V1 and v2b interneurons secure the alternating flexor-extensor motor activity mice require for limbed locomotion

Abstract

Reciprocal activation of flexor and extensor muscles constitutes the fundamental mechanism that tetrapod vertebrates use for locomotion and limb-driven reflex behaviors. This aspect of motor coordination is controlled by inhibitory neurons in the spinal cord; however, the identity of the spinal interneurons that serve this function is not known. Here, we show that the production of an alternating flexor-extensor motor rhythm depends on the composite activities of two classes of ventrally located inhibitory neurons, V1 and V2b interneurons (INs). Abrogating V1 and V2b IN-derived neurotransmission in the isolated spinal cord results in a synchronous pattern of L2 flexor-related and L5 extensor-related locomotor activity. Mice lacking V1 and V2b inhibition are unable to articulate their limb joints and display marked deficits in limb-driven reflex movements. Taken together, these findings identify V1- and V2b-derived neurons as the core interneuronal components of the limb central pattern generator (CPG) that coordinate flexor-extensor motor activity.

Figure 1. A non-V1 Interneuron population is involved in securing reciprocal flexor-extensor activity

(A-B) Left: Representative extracellular recordings from ventral roots of a P0 En1^Cre; R26^{floxstop-TeNT} spinal cord showing locomotor ENG activity prior to and after hemisection. Signals were recorded from the L2, iL5 and cL2 ventral roots. Right: Circular plots showing averaged L2-iL5 phase relationships in whole (upper) and hemisected (lower) preparations. Inner circle indicates an r significance value of p=0.01. Points located near 0.5 represent alternation. Alternating L2-iL5 L2-cL2 locomotor activity is maintained before and after hemisection. The L2-cL2 phase relationship is also shown. The number of cords analyzed is indicated (n). (C-D) Representative sections of FITC-dextran injected spinal cords from wild-type and Pax6^-/- embryos at E18.5, showing FITC-dextran⁺ ells rostral and ipsilateral to the injection site. vGAT in situ hybridization identifies a subset of FITC-dextran⁺ cells as inhibitory neurons in wild-type and Pax6^-/- spinal cords (arrowheads). (E) A section from a F-dextran-labeled E18.5 Pax6^-/- embryo showing F-dextran⁺/Gata3⁺ cells (yellow, arrowheads). (F-H) Comparison of Gata3⁺ V2b cell numbers in E11.5 Gata3^lacZ and Gata3^lacZ; Pax6^-/- spinal cords as assessed by β-galactosidase staining. Quantification is shown in H (mean±SEM.; n=12 sections, 3 cords for each sample).

Figure 2. Neurotransmitter phenotype and axonal projections of V2b INs

(A-B) Gata3 expression in the spinal cords of GlyT2-GFP (A) and Gad1-GFP transgenic mice at E13.5. (C) Co-expression of En1 and Gad1 (D) Co-expression of Pax2 and Gad1. (E) Quantification of Gata3 and En1 co-expression with Gad1 and GlyT2. Bars indicate the number of cells per 30 μm thick hemicord section (mean±SEM, n=3 cords). (F) Quantification of the proportion of Gad1⁺ and GlyT2⁺ neurons that are either V1 or V2b INs at E13.5.

Figure 3. IaINs are derived from V1 and V2b INs

(A) P15 Gata3^Cre; Thy1::flox-stop-YFP spinal cord showing V2b INs are located predominantly in lamina VII. Note the V2b cells close to the border of lamina VII/IX. (B) Schematic showing the protocol for CTB-labeling of specific hindlimb motor neuron pools and visualization of inhibitory synaptic contacts. (C) Section showing the soma and proximal dendrites of two motor neurons (blue) labeled by injecting Cy5-CTB into the posterior biceps muscle. GlyT2 ⁺/vGAT⁺ inhibitory synaptic contacts from V2b INs are indicated (arrowheads). (D) Section showing V2b IN-derived GlyT2⁺/vGAT⁺ inhibitory contacts (arrowheads) on the soma of lamina VII neurons. Inset: Inhibitory contacts (arrowheads) from V2b INs on a ChAT ⁺ V0c IN (blue). (E) Quantification of V1- and V2b-derived synaptic contacts on iliopsoas (IP), gluteus (GL), posterior biceps-semitendinous (PB/St) and quadriceps (Q) motor neurons. Error bars = SEM. (F-G) Sections through the lumbar spinal cord of P15 Gata3^Cre; Thy1::floxstop-YFP (F), and En1^Cre; Thy1::floxstop-YFP (G) mice, showing neurons with the anatomical features of IaINs. Cells decorated with calbindin⁺ (F’, G’) and vGluT1⁺ (F”, G”) contacts were scored as IaINs. Cell counts for V2b-derived IaINs and V1-derived IaINs are given in Table S1.

Figure 4. Abolishing V2b IN transmission degrades flexor-extensor coordination

(A-B) Left; extracellular recordings from L2, iL5 and cL2 ventral roots of wild-type and Gata3^Cre; R26^{floxstop-TeNT} from P0 whole cords following the induction of locomotor activity with 5 μM NMDA and 15 μM 5-HT. Right; circular plots showing the phase of bursts in iL5 and cL2 with respect to L2. Each point in the circular plot represents the average phase for one spinal cord (25-50 steps). The angular vector (arrow) and r-value (r) for each genotype represents the mean phase relationship for all recorded cords (n). The inner circle indicates an r significance value of p=0.01. Alternating flexor-extensor (L2-iL5) activity is observed in wild-type (A) whole cords and in all but one Gata3^Cre; R26^{floxstop-TeNT} (B) whole cord. (C-D) Left: Locomotor-activity following spinal cord hemisection. Left; the normal pattern of flexor-extensor (L2-iL5) alternation in wild-type hemisected cords (C) is markedly reduced in Gata3^Cre; R26^{floxstop-TeNT} hemisected cords (D), with 13/20 cords displaying synchronous L2-iL5 activity. Right; circular plots showing the average L2-iL5 phase and r-values for the cords in each experimental group. n = number of cords analyzed.

