A novel spinal neuron connection for heat sensation

Abstract

Heat perception enables acute avoidance responses to prevent tissue damage and maintain body thermal homeostasis. Unlike other modalities, how heat signals are processed in the spinal cord remains unclear. By single-cell gene profiling, we identified ErbB4, a transmembrane tyrosine kinase, as a novel marker of heat-sensitive spinal neurons in mice. Ablating spinal ErbB4+ neurons attenuates heat sensation. These neurons receive monosynaptic inputs from TRPV1+ nociceptors and form excitatory synapses onto target neurons. Activation of ErbB4+ neurons enhances the heat response, while inhibition reduces the heat response. We showed that heat sensation is regulated by NRG1, an activator of ErbB4, and it involves dynamic activity of the tyrosine kinase that promotes glutamatergic transmission. Evidence indicates that the NRG1-ErbB4 signaling is also engaged in hypersensitivity of pathological pain. Together, these results identify a spinal neuron connection consisting of ErbB4+ neurons for heat sensation and reveal a regulatory mechanism by the NRG1-ErbB4 signaling.

Keywords: ErbB4; NRG1; TRPV1; chemogenetics; dynamic activity; glutamatergic transmission; heat hypersensitivity; heat sensation; population coding; spinal cord.

Figure 1.. Activation of ErbB4⁺ spinal neurons by noxious heat.

(A) Experimental design. (B-C) Immunostaining showing increased shEGFP⁺ cells in superficial layers (laminae I and II) of spinal cord after hot plate exposure. (B) Representative images. Left, low-power, bar, 100 μm; right, high-power of areas in squares; bar, 50 μm. (C) Increased shEGFP⁺ cells in superficial layers (left) but not deep layers (laminae IIi-V, right). T-test, ***P < 0.001, n = 12 slices of 3 mice. (D-E) Expression of intrinsic cFos in shEGFP⁺ cells after 52°C hot plate exposure. (D) Representative images. Arrowheads, shEGFP⁺cFos⁺ cells; arrows, shEGFP⁻cFos⁺ cells; triangles, shEGFP⁺cFos⁻cells; bar, 100 μm. (E) High percentage of double positive of EGFP⁺cFos⁺. (F) Diagram showing single cell RT-PCR strategy. (G) Gene profiles from 50 shEGFP⁺ cells. (H) EGFP⁺ cells positive for Vglut2 or GAD65/67. (I) shEGFP⁺ cells positive of reported IN markers. (J) shEGFP⁺ cells positive of ErbB4. (K) Vglut2+ shEGFP⁺ cells that were positive of ErbB4 and other IN markers. (L-N) Activation of ErbB4⁺ neurons in superficial dorsal horn by noxious heat. (M) cFos expression in tdTomato-labeled ErbB4⁺ cells in area indicated in L. Bar, 30 μm. (N) Increased tdTomato+ and cFos+ cells by exposure to 52°C hot plate; t-test, ***P < 0.001; n = 12 slices of 3 mice. (O) tdTomato+ and cFos+ cells in deep layers. T-test, n = 12 slices of 3 mice.

Figure 2.. Diminished heat sensation by ablating ErbB4⁺ neurons.

(A) Diagram of viral injection and genotype of mice. (B) Diagrams of Cre-dependent expression by AAV. (C) Validation of AAV-mCherry expression in dorsal horn areas of lumbar enlargements. (D) Ablation of EYFP⁺ cells in lumbar enlargements 3 weeks after AAV-dtA injection. Left, example images; right, Quantification result. T-test, ***P < 0.001. n =16 slices of 4 mice. Bar, 250 μm. (E) Reduced Nmur2+ cells and no change in Pax2⁺ and CR⁺ cells by AAV-dtA. Left, example images for RNAscope of Nmur2 and immunostaining of Pax2 and CR. Bar, 200 μm; right, quantitative data. T-test, **P = 0.0017. n = 16 slices of 4 mice. Bar, 250 μm. (F-H) AAV-dtA-injected ErbB4-CreER;LSL-EYFP mice showed increased latency in hot plate test (F, t-test, **P = 0.002; increased latency in Hargreaves test (G, t-test, *P = 0.024) and reduced capsaicin-induced foot licking (H, t-test, *P = 0.0136). (I) Diagram of viral injection and genotype of mice. (J) Ablation of EYFP⁺ cells in lumbar enlargements 3 weeks after AAV-dtA injection. Left, example images; right, Quantification data. Bar, 250 μm. T-test, ***P < 0.001. n = 16 slices of 4 mice. (K-M) Alteration of indicated mice in latency in hot plate test (K) (dash line, the cutoff time; t-test, *P_(SST) = 0.0184, ***P_(tri) < 0.001); latency in Hargreaves test (L) (dash line indicates the cutoff time; t-test, *P_(SST) = 0.039, ***P_(tri) < 0.001); and capsaicin-induced foot licking (t-test, ***P_{(tri, 1st min)} < 0.001, **P_{(tri, 2nd min)} = 0.0029). (N, O) AAV-dtA had little effect on falling time (left) or speed (right) in rotarod test (N) and latency of foot withdrawal induced by cold (O). T-test.

