Tendon-derived progenitor cells: In vitro characterization and clinical applications for tendon repair

Tendinitis/tendinopathies are common performance-limiting injuries. The last decade has seen significant development in mesenchymal stem cell (MSC)based therapies in tendon repair. Using tendon-derived progenitor cells (TDPCs) for tendon healing is based on the rationale that stem cells from tendons are more phenotypically-committed or ‘primed’ for tenogenesis than cells from other tissues. This review article summarizes recent literature addressing the in vitro characteristics of TDPCs and evaluates their in vivo effects in experimental models of tendon repair. TDPCs share many of the common MSC properties; however express some lineage-specific characteristics that are heavily dependent on the extracellular matrix and tensile load stimulation. TDPCs are multipotent cells that undergo aberrant differentiation in an inflammatory milieu. In vivo experimental studies demonstrate that implanting TDPCs, alone or with biologic vehicles, improves the biomechanical and histological characteristics of repair tissue. However, TDPCs, like other MSCs used for in vivo tendon repair do not directly enhance the cellular pool within tendons. Recent evidence suggests that ‘priming’ TDPCs for tenogenesis during in vitro culture and administering these cells along with acellular matrix may augment the efficacy of TDPC-based therapies. Further research focusing on delineating the phenotype of TDPCs, and establishing optimal cell delivery strategies that influence healthy tendon repair are warranted to optimize TDPC-based treatment of tendon pathology. Correspondence to: Matthew C Stewart, 1008 W. Hazelwood Drive, Urbana, IL61802, USA, Tel: 217-265-5206, Fax: 217-244-1475, E-mail: matt1@illinois.edu


Introduction
Tendon injuries range from acute tendon rupture to chronic tendinopathy, and are among the most common orthopedic problems. Achilles tendon injuries account for up to 50% of all sports-related injuries [1,2], while rotator cuff degenerative tendinopathy is a common cause of shoulder pain/disability [3]. Flexor tendon injuries are linked to several occupations in people [4] and in sport/performance horses [5]. Further, long-term disability is a frequent consequence, due to prolonged healing time and high rate of recurrence following the initial injury [6,7]. Consequently, the financial impact of tendon injuries is considerable.
Tendons are highly specialized connective tissues that transmit tensile forces between muscles and bones. Tendons are relatively hypocellular and hypovascular tissues, with little or no intrinsic regenerative capacity. Current therapies involve conservative management and/or surgical debridement and repair, depending on the location and severity of the pathology [8,9]. Irrespective of the approach used, the resultant repair tissue is biomechanically inferior to healthy tendon and re-injury is common despite prolonged rehabilitation. Consequently, cell-based approaches to tendon healing have been widely investigated in experimental models of tendinitis with the goal of improving the quality of repair tissue. Mesenchymal stem cells (MSCs) derived from different sources have been evaluated in both in vitro and in vivo models of tendon repair and the outcomes have been recently reviewed [10][11][12][13][14][15]. This review will specifically focus on tendon-derived stem/progenitor cells, their in vitro and in vivo characteristics, and their efficacy in improving tendon healing in experimental models. Several different terms have been used in the literature to describe the stem/progenitor cell populations within tendons. In this review, the term "tendon-derived progenitor cells" (TDPCs) is used to refer to these cells.

Tendon structure and function
Tendons are dense collagenous tissues that connect muscles to bones and are composed of a hierarchical arrangement of predominantly collagenous subunits ( Figure 1). Morphologically, tendons contain a variable number of fascicles, which are comprised of multiple collagen fiber bundles. The fiber bundles contain many collagen fibrils [16,17]. These collagen units are oriented in the direction of the predominant tensile load. Tendon fascicles are bound together by a loose areolar connective tissue, the endotenon, which becomes confluent with the outer epitenon. The epitenon is surrounded by the peritenon; a fine connective tissue sheath which functions as an elastic sheath to permit free movement of the tendon against the surrounding structures. The collagen molecules are stabilized by intermolecular chemical crosslinks resulting in high tensile strength [18].
