Mature human tendon fibroblasts embedded in a fibrin gel remodel the matrix to form a linear tendon construct within ten days. The cellular network becomes organized in a parallel arrangement concomitant with matrix contraction and with the synthesis and secretion of collagen fibrils with a uniform, narrow diameter (). In the constructs, the fibroblasts are in close proximity to each other, the plasma membranes are extensively convoluted and collagen fibrils in fibripositors have been identified (). The collagen fibrils are aligned along the axis of tension, and we found a colocalization of collagen I with collagen III, XII, and XIV (A–I). Furthermore, a dense network of fibronectin was found and fibroblasts stained positively for integrin α5, suggesting cell–matrix interactions with numerous adhesion structures (J–L). Our findings strongly suggest that adult tendon fibroblasts in a 3D matrix are actively involved in the de novo arrangement of collagen fibrils in the extracellular space and are capable of recapitulating the fibrillogenesis of the developing tendon.
The fact that mature human tendon fibroblasts were found to synthesize collagen fibrils with a homogeneous and narrow diameter () was surprising, since this was thought to be a behavior seen only from embryonic tendon cells. On the other hand, the findings match very well with observations in rat tendon cells obtained from postnatal tissue 
and more recently with observations from human fibroblasts in culture 
. The latter publication examined human fibroblasts from dermis and tendon and demonstrated the formation of fibrils with a narrow-sized diameter in a cell culture system studied 5–18 weeks after cells were embedded in a matrix engineered from polyglyconic acid. In vivo
, Postacchini and Martino 
described the formation of thin immature collagen fibrils in rabbit calcaneal tendon following partial tenotomy with subsequent maturation of fibrils.
In the current study, the constructs were analyzed at day 0, i.e. the day when a continuous linear matrix was formed, and the timepoint might explain the short fibrils with a narrow diameter (42.1 ± 0.2 nm), the presence of cell extensions into the extracellular space, collagen fibrils in fibripositors, and the pronounced convolution of the plasma membrane of tendon fibroblasts, which are all characteristics of tendon during development. In mature tendon, these characteristics have not been found. The distribution and localization of collagen III, XII and XIV also represent processes during tendon development. The distribution pattern of collagen III and XII demonstrates a spatial shift during development [15,27]
. The early phase is characterized by a homogeneous colocalization of collagen III and XII with collagen I fibrils in the tendon proper, in later stages, the localization of the collagen types have only been found in the tendon sheath [11,15,16,27]
. Histological analysis of the constructs showed clearly a colocalization of collagen I and III in the midsection of the constructs, and of type I and type XII throughout the matrix. Collagen XIV has been shown to be inhibiting to lateral growth of collagen I fibrils 
and we found a strong immunoreactivity of collagen XIV with collagen I fibrils suggesting that this process is maintained in 3D culture. Collagen XIV binding to type I collagen fibrils matches to the presence of immature fibrils with a narrow diameter.
The assembly of the collagen and fibronectin network as well as integrin expression is tightly regulated during development [19,20,28,29]
. We found the establishment of a collagen I rich matrix concomitant with fibronectin assembly and integrin expression throughout the tendon construct, which suggests that the cells sense the tension developed through anchoring points in this 3D culture model and translate the stimuli into intracellular signals. Not only is information transmitted through integrin receptors from the ECM into the cells, but also from the intracellular space to the ECM and this interplay most likely has an essential role in expression and assembly of ECM components [19,28,30,31]
, for review see [20,32]
. Although integrin activation and associated signaling pathways were not investigated in the present study, the findings suggest that tendon fibroblasts rapidly synthesize an ECM, to which the cells become linked through adhesion structures. The finding of pronounced integrin α5
expression in the tendon constructs indicates the establishment of focal adhesions, which might be involved in the development of a parallel fibrillar network of collagen.