Figure 3.. Activation of ErbB4⁺ neurons by noxious heat but not mechanical stimulation.

(A-F) Monosynaptic inputs from TRPV1+ fibers onto ErbB4+ neurons. (A) Diagram of viral injection and genotype of mice. (B) EYFP⁺ terminals in close proximity to Homer+ puncta and mCherry⁺ (i.e., ErbB4⁺) cells in spinal slices. (C) Diagram of recording of ErbB4⁺ neurons in spinal slices. SC, spinal cord. (D) Example traces of light-evoked monosynaptic EPSCs in ErbB4⁺ neurons without (top) or with TTX and/or 4-AP (bottom). Bars, upper: 0.5 s, 50 pA; lower: 50 ms, 50 pA. (E, F) ErbB4⁺ neurons with monosynaptic or polysynaptic inputs from TRPV1⁺ DRG neurons. (G-K) Innervation of MrgprD+ neurons onto ErbB4+ neurons. (G) Diagram of viral injection and genotype of mice. (H) Example non-responsive traces of ErbB4⁺ neurons (top, bars, 0.5 s, 50 pA) and light-evoked EPSCs without or with TTX and/or 4-AP (bottom, bars 50 ms, 100 pA). (I) Example traces of light-evoked monosynaptic EPSCs in SST⁺ neurons. EPSC resistance to TTX and 4-AP was shown in lower panel. Scale bars, upper: 0.5 s, 50 pA; lower: 50 ms, 100 pA. (J, K) ErbB4+ and SST+ neurons receiving monosynaptic or polysynaptic inputs from MrgprD+ DRG neurons. (L-P) Activation of ErbB4+ neurons by heat stimulation of skin. (L) Diagram showing recording of ErbB4+ cells in the semi-intact preparation with stimuli being applied to the skin. ST, stimulation. (M) AP traces in ErbB4+ neurons evoked by saline at different temperatures and von Frey filaments. Scale bars, 0.5 s, 50 mV. (N) Heatmap of ErbB4+ neuron responses. (O) Radar plot of responsive ErbB4+ neurons to each modality. (P) Firing rates in response to stimuli. Dash line, threshold of 2 Hz.

Figure 4.. Inhibition of ErbB4⁺ neuron reduces heat response.

(A) Diagram of viral injection and genotype of mice. (B) Diagram of Cre-dependent expression of AAV-hM4Di and mCherry. (C) mCherry expression in injected lumbar enlargements. Bar, 100 μm. n = 16 slices of 4 mice. (D) mCherry⁺ cells in the dorsal horn and LSN. (E) CNO reduced APs in response to current injection in mCherry⁺ (i.e., ErbB4⁺) neurons. Top, representative traces, scale bars, 100 ms, 20 mV; bottom, quantitative data. Slices were incubated with 1 μM CNO for 10 min prior to recording. T-test, ***P_{(40 pA)} < 0.001, **P_{(60 pA)} = 0.0024. n = 10 cells of 3 mice. (F,G) Increased response latency in hot plate (F) and Hargreaves (G) tests by CNO. AAV-hM4Di-injected ErbB4-CreER mice were treated with CNO (i.p., 5 mg/kg) 30 min prior to test. T-test, **P_{(hot plate)} = 0.0013, **P_(Hargreaves) = 0.0074. (H) Decreased capsaicin-induced foot licking by CNO. T-test, **P = 0.0031. (I) Diagram of AAV injection into lumbar enlargement of SST-Cre, CCK-Cre, or ErbB4-CreER;SST-Cre;CCK-Cre (tri-Cre) mice. (J) mCherry expression in injected lumbar enlargements. Top, representative images; bottom, quantitative data. Bar, 100 μm. n = 16 slices of 4 mice. (K) CNO reduced APs in response to current injection in mCherry+ neurons. Left, representative traces, scale bars, 100 ms, 20 mV; right, quantitative data. Slices were incubated with 1 μM CNO for 10 min prior to recording. T-test, ***P_{(40 pA)} = 0.0006, ***P_{(60 pA)} = 0.0003. n = 11 cells of 3 mice. (L,M) Increased response latency in hot plate and Hargreaves test by CNO in AAV-hM4Di-injected SST-Cre and tri-Cremice. Dash lines indicate the cutoff time. T-test. *P_{(SST, hot plate)} = 0.0373, *P_{(SST, Hargreaves)} = 0.0229; ***P_{(tri, hot plate)} < 0.001, ***P_{(tri, Hargreaves)} < 0.001. (N) Decreased capsaicin-induced foot licking by CNO in AAV-hM4Di-injected SST-Cre and tri-Cre mice. T-test. ***P_{(tri, 1st min)} < 0.001, *P_{(tri, 2nd min)} = 0.0109. (O) Little changes in falling time (left) or speed (right) in rotarod test. T-test.

