All tissue samples analyzed were gifted by Professor J Okazaki, representing residual tissue arising from a separate study. A total of 12 male diabetic GK rats (six aged 10 weeks and six aged 18 weeks) and 12 age-matched male Wistar rats (Shimizu Laboratory Supplies Co. Ltd., Kyoto, Japan) were used. Raised levels for blood glucose and HbA1c were confirmed in the GK rats from the age of 3 weeks (11). All animal experiments performed were reviewed and approved by the Animal Committee of Osaka Dental University (approval number 08-03009). Sterile titanium alloy (Ti–6A1–4V) implants (length 17.0 mm, diameter 1.2 mm; SNK 123 screwpost Ti-tan R, Dentsply-Sankin K.K., Tokyo, Japan) were placed into the socket of a freshly extracted lower incisor, as previously described (11, 12). At 3 and 9 weeks after implant placement, three rats from each experimental group were euthanized by an intraperitoneal injection of sodium pentobarbital 30 min after an intraperitoneal injection of the anticoagulant sodium heparin (Novo-Heparin Injection 1,000®; Mochida Pharmaceutical, Tokyo, Japan). Tissues were fixed by perfusion with 10% neutral buffered formalin (11) and the mandible tissues containing the implants dissected.

Implants were gently unscrewed from mandibles. Using a bone saw, tissue blocks with a width of 2 mm were cut along the entire length of the mandible in an axis running perpendicular to the implant socket. Tissue pieces were demineralized with 10% formic acid for 72 h and then dehydrated by passing them through 70%–100% graded alcohol and finally into xylene, before embedding them in paraffin. Sections of 5 µm were cut and mounted on poly-L-Lysine coated glass slides (Sigma-Aldrich, Poole, UK), stained with hematoxylin and eosin (H&E), and mounted using a DPX mounting medium (ThermoFisher Scientific, Paisley, UK). Sections were examined using light microscopy.

Alternatively, tissue blocks were further cut in the sagittal and coronal planes to produce four pieces, each containing tissue that had been adjacent to the implant. Tissue blocks were demineralized using 6% ethylenediaminetetraacetic acid (EDTA) solution (pH 7.4), for 4 weeks. Specimens were washed in 0.01 M phosphate buffered saline (PBS) before dehydration through increasing concentrations of ethanol and then embedded in Lowicryl HM20 (Agar Scientific Elektron UK Ltd., Stansted, UK), using a progressive lowering of temperature (PLT) protocol. Full polymerization of the resin was achieved by using indirect ultraviolet (UV) light at 35°C for 24 h and direct UV light for another 72 h at room temperature (48).

Immunogold labeling of antigens was visualized using transmission electron microscopy (TEM), following the previously reported method of Hobot and Newman (48). Sections of 4 µm were cut from the Lowicryl embedded tissue blocks and collected onto 400 mesh nickel grids. Sections were washed with 0.01 M PBS (50 µl) for 10 min, blocked with 0.6% w/v bovine serum albumin (BSA), diluted in 0.01 M PBS (pH 7.4, 50 µl) for 10 min, and then incubated with an affinity-purified polyclonal rabbit IgG anti-rat TGF-β₁ antibody (diluted 1:10; Santa Cruz Biotechnology, Dallas, TX, USA) for 1 h. Goat anti-rabbit IgG (Sigma-Aldrich), conjugated to 10 nm colloidal gold, GAR-10 (Insight Biotechnology, Wembley, UK), was prepared as previously described (48), diluted 1:5 in filtered 20 mM Tris-HCl buffer (pH 8.2) and centrifuged at 10,000 g for 4 min. This secondary antibody was incubated with sections for 1 h. Replacement of the primary antibody with 0.6% BSA served as a negative control. Sections were washed for 1 min with 20 mM Tris–HCl buffer (2 × 2 min) and stained with 4% w/v aqueous uranyl acetate (VWR International, Lutterworth, UK) for 20 min. The stain was removed with deionized water and sections were air-dried and viewed using a CM12 Transmission Electron Microscope (Phillips Ltd., Cambridge, UK) operating at 80 kV. A total of 24 images were obtained from the tissue sections of three animals within each experimental group. To provide a semi-quantitative measure of TGF-β₁ levels within the respective healing collagenous matrices, gold-labeled particles were counted within randomly selected fields of view of 30 µm² within each image. The scatter of data was examined to determine mean, median, and quartile ranges. The normality of data was analyzed using the Shapiro–Wilk test. Only data derived from the Young Normal Week 3, Aged Diabetic Week 3, and Young Diabetic Week 9 groups indicated normality. Therefore, data were analyzed using a mixed-parametric one-way ANOVA incorporating the Tukey method (Graphpad Prism statistical software).

