Fracture healing and lipid mediators

Abstract

Lipid mediators regulate bone regeneration during fracture healing. Prostaglandins and leukotrienes are well-known lipid mediators that regulate inflammation and are synthesized from the Ω-6 fatty acid, arachidonic acid. Cyclooxygenase (COX-1 or COX-2) and 5-lipoxygenase (5-LO) catalyze the initial enzymatic steps in the synthesis of prostaglandins and leukotrienes, respectively. Inhibition or genetic ablation of COX-2 activity impairs fracture healing in animal models. Genetic ablation of COX-1 does not affect the fracture callus strength in mice, suggesting that COX-2 activity is primarily responsible for regulating fracture healing. Inhibition of cyclooxygenase activity with nonsteroidal anti-inflammatory drugs (NSAIDs) is performed clinically to reduce heterotopic ossification, although clinical evidence that NSAID treatment impairs fracture healing remains controversial. In contrast, inhibition or genetic ablation of 5-LO activity accelerates fracture healing in animal models. Even though prostaglandins and leukotrienes regulate inflammation, loss of COX-2 or 5-LO activity appears to primarily affect chondrogenesis during fracture healing. Prostaglandin or prostaglandin analog treatment, prostaglandin-specific synthase inhibition and prostaglandin or leukotriene receptor antagonism also affect callus chondrogenesis. Unlike the Ω-6-derived lipid mediators, lipid mediators derived from Ω-3 fatty acids, such as resolvin E1 (RvE1), have anti-inflammatory activity. In vivo, RvE1 can inhibit osteoclastogenesis and limit bone resorption. Although Ω-6 and Ω-3 lipid mediators have clear-cut effects on inflammation, the role of these lipid mediators in bone regeneration is more complex, with apparent effects on callus chondrogenesis and bone remodeling.

Figure 1

Pathways for synthesis and signaling of arachidonic acid, docosahexaenoic acid and eicosapentaenoic acid-derived lipid mediators. The diagram summarizes some of the synthetic pathways for lipid mediator synthesis. Enzyme activities are shown in red text and include cyclooxygenase (COX), 5-lipoxygenase (5-LO or 5-LO·FLAP), lipoxygenase, cytochrome P450, LTA4 hydrolase, LTC4 synthase, 5-hydroxy-eicosanoid dehydrogenase and COX-2 covalently modified by acetylsalicylic acid. Receptors are shown in green text. Lipid mediators and synthetic intermediates are shown in black text and include resolvins, protectin D1, maresin-1, prostaglandins, leukotrienes, lipoxins, 5-oxo-eicosatetraenoic acid, 20-hydroxy-eicosatetraenoic acid, various epoxyeicosatrienoic acids and others. Known effects of receptor activation on intracellular cyclic adenosine monophosphate or calcium levels are shown at the bottom.

Figure 2

Histomorphometric analysis of fracture healing in COX-deficient mice (Ptgs1^tm1Unc and Ptgs2^tm1Unc). Using plastic-embedded,calcified sections, femur fracture callus area (a), callus percent cartilage area (b) and callus percent mineralized tissue area (c) were measured and calculated at 7, 10, 14 and 21 days after fracture. Shown are mean values and standard errors (n≥3 for all time points and genotypes). Callus area, percent cartilage area and percent mineralized tissue area were reduced in COX-2-null mice as expected. However, no significant differences were detected between COX-1-null, wild-type or mice heterozygous for the COX-2-null allele.

Figure 3

Mechanical testing analysis of fractured femurs from COX-deficient mice. Fractured femurs and contralateral control femurs were collected from COX-1-null, COX-2-null, wild-type and mice heterozygous for the COX-2-null allele after 4 weeks of healing. The femurs were analyzed for structural (peak torque and maximum rigidity, a and b) and material properties (maximum shear stress and shear modulus, c and d) by torsional mechanical testing to failure. Each fractured femur value was normalized to its contralateral femur value as a percentage and the means with standard errors for each genotype are shown (n≥9). For the COX-2-deficient mice, all values were significantly less than the values from other genotypes (P<0.05). However, no significant differences were detected between the COX-1-null mice and controls.