Extending the realm of membrane capacitance measurements to nerve terminals with complex morphologies

Almost thirty years ago, Neher & Marty (1982) showed that time-resolved membrane capacitance (C_m) measurements have the sensitivity to monitor single vesicle fusion (exocytosis) events and subsequent membrane retrieval (endocytosis) in living neuroendocrine cells. Almers & Neher (1987) then showed that while C_m measurements are exquisitely sensitive to exocytosis, they are not a very specific measure of secretion, since non-exocytotic processes like membrane pinocytosis and/or gating-charge movements of transmembrane proteins can produce C_m changes, and moreover, these changes may be insidiously Ca²⁺ dependent. Since those pioneering papers, C_m measurements have been extended to several different types of excitable and non-excitable secretory cells, and also to sensory neurons, like photoreceptors (Rieke & Schwartz, 1994; Bartoletti et al. 2010), and to epithelial cells, like hair cells (Parsons et al. 1994). Most of this work has employed quasi-spherical, isopotential and one-compartment secretory cells, which are ideally suited for this measurement technique. Conventional neurons, however, have nerve terminals with complex geometries and multiple bouton-type endings. The first paper to extend C_m measurements to nerve terminals in brain slices was by Hsu & Jackson (1996) using pituitary nerve varicosities on an unmyelinated axon (beads-on-a-string morphology). C_m measurements were then extended by Mennerick et al. (1997) to a more complex secretory cell: the intact goldfish retinal bipolar neuron, which contains multiple ribbon-type active zones in its large synaptic terminal. A two-compartmental RC circuit model was developed for bipolar cells with soma and terminal compartments connected by a thin cylindrical axonal resistance. It was shown that C_m measurements were possible with the patch pipette located in either the soma or terminal, provided one followed certain precautions to avoid potential artifacts. Capacitance measurements are also feasible in another giant and specialized nerve terminal with a large readily releasable pool of vesicles, the calyx of Held (Sun & Wu, 2001; Taschenberger et al. 2002). More recently, Hallermann et al. (2003) showed how to successfully perform C_m measurements at hippocampal mossy fibre bouton-type terminals (∼3 μm diameter), which contain conventional active zones. This seminal paper showed that a careful reconstruction of nerve terminal and axonal morphology, followed by a morphology-based equivalent electrical model of the nerve terminal and axon, was necessary to implement reliable C_m measurements in more conventional CNS nerve terminals.

Now a paper in The Journal of Physiology by Oltedal & Hartveit (2010) advances this steady progress in extending the realm of C_m measurements by showing that it is possible to record C_m changes directly from small mammalian rod bipolar cell synaptic terminals. The authors use rat rod bipolar cells that have ∼45-μm-long axons and two to three axon terminal boutons (a grapes-on-a-vine morphology; the size of the terminal bouton attached to the recording pipette is ∼2–4 μm). Their computer simulations suggest that in order to detect a C_m increase at the axon terminal with high reliability, one should either use soma-end recordings with low lock-in sine wave frequency (∼100 Hz), or the more technically challenging direct terminal-end recordings with high lock-in sine wave frequency (∼2 kHz). An effective way to minimize noise in C_m measurements is to increase the amplitude of the lock-in voltage-clamp sine wave (Lindau & Neher, 1988). But the higher the amplitude of the sine wave, the more likely it is to activate voltage-gated currents, and the resulting non-linearities may produce C_m artifacts. Interestingly, the authors find that a high-frequency sine wave (∼2 kHz) at a V_hold of −65 mV with an amplitude of ±30 mV (peak to peak) produces very little activation of voltage-gated Ca²⁺ channels. This type of lock-in stimulus may thus be used for reliable C_m measurements in retinal bipolar cells. Using this approach the authors find a high rate of exocytosis following the activation of Ca²⁺ current and strong short-term synaptic depression mediated by the rapid depletion of a relatively small vesicle pool. Glutamate release thus has a highly transient nature at this synapse following a strong stimulus. These results are in remarkable quantitative agreement with previous paired recordings between rat rod bipolar cells and amacrine cells (Singer & Diamond, 2003).

Nevertheless, rather contradictory results were also reported from acutely dissociated mouse rod bipolar cells (Zhou et al. 2006), which showed no statistical difference of C_m changes in soma-end and terminal-end recordings using a 1 kHz lock-in sine wave. In order to understand the apparent contradiction, the authors also built a model for the mouse rod bipolar cell based on its morphology (a relatively short and thick axon: 32 μm length, 1 μm diameter) and showed the feasibility to C_m measurements from soma-end recordings. Taken together, geometric considerations and a judicious choice of lock-in sine wave parameters should become necessary prerequisites for future C_m measurements in more complex nerve terminals.

Remarkably, in the same issue of The Journal of Physiology, Palmer (2010) also reports C_m measurements from bipolar cells, but couples these with paired recordings between bipolar cell terminals and ganglion cell layer neurons in goldfish retina. These tour-de-force and elegant experiments reveal that a presynaptic depolarizing step to the bipolar cell terminal evokes both AMPA/kainate receptor- and NMDA receptor-mediated excitatory postsynaptic currents (EPSCs) in connected ganglion cell layer neurons. A linear relationship between bipolar cell terminal exocytosis (C_m jumps) and total EPSC charge transfer indicates that glutamate receptor saturation and desensitization are not present in this ribbon-type synapse. Finally, the quantal content of the evoked EPSCs was estimated by the analysis of miniature EPSCs in ganglion cell layer neurons. The readily releasable pool of vesicles was composed of about 23 vesicles, which corresponds to the number of vesicles docked at the bottom row of the synaptic ribbon. The ganglion cell layer neuron is thus apparently able of sense the full complement of fusing synaptic vesicles in the dyad synapse formed by a single synaptic ribbon and two postsynaptic neurons.

In summary, the careful and thorough analysis of Oltedal & Hartveit (2010) provides us now with a detailed roadmap for future C_m measurements of nerve terminals with more complex morphologies, while the paired recordings of Palmer (2010) provide an independent confirmation that these C_m measurements are indeed bona fide proxies of transmitter secretion. So twenty years after a lecture at Stockholm where Erwin Neher challenged us to extend C_m measurements to nerve terminals by stating: ‘Unfortunately, nerve terminals are usually not accessible to the kind of biophysical investigations described here (i.e. C_m measurements)’, we are finally beginning to meet that challenge.