MSCs were isolated by explant culture from the rat compact bone chips (CB-MSCs) obtained from the diaphysis of femur and humerus long bones of 28-day-old male Wistar rats, as previously described (43). Tissue samples were collected in accordance with Basel Declaration guidelines and the UK Animals (Scientific Procedures) Act 1986, following Schedule 1 Code of Practice for the Humane Killing of Animals. Bone marrow was flushed from the long bones and dissected into 1–3 mm² bone chips, which were then digested with a source of collagenase II (sigma-Aldrich) for 2 h at 37°C. Mesenchymal cells from the endosteal and periosteal lining (CB-MSCs) were obtained from explants derived from the culture of the bone chips. CB-MSCs were expanded in complete culture medium (CCM), consisting of αMEM with ribonucleosides and deoxyribonuclosides (ThermoFisher Scientific), 20% heat-inactivated fetal bovine serum (FBS; ThermoFisher Scientific), 1% antibiotics–antimycotics (10,000 units penicillin, 10 mg streptomycin, and 25 µg amphotericin B per ml in original solution; Sigma-Aldrich), and 100 µM L-ascorbic acid 2-phosphate (Sigma-Aldrich), supplemented to achieve normal (5.5 mM) or high (25 mM) glucose concentrations. CB-MSCs were subsequently expanded under normoglycemic or hyperglycemic conditions at 37°C / 5% CO₂ for approximately 350 days in culture (44). The culture medium was changed every 2–3 days. Population doublings (PDs) were calculated as previously reported, with CB-MSCs expanded to reach PD15 (representing early PDs) or PD150 (late PDs) (44).

CB-MSCs at PD15 and PD150 were seeded at 4,000 cells/cm² and cultured in basal αMEM containing normal (5.5 mM) or high (25 mM) glucose concentrations. At 24 h, the culture media was replaced with respective media supplemented with 100 µM L-ascorbic acid 2-phosphate, 10 nM dexamethasone, and 100 µM β-glycerophosphate (all Sigma-Aldrich). CB-MSCs cultured in the absence of osteogenic factors served as negative controls. At days 2, 7, 14, and 21, the conditioned media and cells were recovered for analysis of gene expression by quantitative PCR (qPCR) or protein levels by western blot analysis (described below). At day 28, cells were fixed with 4% paraformaldehyde solution (Santa Cruz) and mineral-containing nodules visualized by staining with Alizarin red S (Sigma-Aldrich), pH 4.2, for 30 min at room temperature, with images captured by light microscopy (Eclipse TS100 Inverted Microscope; Nikon UK Ltd., Kingston upon Thames, UK).

A single-cell suspension of bone marrow cells was flushed from the long bones of 4-week-old male Wistar rats using ice-cold PBS by passing it through a 21-gauge needle and then passing it through a 70-µm cell strainer. Cells were centrifuged at 1,800 rpm for 5 min and re-suspended in BD Pharm Lyse™ lysing solution (2 ml; ammonium chloride-based reagent for lysis of red cells; BD Biosciences, Oxford, UK) for 2 min. Viable monocytes were recovered by centrifugation, re-suspended for seeding at 1 × 10⁶ cells/cm² in RPMI 1640 (ThermoFisher Scientific), and supplemented with 10% FBS, 1% antibiotics–antimycotics, and normal (5.5 mM) or high (25 mM) glucose concentrations. RPMI 1640 media was further supplemented with either rhGM-CSF (10 ng/ml) or rhM-CSF (10 ng/ml) and rhIL-4 (10 ng/ml) (all R&D Systems, Abingdon, UK) to promote monocyte differentiation into M1 or M2 macrophages, respectively. Monocytes were maintained at 37°C, 5% CO₂ for 7 days, with conditioned media collected on days 3 and 5 for analysis of TGF-β₁ protein levels by ELISA (described below). Cellular morphology was examined by light microscopy at day 7, before harvesting for analysis of M1 and M2 macrophage marker gene expression by qPCR at day 7.

