In an editorial published in May 2006, Loeser discussed how excessive forces in combination with aging and damaged cartilage can promote inflammatory changes and tip the delicate balance between anabolism and catabolism in the chondrocyte toward catabolism (1). In this issue of Arthritis & Rheumatism, Dossumbekova et al (2) report findings that offer a possible explanation of the mechanism of a positive role of moderate exercise in counteracting inflammatory cytokines in osteoarthritic (OA) joints.
Mild to moderate exercise is necessary for maintaining all structures of an articulating joint and maintaining mobility. There is ample clinical evidence that structured muscle strengthening and joint mobility exercises improve symptoms and function in patients with OA, with the caveat that care should be taken in exercising joints with injuries and those with bad mechanics. The OA Systematic International Review and Synthesis group (3) chose 72 studies that investigated the effects of activities of daily life, exercise, sports, and occupational activities in patients with knee or hip OA and met standardized epidemiologic criteria. The group developed recommendations based on those studies and ranked the recommendations according to the level of scientific evidence available, with A indicating a high level of scientific evidence, B indicating a moderate level, and C indicating clinical consensus.
The group found that healthy subjects and OA patients can perform a high level of physical activity, as long as the activity is not painful and does not predispose to trauma (grade B). Physical activity is not contraindicated in sedentary patients with radiographic or clinical OA (grade C). There is a high level of scientific evidence indicating that exercise and other structured activities ameliorate pain and improve function in sedentary patients with OA of the knee (grade A). The criteria for choice of exercise include availability, preference, and tolerance, and static exercises are not favored over dynamic exercises (grade A). Because results deteriorate when activity is stopped, exercise should be undertaken 1–3 times per week (grade B) (3).
Cells within all joint tissue (bone, ligament, meniscus, joint capsule, synovium, and articular cartilage) respond to biomechanical stimuli and require a certain minimum level of continuous biomechanical stimulation for proper development and maintenance of the tissue. These levels may change with maturation and aging as well as during the development and progression of OA. This is illustrated by an in vivo model in which joint immobilization without compression or distraction with motion results in an ∼40% loss of proteoglycan, primarily aggrecan, in the weight-bearing areas of articular cartilage. Proteoglycan content is largely restored, however, after careful remobilization of the joint (4). In addition, in middle-aged women the frequency of relatively moderate exercise (20 minutes with tachypnea and increased pulse) was correlated with an increase in the volume of medial tibial cartilage, as measured by magnetic resonance imaging (5). Thus, joint activity within normal range is a factor in maintaining articular cartilage. Few precise details are known about how this response changes with age and during the development and progression of OA or about the impact of inflammation.
With regard to articular cartilage, in vitro studies of the exact mechanisms of translation of a biomechanical signal(s) into possible pathways from the cell's plasma membrane to the nucleus and back are under way, but are less advanced than studies of many other cell types, such as osteoblasts, fibroblasts, and endothelial cells. Chondrocytes in articular cartilage sense and respond to mechanical stimuli through many pathways, including downstream signaling pathways and mechanisms that may lead to direct changes at the level of transcription or translation, posttranslational modifications, and cell-mediated extracellular assembly and degradation of the tissue matrix (6).
How biomechanical signals are translated to the level of the cell is best illustrated in explants and cell cultures in which the biomechanical and biochemical environment can be controlled. These culture conditions should also assure that chondrocytes are surrounded by a well-organized pericellular matrix (chondron) with a different composition than the interterritorial matrix and can modulate the signal transmission to the cell and provide those matrix proteins and glycosaminoglycans that are in direct contact with cell receptors (7).