Histologically, tendons have a highly organized and anisotropic structure ( Figure 2). Collagen fibers are aligned along the longitudinal axis in each fascicle. Tenocytes are located both within and between the fascicles, arranged in rows along the direction of the collagen fibers [19]. A characteristic crimp pattern of collagen fibers is a typical ultrastructural feature of tendons ( Figure 3A). The crimps function as a buffer to provide immediate longitudinal elongation in response to physiological tensile loads [17].
Tendons can be classified as positional or energy-storing tendons.
All tendons transmit forces from muscle to bone; however, energystoring tendons have the additional function of extension and recoil to increase the efficiency of locomotion [19]. Tendons respond to tensile loads at multiple structural levels [20,21]. Crimp elongation provides a relatively modest (approximately 3%) strain response to load. Elastic 'sliding' between adjacent fibers, fibrils and fascicles, rather, than direct 'unit' extension, provides the majority of tensile strain. In addition, recent evidence suggests the presence of helical substructures within the tendon fascicles, which provides a mechanism for efficient extension under load and recoil when unloaded [22]. Therefore, the collective mechanical properties of both the interfascicular and intrafascicular matrices are responsible for the functional capacity of tendons.

Endogenous tendon healing
Healing tendons undergo the traditional phases of an initial inflammatory response, a proliferative phase and a remodeling phase [5]. The reactive inflammatory phase lasts for about a week, in the absence of ongoing injury, and is characterized by a marked increase in cross-sectional area at and around the site of injury, consequent to local hemorrhage and edematous swelling. This is followed by inflammatory cell infiltration, primarily neutrophils and macrophages. The proliferative phase overlaps with the latter half of the inflammatory phase and peaks in about 2-3 weeks. Neovascularization, local synthesis of chemokines, trophic factors and proliferation of fibroblasts are dominant features of this phase. These processes culminate in tenocyte proliferation and collagen synthesis leading to formation of immature fibrovascular tissue. The remodeling phase is characterized by formation of fibrous tissue. During this stage, the healing tissue undergoes changes in size and shape. This phase is divided into consolidation and maturation processes [23]. The repair tissue transitions from predominantly cellular to fibrous in nature, as new matrix is synthesized at the injury site. However, increased vascularity, neuronal infiltration and cellularity persist for up to 3 months postinjury [6,24,25] and the collagen architecture remains disorganized for several months ( Figure 3B). This results in reduced extensibility and elasticity of the repair tissue and predisposes the site to re-injury.

Tendon-derived progenitor cells
The focus on TDPCs for tendon regeneration is based on the rationale that cells derived from the target tissue (in this case, tendon) will be phenotypically and biosynthetically more capable of stimulating functional repair than MSCs derived from other tissue sources. The existence of MSCs in tendon tissue was first reported in 2003 [26]. Tendon-derived cell lines expressing tendon phenotype-related genes such as scleraxis, cartilage oligomeric matrix protein (COMP), and type I collagen (in addition to osteopontin and Runx2), were developed from transgenic mice. Bi et al. (2007) identified and characterized a unique cell population, termed tendon stem/progenitor cells, from mouse and human tendon samples [27]. These cells demonstrated universal stem cell characteristics of clonogenicity, multipotency and selfrenewal capacity. This study also showed that the bioactivity of TDPCs, like other MSCs, is heavily dependent on their local environment/ matrix interactions ('niche'). Comprehensively characterizing stem/ progenitor cells in tendons and their role in tendon responses to injury is paramount for developing effective regenerative therapies. Donor age may influence the number, proliferative and multi-   lineage capacities of TDPCs. The total number of TDPCs in aged rats decreased by 70% compared to young rats [37]. In human isolates, proliferation and clonogenicity of TDPCs from aged tendons was decreased although their multi-lineage potential was retained [38]. Age did not affect in vitro characteristics in equine TDPCs [31], suggesting that species-specific differences in age-dependent in vitro characteristics of TDPCs exist.

