Extended Data Fig. 4: Epidural electrical stimulation induces pressor responses through the recruitment of posterior afferents and excitatory interneurons.
From: Neuroprosthetic baroreflex controls haemodynamics after spinal cord injury
Step 1(a): We developed a hybrid computational model based on real anatomical structures generated through high-resolution computed tomography and MRI scans. This model combines a geometrically realistic 3D finite element model of the spinal cord with realistic compartmental cable models of all afferent neurons, efferent neurons and some interneurons. We established a computational pipeline to obtain anisotropic tissue property maps, discretize the model, perform simulations using an electro-quasi-static solver and couple these simulations with NEURON-based electrophysiology models (Sim4Life by ZMT, www.zurichmedtech.com). We investigated the recruitment patterns of various afferent and efferent fibres within the spinal cord structure. We found that stimulation over the dorsal aspect of the spinal cord led to high levels of recruitment of major afferents, before any recruitment of efferent neurons directly from the stimulation. This suggested that epidural electrical stimulation activates pressor responses by recruiting afferents. Step 2(b): Next, we experimentally tested the hypothesis that pressor responses induced by EES were dependent on afferent activation. We completed successive dorsal rhizotomies at T11, T12 and T13 and found a graded reduction in the response to stimulation (one-way ANOVA; all P < 0.001; post hoc results indicated), with the largest decrease when removing T12, consistent with our functional and anatomical mapping results. Grey box indicates stimulation. Bar charts represent the mean with raw data overlaid. Step 3(c): Next, we developed a NEURON-based spiking neural network model composed of integrate-and-fire neurons to predict the presence of direct, indirect excitatory, and indirect inhibitory connections. Indirect inhibitory connections resulted in poor sympathetic pre-ganglionic neuron recruitment (left) and in the minimization of membrane potentials in response to increasing stimulation amplitude (right; various stimulation amplitudes indicated by alpha; action potential threshold indicated by horizontal dotted line; stimulation onset indicated by vertical dotted line). This suggested that pressor responses to EES likely are mediated by either direct, monosynaptic connections between afferents and sympathetic pre-ganglionic neurons or by indirect circuits including excitatory interneurons. Step 4(d): We completed anterograde tracing of the dorsal root ganglia. Using dynamic image registration we generated a digital dorsal horn whereby we could select a region of interest (ROI; grey box) and determine the mean intensity (‘Observed ROI’) of either axons (orange) or synapses (red). Using 1000 bootstraps of random ROIs as a null distribution we found a depletion of axons (empirical P = 0.019) and synapses (empirical P = 0.001) in the intermediolateral column. We confirmed this result by counting neurons with appositional synapses on ChATON neurons in the ventral horn versus the lateral horn and found a similar statistical depletion (n = 10 images, 294 neurons; Fisher’s exact test; odds ratio (OR): 0.082; P < 2.2 × 10−16). This suggested that the most likely circuit mediating these responses instead included an excitatory interneuron. We therefore completed retrograde trans-synaptic tracing and found interneurons trans-synaptically connected to splanchnic ganglia that were SLC17A6 positive, and had VGLUT1 synaptic puncta in their immediate vicinity (see Fig. 2), suggesting direct connections with large diameter afferents. *P < 0.05; **P < 0.01; ***P < 0.001. Rz, rhizotomy.