The possible forces that act on chondrocytes are currently best understood when tissue is compressed or in shear, since cartilage experiences either static compression, cyclic compression, cyclic shear, or a combination of the latter two. Even in simple compression chondrocytes are subjected to a complex local mechanical environment consisting of tension, compression, shear, and fluid pressure and are exposed to a number of potential biochemical signals, including hydrostatic pressure, streaming potential, and fluid flow, and physiochemical changes, including matrix, cell, and membrane deformation and volume changes (6). This is especially true for surface articular chondrocytes where there is significant lateral expansion of the cell in the direction perpendicular to the cartilage surface split-line (7). In shear without water loss, cells are largely deformed along with the matrix, but with water loss there is also a change in hydrostatic pressure, streaming potential, and fluid flow (6, 8). It has been suggested that transducers for translating forces into cellular signals include specific matrix receptors such as integrins, specifically α5β1, annexin V, stretch receptors, and voltage-gated calcium and potassium channels, as well as cell volume change and membrane deformation (6).
In cartilage slices chondrocytes have a biphasic response to cyclic compression. At low levels of compression and relatively slow cycling there is early up-regulation of a cluster of genes, including several matrix proteins (8). Up-regulation of proteoglycan synthesis by cyclic compression involves signals from Ca+2 fluxes and cyclic AMP (9) as well as extracellular ATP and adenylate receptor (10). At high levels of cyclic compression or static compression, the signals become inhibitory. In shear, low levels of oscillatory strain (1–5% dynamic strain amplitude at frequencies ranging from 0.10 to 1.0 Hz) can stimulate synthesis of proteoglycan, collagen, and other matrix proteins, and messenger RNA (mRNA) for many of the same proteins is up-regulated, as in cyclic compression. Higher levels of strain and/or strain rates can cause cell death and matrix damage (8).
Inflammation is increasingly recognized as an element of OA and a contributor to the progression of OA. Tumor necrosis factor α (TNFα), interleukin-1α (IL-1α), and IL-1β all cause a concentration-dependent decrease in many matrix proteins, such as aggrecan, at the mRNA level. At higher levels, TNFα, IL-1α, and IL-1β also up-regulate catabolic proteins, including proteinases such as matrix metalloproteinase 1 (MMP-1), MMP-2, and MMP-13. Similarly, advanced glycation end products (AGEs), which are formed by the nonenzymatic reaction of glucose with side chains of lysine or arginine in proteins, and which increase with age, can signal through the receptors for AGEs to add to the proinflammatory environment. One of the responses is a shift of some chondrocytes toward the hypertrophic pathway (11). Signals from AGEs, TNFα, and IL-1 originate through different receptors but eventually propagate signals in large measure through the canonical NF-κB pathway.
NF-κB activation can occur in several ways, including via canonical, noncanonical, and atypical pathways (Figure 1). The canonical NF-κB pathway in chondrocytes has been well characterized. In the canonical pathway, there is rapid activation and nuclear localization of cytoplasmic NF-κB that is released after degradation of its inhibitor IκBα as a result of phosphorylation by the activated IKK complex (Figure 1). In this pathway, the IKK complex consists mainly of the 2 catalytic kinase subunits (IKKα and IKKβ) and multiple copies of IKKγ (NF-κB–essential modulator [NEMO]). NEMO is required for integration of signals from upstream kinases and receptor-associated proteins. The canonical pathway is activated in response to inflammatory cytokines, such as IL-1α, IL-1β, and TNFα (12).
First, the components of the IKK complex are further modified, including ubiquitination and phosphorylation of IKKγ and 2 serine hydroxyls in the activation loop of IKKβ. IKKβ then phosphorylates IκBα at serines 32 and 36. IκBα is now a substrate for the ubiquitin ligase complex Skp1/Cul1/F-box protein β-TrCP. After ubiquitination IκBα is rapidly degraded by the proteasome, and the released NF-κB is relocated to the nucleus. The most common NF-κB dimer (p50-RelA) is transported to the nucleus in minutes (12). For full activation of the dimer, frequent modification of the subunits, including acetylation and phosphorylation, occurs. Prolyl isomerization of RelA may also take place (12).
