A multi-chamber tissue culture device for load-dependent parallel evaluation of tendon explants

Semitendinosus tendons from New Zealand white rabbits (mean + sd weight of 3.62 + 0.27 kg) were donated by Mayo Clinic researchers after the conclusion of their studies that were independently approved by the Institutional Animal Care and Use Committee (IACUC). Tissues were harvested immediately following sacrifice to ensure consistency and maximize similarity to an in vivo setting. Tendons, weighing 84.7 + 23.9 mg, were excised proximally at the tendon-muscle junction site and distally at the tibia insertion site under aseptic conditions. Any remaining muscle or other soft tissue was carefully removed prior to immersion in transport media comprised of advanced Modification of Eagle’s Media (aMEM) (Invitrogen, Carlsbad, CA), supplemented with 10% fetal bovine serum (FBS), 1% penicillin/streptomycin, and L-glutamine (Invitrogen, Carlsbad, CA), standardized to a pH of 7.2. All tissues were divided into five experimental groups: (1) Control samples snap frozen immediately after collection, (2) tissues cultured in a 60 mm dish w/mesh, (3) tissues cultured in the explant device without tension, (4) with ~ 12 g of tension, and (5) with ~ 21 g of tension applied to the tissues using a suture system (Fig. 1).

Cylindrical wells were fabricated from Poly (methyl methacrylate) (PMMA) rods (5 × 1 cm) by removing a central core of 5 mm in diameter. Both ends of the cylinders were threaded to allow attachment of cylinders to the base and top cap. Two-millimeter central openings were added to cylinder caps to ensure sufficient gas exchange and facilitate passage of a holding suture. The bottom between the base and the cylindrical well was sealed with a rubber washer to prevent leakage. A stainless steel hook securely fixed with a screw was placed to the bottom of each well. A suture was then placed between the tissue specimen and a specified load was guided over two crossbars attached to the device base. Twelve independent cylindrical tissue culture wells were fixed to the base of the device (Fig. 2). All parts of the device were sterilized prior to conducting the experiments by Ethylene Oxide (EtO) gas to prevent contamination without causing damage to the PMMA.

Following removal of transport media and thorough rinsing of the tendons in phosphate buffered saline (PBS), an Ethibond 3–0 (Ethicon, Sommerville, NJ, USA) holding suture was added to the proximal end, under aseptic conditions. The distal end of the tissue specimen was then attached to the stainless steel hook in each well bottom before the cylinder was passed over the graft and securely attached to the device base. The suture was passed through the opening in the top of the cap and a load was applied (or not applied in the case of control samples). To assess correlation between applied tension to the tissue and gene expression, two different loads (~ 12 g and ~ 21 g), in addition to unloaded (0 g) control specimens, were compared. Applied loads were equivalent to 0.12 Newtons (N), 0.21 N and 0 N, respectively. Even though the applied loads were multiple times the weight of the tendon tissues itself, our main intention was to maintain the tissue under tension in culture while avoiding micro damage to the sub-synovial connective tissue, including microtears and rupture of thin fibrils. According to Morizaki et al. [20], the threshold of applied force for damage to rabbit tendons appears to be ~ 200 mN. Similar tensions (0.04 N – 0.2 N) were applied by Arnoczky et al. [17] when assessing the response of rat tail tendons to static tensile loads. We therefore set our maximum load to 0.21 N, with a group receiving half that load (0.12 N) and a third group receiving no load (0 N). None of the tissue samples were pre-tensioned. Growth media was added to each well in the following formulation: 900 μL of aMEM, 5% platelet lysate, 1% penicillin/streptomycin, and L-glutamine.

Tissue specimens for the external control group were cultured in 60 mm dishes, placed in a custom-made mesh to keep tendons straight, but without any longitudinal or compressive forces applied to the tissue. Tissues were maintained in 5 mL per dish of the same 5% platelet lysate media. The explant culturing system and dishes were placed in a controlled incubator maintained at 37.5 °C, 95% humidity and 5% CO₂. Media were changed every 24 h. All tissues in group 1 (day 0) were harvested from animals immediately after euthanasia, debrided of non-tendon tissue, rinsed in PBS, and snap frozen in liquid nitrogen. For specimens in treatment groups 2 thru 5, harvested tissues were rinsed in PBS, debrided of non-tendon tissue, cultured for seven days, rinsed in PBS, and snap frozen in liquid nitrogen for further analysis by RNA extraction and real time quantitative PCR (RT-qPCR).