In vitro characteristics of TDPCs
TDPCs have been isolated from rat and murine Achilles and patellar tendons, lapine Achilles tendon, human patellar, Achilles, rotator cuff and biceps tendons, and equine extensor and flexor tendons. Clear differences in the characteristics of TDPCs isolated from different tendons have not been demonstrated, but the number and biosynthetic activity of TDPCs are increased in response to physiologic loading and exercise, [39], and the stem/progenitor cell populations derived from the peritenon and tendon proper are distinct [40][41][42]. Tendon proper-derived TDPCs were more permissive for in vitro tenogenic differentiation than peritenon-derived progenitor cells. Further investigation into this aspect of TDPC biology is warranted, since these sub-populations may have distinct functions and efficacies in intrinsic and extrinsic tendon healing.
In vitro expansion and proliferation: Using TDPCs to treat tendon injuries is dependent on efficiently expanding these cells to clinically relevant numbers while maintaining their 'stemness' and therapeutic value during multiple passages. TDPCs, like other MSCs, proliferate more rapidly than terminally differentiated tenocytes during in vitro expansion [30]. Culturing human TDPCs in reduced oxygen conditions increases proliferation [43-45], metabolic rates and biosynthetic activities [40] but in one of these studies [43], 2% oxygen levels reduced their multi-lineage potency. Similarly, equine TDPCs isolated via low-density plating and cultured under hypoxic conditions had higher proliferation rates than TDPCs cultured in normoxic conditions [31]. The effect of passage on the in vitro characteristics of TDPCs has received little attention, and the existing data is not consistent. Zhang et al. (2010) reported that the proliferation of porcine TDPCs decreased as the passage number increased [44]. In contrast, Tan et al. (2012) found that the proliferation rate and colony-forming ability of rat TDPCs increased with subsequent passages [46]. More research will be necessary to clarify this issue and determine speciesspecific consequences of multiple passages on subsequent activities.
Immunophenotypic profile: TDPCs share common stem cell markers identified in MSCs from other tissue sources. Specific markers that distinguish TDPCs from terminally differentiated tenocytes are poorly defined. TDPCs express CD44, CD90, CD146, CD73, Sca-1, Stro-1, nucleostemin, Oct-4, SSEA-4 and are negative for CD31, CD34, CD45, CD144, CD106 [27-31,40,44]. These markers cannot distinguish TDPCs from other MSCs and some these MSC markers are, in fact, fibroblast markers (CD44 and CD90). On the other hand, certain markers that distinguish TDPCs from bone marrow MSCs have been identified. Murine and human TDPCs lacked CD18 expression, which is expressed by bone marrow MSCs [27]. CD106 is expressed by human and rat TDPCs but is absent in bone marrow MSCs. Phenotypic markers expressed in TDPCs isolated from different species are summarized in Table 1.
The specific source of TDPCs may also affect their immunophenotype. Approximately 90% of TDPCs from tendon proper expressed Sca-1 whereas only 70% of peritenon-derived TDPCs expressed this marker [41]. As expected, TDPCs isolated from peritenon had higher expression of CD133 (a pericyte marker) than TDPCs from tendon proper. A similar study found that TDPCs from peritenon were CD146 + , CD34 -; TDPCs from interstitial tissue were CD146, CD34 + ; and TDPCs from tendon proper were CD146 -, CD34 -. All three populations were CD44 + , CD31and CD45 - [40]. These findings must be interpreted with caution as these results were largely derived from rat patellar tendon-derived TDPCs and the study sample sizes were low. Further research addressing the regional immunophenotypic characteristics of TDPCs is required to understand the link between immunophenotype and clinical efficacy.