Total RNA was extracted using the RNeasy® Mini Kit and QIAShredder (Qiagen, Manchester, UK). RNA purity and concentration were determined by absorbance at 260 nm/280 nm (NanoVue™; GE Healthcare, Little Chalfont, UK). Complementary DNA (cDNA) was synthesized from 500 ng total RNA, using 5 µl 5× Moloney murine leukaemia virus (M-MLV) buffer, 0.5 µg Random Primers, 0.6 µl RNasin, 1.25 µl deoxynucleotide triphosphates (dNTPs; 10 mM), and 1 µl M-MLV reverse transcriptase, reconstituted in 10 µl nuclease-free water (all Promega, Southampton, UK). cDNA was diluted 1:5 (oseteogenic markers) or 1:10 (macrophage markers) with RNA-free water. qPCR reactions were performed using a 5 µl diluted cDNA sample, 10 µl SYBR Green Precision qRT-PCR Master Mix (Primer Design Ltd., Southampton, UK), 2.5 µl 3 µM primers, and 2.5 µl RNA-free water. Osteogenic and M1/M2 macrophage marker gene primers were purchased from Primer Design Ltd. and Eurofins MWG Operon (Ebersberg, Germany), respectively (Table 1). Reactions were performed using an ABI Prism 7000 Sequence Detection System and ABI Prism 7000 SDS Software V1.0 (ThermoFisher Scientific). All reaction conditions were as follows: 1 cycle of 95°C for 10 min, 40 cycles of 95°C for 15 s, 55°C for 30 s, and 72°C for 30 s. Relative fold changes in marker gene expression (RQ) were calculated using the 2^–ΔΔCt method (49), normalized against the β-actin housekeeping gene.

Conditioned media was collected at days 2, 7, 14, and 21, centrifuged at 8,000 rpm for 5 min, and stored at −80°C, until required. At these same time points, adherent cells were washed in ice-cold PBS (×2) and cellular-associated protein extracted by treatment with 0.1% Triton X-100, 0.05 M sodium acetate buffer, pH 6.8, containing cOmplete™ Protease Inhibitor Cocktail (Roche, Burgess Hill, UK) at 37°C/5% CO₂ for 15 min. The residual ECM was washed in ice-cold PBS (×2) before treatment with 2% Triton X-100, 4 M guanidinium hydrochloride, 0.05 M sodium acetate buffer, pH 6.8, containing cOmplete™ Protease Inhibitor Cocktail at 37°C/5% CO₂ for 15 min. Conditioned media, cell lysates, and ECM extracts were exhaustively dialyzed against double-distilled water, containing protease inhibitors, before lyophilization. Extracts were re-suspended in PBS and protein concentrations quantified (Pierce® BCA Protein Assay Kit; ThermoFisher Scientific).

Protein samples (20 µg) were separated under reducing conditions by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) on pre-formed 4%–15% Mini-PROTEAN® Precast Gels (Mini-Protean® Tetra Cell System; BioRad, Hemel Hempstead, UK) and electroblotted onto polyvinylidene difluoride membranes (Hybond™-P; ThermoFisher Scientific) using a Semi-Dry Trans-Blot System (BioRad), as per the manufacturer's instructions. Membranes were blocked with 5% semi-skimmed milk/1% Tween 20 in Tris-buffered saline (TBS) at 4°C overnight. Membranes were immuno-probed with primary antibodies specific to TGF-β₁ (polyclonal anti-goat, 1:100; Santa Cruz), total Smad 2/3 (mouse monoclonal anti-rabbit, 1:1,000; New England Biolabs, Hitchin, UK), and using mouse monoclonal IgG against the core protein of decorin (70.6) and biglycan (PR8A4) (both used at 1:50; gifts from Professor Bruce Caterson, School of Biosciences, Cardiff University, UK), diluted in 5% semi-skimmed milk/1% Tween 20 at room temperature for 1 h. Membranes were washed (×3) in 1% TBS-Tween and incubated with either HRP-conjugated, rabbit anti-goat IgG (to detect TGF-β₁, 1:3,000; Santa Cruz), HRP-conjugated, rabbit anti-mouse IgG (to detect decorin or biglycan, total Smad 2/3, 1:50,000; Abcam), or HRP-conjugated, polyclonal swine anti-rabbit IgG (to detect β-actin Loading Control, 1:3,000; Dako UK Ltd., Cambridge, UK). Secondary antibodies were diluted in 5% semi-skimmed milk/1% Tween 20 for 1 h at room temperature. Membranes were washed (×3) in 1% TBS-Tween and TBS and incubated in ECL™ Prime Detection Reagent (VWR International) and autoradiographic films (Hyperfilm™-ECL; ThermoFisher Scientific) developed as per the manufacturer's instructions.

