Spindles are doin’ it for themselves: Glutamatergic autoexcitation in muscle spindles

Abstract

Mention glutamate secretion to most physiologists, even to most neuroscientists, and they’ll think only of excitatory synaptic neurotransmission in the central nervous system. Although by far its best known function, evidence from the last 15 years shows it is also vital for controlling peripheral mechanosensory terminal responsiveness, via autogenic excitation. In an article appearing in a previous issue of The Journal of Physiology, investigating rat muscle spindles (Bewick et al. 2005), we proposed that glutamate was released from synaptic-like vesicles (SLVs), given the ubiquitous presence of SLVs in mechanosensory terminals, their constitutive but stretch-modulated recycling, the high levels of glutamate and vesicular glutamate transporter 1 (vGluT1), and that glutamate receptor exogenous ligands can alter stretch-evoked firing over a wide range, even abolish it altogether. In this issue of The Journal of Physiology, a study conducted by Katherine Wilkinson’s group in muscle spindles provides the first direct evidence that glutamate secretion is indeed vesicular (Than et al. 2021). Than et al. (2021) focus on manipulating vGluT1 activity and monitor responses of single afferents from muscle spindles of isolated mouse extensor digitorum longus (EDL) muscle during sustained (ramp and hold: static) and rapidly changing (vibration: dynamic) stretches. Exogenous glutamate increased firing, whereas reducing glutamate loading into SLVs by inhibition with xanthurenic acid (XA) or vGluT1 haploinsufficiency greatly reduced the ability of spindles to maintain firing during sustained stretch, often abolishing it. The dynamic response to rapid stretch or vibration was usually much less affected, although that too was occasionally blocked. It was concluded that the endogenous glutamate release essential for normal stretch sensitivity of the sensory endings, particularly maintained stretch, is vesicular. Although our experiments used whole-nerve electroneurograms and glutamate receptor pharmacology, their single unit recordings are spindle-specific and their focus on impairing vesicle filling provides the first direct evidence the autogenic excitation is via vesicular glutamate. Interestingly, Than et al. (2021) report substantial between-spindle variability in sensitivity to experimental manipulations. In some spindles, maintained stretch-evoked firing was largely silenced, whereas others were much less affected. In a few spindles, dynamic firing showed time-dependent inhibition, although this did not reach significance overall. This was greater variability than in our studies. Although it may reflect species or muscle differences (rat vs. mouse, deep lumbrical vs. EDL), for XA it may reflect the shorter (40 min) drug exposures. Than et al. (2021) suggest inconsistent drug access might underlie this variability. In our protocol, pharmacological agents were applied for at least 1 h to ensure maximal repeatability. Thus, both the overall variability and the relative insensitivity of dynamic responses to XA may reflect slow drug access to these deeply embedded, highly encapsulated organs. A less obvious, but probably as important, source of variability is the constitutive, stretch-modulated SLV recycling. Than et al. (2021) set baseline muscle length for the duration of the experiment to L0, at which it produces maximal contractile tension. This tonic stretch will raise baseline SLV recycling, elevating constitutive vesicular secretion and hence initial excitability. This may blunt responses to manipulations, even saturate some spindles, because even vesicles in haploinsufficient mice will have some loading. Combining these vesicular glutamate findings and their previous findings of a vital role for the non-selective cation channel Piezo2, Than et al. (2021) propose a linear model (Fig. 1A), in which lengthening triggers Piezo2-dependent Na+ and Ca2+ influx, possibly involving other channels. This produces both initial dynamic depolarization and the Ca2+-triggered secretion that directly initiates static firing. This raises two immediate areas for further study: if significant dynamic current flow occurs through Piezo2, there is the need to explain the receptor potential’s lack of a directly stretch-gated K+ component (Hunt et al., 1978), as well as how the glutamatergic activation of ‘static-stretch’ channels can be simultaneously prolonged (seconds during constant stretch) and rapidly stopped (cessation within milliseconds at onset of shortening). We recently proposed an alternative model where these same elements follow parallel but interdependent pathways with different timecourses (Bewick & Banks, 2021) (Fig. 1B and C). Here, Na+ and Ca2+ channels produce both the dynamic and static firing. Ca2+ influx through Piezo2 again enhances vesicular glutamate release but glutamate signalling is slow, setting the steady-state population of stretch-activated Na+ and Ca2+ channels, opposing their slow, continuous removal by constitutive SLV internalization. However, this also has issues: how are dynamic and static firing differentiated, and why is static firing more sensitive to glutamate manipulation? Thus, Than et al. (2021) have produced the most direct evidence it is vesicular glutamate release that autoregulates stretch-evoked firing in spindle primary mechanosensory terminals. These models highlight new experimental avenues for understanding what that glutamate signalling does. It will be intriguing to see which model, if either, most closely approximates to the actual mechanism that will eventually emerge. Watch this space.