Due to a relatively small proportion of progenitor cells in tendon, most studies analyze the immunophenotype of TDPCs after a short period of sub-culture. The immunophenotype of TDPCs, as with other MSCs, changes during in vitro passaging. One study reported the Expression of CD146, CD73 and CD90 in freshly isolated rat patellar TDPCs is lost after in vitro culture [46,47]. Kowalski et al. (2015) reported that although in vitro passaging altered the expression of CD34 and CD44 in the three sub-populations of TDPCs, their overall pattern of expression was unchanged [40]. These findings suggest that any protocol for TDPC selection based on immunophenotype will need to accommodate the alterations that occur with in vitro culture. Data on the influence of donor age on TDPC multipotency is not consistent. TDPCs isolated from aged rat tendon underwent adipogenesis more readily and expressed higher levels of adipogenic markers (PPARγ, leptin) than their younger counterparts, whereas the osteogenic and chondrogenic capacity of TDPCs was unchanged [37]. However, more recent study that compared activities of human hamstring and Achilles tendon-derived TDPCs did not identify an effect of donor age on tri-lineage differentiation characteristics of TDPCs [38,51]. It is likely that species-and donor site-specific differences in the respective TDPC populations contributed to the disparities in outcome. Regardless, the influence of age on TDPC activity requires further investigation, given that degenerative tendinopathy is more prevalent in older individuals.

Multipotency of tendon-derived progenitor cells
The in vitro differentiation potential of TDPCs isolated from healthy tendon tissue and pathological tissue are markedly different. TDPCs isolated from injured tendon had a higher in vitro chondrogenic potential than TDPCs from normal tendon [52]. These cells were implicated in chondro-degeneration noted during tendon healing and were characterized as CD105cells. A follow-up study by the same group demonstrated that in vitro culture of TDPCs isolated from injured tendon treated with IL-1b decreased their trilineage differentiation potential. Further research on this aspect of TDPC pathobiology is required to determine whether cells isolated from pathological tissue during reparative surgeries can be used for consequent cell-based therapies [53]. Further, given the detrimental effects of inflammatory cytokines on TDPC activities, it will be critical to define the optimal time (following injury) for cell delivery to avoid aberrant responses of stem cells implanted in an active inflammatory milieu.

Aberrant TDPC differentiation in tendinopathy:
Fatty degeneration, chondrogenic dysplasia and ectopic calcification within the repair tissue of chronically injured tendons are well-documented [54][55][56][57][58][59][60][61][62][63]. Ectopic chondro-ossification in the mid-substance of Achilles and patellar tendons occurred as a consequence of endochondral ossification [63,64], reflecting a major phenotypic shift within the tendon cell population. The underlying pathogenesis for these metaplastic changes in chronic tendonopathy is poorly understood. Abnormal matrix deposition likely occurs from extrinsic cells that migrate to the site of injury [65] or from native tenocytes and/or TDPCs that undergo trans-differentiation to non-tenogenic phenotypes.
Experimental evidence indicates that alterations in matrix components within tendon repair tissue can profoundly impact the phenotype of TDPCs. TDPCs isolated from the biglycan-fibromodulin double knockout mice had increased collagen type II and aggrecan expression compared to wild-type TDPCs [27]. In vivo, TDPCs isolated from these mice formed bone in addition to tendon-like tissue, while the wild type TDPCs formed tendon-like tissue only. Asai et al. (2014) showed that TDPCs in injured tendons trans-differentiate into chondrogenic cells and induce chondro-degenerative lesions [52]. The molecular pathogenesis of aberrant TDPC differentiation in tendinopathy has not yet been fully elucidated, but several recent studies have implicated inflammatory cytokines, TGF-β/BMP signaling, extracellular matrix changes and altered biomechanical stimuli in these phenotypic shifts.