Conditioned media was collected during monocyte differentiation into the M1 or M2 macrophages under normoglycemic and hyperglycemic conditions at days 3 and 5, as described above. TGF-β₁ concentrations in the conditioned media were quantified using TGF-β₁ Platinum ELISA Kits (eBioscience, Hatfield, UK), as per the manufacturer's instructions. Absorbance values were measured using a FLUOstar® Optima Microplate Reader (BMG Labtech, Aylesbury, UK) at 450 nm, with data expressed as pg/ml.

Data are expressed as mean ± standard deviation (SD). The normality of data was confirmed using the Shapiro–Wilk test. The statistical significance of the differences between the experimental groups was evaluated using a parametric one-way ANOVA, followed by multiple comparisons using Tukey's test with Graphpad Prism statistical software. All experiments were performed in triplicate and repeated on three separate occasions.

Implants were inserted into freshly excised tooth sockets of 10-week-old (young) and 18-week-old (aged) diabetic male Goto-Kakizaki (GK) rats and age-matched, non-diabetic male Wistar rats; bone deposition was monitored using histological analysis (Figure 1). At 3 weeks after insertion, the non-diabetic (control) tissues of the young (Figure 1A) and aged (Figure 1C) animals exhibited cell-rich, granulation tissue formation around the implant insertion sites, with additional evidence of distance osteogenesis, identifiable as the formation of new bone-like tissue extending from the original bone tissue of the tooth sockets. At 9 weeks after insertion, new bone formation was noted to have progressed in both the non-diabetic, young (Figure 1B) and aged (Figure 1D) tissues, although the deposition of fully mineralized bone was observed to be more advanced in the younger tissues, with greater contact osteogenesis evident close to the implant surface, compared to the aged animals. The new bone was predominantly cancellous bone, which precluded accurate measurement of new bone formation by image analysis.

When comparing the young diabetic animals (Figures 1E,F) with their age-matched non-diabetic equivalents (Figures 1A,B), the rate of bone deposition was slightly reduced by the hyperglycemic wound-healing environment, with greater levels of mineralized bone via distance osteogenesis evident within the young non-diabetic group. However, and in stark contrast, significant areas of soft granulation tissue were still evident at 3 and 9 weeks after implant insertion with very little contact osteogenesis apparent within the aged diabetic rats (Figures 1G,H), indicating that bone healing was significantly impaired compared to both the aged normal and the young diabetic animals.

After visualization by TEM, representative images of immunogold labeling for TGF-β₁ obtained from reparative tissues close to the implant sites are shown in Figure 2. Antibody-free controls exhibited no gold labeling (Figure 2A). Initial observations showed that immunogold labeling, representing TGF-β₁ sequestration, was predominately associated with the pericellular regions (arrows, Figure 2B) and collagen fibers throughout the healing sites (arrows, Figures 2C,D). However, labeling did not conform to any regular patterning that matched the banding of the collagen fibers. Similar staining patterns were obtained for all experimental groups.

A semi-quantitative analysis of TGF-β₁ levels within the respective healing collagenous matrices was calculated and presented as box and whisker plots (Figure 2E). The data obtained demonstrated moderately widespread particle counts overall, suggesting an uneven distribution of labeling throughout the healing matrices. As the data point distributions did not exhibit normality for all groups, a Kruskal–Wallis non-parametric ANOVA analysis with Dunn's test for multiple comparisons was performed, which identified no significant differences between diabetic and normal, or aged and young tissues at 3 weeks after implant insertion (all p > 0.05) (Figure 2E). However, significant increases in immuno-labeling levels for TGF-β₁ sequestered within the healing matrices of young diabetic tissues at 9 weeks after implant insertion were observed, compared to the TGF-β₁ immuno-labeling counts in aged and non-diabetic tissues (all p < 0.01) (Figure 2E).

