Hypoxia mimetics restore bone biomineralisation in hyperglycaemic environments

Abstract

Diabetic patients have an increased risk of fracture and an increased occurrence of impaired fracture healing. Diabetic and hyperglycaemic conditions have been shown to impair the cellular response to hypoxia, via an inhibited hypoxia inducible factor (HIF)-1α pathway. We investigated, using an in vitro hyperglycaemia bone tissue engineering model (and a multidisciplinary bone characterisation approach), the differing effects of glucose levels, hypoxia and chemicals known to stabilise HIF-1α (CoCl₂ and DMOG) on bone formation. Hypoxia (1% O₂) inhibited bone nodule formation and resulted in discrete biomineralisation as opposed to the mineralised extracellular collagen fibres found in normoxia (20% O₂). Unlike hypoxia, the use of hypoxia mimetics did not prevent nodule formation in normal glucose level. Hyperglycaemic conditions (25 mM and 50 mM glucose) inhibited biomineralisation. Interestingly, both hypoxia mimetics (CoCl₂ and DMOG) partly restored hyperglycaemia inhibited bone nodule formation. These results highlight the difference in osteoblast responses between hypoxia mimetics and actual hypoxia and suggests a role of HIF-1α stabilisation in bone biomineralisation that extends that of promoting neovascularisation, or other system effects associated with hypoxia and bone regeneration in vivo. This study demonstrates that targeting the HIF pathway may represent a promising strategy for bone regeneration in diabetic patients.

Figure 1

Hypoxia (1% O₂) and hyperglycaemia inhibited bone nodule formation. After 21 days, Alizarin Red stained (ARS) dense nodules were observed in (a) normal (5.5 mM) glucose, whilst (b) moderate (25 mM) and (c) high (50 mM) glucose conditions inhibited nodule formation. (d) Hypoxia normal glucose showed discrete biomineralisation that was not associated with collagen fibres. (e) Moderate and (f) high glucose inhibited biomineralisation. Transmission electron microscopy (TEM) micrographs of (g) normoxia normal glucose, (h) moderate glucose and (i) high glucose showed a glucose concentration dependent inhibition of bone nodule formation. COL fibres were not observed in (j) hypoxia normal glucose, but (k) moderate and high glucose environments in hypoxia showed some collagen fibres. Scale bar for (a–f) is 200 µm and for (g–i) is 2 µm. (n = 5) (N: nodule, COL: collagen fibres, OB: osteoblast).

Figure 2

Inhibition of bone nodule formation in hypoxia and hyperglycaemia quantified by (a) interferometry (area above 30 µm) and (b) image analysis of Alizarin Red staining (total area). Nodules cultured in normoxia normal (5.5 mM) glucose covered a substantially larger area (a and b total area) than both normoxia moderate (25 mM) and high (50 mM) glucose. (c) ALP activity per unit protein revealed that all normoxic conditions had higher ALP production than hypoxic conditions. Error bars represent the SD from the mean values. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001 (n = 4).

Figure 3

CoCl₂ and DMOG restored nodule formation in hyperglycaemic cultured osteoblasts. After 21 days, Alizarin Red staining (ARS) showed bone nodule formation in untreated cultures in (a) normal glucose (5.5 mM), whilst the addition of (b) moderate (25 mM) and (c) high glucose (50 mM) inhibited nodule formation. (d–o) Cultures treated with hypoxia mimetics CoCl₂ and DMOG decreased bone nodule formation in normal glucose levels but increased bone nodule formation in moderate and high glucose levels compared to untreated controls. Scale bar for all images is 200 µm. (n = 5).

Figure 4

The effects of HIF-1α mimetics and differing glucose environments on the ultrastructure of bone nodules. Transmission electron microscopy (TEM) of (a–c) moderate and high glucose levels showed reduced extracellular mineralised collagen fibres compared to normal glucose in normoxia. The HIF-1α mimetics (d,j) CoCl₂ and (j,m) DMOG, reduced extracellular mineralised collagen in normal (5 mM glucose conditions) but restored extracellular (bone-like) mineral in (e,k,h) moderate (25 mM) and (f,i,l) high (50 mM) glucose conditions compared to (a–c) untreated controls (with the exception of o) 500 mM DMOG where no extracellular mineralised collagen fibres were observed. Scale bar for all images is 200 µm. (n = 5) (N: nodule, COL: collagen fibres, OB: osteoblast).

Figure 5

Quantification of bone nodules restored by CoCl₂ and DMOG treatment in hyperglycaemic conditions using interferometry and ImageJ. (a) The percentage change (percentage change compared corresponding glucose levels in atmospheric O₂) in the surface covered with nodules (above 30 µm compared to untreated control (similar level of glucose) without CoCl₂ as quantified with interferometry and (b) 2D image analysis of area of surface covered by mineral. Both 12.5 µM CoCl₂ and 25 µM CoCl₂ recovered nodule formation in moderate (25 mM) and high (50 mM) environments. (c) DMOG interferometry and (d) Image J analysis also revealed that both 250 µM DMOG and 500 µM DMOG restored nodule formation, however, DMOG exhibited a reduced restoration capacity compared to CoCl₂. Error bars represent the SD from the mean values. *P ≤ 0.05; **P ≤ 0.01; ****P ≤ 0.0001 (n = 4).

Figure 6

Raman spectra of rat calvarial native bone and rat calvarial osteoblasts in normoxia, hypoxia and 12.5 µM CoCl₂ all in normal glucose conditions (5.5 mM). (a) Average Raman peaks associated with proteins (Amide III, CH₂, Amide I) are much reduced in hypoxia compared to normoxia, (b) with a much higher mineral to matrix ratio. The hypoxia mimetic CoCl₂ had a different effect than hypoxia and had a biochemical composition more similar to normal culture conditions and bone. **P ≤ 0.01; ***P ≤ 0.001*; ****P ≤ 0.0001 (the minimum number of bone nodules per treatment = 10).

Figure 7

HIF-1α stabilisation on day 3 and VEGF response on day 1 and 7 of osteoblast to hypoxia and hypoxic mimetics CoCl₂ and DMOG in hyperglycaemic conditions. (a) Hyperglycaemia impaired HIF pathway and (b) decreased VEGF expression. (c–f) Both CoCl₂ and DMOG restored HIF-1α stabilisation and VEGF expression in hyperglycaemic conditions (percentage change compared corresponding glucose levels in atmospheric O₂). Error bars represent the SD from the mean values. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001*; ****P ≤ 0.0001 (n = 4).