Inflammatory cytokines and biological factors:
The proinflammatory cytokines IL-1β, IL-6 and TNFα are up-regulated in both acute and chronic strain type injuries [66][67][68], inducing expression of inflammatory mediators Cox-2, PgE2, and collagenases MMP-1 and -13. All these factors are known to be involved in tendon matrix degradation [69]. TDPCs isolated from injured tendon and cultured in the presence of IL-1β irreversibly lose their tenogenic identity and increase their chondrogenic and osteogenic capacities [53].
TGFβ signaling is critical to fibrosis and scar formation in connective tissues and has also been implicated in pathogenesis of tendon injury [57, 70,71]. TGFβ signaling activity, from TGFβ-1 in particular, is increased in injured tendon, particularly in regions of chondrogenic metaplasia and heterotopic ossification [63,64]. Excessive TGFβ signaling also stimulates pro-inflammatory effects and tenocyte apoptosis [70]. TGFβ alters the metabolic activities of tenocytes during healing, increasing collagen secretion and consequent scar tissue formation, providing a therapeutic rationale for TGFβ signaling blockade. In support of this concept, attenuation of TGFβ signaling by targeting TGFβ-1, CTGF and Smad 3 with anti-sense oligonucleotides reduced scarring and adhesion formation in a murine flexor tendon repair model [72]. Further, the chondro-degenerative lesions induced by injured TDPCs in healing tendons are dependent on TGFβ signaling [52].
Chondro-osteogenic BMPs, such as BMP-2, BMP-4, BMP-6 and BMP-7, promote cartilage, bone and bone-tendon junction repair [73][74][75][76][77]. Several lines of evidence implicate dysregulation of BMP activity in tendinopathy as a cause of ectopic calcification. Ectopic overexpression of BMPs is observed in naturally occurring calcifying tendinopathy and experimental models of tendinitis [78,79], and intra-tendinous administration of rhBMP-2 results in ectopic calcification [80]. Murine TDPCs exposed to BMP-2 during in vitro culture prior to subcutaneous Altered tendon matrix composition: Disorganized collagen matrix, increased non-collagenous ground substance and an increased number and rounded morphology of the tenocytes are hallmarks of injured tendon [81,82]. During the healing process of experimental and naturally occurring tendinitis, the levels of large proteoglycans and sulfated glycosaminoglycans (characteristic of cartilage matrix) increase within tendon matrix [62,[83][84][85]. These changes reduce the elasticity and tensile strength of the repair tissue. Proteoglycans such as biglycan, decorin, fibromodulin and lumican, although constituting a very small portion of tendon ECM, are active participants in collagen fibrillogenesis [86][87][88][89] and can also bind and sequester growth factors such as TGFβ and IGF-I [52,70,90-92] to modulate TDPC activities.
Altered tendon biomechanics: TDPCs, like terminally differentiated tenocytes, are sensitive to repetitive tensile loading in vitro [30,93]. Short-term treadmill exercise increased the yield of TDPCs from murine Achilles and patellar tendons. Further, the TDPCs isolated from exercised mice had higher biosynthetic activities than control mice. In vitro exposure to 4% tensile strain promoted tenogenic differentiation, whereas 8% tensile strain induced osteogenic differentiation [30]. TDPCs exposed to in vitro mechanical loading increased BMP-2 expression and had a higher osteogenic potential compared to unloaded TDPCs [93]. A recent study demonstrated that applying in vitro biaxial mechanical stress induces the expression of the proteoglycans, fibromodulin, lumican and versican in TDPCs [94]. These findings provide a mechanistic explanation for ectopic calcification that occurs as a result of mechanical overloading. The combination of excessive loading, BMP up-regulation and alterations in non-collagenous protein expression could generate conditions favoring TDPC chondro-osteogenesis, at the expense of tenogenic differentiation.

