To alkalinize or acidify, that is the question

Main text

Neurons are engineering marvels that respond to environmental stimuli with both stochastic and predictable actions. One stimulus that has been intensely studied over the years is extracellular pH. Extracellular pH, or more specifically synaptic pH, is predicted to decrease after neural activity because of the release of neurotransmitters that are packaged into acidic vesicles (1). Although this hypothesis is biologically sound, the experimental data to support it have been mixed.

For example, in the 1980s, studies in rats demonstrated that cerebellar neural activity produced alkalinization followed by an acidification in the bulk interstitial space (2). Several years later, a similar approach in the turtle cerebellum also revealed that neural activity produced an alkalinization followed by acidification (3). However, when a different method of measuring pH was applied (pH-sensitive dye) and a different brain region was examined (hippocampus), pH was reported to rapidly decrease and then alkalinize (4). A similar pattern of contrasting results emerged in the subsequent decades, with some individuals reporting that neural activity acidified the microenvironment (5,6), whereas others reported it alkalinized the microenvironment (7,8). One general trend was that as spatial resolution of the methodology used to measure synaptic pH improved, acidification was more commonly observed. Indirect evidence of synaptic acidification was also reported by several groups (9,10).

Yet, reconciling these data have been difficult because different methods were used to measure extracellular pH, including some with greater anatomical resolution, as highlighted above, and thus greater ability to approximate synaptic pH. In addition, differences in the brain regions assessed as well as the model systems examined complicated generalizations. Last, a variety of stimuli were used to evoke neural activity, further limiting the ability to draw global conclusions.

In a recent issue of Biophysical Journal, Feghhi and colleagues weighed in on this debate by using a computational modeling approach to predict the impact of neural activity on synaptic pH at the Drosophila neuromuscular junction. The authors first detailed key anatomical and physiological properties of the Drosophila neuromuscular junction, including changes in synaptic pH, using conventional pH-sensitive fluorophores. They then used this information to construct a computational model. Using this computational model, the authors found that synaptic pH was predicted to transiently acidify after (simulated) neurotransmission, followed by alkalinization. The lowest predicted pH under simulated glutamatergic transmission in the presence of buffering systems (e.g., bicarbonate and phosphate) and ATP at the Drosophila neuromuscular junction was 5.5 and occurred on a microsecond timescale. Although the authors found alkalization and not acidification in the Drosophila neuromuscular junction using conventional pH-sensitive fluorophores and live imaging, the transient acidification predicted by their model was consistent with more recent work demonstrating that glutamatergic signaling transiently acidifies the synapse in the mouse amygdala (6).

The authors also investigated the impact of buffering systems and ionic species movement on synaptic pH using their computation model. They found that phosphate buffering, and not bicarbonate buffering, could prevent initial acidification. Further, mimicking a closed synaptic cleft, in which free movement ionic species is limited, revealed that return to neutral pH after synaptic activity was greatly impaired compared with an open synaptic cleft. These data establish conceptual feasibility that both buffering systems and synaptic morphology have an important role in influencing pH homeostasis within the synaptic cleft.

Although the computational model generated by Feghhi and colleagues provides powerful new insight into the regulation of pH at the synapse, there are some considerations when applying it to other synapses. For example, because detailed information from the Drosophila neuromuscular junction was used to construct the computational model, predictions about central nervous system synapses may require modification and/or additional details. Such modifications may entail synaptic cleft distance as well as the key exchangers and/or pumps that regulate synaptic pH. A similar rationale could also be applied when making predictions about other neurotransmitter systems and synapses in other species. However, it is likely that with additional details and slight modifications, reliable predictions about other synapses could be made.

In summary, the tool generated by Feghhi and colleagues sheds further light onto a long-standing debate in the field and provides a new approach to predict whether a synapse will alkalinize or acidify or do both.

Editor: Gabriela Popescu.