The present study used CB-MSCs, where long-term expansion in normoglycemic conditions has previously been characterized to indicate that isolated cell populations, expanded to PD15 in culture in a normoglycemic basal media, contain a high population of low colony-forming, pre-osteoblast cells that contribute to the bone lining cell population (44). However, following expansion beyond PD50, the heterogeneous cell population becomes dominated by high colony-forming, multipotent cells (44). Within the present study, long-term cultures in normoglycemic (5.5 mM) and hyperglycemic (25 mM) conditions were demonstrated to have a limited impact on CB-MSC expansion capabilities over 350 days in culture. CB-MSCs expanded in normoglycemic conditions took 156 days to reach PD150, while in hyperglycemic conditions, CB-MSCs took a further 12 days (178 days). Analyses were performed on CB-MSCs at PD15 and PD150 to allow some comparison with in vivo findings, as CB-MSCs in the GK rats were exposed to a hyperglycemic environment for 10–27 weeks (189 days).

Within the present study, we provided further data relating to the effects of high glucose (25 mM) supplementation on osteogenic differentiation and associated marker gene expression by CB-MSCs (Figure 3). When considering the effects of short-term exposure to high glucose concentrations on CB-MSCs originally cultured to PD15 under normoglycemic conditions, subsequent culture in osteogenic media supplemented with 25 mM glucose resulted in significant increases in the expression of Sp7/osterix (Figure 3A), along with lesser, yet significant increases in gene expressions of osteopontin (Figure 3B) and osteocalcin (Figure 3C), which correlated with a moderate increase in Alizarin red staining (Figure 3D). For CB-MSCs cultured to PD150 under normoglycemic conditions, subsequent culture in osteogenic media supplemented with 25 mM glucose resulted in significant decreases in Sp7/osterix (Figure 3A) and osteopontin (Figure 3B) but an increase in osteocalcin (Figure 3C). There appeared to be little further effect on mineral nodule formation (Figure 3D), which was already limited due to long-term culture expansion, leading to a predominance of transit-amplifying cells that would be expected to take longer to differentiate in vitro. Similar results were observed for CB-MSC cultures expanded in hyperglycemic conditions, where continued culture of these cells in hyperglycemic osteogenic induction media caused a decrease in Sp7/osterix (Figure 3A) and osteopontin (Figure 3B) and a slight delay in osteocalcin gene expression (Figure 3C) when compared to CB-MSCs culture expanded in hyperglycemic but then cultured in 5.5 mM normoglycemic conditions during osteogenic induction. Likewise, the inclusion of high glucose during osteogenic induction of PD150 expanded in hyperglycemic conditions elicited little further effect on nodule formation, which was low due to culture expansion.

Using the CB-MSC populations that had initially been expanded to PD15 or PD150 in either normoglycemic or hyperglycemic basal media, the subsequent effects of high glucose concentrations on TGF-β₁ during osteogenic induction were analyzed, examining levels localized intracellularly, sequestered within the ECM, or released into the culture medium (Figure 4). When considering the bioavailability and the signaling activity of TGF-β₁, it is important not only to assess the levels of growth factor but also to identify the different isoforms. Western blot analysis facilitated the detection of TGF-β₁ in several processed isoforms: the ∼55 kDa pre-pro-TGF-β₁ monomers; the ∼45 kDa precursor protein; and, when attached to its latency associated protein (LAP), the ∼110 kDa TGF-β₁-LAP homodimers; and the 25 kDa active TGF-β₁ dimer (19). For most blots, the inactive ∼55 kDa pre-pro-TGF-β₁ monomers and the ∼45 kDa precursor protein predominated. In order to identify the 25 kDa active dimer and the TGF-β₁ form attached to LAP, it was necessary to load gels with a high level of protein (20 µg). This did have the consequence of producing blots where protein bands with high intensity could not be measured by densitometry due to their supersaturated pixel density, which would produce false results. Further, due to the high number of samples analyzed, the detection of TGF-β₁ was necessarily performed on a separate gel and, although protein-loading levels were equal for each gel, this hampered any direct quantification when comparing groups. However, a comparison of the respective blots still presents clear observations for monitoring changes in the levels of the various iosforms. The ∼55 kDa pre-pro-TGF-β₁ monomers and the ∼45 kDa precursor protein were detected in the cell lysate, incorporated into the ECM, and released into the culture media by CB-MSCs during the osteoinduction of CB-MSCs expanded to PD15 in normoglycemic basal media (Figure 4A). The addition of 25 mM glucose to the osteoinduction media had little effect on the levels of the TGF-β₁ monomer/precursor protein, but a substantial increase in the level of the 110 kDa TGF-β₁-LAP homodimers was observed in the protein isolated from the cell extract and that released into the culture media (Figure 4A).

