Development of a novel in vitro insulin resistance model in primary human tenocytes for diabetic tendinopathy research

Sample procurement

This study was conducted in accordance with the recommendations and approval by the University of Malaya Medical Center (UMMC) Medical Research Ethics Committee (MREC reference number: 20157-1486 and 20164-2398). Tendon specimens were collected from donors with informed consent. Remnants of hamstring tendon grafts were obtained from patients undergoing anterior cruciate ligament reconstruction (Table S1) for human primary tenocyte (hTeno) culture. For immunostaining of insulin receptor, normal Achilles tendons were collected from patients requiring a major lower limb amputation due to traumatic injury or oncology and were free from T2DM and infection.

hTeno Isolation and Culture

Hamstring tendon specimens were collected into sterile phosphate-buffered saline (1X PBS; Worthington Biochemical Corporation, United States), supplemented with penicillin-streptomycin (1% v/v; Gibco, USA), stored at 4° C for same day processing. Tendon explant cultures were prepared as previously described (Tan et al., 2012), the primary outgrowth fibroblastic cells were cultured to 80–90% confluence, trypsinized and sub-cultured to P2 or P3.

Immunofluorescence Staining

For immunocyto-staining, hTeno were seeded at 2000 cells per chamber in the 8-well chamber slide (Thermo Scientific™, Singapore), and cultured for 24 h. Then, the cells were fixed with ice-cold 100% methanol for 5 min at room temperature, and proceed to immunofluorescence staining (Methods S1). Antibodies used were mouse monoclonal anti-insulin receptor beta (anti-INSR-β; 1:20, Thermo Fisher, Singapore), rabbit monoclonal anti-glucose transporter type 1 (anti-GLUT1; 1:50, Sigma, Singapore) and rabbit polyclonal anti-glucose transporter type 4 (anti-GLUT4; 1:50, Sigma, Singapore). For immunohisto-staining, snap frozen Achilles tendons were cryo-sectioned, fixed with 4% buffered formaldehyde for 10 min at room temperature and proceed to immunofluorescence staining (Methods S1). Tissue sections were stained with mouse monoclonal anti-insulin receptor beta (anti-INSRβ; 1:20, Thermo Fisher, Singapore).

For both tendon tissue or hTeno cells, the secondary antibodies used were from Abcam, USA, namely Alexa Fluor^® 488 donkey anti-mouse IgG H&L (1:1000; ab150105), Alexa Fluor^® 647 goat anti-rabbit (1:1000; ab150079) and Alexa Fluor^® 555 donkey anti-rabbit IgG H&L (1:1000; ab150074). All images were captured using the Leica TCS SPII confocal laser scanning microscope (Leica Microscopy, Mannheim, Germany) with LAS AF Lite software. Images were captured with sequential scanning to avoid signal cross-talk and enhance image quality.

Optimization of Human TNF-α Concentrations

To determine the effects of the different concentrations of recombinant human TNF-α, hTeno were plate at a cell density of 210 cells per mm² and cultured for 24 h with low glucose DMEM supplemented with 10% FBS, 1% penicillin-streptomycin and 1% GlutaMAX. Then, the cells were synchronized with 0.5% FBS supplemented DMEM for 24 h. Next, different concentrations of recombinant human TNF-α (R&D Systems, USA) were added to the hTeno culture: 0.008, 0.08, 0.8 and 8 µM, and cultured for another 24 h (This will be the endpoint, if not indicated differently) at 37 °C. For hTeno cultured with insulin supplement, human recombinant insulin (10 µg/mL; Thermo Fisher, Singapore) was added into the culture at 30 min before the end point. In all the assays, the hTeno without TNF-α and insulin was used as the untreated control (basal group), and the positive control (the insulin-stimulated basal group which is supplemented with 10  µg/mL insulin and without TNF-α) was also included in each experiment. For glucose uptake assay, total collagen assay and type I collagen assay, the end points were at 24 h after adding the TNF-α. For the apoptosis assay and gene expression analysis, the end points were at 24, 48 and 72 hr.

For glucose uptake assay, the hTeno cells were seeded in a 24-well plate and analysed using the 2-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl) amino]-2-deoxy-d-glucose (2-NBDG) glucose uptake assay kit (BioVision, San Francisco), as described in Methods S2. For the total collagen assay and type I collagen enzyme-linked immunosorbent assay (COL-I ELISA) assay, hTeno were seeded in the 24-well plate and analysed using Sircol™ Soluble Collagen Assay kit (Biocolor, Carrickfergus, United Kingdom) and human collagen type I alpha (COL1A1) ELISA kit (Cusabio Biotech, China) respectively (Methods S3 and S4).