Applications of tendon-derived progenitor cells in tendon regeneration
Stem cell implantation has improved tendon healing in most studies [reviewed in 10,11,14,15]. Accepting the experimental benefits of stem cell-based therapies, several factors must be considered while choosing a particular cell type to treat tendon injuries. The tissue source must be easily accessible with acceptable donor site morbidity. The requirements for vitro manipulation (expansion and phenotypic modulation) should be minimized, to mitigate the risks of contamination and 'chain of custody' lapses. Finally, the phenotypic and reparative activities of the therapeutic cell type should closely match the target tissue. In this respect, tenogenically-committed stem/progenitor cells derived from tendons seem more appropriate for tendon regeneration than MSCs obtained from other tissue sources.

In vitro evidence supporting TDPCs for tendon regeneration
Several recent studies have evaluated the tenogenic potential of TDPCs under in vitro conditions. Acellular tendon has been used in tissue-engineering studies as a scaffold in cell-based approaches for flexor tendon injuries in murine models [95,96]. Decellularization of tendons was carried out by freeze-thaw cycles followed by trypsin digestion. Co-culturing TDPCs with acellular tendon matrix in vitro significantly increased their tenogenic marker expression and subsequent tenogenic differentiation. This acellular tendon-matrix model has been used by our group for comparative analyses of equine TDPCs and bone marrow for tendon healing [97,98]. TDPCs were more viable and showed superior integration into acellular tendon matrices than bone marrow MSCs ( Figure 5). In addition, TDPCs had significantly higher collagen and proteoglycan synthesis levels than bone marrow MSCs. A follow-up study showed that supplementing FGF-2 during monolayer expansion of TDPCs potentiated the biosynthetic activities of TDPCs compared to bone marrow MSCs during in vitro culture with pulverized acellular tendon matrix [35]. Human fetal TDPCs cultured in aligned nanofibrous scaffold supported tenogenesis and suppressed osteogenic differentiation [28]. Collectively, these results indicate that TDPCs respond to tendon matrix components by adopting a biosynthetically active tenogenic phenotype, supporting the strategy of implanting these cells into healing tendon lesions.

In vivo evidence supporting TDPCs for tendon regeneration
The benefits of cell-based treatments for experimental tendon defects were first reported in 2002. Autologous tenocyte constructs were used to bridge partial flexor tendon defects in adult chickens [99]. Fourteen weeks following implantation of tenocytes, the histologic structure and biomechanical properties of the tenocyte-treated tendons were significantly improved compared to the untreated controls. Since the discovery and characterization of TDPCs, several studies have evaluated the reparative activity of TDPCs in in vivo models of tendon injury. These studies are summarized in Table 1. Ni et al. (2012) were the first to investigate the in vivo healing characteristics of TDPCs in a rat patellar tendon window defect model [100]. GFP-labeled TDPCs in a fibrin glue matrix were injected into the defect. TDPC numbers at the site of injection decreased over time and were completely absent by four weeks after implantation. TDPC implantation improved the histologic, biomechanical and ultrasonographic characteristics of patellar tendon healing. No ectopic bone formation was detected at 4 and 16 weeks post-injury, which were the end-points of this two-part study. A subsequent study by the same group evaluated the effect of rat patellar tendon-derived TDPCs transduced with scleraxis (SCX) in the same model [101]. The histological and biomechanical characteristics of TDPC-Scx treated tendons were significantly better than tendons treated with TDPCs transduced with empty viral vector, suggesting that genetically priming TDPCs for tenogenic differentiation is clinically beneficial, although an alternative to viral delivery will likely need to be developed prior to approval for clinical applications.
A recent study by our group evaluated the effect of autologous TDPCs in a collagenase-induced equine flexor tendinitis model [102]. The TDPCs were implanted four weeks after tendinitis induction, and improved the biomechanical and histological characteristics of the tendons, 12 weeks following administration. The biochemical and transcriptional outcomes were not significantly influenced by TDPC injection. Similar to the findings of Ni et al. (2012) [100], TDPCs were not detected at the injection sites beyond 4 weeks after treatment. Significant improvements in collagen micro-architecture have also been reported in this equine model, following bone marrow MSC [103], adipose-derived MSC [104] and fetal-derived embryonic stem cell [105] administration. The biomechanical consequences of these other progenitor types have yet to be determined and the relative merits of these stem cell populations for tendon repair have yet to be defined in direct comparative studies.