Figure 5.. Involvement of spinal ErbB4 and its kinase activity in heat sensation.

(A-G) Requirement of ErbB4. (A) Diagram of viral injection and genotype of mice. (B) Diagram of AAV-Cre and AAV-GFP. (C) AAV-Cre expression in lumbar enlargements. Bar, 250 μm. (D) Reduced ErbB4 expression in dorsal horns of lumbar enlargements. Tissues were dissected three weeks after viral injection. T-test, ***P < 0.001. (E-F) Increased response latency in hot plate (E) and Hargreaves (F) test in AAV-Cre-injected ErbB4f/f mice. T-test, **P_{(hot plate)} = 0.0087, *P_(Hargreaves) = 0.0284. (G) Reduced capsaicin-induced foot licking in AAV-Cre-injected ErbB4f/f mice. T-test, *P = 0.0351. (H-L) Inhibition by afatinib. (H) Diagram of intrathecal injection. (I) Reduced pErbB4 by afatinib (2 μM, 8 μL) in lumbar enlargements of WT mice. Tissues were collected 30 min after injection. T-test, ***P < 0.001. Behavioral tests were performed 30 min after intrathecal injection. (J,K) Increased response latency in hot plate (J) and Hargreaves (K) tests in afatinib-injected WT mice; t-test, *P_{(hot plate)} = 0.0181; *P_(Hargreaves) = 0.0206. (L) Reduced capsaicin-induced foot licking in afatinib-injected WT mice; t-test, *P = 0.0286. (M-Q) Inhibition by 1NMPP1. (M) Diagram of enlarged ATP binding pocket of ErbB4 T796G mutant. (N) Reduced pErbB4 in lumbar enlargements of T796G mice by 1NMPP1 (2 μM, 8 μl, intrathecally injected 30 min prior to tissue dissection). T-test, t₁₆ = 7.657, ***P < 0.001. Behavioral tests were performed 30 min after intrathecal injection of 1NMPP1. (O) Increased latency of foot licking and flinching in hot plate test in 1NMPP1 - injected T796G mice; t-test, *P = 0.0119. (P) Increased latency of foot withdrawal in Hargreaves test in 1NMPP1-injected T796G mice; t-test, *P = 0.0254 mice. (Q) Reduced capsaicin-induced foot licking in 1NMPP1-injected T796G mice; t-test, *P = 0.025.

Figure 6.. Attenuated response to noxious heat by neutralizing endogenous NRG1 in spinal cord or deleting NRG1 in DRGs.

(A) Diagram of intrathecal injection. (B) Reduced pErbB4 by ecto-ErbB4 in lumbar enlargements. Ecto-ErbB4 (5 μg in 8 μl) was intrathecally injected 1 hr prior to tissue dissection. T-test, ***P < 0.001. (C-E) Behavioral tests were performed 1 hr after intrathecal injection. (C,D) Increased response latency in hot plate (C) and Hargreaves (D) test in ecto-ErbB4-injected mice. T-test, *P_{(hot plate)} = 0.0258; *P_(Hargreaves) = 0.0232. (E) Reduced capsaicin-induced foot licking in ecto-ErbB4-injected mice; t-test, *P = 0.0146. (F) More NRG1 mRNA in DRG than spinal dorsal horn. SC, spinal cord. T-test, ***P < 0.001. The content of NRG1 was calibrated with GAPDH. (G) Breeding paradigm. (H) ELISA showing reduced NRG1 protein in the spinal dorsal horn of advillin-CreER;Ai9 mice. T-test, ***P < 0.001. (I,J) Reduced NRG1 protein and pErbB4 in the spinal dorsal horn of advillin-CreER;NRG1f/f mice. T-test, ***P < 0.001. (K) Reduced NRG1 mRNA in DRG, but not spinal cord, of advillin-CreER;NRG1f/f mice. T-test, ***P < 0.001. (L-N) Increased response latency in hot plate (L) and Hargreaves (M) tests and reduced duration of capsaicin-induced foot licking (N) in advillin-CreER;NRG1f/f mice. T-test. *P_{(hot plate)} = 0.011; *P_(Hargreaves) = 0.0119; **P_(capsaicin) = 0.0012.