RNA was isolated using the miRNeasy kit (Qiagen, Hilden, Germany). Isolated RNA was reverse transcribed into cDNA using the SuperScript III first strand synthesis system (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA). Gene expression for selected gene markers was quantified using RT-qPCR whereby each reaction was performed with 2.5 ng of cDNA per 10 ul, QuantiTect SYBR Green PCR kit (Qiagen, Hilden, Germany), and the CFX384 real time system machine (Bio-Rad, Hercules, California, USA). Transcript levels were quantified using the 2^ΔΔCt method and normalized to the housekeeping gene Actb (set at 100).

Heat maps of the RT-qPCR data for (i) culture conditions and (ii) device conditions were generated using GENE-E v.3.0.215. Analysis of variance (ANOVA) was used to test for significant differences between (i) the three culture conditions or (ii) the three device conditions using SAS v.9.4 (SAS Institute Inc., Cary, NC). To account for multiple comparisons, we adjusted the false discovery rate (FDR) using the Benjamini-Hochberg procedure. Box and whisker plots identify the median and interquartile range of the data sets, while outliers were plotted as individual points. In all cases, significance was defined as p < 0.05. Venn diagrams made within FunRich v3 [21] were used to show the distribution of genes with increasing or decreasing fold changes for the mesh dish tendons and the 0 g tendons relative to frozen controls. The areas of each circle were proportional to the number of genes in each group.

A total of 37 fresh frozen tendons were collected from rabbits donated to our study and divided into five groups (1 thru 5) and three experiments (A, B, and C) (Figs. 1, 2). Tissues in groups 1 and 2 were not cultured in the device, but served as unloaded controls cultured independent of our novel tensioning system. We did not observe any errors or complications related to the use of the culturing system for explant tendons in groups 3, 4, or 5.

A set of 20 rabbit-specific qPCR primer pairs (see Additional file 1 for a complete list) were used to characterize biological effects of different culture conditions (substrate type, or volume of media). Analysis of these data by hierarchical clustering revealed clear patterns of cladistic grouping among control samples, samples cultured in a custom mesh sandwich in dishes, and samples cultured in the device without weight. Interestingly, Col1a1, Col3a1, and Fn1 markers were elevated in both the mesh dish- and device- cultured tissues, whereas Dcn, Spp1, Col5a1, and Acta1 were upregulated in control tissues (Fig. 3). Further, compared to control samples, gene expression of Col1a1 was significantly increased in both mesh dish and unloaded samples. While mesh samples demonstrated significantly increased expression of Col3a1, 0 g samples had increased expression of Col10a1. Mmp1 and Mmp13 expression was significantly increased in 0 g samples (Fig. 4).

Overall, adjusted FDR values for collagen (Col1a1, Col10a1, Col3a1) and matrix metalloproteinase markers (Mmp1, Mmp13) were significantly different among the three culture condition groups. The group differences for Mmp10 data were nearly significant (p < 0.053; Table 1). Additionally, Fn1 and Alpl values were statistically different among the groups (Table 1). Further, the noted findings were confirmed by the FunRich diagram, demonstrating an increased fold-change for both Col10a1 and Col5a1 and matrix metalloproteinase markers (Mmp1, Mmp3, Mmp10 and Mmp13) in the 0 g samples compared to the flash frozen controls (Fig. 5).

Hierarchical clustering analysis did not reveal a clear pattern of distinction between samples cultured in the device (with or without tension). Of note, however was the clustering of molecular markers into distinct clades formed primarily by (i) an Mmp and Collagen cluster, (ii) a cluster of mineral deposition proteins, and (iii) a cluster comprised of Dcn and Fn1 (Fig. 6). There was no difference in expression between zero loading and the loaded samples for the various collagen makers. We did note a significant increase in Mmp1 expression in 0 g load compared to both 12 g and 21 g. In addition, Mmp10 was increased in 0 g tendons, however the relationship was significant only when compared to 21 g samples. Zero load samples also demonstrated increased expression of Adamts4 and Hapln1 (Fig. 7). Importantly, differences among device-cultured samples for the following primer sets were significant (p < 0.05): Hapln1, Mmp1, Mmp10, and Adamts4. Nearly significant (p < 0.06) differences identified Ibsp and Mmp13 as interesting molecular markers related to the biological response of tendon tissues exposed to different levels of strain (Table 2). Increasing and decreasing fold changes (relative to 0 g controls) are represented by Venn diagrams in Fig. 8.