Long-term expansion of CB-MSCs to PD150 in normoglycemic conditions indicated that the TGF-β₁-processed isoforms, synthesized following osteogenic differentiation under normoglycemic conditions, were generally comparable to CB-MSCs at PD15, with the immuno-detection of the TGF-β₁ monomer (∼45–55 kDa) and TGF-β₁-LAP homodimers at ∼110 kDa within the cell lysates, the associated ECMs, and the culture media. When these normoglycemic PD150 CB-MSCs were driven down an osteogenic lineage under high glucose conditions, few changes in the TGF-β₁ isoforms forms were noted (Figure 4A). However, long-term CB-MSC expansion under hyperglycemic conditions to PD150 led to significant reductions in detectable TGF-β₁ overall, with the notable loss of detectable TGF-β₁ at ∼110 kDa in the cell and ECM extracts and overall loss of the active TGF-β₁ bands at ∼25 kDa (Figure 4A). Such responses were further exacerbated when these CB-MSCs were cultured in hyperglycemic (25 mM glucose) osteoinductive medium, with further reductions in cell-, ECM-, and culture medium-associated TGF-β₁ immuno-reactivity observed (Figure 4A).

To corroborate our observed results, the gene expression for TGF-β₁ was also examined using qPCR (see Supplementary Material). The results, however, suggested overall increases in TGF-β₁ gene expression after the long-term expansion of CB-MSCs in normoglycemic basal culture media to PD150, and smaller increases in TGF-β₁ were observed for CB-MSCs expanded to PD150 in hyperglycemic media. Thus, although increases in TGF-β₁ gene expression were observed, the subsequent translation and synthesis of TGF-β₁ proteins did not appear to be evident. In considering our research aim for investigating the bioavailability and bioactivity of TGF-β₁, the focus was therefore placed on the results obtained for the analysis of protein levels within the extracellular matrix and culture media.

To assess whether the signaling potential of TGF-β₁ was disrupted in CB-MSCs subjected to such contrasting levels of hyperglycemic exposure, we also evaluated the impact on total Smad2 and Smad3 expression levels in considering their roles as key regulators of TGF-β₁ signaling (50, 51). Both Smad proteins were detectable in the cell lysates obtained from CB-MSCs (Figure 4B), but long-term CB-MSC expansion, whether in normo- or hyperglycemic conditions, was observed to have minimal effects on total Smad2/3 levels.

Protein extracts from the CB-MSC populations analyzed above were also assessed to identify the effects of high glucose (25 mM) on the protein levels of decorin and biglycan, as potential binding partners for TGF-β₁ (Figure 5). Western blot gels were loaded with the same amount of protein (20 µg) to allow for comparison with the TGF-β₁ levels reported above and to identify processed products that would influence their ligand binding properties. As discussed above, this prevented a quantitative analysis but did provide observational trends in the data. Immuno-detection for biglycan produced strongly staining bands at ∼45 kDa, which correspond to its core protein, while immuno-detection for decorin synthesized and secreted by the CB-MSC populations revealed strongly staining bands at ∼45 kDa and ∼120 kDa, corresponding to the molecular weights of the decorin core protein and decorin conjugated to its glycosaminoglycan chain, respectively (52).