IKK is activated by the kinase associated with the particular activated cytokine receptor. The whole activated system is in a continuous state of flux, in which NF-κB has a lifespan of 30 minutes and is rapidly down-regulated by a number of mechanisms. One such mechanism occurs when newly synthesized IκBα, the synthesis of which is up-regulated by NF-κB, enters the nucleus, binds NF-κB, and returns it to the cytoplasm. Though NF-κB does not regulate the transcription of IκBβ or IκB∈, they appear to function through a similar mechanism of IKK-dependent phosphorylation and degradation, and these proteins attenuate the oscillations of NF-κB activation seen with IκBα alone (12). Other feedback loops also exist (12).
Using their somewhat unconventional chondrocyte culture system of well-differentiated primary rat chondrocytes grown on an elastic membrane coated with fibronectin, which can be subjected to equibiaxial tensile stresses, Agarwal and colleagues (13) (see Dossumbekova et al  for additional references) have defined 2 regions of chondrocyte response to cyclic tensile strain. At low levels of equibiaxial elongation (3%) and low cycle rates the chondrocytes responded by up-regulation of matrix synthesis, while greater elongation led to expression of catabolic enzymes and extensive cell death.
Interestingly, a similar response has been found with compression both in explants (8) and in alginate constructs (10). While tension is not typically experienced by normal cartilage chondrocytes, these cells may experience tension in their local environment, and tension may develop in OA cartilage during joint motion when the surface is damaged or the cartilage surface loses proteoglycan. Therefore, the findings reported by Dossumbekova et al are very informative and add insight into the mechanism by which cyclic tensile strain inhibits the NF-κB–mediated response to IL-1β (12), an inhibition that has also been observed in chondrocytes in compression in alginate (14).
Dossumbekova and colleagues traced the effect of cyclic tensile strain on the NF-κB signaling pathway to several points of intervention. Consistent with their observation that cyclic tensile strain markedly inhibited NF-κB nuclear translocation, the authors report that cyclic tensile strain substantially suppressed IL-1β–dependent IKK activation, inhibited IκBα and IκBβ transcription, and down-regulated NF-κB binding to DNA. Because IκBα, but not IκBβ, expression is controlled by NF-κB, and the pathways that control IκBβ synthesis after its degradation are incompletely characterized, the main conclusion is that cyclic tensile strain abrogates IL-1β–dependent activation of IKK.
Furthermore, the signals generated by cyclic tensile strain–regulated proinflammatory gene expression involve proteins other than those controlled by NF-κB. Because the IκBα promoter is under the control of NF-κB, it is not surprising that inhibition of NF-κB nuclear translocation leads to suppression of IκBα mRNA expression and synthesis. NF-κB directly regulates several genes that are essential for the regulation of its own signaling cascade. Thus, cyclic tensile strain, by repressing IKK activity, inhibits transcriptional up-regulation of such genes as IL-1 receptor 1 (IL-1R1), IL-1R2, TNF receptor–associated factor 1 (TRAF1), and TRAF2, which are required for IKK activation. This leads to multidimensional inhibition of IL-1β activation, and presumably TNFα activation, in chondrocytes.
Dossumbekova et al suggest that signals from tensile tension are transferred into the cell via integrins through stress fibers and focal adhesion kinase, but this notion is based on inhibition of the biosynthetic response by peptides that inhibit α5β1 integrins (8) and so should be investigated in more detail. How cyclic tensile strain signals inhibit IKK activation at the level of transcription, and why this inhibition can persist for hours after the end of cyclic tensile strain, is still unexplained. Understanding the mechanisms of action of cyclic tensile strain in inhibiting NF-κB could provide new insights into the responses of other cell types to cyclic compression and into whether the findings are applicable to OA chondrocytes, which may express a different set of genes. Such information could also help identify antiinflammatory drug targets other than NF-κB and IKK. Meanwhile, keep on walking!