Of the nonspiking neurons, 61% (11/18) exhibited membrane potential oscillations in phase with ventral root bursts ( Figures 5C and 5F). Thus, Shox2 INs are rhythmically active during drug-evoked locomotor-like activity. We next analyzed, separately, the set of Shox2 INs located in predominantly flexor-related (L2 and L3) or extensor-related (L4 and L5) segments.

We found that EGFR phosphorylation 20/27 of the Shox2 INs in L2/L3 spiked rhythmically whereas only 12/25 of the L4/L5 Shox2 INs spiked rhythmically. For both rhythmic firing and membrane oscillations, there was a clear flexor dominance. We found that 70% of neurons in L2/L3 (14/20) spiked in phase with local flexor-related ventral root activity, whereas in L4/L5, spiking was equally divided into flexor- and extensor-related neurons. Approximately 60% of L2/L3 (15/26) and of L4/L5 (13/21) Shox2 INs had the depolarizing peak in the flexor phase. Therefore, regardless of segment, most Shox2 INs are rhythmically active in the flexor-phase. This finding is in contrast to rhythmic Chx10-GFP neurons (a mix of Shox2+ and Shox2off V2a neurons), where flexor- and extensor-related neurons were evenly distributed throughout the lumbar cord (Dougherty and Kiehn, 2010a and Dougherty and Kiehn, 2010b). Our electrophysiological findings on Shox2 INs reveal preferential activation

of Shox2 INs during the flexor phase of locomotion. To determine whether this feature is correlated with connectivity profiles detected at the premotor level, we evaluated connectivity between Shox2 INs and motor neurons. First, we injected a floxed-synaptophysin-GFP adeno-associated http://www.selleckchem.com/products/birinapant-tl32711.html virus into the spinal cords of P3 Shox2::Cre mice and monitored the presence of GFP-labeled Shox2 IN terminals on motor neurons at P17 ( Figures 6A–6D). We observed many Shox2 IN terminals in lamina IX, often in apposition to motor neuron somata and proximal dendrites ( Figures 6A–6C). Shox2 IN terminals

were also detected in the intermediate zone Phosphatidylinositol diacylglycerol-lyase and in lamina VIII, the settling position of other CPG interneurons. High-resolution reconstructions of synaptic input from Shox2 INs to motor neurons innervating ankle flexor tibialis anterior (TA) or ankle extensor gastrocnemius (GS) muscles, specifically marked by retrograde labeling from the muscle, revealed a Shox2 IN synaptic bias toward flexor (TA) motor neurons ( Figure 6D). To determine the position of Shox2 INs that are monosynaptically connected to motor neurons, we performed retrograde transsynaptic labeling by the application of rabies viruses with monosynaptic restriction (Stepien et al., 2010 and Tripodi et al., 2011). We carried out unilateral virus injections coincidently into several hindlimb muscles to target many premotor neurons. We found that ∼50% of Shox2 INs in the rostral lumbar spinal cord were transsynaptically marked from hindlimb innervating motor neurons, whereas this fraction decreases caudally (Figures 6E–6G).