When analyzing CB-MSCs at PD15, supplementation of the osteogenic induction media with 25 mM glucose led to a moderate increase in biglycan immuno-detection within the cell lysate and the ECM extract (Figure 5A). High levels of biglycan were released into the culture media by PD15 CB-MSCs; however, there were no discernible differences in the levels observed when comparing cells cultured in normo- or hyperglycemic osteogenic induction media (Figure 5A). For these same cells, the presence of high glucose in the osteogenic induction media resulted in considerable increases in decorin extracted from the cell lysate and the culture media, but few differences were identifiable in the levels incorporated into the ECM extract (Figure 5B).

The long-term expansion of CB-MSCs to PD150 in either normoglycemic or hyperglycemic culture conditions resulted in subtle decreases in the levels of biglycan core protein in the cell culture media (Figure 5A). However, the long-term expansion of CB-MSCs in normoglycemic conditions indicated a moderately increased level of biglycan detected in the cell lysate of CB-MSCs after culture expansion in normoglycemic conditions, but reduced levels of biglycan in cell lysates were observed after culture expansion in hyperglycemic culture media (when compared to PD150 in normoglycemic conditions). Low levels of biglycan were incorporated into the ECM matrix whether expanded in normo- or hyperglycemic conditions.

The immuno-detection of decorin after long-term culture expansion presented different results (Figure 5B). Culture expansion of CB-MSCs to PD150 in normoglycemic conditions was associated with moderately increased levels of decorin isolated from the cell lysate, ECM, and culture media. Supplementation of the osteoinductive media with 25 mM glucose increased the levels of decorin detected in the culture media and cell lysate further. When CB-MSCs were cultures expanded to PD150 in hyperglycemic conditions, similar increases in decorin were noted in the culture media and cell lysate, but little decorin was incorporated into the ECM. The inclusion of high glucose in the osteoinductive media had little further discernable influence on decorin levels contained within any of the protein extracts from the cell, ECM, or released into the culture media (Figure 5B).

These cells were also analyzed using qPCR to examine the effects of high glucose on the gene expression of decorin and biglycan by the cells after culture expansion and the presence of high or normal glucose levels (Supplementary Material). These results partially corroborated with the raised protein levels for decorin and biglycan reported above.

For both macrophage phenotype induction conditions in basal (5.5 mM) glucose concentrations, full macrophage differentiation was achieved after 7 days, as judged by the development of a heterogeneous population containing round or oval cell morphologies (Figure 6A). The effect of high glucose conditions on the formation of the M1 or M2 phenotype was subsequently examined by the quantification of established M1 and M2 macrophage marker genes using qPCR analysis (44, 45). In accordance with these prior published data, M1 macrophage formation in GM-CSF-stimulated cultures exhibited significant increases in the gene expression of pro-inflammatory cytokines, TNF-α (p < 0.001) and IL-6 (p < 0.01), in addition to inducible nitric oxide synthase (iNOS; p < 0.05) expression, compared to isolated monocytes at day 0 (Figure 6C). However, under hyperglycemic conditions, significant decreases in IL-6 (p < 0.01) were observed, although TNF-α and iNOS expressions were unaffected (p > 0.05). In contrast, after M-CSF/IL-4 induction, M2 macrophages displayed an increased expression of classical markers TGF-β₁ (p < 0.001), arginase 1 (Arg1; p < 0.01), vascular endothelial growth factor (VEGF; p < 0.001), and CD163 (p < 0.01) compared to isolated monocytes at day 0 (Figure 6C). When M2 induction was performed in high (25 mM) glucose conditions, the expression of each of these markers was significantly downregulated (p < 0.001 for TGF-β₁, Arg1, and VEGF; p < 0.01 for CD163), suggesting the significant attenuation of M2 macrophage formation by hyperglycemic environments.

The quantification of TGF-β₁ released into the culture media by differentiating M1 and M2 macrophages by ELISA demonstrated that secreted TGF-β₁ levels rose significantly for developing M2 macrophages by day 5 in culture (Figure 6B). The presence of high glucose (25 mM) in the M2 inductive media resulted in a reduction in TGF-β₁ secretion to basal levels that were consistent with secretions from M1 developing macrophages. It is noted that these findings on the secreted levels of TGF-β₁ were corroborated by the significant decreases in TGF-β₁ gene expression (Figure 6C).