Accepting the multipotency of TDPCs, priming TDPCs for tenogenic differentiation during in vitro culture and expansion has been investigated to avoid abnormal matrix deposition within the healing tissue [106]. Connective tissue growth factor (CTGF) is highly expressed during early stages of tendon repair [107]. In addition, CTGF stimulates tenogenic differentiation of TDPCs in vitro, when supplemented with ascorbic acid [108]. Rat patellar tendon-derived TDPCs were cultured with CTGF and ascorbic acid for two weeks to produce a thin cellular sheet, before being transplanted into patellar tendon window defects [109]. The biomechanical and histologic characteristics of the TDPC-treated tendons at 2, 4, and 8 weeks were improved compared to un-treated controls. A recent follow-up study by the same group included an additional experimental group in which untreated TDPCs in a fibrin glue matrix (i.e., cultured without CTGF and ascorbic acid) were implanted into the patellar tendon window defect [110]. The biomechanical, histological and ultrasonographic characteristics of tendons defects treated with TDPCs that were exposed to CTGF and ascorbic acid were significantly better than defects treated with control TDPCs, suggesting that pre-implantation 'priming' of TDPCs will be clinically beneficial.
Autologous products like platelet-rich plasma (PRP) can also be administered with TDPCs to enhance tendon repair. Intralesional PRP administration alone has improved the overall healing characteristics of repair tissue in experimental models of tendinitis [111,112]. In vitro, platelet-rich clot releasate stimulates tenogenic differentiation of TDPCs while inhibiting osteogenic differentiation [113]. TDPCs and PRP synergize to stimulate collagen gene expression of healing rat Achilles tendons [114]. In a follow up study by the same group, passage 2 TDPCs were cultured with platelet-rich clot releasate for 3 days prior to in vivo administration [115]. TDPCs supplemented with PRP prior to intralesional administration significantly improved the overall healing characteristics of the Achilles tendon compared to control TDPCs. These findings support the hypothesis that PRP likely augments the local trophic factor synthesis and cytokine modulatory effects of TDPCs and improves reparative effects, while also acting as a bio-compatible delivery vehicle.

TDPCs in tissue engineering
TDPC activities are heavily influenced by extracellular matrix. Given that there is gross disruption of tendon matrix in acute injuries and major changes in extracellular matrix composition in chronic tendinopathy, incorporating TDPCs in an appropriate 'teno-inductive' scaffold may improve tissue repair, in comparison to direct injections of TDPCs. Biological scaffolds seeded with differentiated tenocytes and bone marrow MSCs have improved repair of tendon defects in several in vivo models [reviewed in 13]. Similar approaches using TDPCs have been evaluated in a few studies. Zhang et al. (2009) prepared decellularized matrix by pulverizing and nuclease-digesting tendon. This matrix promoted proliferation and tenogenic differentiation of human and rat patellar tendon-derived TDPCs in vitro [95]. Subsequently, TDPCs cultured with decellularized tendon, or TDPCs alone were implanted subcutaneously, along the dorsal midline, and into patellar tendon window defects of nude mice. Interestingly, TDPCs cultured with tendon matrix synthesized neotendon tissue whereas naïve TDPCs did not form recognizably tendonlike tissue at either site. A decellularized matrix prepared from dermal fibroblasts was used to support rat and human patellar tendon-derived TDPCs for one week prior to in vivo implantation in a similar in vivo model [116]. Co-culturing TDPCs with dermal fibroblast-derived matrix promoted tenogenic differentiation in vitro and neo-tendon formation in vivo, whereas these effects were not seen with control TDPCs. Given that matrices from both tendinous and non-tendinous sources support proliferation and tenogenic differentiation of TDPCs, an appropriate bio-matrix could optimize the therapeutic value of these cells after in vivo delivery, although implantation of semi-solid cell-matrix composites is clearly more invasive and complicated than percutaneous injection.

