Myelinating satellite oligodendrocytes are integrated in a glial syncytium constraining neuronal high-frequency activity

Satellite oligodendrocytes (s-OLs) are closely apposed to the soma of neocortical layer 5 pyramidal neurons but their properties and functional roles remain unresolved. Here we show that s-OLs form compact myelin and action potentials of the host neuron evoke precisely timed Ba2+-sensitive K+ inward rectifying (Kir) currents in the s-OL. Unexpectedly, the glial K+ inward current does not require oligodendrocytic Kir4.1. Action potential-evoked Kir currents are in part mediated by gap–junction coupling with neighbouring OLs and astrocytes that form a syncytium around the pyramidal cell body. Computational modelling predicts that glial Kir constrains the perisomatic [K+]o increase most importantly during high-frequency action potentials. Consistent with these predictions neurons with s-OLs showed a reduced probability for action potential burst firing during [K+]o elevations. These data suggest that s-OLs are integrated into a glial syncytium for the millisecond rapid K+ uptake limiting activity-dependent [K+]o increase in the perisomatic neuron domain.

A non-satellite oligodendrocyte labeled with the same antibodies. OLs was on average 7.4 ± 1.4 µm away from the onset of the AIS as defined by the start of beta-IVspectrin labeling (n = 26). When located at the neuron soma base the s-OLs were sometimes touching the AIS (6 out of 26), but oriented more to the somatic domain (see Supplementary Movie 1).
(d) Top: Satellite cell that was morphologically identified as OPC after filling with Alexa 594. Bottom: The cell shown in the live image was fixed with PFA after the recording and subsequently labeled for the OPC marker NG2 (polyclonal rabbit antibody, 1:250, AB5320, Millipore) that gave a positive signal.

Supplementary Figure 2.
Fluorescence and ultrastructure analysis of single identified s-OL internodes.
(a) A process of a single filled s-OL co-localizes with myelin structures that were identified by two dark bands in the transmitted light. Scale bar 10 µm.
(b) Maximum z-projected bright field image of a HRP filled s-OL after DAB reaction and embedding in epon.
Several internodes and the cell body are visible as black staining. Scale bar 20 µm. (f) Inward current amplitudes of s-OLs in relation to different holding voltages normalized to the maximum recorded inward current (n = 9 cells). The inward current was strongly attenuated at more depolarized holding potentials, but was never completely abolished probably due to gap-junction coupling. Membrane areas that are not occupied by the nucleus show the strongest signal. Based on the DAPI labeling the OL was assumed to be a s-OL as its nucleus can be seen next to a larger cell nucleus, presumably from a neuron (white arrow). (c) With a 25 nm distance (or α = 0.01 uniform) the extracellular K + is poorly buffered causing seizure generation. Simulations were run with a Kir peak conductance density of 0.05 pS µm -2 .

Supplementary
(d) Simulations with a 110 nm intercellular distance (or α = 0.01 uniform) using a 700 ms current injection at 57.6 pA to compare model voltage responses. Reducing Kir conductance density from of 0.1 to 0.001 pS µm -2 primarily reduces the K + uptake during the high-frequency AP generation but has little to no impact on AP-evoked K + release during low-frequency steady firing (~10 Hz). .0 pA, n = 25, vs. without s-OL 155.9 ± 24.5 pA, n = 9, unpaired t-test P = 0.31) nor the slope (with s-OL 0.1 ± 0.003 Hz pA -1 , n = 25, vs. without s-OL 0.098 ± 0.005 Hz pA -1 , n = 9, unpaired t-test P = 0.80). No difference of the firing rate was found between the two groups for the same current injection step (ANOVA post-hoc Bonferoni, P = 0.20).

Supplementary
(c) Summary plot of the average firing rate obtained from paired experiments in 3 mM K + and 8 mM K + for neurons with and without s-OL (n = 7 each group, see data in Figure 8). As expected by the application of high K + the F-I curve is shifted leftward.