliferation compared to both ACLOUT and ACLDIG populations at all time points (Fig. 1b). Further, we found the BMSC to exhibit a higher chondrogenic potential and similar osteogenic potential compared to both ACL cells types (Figs. 3–5), whereas, in contrast, Cheng and colleagues found the BMSCs to have higher osteogenic capacity and similar chondrogenic capacity compared to ACL digest cells.43 We attribute the discrepancies in both studies to the different protocols that were used for cell isolation, maintenance, and differentiation as well as in differences in growth factor and FBS supplementation used to the respective cultures. However, the data in both studies correspond in that the adipogenic and ligamentogenic potential in all cell types examined were at comparable levels,43 undermining the relevance of ligament inherent stem cell populations and their implications over a possible use of this technology toward biological ACL repair. The fact that ACL outgrowth cells, the primary cell source that mediates any direct biological repair of the ACL,44 share these features further adds to this conclusion.
Putative stem/progenitor cells have recently been identified in similar locations of the tendon, suggesting that this may be a general phenomenon within fibrous tissues and several extracellular matrix components are important organizers of the stem cell niche.24 Not all fibroblasts within the ACL stained positively for MSC surface epitopes (Fig. 7), suggesting the presence of a mixed population of MSCs and ligament fibroblasts within the tissue. The heterogeneity of the cell population may be reflected in the nature of the cartilaginous pellets formed during chondrogenesis assays, where the inner core was much less cartilaginous than the outer component (Fig. 3a).
While the information gained in this study provided insight into the cellular characteristics of ACL outgrowth cell populations, the study is limited to its in vitro nature, and further in vivo work is required to determine its relevance to ligament repair including evaluation for structural and mechanical properties of the repair tissue.
The presence of MSCs within the ACL opens new opportunities for biological repair of this ligament. As noted above, biological repair has been largely ignored because of the belief that the ACL lacks regenerative capacity. Our data suggest, on the contrary, that the ACL houses a rich population of progenitors that are rapidly mobilized in response to injury. These cells migrate from the injured ligament, and can colonize a suitable, adjacent scaffold to synthesize repair tissue.9,10 Recent work by Cheng and colleagues revealed a vivid response of collagenase released ACL progenitors to several growth factors with respect to cell proliferation and differentiation.45 In animal models, the repair of tendons and ligaments can be enhanced by various morphogens and growth factors,38 such as BMP12,46 TGF-β,47 growth and differentiation factor (GDF)5,48 and platelet derived growth factor (PDGF).49 Gene transfer is an attractive technology for delivering these factors, especially if the vectors are associated with the matrix into which the ACL outgrowth cells migrate. Under these conditions, selective gene transfer to the ACL outgrowth cells occurs very efficiently in situ.50,51 This suggests one way in which these newly identified cells can be manipulated to improve healing of the ACL.
We are grateful to Nadja Karl, Viola Monz, Martina Regensburger, and Christa Amrehn for their excellent technical assistance. We thank Mr. Thomas Peter (Fraunhofer IGB, Stuttgart, Germany) for his expertise and help with the FACS analyses. Financial support was given by grants from BayFor (FORZEBRA TP1/WP1 to F.J., TP2/WP5 to A.F.S. and U.N.), BMBF (Grant No. 0313386E to U.N.), and NIH (RO1 AR052809 from NIH/NIAMS to C.H.E.; R01 AR054099 to M.M.M.; F32 EB005566 to R.M.P.).
No competing financial interests exist.