Future directions and conclusions
Understanding the basics of TDPC biology is critical for their successful application in tendon repair/regeneration. To date, markers specific to TDPCs and tenocytes in general are poorly defined, and a reliable in vitro tenogenic differentiation assay is still lacking, although research efforts in this area are ongoing. TDPCs are heterogeneous cells and therefore developing a single marker that can definitively identify TDPCs is likely not feasible. Studies focusing on identifying a panel of co-expressed markers to define TDPCs are more realistic. Selfevidently, determining whether any given immunophenotypic tendonderived sub-population holds any therapeutic advantage will also need to be addressed in rigorous in vivo models.
TDPCs and their fate in tendon tissue can vary considerably on the physiological or pathological status of the tissue. As our current understanding of the mechanisms controlling tendon pathology and repair are limited, identifying the cellular phenotype(s), and delineating the biological and molecular processes influencing healthy tendon repair will be critical. Fetal tendon defects heal via an authentic regenerative process, unlike adult tendons [117]. Elucidating the reparative mechanism of fetal healing may provide insights to enable the recapitulation of regeneration in adult tendons. As detailed above, this will likely involve modulating TGFβ signaling in the healing tendon.
Tendon healing in experimental models has been substantially improved with cell-based and other biologic approaches, although these therapies do not completely restore the tissue microarchitecture. With the data from recent cell-tracking studies, it is well established that exogenous stem cells are cleared from the injection site within a few weeks and do not directly contribute to the pool of tenocytes and/or progenitor cell participating in tendon repair/regeneration [32,118,119]. Future studies focusing on cytokines and/or trophic factors secreted by TDPCs that mediate their therapeutic effects could simplify therapy considerably. Optimizing stem cell delivery by combining cells with teno-inductive scaffolds may retain cells at the implantation site for longer periods of time, with correspondingly longer therapeutic actions.
Restoring the biomechanical function of repair tissue should be the ultimate goal of any regenerative therapy for treating tendon injuries. Tendons respond to tensile loads via elongation and sliding between each element of the hierarchical structure. As the gross and microscopic structure of tendons is disrupted in tendon injury, the ideal regenerative therapy must restore the hierarchical structure of tendons and the sliding mechanisms of the tendon components, in addition to improving collagen alignment to regain full biomechanical function. Currently, collagen fiber pattern along the long axis of the tendon is the major outcome parameter used to assess tissue morphology but this does not address restoration of tertiary structure. Developing non-invasive and histological techniques that comprehensively assess tendon histology at multiple levels of matrix organization will be vital to comprehensively evaluate new therapies.
Finally, the clinical use of TDPCs for treating tendon injuries is dependent on identifying a suitable tissue source, with minimal donor site morbidity. In general, a tenectomy procedure for isolating autologous TDPCs is more invasive than bone marrow aspiration. In our in vivo study, autologous equine TDPCs were derived from the lateral digital extensor tendon with minimal post-operative morbidity [32]. In human patients, healthy tendon autografts are routinely obtained from patellar [120,121] and palmaris longus tendons [122], for cruciate and ulnar collateral ligament reconstruction surgeries, respectively. These sites could also be used for autogenous TDPC isolation. Further work is needed in this regard. Tendon tissues excised during reparative surgeries of Achilles tendon [123] and rotator cuff injuries [124] could also be used to isolate autogenous TDPCs; however, given the altered characteristics of TDPCs from pathological tissues, further research on these TDPCs is required before this possibility can be considered feasible. Allogeneic TDPCs can also be considered, as they avoid donorsite complications and culture delays that are required to generate clinically relevant autologous cell numbers. Although several in vivo experimental studies demonstrate the therapeutic benefits of allogeneic TDPCs, major regulatory and immunogenic concerns exists to prevent translating this option to human tendon injuries.