Figure 7.. NRG1-ErbB4 signaling promotes glutamatergic transmission

(A-I) Glutamatergic transmission by ErbB4+ neurons to downstream neurons. (A) Diagram of optogenetic activation of ErbB4⁺ neurons and EPSC recording in downstream neurons in spinal cord slices. (B) Expression of ChR2 and EYFP in ErbB4+ neurons. Bar, 10 μm. (C) Example traces of light-evoked photocurrents (upper) and APs (lower) in ErbB4⁺ neurons under voltage-clamp and current-clamp mode, respectively. Scale bars, upper: 0.5 s, 100 pA; lower: 0.5 s, 20 mV. (D-H) Light-evoked glutamatergic EPSCs in neurons downstream. (D,E) Light (473 nm)-evoked EPSC traces without (top, bars, 0.2 s, 50 pA) and with glutamate antagonists (E, bars, 20 ms, 50 pA). (F) Short onset latency of EPSCs recorded in downstream neurons. Scale bars, 10 ms, 100 pA. (G) EPSCs resistant to TTX and 4-AP. Scale bars, 20 ms, 100 pA. (H) Quantitative data in D-G. (I) Innervation of biocytin-labeled neurons by ErbB4⁺ neurons. Recorded neurons were labeled by biocytin in recording pipette and examined for EYFP puncta (encoded by Cre-dependent AAV). Bars, 10 and 2 μm for lower- and higher-power images, respectively. (J) Diagram of recording. (K) Increased amplitudes of light-evoked EPSCs in downstream neurons by NRG1 (10 nM). Upper, representative traces, scale bars, 20 ms, 40 pA; lower, quantitative data. Repeated measures One-way ANOVA, **P_{(baseline vs. NRG1)} = 0.0065, *P_{(wash vs. NRG1)} = 0.0186; n = 10 cells of 3 mice. (L) Afatinib (10 nM) reduced the amplitude of light-evoked EPSCs in downstream neurons. Upper, representative traces, scale bars, 20 ms, 40 pA; lower, quantitative data. Repeated measures One-way ANOVA, ***P_{(baseline vs. afatinib)} < 0.001, **P_{(wash vs. afatinib)} = 0.0074; n = 10 cells of 3 mice.

Figure 8.. Involvement of NRG1-ErbB4 signaling in heat hypersensitivity induced by peripheral inflammation and nerve injury.

(A) Diagram of a CFA model. i.p., intraplantar injection. (B) Reduced withdrawal latency of CFA-injected hindpaw in response to radiant heat in Hargreaves test. Veh, vehicle; ipsi, ipsilateral hindpaw; contra, contralateral hindpaw; pre, before CFA injection. T-test, ***P_{(Veh vs CFA)} < 0.001, ^##P_{(Contra vs Ipsi)} = 0.002; ^$$$P_{(Pre vs Post)} < 0.001. (C) Dissection of the spinal dorsal horn ipsilateral to the CFA injection. (D) ELISA showing increased NRG1 protein in the ipsilateral dorsal horn of CFA mice. T-test, *P = 0.025. (E,F) Western blot showing increased NRG1 protein and pErbB4 in the ipsilateral dorsal horn. E, representative bands; F, quantitative data. T-test, **P_(NRG1) = 0.0011, **P_(p-ErbB4) = 0.0064. (G) Diagram of intrathecal injection. (H-J) Increased withdrawal latency of the CFA-injected hindpaw in mice injected with ecto-ErbB4 (H), afatinib (I), or 1NMPP1 (J). T-test, ***P_(ecto-ErbB4) = 0.0008; **P_(afatinib) = 0.0028; **P_(1NMPP1) = 0.0038. (K) Diagram of a CCI model. (L) Reduced withdrawal latency of the CCI hindpaw. Sham, sham surgery; ipsi, ipsilateral hindpaw; contra, contralateral hindpaw; pre, before CCI surgery. T-test, **P_{(Sham vs CCI)} = 0.0097; #P_{(Contra vs Ipsi)} = 0.0172; ^$$P_{(Pre vs Post)} = 0.0016. (M) ELISA showing increased NRG1 protein in the ipsilateral dorsal horn. T-test, **P = 0.0043. (N,O) Western blot showing increased NRG1 protein and pErbB4 in the ipsilateral dorsal horn. N, representative bands; O, quantitative data. T-test, **P_(NRG1) = 0.0055, *P_(p-ErbB4) = 0.0123. (P) Diagram showing intrathecal injection. (Q-S) Increased withdrawal latency of the CCI hindpaw in response to radiant heat stimulation, in mice injected with ecto-ErbB4 (Q), afatinib (R), or 1NMPP1 (S). T-test, *P_(ecto-ErbB4) = 0.0223, *P_(afatinib) = 0.0124, **P_(1NMPP1) = 0.0071.