We suggest that the experimental conditions in this study had a stimulating effect on the tendon fibroblasts to initiate processing and deposition of collagen fibrils that shares features of immature tendon. Starting with a high cell-to-ECM and a high cell-to-cell ratio might be potent activators of developmental tendon fibrillogenesis. This hypothesis supports the present finding that mature tendon fibroblasts have the intrinsic capacity to behave like cells in the developing tendon, and suggests that the environment in developing and mature tendon determines mechanisms of fibril synthesis, deposition and alignment.
A high cell number could stimulate embryonic fibrillogenesis. As seen in , there is a high cell-to-matrix ratio in the present study, in contrast to cell–matrix ratio seen in mature human tendon 
. Cell–cell junctions allow cells to establish cell polarity and guide the cellular responses to the local environment [33,34]
. Richardson and co-workers 
found that the structural integrity of the developing tendon in the embryo is dependent on cell condensation through cell–cell interactions mediated by the junctional protein cadherin-11. A loss of cell contacts resulted in a concomitant loss of structural alignment of collagen fibrils. Also in the mature tendon, fibroblasts demonstrate immunoreactivity for cadherin-11 in vivo
(Bayer ML, unpublished observation) and isolated human tendon fibroblasts retain the ability to form cadherin-11 mediated cell junctions in 2D 
. This indicates that tendon fibroblasts form multiple cell junctions in the tissue, but also when isolated from the natural environment. In the tendon constructs, we found high immunoreactivity for cadherin-11 and connexin43 (data not shown), which suggests that the cells are interconnected and actively interacting during matrix formation.
Culture conditions might have a substantial impact on the results presented here. Fibrin is a substrate which is used extensively in tissue engineering 
, but it is not known, whether the human mature tendon fibroblasts respond with an injury response to the fibrin, given its natural occurrence during wound healing, and this reaction would explain the immature narrow-sized collagen fibrils. However, Deng and collaborators 
reported narrow-size fibril diameter in the range of ~20 nm after 5–9 weeks culturing in a 3D matrix made of polyglycolic acid. This suggests that the results seen in our study are not a direct response to fibrin. Postacchini and Martino 
described the synthesis of narrow-sized fibrils in an animal model following tendon injury. This might appear as an injury response, however, the fibrils underwent maturation in the weeks following injury, which indicates that collagen fibrillogenesis begins with immature thin collagen fibrils that subsequently mature. Moreover, it appears as if the fibrillogenesis we report is a well-ordered process. The appearance of collagen-associated proteins, the fibronectin network and the development of fibripositors are considered as support for this assumption. Further, we report the characteristic toe and linear region of the force-elongation curve when constructs were mechanically tested (). Accordingly, our findings are in contrast to what would be expected in unstructured scar formation after injury.
The choice, amount and composition of the serum have effects on cell proliferation, ECM production and force production of engineered tissue [37,38]
. Earlier studies on tendon and muscle engineering have applied a reduction from high to low serum content when cells in the model reached confluency [24,38]
, whereas other researcher have not altered serum concentration throughout the culturing period [21,25]
. In the present study, we chose not to change serum concentration over the culturing period. It cannot be excluded that some of the findings in collagen fibrillogenesis were triggered by creating a somewhat fetal
environment by the supplementation of a multitude of growth factors in high amounts throughout construct formation. Even if this is so, the findings demonstrate the capacity of fibroblasts to recapitulate embryonic mechanisms in a proper environment. This is in line with the idea that the hormonal and mechanical environment rather than the intrinsic cell activity determines cell behavior. This fits with findings in both animal and human skeletal muscle satellite cells, where isolated cells in culture demonstrated a behavior according to the characteristics of the growing medium, and where older cells showed rejuvenation when subjected to serum from young donors [39,40]
. This view is interesting from a clinical perspective, as it is well-described that mature human tendon has a poor regenerative capacity 
. Potentially, the regenerative problem may not relate to any intrinsic cell deficiency, but rather relates to an unfavorable environment, that does not allow for collagen fibrillogenesis.