Gene expression analysis was performed for candidate tenogenic markers [Scleraxis (SCX) and Mohawk (MKX)] and extracellular matrix (ECM) metabolism-related markers [type I collagen (COL1A1), type III collagen (COL3A1), matrix metalloproteinases 9 (MMP-9) and matrix metalloproteinases 13 (MMP-13)]. The qPCR analysis was completed using the Integrated Fluidic Circuit (IFC) plate with Biomark HD real-time PCR system (Fluidigm, South San Francisco, California, USA) (Methods S5).

Apoptosis assay was performed using the FITC Annexin V/Dead cell apoptosis kit (Invitrogen, California, USA; Methods S6). The stained and unstained cells were analyzed with the BD FACSCanto™ II flow cytometer (BD Biosciences, USA). The samples were acquired at fluorescence emission at 530 nm and >575 nm. The populations were analyzed in three groups: live healthy cells (with a low level of fluorescence signal), apoptotic cells (annexin V positive cells) and the dead cells (annexin V and propidium iodide double positive cells). The percentage of healthy cells and apoptotic cells were recorded and analysed.

Statistical analysis

At least 3 independent experiments were performed for each assays. Data were analyzed using SPSS software (version 22). All raw data were normalized with their basal group. Either parametric test (independent t-test, ANOVA or Turkey Post-Hoc test) or non-parametric tests (Kruskal–Wallis or Mann–Whitney U) were used based on the normality of the data sets (Methods S7, Table S2). Data were presented as either mean ± standard deviation or median ± interquartile ranges. A p-value of less than 0.05 (p < 0.05) was considered statistically significant. Different significance symbols were applied, where:

1. For the comparison between the treatment groups (with different concentrations of TNF-α) versus the basal group (untreated control; without both TNF-α and insulin), * indicates p < 0.05 and † indicates p < 0.01, versus basal group.

2. For the pairwise comparison between the insulin-stimulated groups versus their respective paired-treated groups without insulin stimulation, ‡ indicates p < 0.05 and § indicates p < 0.01, versus paired-treated groups without insulin stimulation.

3. For the comparison between fold change of different concentrations of TNF-α versus the non-TNF-α treated basal group, ∥ indicates p < 0.05 and ¶ indicates p < 0.01, versus non-TNF-α treated basal group.

Primary Human Tenocytes Culture

In this study, the primary human tenocytes (hTeno) were cultured in monolayer culture. At day-0 when the tendon explants were placed in the cell culture flask, the cells were floating in suspension and some were within the digested tendon collagen fibers (Fig. 1A). The fibroblastic cells started to attach to the plastic surface of the cell culture flask on day-14 of the explant culture. The hTeno cells grow in colonies (Figs. 1B and 1C) and proliferate to reach confluence within about 30 days after the cells attached to the culture flasks (Fig. 1).

INSR, GLUT1 and GLUT4 Expression in Tendon and hTeno

The single transmembrane domain of the INSR-β was positively stained in the tendon cells in the human Achilles tendons (Fig. 2) as well as in the hTeno in monolayer culture (Figs. 3A–3C). The GLUT1 and GLUT4 were both positively expressed in the hTeno (Figs. 3D–3F and Figs. 3G–3I). The GLUT4 were distributed on the hTeno plasma membrane (Fig. S1).

TNF-α reduced insulin-mediated 2-NBDG glucose uptake activity

The hTeno treated with TNF-α showed a significant increase in the relative 2-NBDG uptake compared to the basal group regardless of the TNF-α concentrations, with and without insulin supplement (Figs. 4, 5A, Table S3A). In the pairwise comparisons between treatment groups with and without insulin stimulation, only the basal group with insulin-stimulation showed a significant increase in the relative 2-NBDG uptake (1.414 ± 0.490; Fig. 5A and Table S3B) compared to its corresponding group without insulin stimulation (1.000 ± 0.000). There was a significant decreased in the fold change of insulin-mediated 2-NBDG uptake in the hTeno treated with all the different concentrations of TNF-α compared to non-TNF-α treated basal group (Fig. 5B and Table S3C). The hTeno treated with 0.008 µM TNF-α showed the greatest reduction in the 2-NBDG uptake (0.963 ± 0.13) compared to the other concentrations of TNF-α (1.041 ± 0.030; 0.968 ± 0.140; 1.018 ± 0.070 for 0.08, 0.8 and 8 µM TNF-α respectively), in relative to the based line (1.414 ± 0.490).