Figure 5. V1 and V2b INs function cooperatively to establish flexor-extensor alternation

(A) Left; Example of extracellular recordings from L2, iL5 and cL2 ventral roots of a P0 Gata3^Cre; En1^Cre; R26^{floxstop-TeNT} spinal cord following application of 5 μM NMDA and 15 μM 5-HT. Right: Phase relationship between L2 and iL5 (L2-iL5), and between L2 and cL2 (L2-cL2) ventral roots. The inner circle indicates an r significance value of p=0.01. (B-C) Extracellular recordings from L2, iL5 and cL2 ventral roots of P0 wild-type (B) and P0 Gata3^Cre; En1^Cre; R26^{floxstop-TeNT}; (C) spinal cords following sacral dorsal root stimulation. Normal flexor-extensor (L2-iL5) and left-right (L2-cL2) alternation is observed in wild-type cords (B). Flexor-extensor (L2-iL5) alternation is absent in Gata3^Cre; En1^Cre; R26^{floxstop-TeNT} cords, however, left-right (L2-cL2) alternation remains intact. n = number of cords analyzed.

Figure 6. Loss of flexor-extensor alternation in Gata3^Cre; En1^Cre; R26^{floxstop-TeNT} spinal cords is not due to changes in the balance of excitation and inhibition

(A) Extracellular recordings from the L2, iL5 and cL2 ventral roots of Gata3^Cre; En1^Cre; R26^{floxstop-TeNT} whole-cord preparations at P0 after induction of locomotor activity with 5 μM NMDA and 15 μM 5-HT. (B-C) Synchronous flexor-extensor (L2-iL5) activity is maintained at lower NMDA concentrations (B) and after reducing AMPA-dependent excitatory transmission with 2.5 μM CNQX (C). (D) Enhancing inhibitory transmission with the glycine uptake inhibitor sarcosine (500 μM) does not rescue L2-iL5 alternation. Note that left-right (L2-cL2) alternation is preserved under all conditions.

Figure 7. Defective limb movement in newborn Gata3^Cre; En1^Cre; R26^{floxstop-TeNT} mice

(A-B) EMG recordings from knee extensor (quadriceps, Q) and knee flexor (posterior biceps, PB) muscles of newborn P0 mice following induction of locomotor activity in a limbs-attached spinal cord preparation with a cocktail of 10 μM NMDA, 25 μM 5-HT and 25 μM dopamine. Quadriceps (Q) and posterior biceps (PB) muscle activity is alternating in wild-type animals (A), and synchronous in Gata3^Cre; En1^Cre; R26^{floxstop-TeNT} mice (B). (C) A series of time-lapse images of P0 wild-type pups showing flexion-extension movements of the hindlimbs in response to tail pinch. (D) Gata3^Cre; En1^Cre; R26^{floxstop-TeNT} mutant animals show no flexion-extension movements and their limbs are immobile. (E) Number of flexion-extension movements in response to a single tail pinch in wild-type pups and age matched En1^Cre; R26^{floxstop-TeNT}, Gata3^Cre; R26^{floxstop-TeNT} and Gata3^Cre; En1^Cre; R26^{floxstop-TeNT} animals.

Figure 8. Loss of reciprocal inhibitory reflexes in mice lacking V1 and V2b INs

(A) Schematic showing disynaptic pathway from quadriceps (Q) afferent nerve to posterior biceps-semitendinosus (PBSt) motor neurons. (B-E) Intracellular recordings of synaptic responses in PBSt motor neurons were elicited by stimulating the Q nerve (3-5x threshold). Hyperpolarizing inhibitory potentials often become depolarizing as the resting potential (indicated at the beginning of each trace) increased. An inhibitory potential (IPSP) in a wild-type mouse that is disynaptic (13-17 ms) is shown in B. In Gata3^Cre; En1^Cre; R26^{floxstop-TeNT} mice, synaptic potentials with disynaptic latencies (13-17 ms) were either depolarizing and could not be reversed with current injection (C), or were long latency (D and E). A hyperpolarizing response (23 ms) that is inconsistent with disynaptic inhibition is shown in E. (F-G) Frequency of disynaptic inhibitory potentials in control and Gata3^Cre;En1^Cre; R26^{floxstop-TeNT} pups before (F) and after (G) current injection. The significance between thenumber of disynaptic IPSPs in wild-type versus Gata3^Cre; En1^Cre; R26^{floxstop-TeNT} samples was determined using the Mann-Whitney two-tailed test. The sample size is indicated below each graph.

Figure 9. Model for V1 and V2b IN control of flexor-extensor motor activity

Prospective model for how the limb CPG is functionally organized with respect to V1 and V2b interneurons. Arrows indicate excitatory pathways. Lines with filled spheres indicate inhibitory pathways. Weak excitatory connections between presumptive rhythm-generating excitatory neurons result in the synchronous rhythmic excitation of flexor and extensor motor neurons when V1 and V2b inhibition is absent.