Numerous animal and human studies have demonstrated that mechanical strain is an important determinant of skeletal architecture and bone mass, although the precise mechanotransductory events that direct this biological phenomenon are unclear. Like neurones, bone cells express complex glutamate signalling machinery including functional ionotropic and metabotropic receptors that share comparable basic pharmacological and electrophysiological properties to their neuronal counterparts [1-7]. Osteoblasts also spontaneously release glutamate and express functional glutamate transporters, providing evidence for endogenous receptor activation and agonist recycling in the bone microenvironment [8-10]. Together these data support the existence of a complex, highly orchestrated and functionally significant multicellular glutamate-signalling network in bone. But what role does glutamate play in musculoskeletal physiology and bone cell function? One possibility is the regulation of cell differentiation. Glutamate receptor antagonism has been reported to inhibit differentiation of osteoblasts and osteoclasts in vitro, although the in vivo significance of these findings remains unclear [4,11].

Over recent years the excitatory amino-acid glutamate has emerged as an important molecular determinant of plasticity and synaptic enhancement in the central nervous system responsible for the processes of learning and memory – the best studied of which is NMDA receptor-dependent long-term potentiation (LTP). LTP can be crudely defined as a long-lasting activity dependent change in synaptic efficacy in which synaptic strength is persistently increased in response to brief periods of repetitive electrical stimulation lasting only seconds. At a molecular level, LTP is initiated by NMDA receptor activation and calcium influx, which mediates increased delivery to, and/or conductance of existing AMPA-type glutamate receptors at the postsynaptic membrane. Further episodes of signalling are subsequently enhanced as a result of increased probability of receptor activation and glutamate mediated synaptic conductivity. NMDA receptor mediated calcium influx and activation of CaMKII plays a key role in the expression of LTP [12,13]. Inhibition or knockout of CaMKII has been reported to prevent LTP induction [14-16] while constitutively activated CaMKII increases synaptic strength [17,18]. Studies have demonstrated that CaMKII activation occurs immediately after induction of LTP, increasing single channel conductance of AMPA receptors through serine phosphorylation of GluR1 and mediating delivery and clustering of new AMPA receptors at synapses through the regulation of PDZ interactions between GluR1, Stargazin protein and PSD-95 [19-22].

Evidence exists for conventional memory formation in bone in which responses to mechanical loading are affected by preceding episodes of loading, at least fulfilling the conceptual requirement for LTP. For instance, it has been demonstrated that mechanical loading is more osteogenic when separated into discrete episodes, compared to an equivalent period of uninterrupted loading. Using a well-characterised model of loading induced bone formation in vivo, several studies have demonstrated that the osteogenic potential of mechanical loading is enhanced in terms of bone formation and biomechanical and structural properties, by separation of loading cycles into discrete episodes with varying degrees of rest between periods of stimulation [29-31]. In these studies, the order of osteogenic potency was such that 6 periods of 60 cycles > 4 periods of 90 cycles > 360 cycles at once. These data were interpreted as bone cells becoming increasingly less responsive or 'deaf' to prolonged periods of mechanical stimulation, a form of depression or desensitisation to the stimulus. Based on a hypothesis of LTP in bone described here, in contrast to losing responsiveness, bone cells could be considered primed by earlier periods of loading, so that further episodes are increasingly efficacious and subsequent responses potentiated. This would account for the extrapolatory deduction that less total loading cycles should be required to produce the equivalent level of osteogenic response if loading is separated into discrete bouts compared to a single sustained, and less effective period of loading. Moreover, other studies have reported that the osteogenic response of mechanical strain increases with loading frequency [32], which has clear similarities to LTP in central neurones, which is more effectively induced by high frequency stimulation. Lower frequency stimulation on the other hand is associated with short-term potentiation (STP) or long-term depression (LTD). Perhaps most significantly, LTP in bone would also account for the observation that levels of osteogenic stimulus saturate rapidly so that only a few loading cycles are required to achieve the maximum osteogenic effect [33]. In support of LTP-like mechanisms mediating the osteogenic response of bone cells, key events involved in LTP are observed in osteoblasts following mechanical stimulation in vitro, including specific and rapid initiation of calcium transients and activation of neuronal CaMKII. Furthermore, CaMKII activity appears to be required for normal osteoblast differentiation, which temporally precedes new bone formation post-loading [34,35]. In vivo evidence also exists for the involvement of an LTP-like mechanism in loading induced osteogenesis; potent inhibition of AMPA/Kainate receptor activation with DNQX causes a significant decrease in loading induced bone formation in the 4-point bending rat ulnae model [36]. This could be interpreted as functional antagonism of potentiated AMPA receptor responses induced by LTP-like mechanotransduction in bone.

Our hypothesis could be tested using a combined in vitro electrophysiological, and in vivo approach. One would expect electrophysiological responses of bone cells to glutamate receptor agonists to be potentiated following mechanical strain in vitro coincident with an increase in AMPA receptor number and/or conductance. The signalling pathways responsible for mediating these effects could be investigated using known agonists and antagonists of neuronal LTP. Similarly pharmacological intervention could be used to investigate the in vivo significance of LTP in which antagonists of NMDA receptors and CaMKII should diminish the potentiated osteogenic responses of segregated loading.

Naturally this hypothesis raises the question as to why bone cells would require such an exquisitely sensitive and rapid signalling system as fast excitatory neurotransmission. The millisecond time-scale through which these events are mediated may be required to detect and respond to the very rapid strain changes experienced during forms of exercise with high osteogenic potential. Perhaps the most significant implication of this would be the ability to pharmacologically prime the skeleton to retrieve or potentiate the osteogenic capacity of mechanical strain, an effect desirable for the treatment of bone disorders associated with low bone mass and architectural abnormalities such as osteoporosis.

AMPA; α-amino-3-hydroxy-5-methyl-isoxazole propionate, CaMKII; Calcium/calmodulin-dependent protein kinase II, LTP; Long-term potentiation, NMDA; N-methyl-D-aspartate.

GJS and PGG discussed this hypothesis, GJS drafted and PGG critically reviewed the paper. Both authors read and approved the final manuscript.