TNF-α suppressed insulin-mediated total collagen expression at concentrations lesser than 8 µM

The hTeno treated with TNF-α showed a significant increase in relative total collagen expression compared to basal group (1.000 ± 0.000), regardless of the TNF-α concentrations with and without insulin (Fig. 6A, Table S4A). The hTeno treated with 0.008 µM TNF-α with insulin supplement showed smallest increment in the total collagen expression (1.100 ± 0.090), whereas the 8 µM TNF-α with insulin group showed the highest increment (2.332 ± 0.510), compared to the basal group. Interestingly, in pairwise comparison, between treatment groups with insulin stimulation and without insulin stimulation, hTeno treated with 0.008 µM TNF-α supplemented with insulin, showed a significant reduction in the relative total collagen expression level (1.100 ± 0.090; Fig. 6A and Table S4B), compared to hTeno treated with 0.008 µM TNF-α without insulin (1.393 ± 0.080), whereas for the other pairwise comparisons, a significant increase in the relative total collagen expression were observed only in the insulin-stimulated basal group (1.389 ± 0.110) and 8 µM TNF-α with insulin group (2.332 ± 0.510). A significant decrease in the fold change of insulin-mediated total collagen expression in the hTeno treated with 0.8 µM TNF-α or lower (Fig. 6B and Table S4C). At concentration higher than 0.8 µM TNF-α, a significant increase in the fold change of insulin-mediated total collagen expression was observed (1.980 ± 0.106) compared to the basal group.

TNF-α suppressed insulin-mediated Type I collagen (COL-I) expression

The hTeno treated with TNF-α showed a significant reduced in relative COL-I expression compared to basal group (1.000 ± 0.000) when treated with lesser than 8 µM TNF-α without insulin supplement, or at 0.008 µM and 0.8 µM TNF-α with insulin supplement (Fig. 7A, Table S5A). The hTeno supplemented with insulin showed a significant increase in the COL-I expression levels when treated with higher concentrations of TNF-α at 0.08 µM (1.135 ± 0.060) and 8 µM (1.174 ± 0.210). In the pairwise comparisons between treatment groups with and without insulin stimulation, only hTeno treated with 0.008 µM TNF-α (0.240 ± 0.180) showed no significant changes in the COL-I expression levels compared to its non-insulin treated 0.008 µM TNF-α group (0.257 ± 0.170). All the other pairwise comparisons, showed a significant increase in the COL-I expression levels. In the fold change of insulin-mediated COL-I expression levels, only hTeno treated with 0.008 µM TNF-α (0.968 ± 0.110) showed a significantly reduction compared to the basal group (1.324 ± 0.190; Fig. 7B and Table S5C).

TNF-α suppressed candidate tenogenic markers gene expressions

The hTeno treated with TNF-α showed a significant downregulation in their insulin-mediated normalized SCX mRNA gene expression levels compared to non-TNF-α treated basal group (1.000 ± 0.000), regardless of the TNF-α concentrations (Fig. 8A and Table S6A). A significant downregulation was also detected in insulin-mediated normalized MKX mRNA gene expression levels compared to non-TNF-α treated basal group (1.000 ± 0.000), regardless of the TNF-α concentrations (Fig. 8B and Table S6B).

TNF-α disrupted the balance between ECM metabolism in hTeno

The hTeno treated with TNF-α showed a significant downregulation in their insulin-mediated normalized COL1A1 mRNA gene expression levels compared to non-TNF-α treated basal group (1.000 ± 0.000), regardless of the TNF-α concentrations (Fig. 9A and Table S7A).

In contrast, insulin-mediated normalized COL3A1 mRNA gene expression levels were significantly upregulated at all the different TNF-α concentrations (Fig. 9B and Table S7B). Similarly, for both the insulin-mediated normalized MMP-9 (Fig. 9C and Table S7C) and MMP-13 (Fig. 9D and Table S7D) mRNA gene expression levels, a significant upregulation were observed at all the different TNF-α concentrations.

TNF-α increased apoptosis events in hTeno

Based on the findings from the 2-NBDG analysis, total collagen expression levels, COL-I expression levels and gene expression levels analysis, 0.008 µM TNF-α is selected as the optimum concentration to induce the IR condition in hTeno. To further analyzed the effect of 0.008 µM TNF-α in hTeno viability, apoptosis assay was performed in the hTeno treated with 0.008 µM TNF-α for 24 h, 48 h and 72 h. The results showed that 0.008 µM TNF-α increased the total number of apoptotic cells with increase of exposure time to TNF-α for 24 h, 48 h and 72 h (Figs. 10A–10D). There was a significant gradual decrease in the percentage of live cells in hTeno treated with TNF-α for 24 h (75.950 ± 5.080), 48 h (59.150 ± 27.800) and 72 h (41.650 ± 14.270), compared to the basal group (88.400 ± 10.100) (Fig. 10E and Table S8A). As for the apoptotic cells, there was a significant increase in the percentage of apoptotic cells with the increase of exposure time to TNF-α, from 24 h (23.700 ± 5.100) to 48 h (40.850 ± 27.800) and 72 h (58.350 ± 14.200), compared to basal group hTeno not treated with TNF-α (11.500 ± 9.550) (Fig. 10F and Table S8B).