Stimuli that elicit itch are detected by sensory neurons that innervate the skin. This information is processed by the spinal cord; however, the way in which this occurs is still poorly understood. Here we investigated the neuronal pathways for itch neurotransmission, particularly the contribution of the neuropeptide somatostatin. We find that in the periphery, somatostatin is exclusively expressed in Nppb+ neurons, and we demonstrate that Nppb+somatostatin+ cells function as pruriceptors. Employing chemogenetics, pharmacology and cell-specific ablation methods, we demonstrate that somatostatin potentiates itch by inhibiting inhibitory dynorphin neurons, which results in disinhibition of GRPR+ neurons. Furthermore, elimination of somatostatin from primary afferents and/or from spinal interneurons demonstrates differential involvement of the peptide released from these sources in itch and pain. Our results define the neural circuit underlying somatostatin-induced itch and characterize a contrasting antinociceptive role for the peptide.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank the NIDCR Gene targeting facility for help generating chimeric Sstf/f mice. We also thank C. Birchmeier, Max Delbrück Center, Berlin; M. Krashes, NIH, Bethesda; and Q. Ma, Dana-Faber Cancer Institute, Boston, for generously providing mice. We are very grateful to T. Furuta, Kyoto University, Kyoto, for the gift of PPD antibody, to A. Bell and D. Hughes for comments on the manuscript, to X. Gu for help in some of the experiments and to R. Kerr and C. Watt for expert technical help. This work was supported by the intramural research program of the National Institute of Dental and Craniofacial Research (NIDCR)-National Institutes of Health (MAH) and grants from the Medical Research Council (MR/L003430/1), the Biotechnology and Biological Sciences Research Council (grant N006119), the Wellcome Trust (102645) (A.J.T.), the Swiss National Science Foundation (156393) (H.U.Z.) and the Natural Science Foundation of China (31671247) (J.H.). The Sstf/f mice used in this study were generated from ES cells obtained from the National Center for Research Resources (NCRR)-NIH-supported KOMP repository and engineered by the Welcome Trust Sanger Institute and the Mouse Biology Program.
Integrated supplementary information
Supplementary Figure 1 ISH analysis of the expression of Nmb, Gal, Tac1, Calca, Nppb and Sst in the DRG.
Triple label in situ hybridization was used to assess the numbers of neurons expressing genes for the neuropeptides: A, neuromedin B (Nmb) and galanin (Gal); B, tachykinin 1 (Tac1, substance P) and calcitonin gene related peptide (Calca); C, natriuretic polypeptide b (Nppb) and somatostatin (Sst) relative to the pan-neuronal gene β3-tubulin (Tubb3). The fractions of neuropeptide expressing neurons were determined relative to Tubb3-positive cells, D. Data are means ±SEM (n=3 animals, 3 sections per mouse). Mean cell counts for each labeling experiment were, Nmb 855, Gal 215, and Tubb3 1400; Tac1 457, Calca 476, and Tubb3 1006; Nppb 92, Sst 89, and Tubb3 1059.
Supplementary Figure 2 Sstcre neurons innervate the skin and can be reliably activated using optogenetics.
A, double label ISH reveals that in SstCre;Ai9 mice expression of tdTomato (tdT, magenta) occurs largely in Nppb-neurons (green). White arrows indicate Nppb-expressing neurons not expressing tdTomato, and the grey arrow indicates a tdTomato-positive neurons that does not express Nppb. Similar results were obtained from 3 mice. B, Somatostain-neurons innervate superficial layers of the skin. A section through the ear of a Sstcre;Ai9 mouse reveals tdTomato fibers (red) that are close to the surface of the skin (image is superimposed on a DIC-image to highlight the location of fibers); also see1,2 for evidence of cutaneous segregation of somatostatin-expressing afferents. C, Neurons from Sstcre;Ai32 mice can be optogenetically activated with a variety of frequencies of light stimulation. Cultured DRG-neurons from Sstcre;Ai32 were stimulated at 1, 5, 20, and 40 Hz and electrophysiological responses recorded under current (left panel), or voltage clamp (right panel) conditions. n=11/11 neurons responding, all neurons exhibited non-saturating responses up to 20Hz. 1. Molander, C., Ygge, J. & Dalsgaard, C.J. Substance P-, somatostatin- and calcitonin gene-related peptide-like immunoreactivity and fluoride resistant acid phosphatase-activity in relation to retrogradely labeled cutaneous, muscular and visceral primary sensory neurons in the rat. Neurosci Lett 74, 37-42 (1987). 2. O’Brien, C., Woolf, C.J., Fitzgerald, M., Lindsay, R.M. & Molander, C. Differences in the chemical expression of rat primary afferent neurons which innervate skin, muscle or joint. Neuroscience 32, 493-502 (1989).
Supplementary Figure 3 Chemogenetic activation and transmitter phenotype in mCherry-labeled neurons.
A. Immunostaining for Pax2 was used to define inhibitory cells among the mCherry-labelled neurons in the PdynCre mice and a mean of 97 (range 82-120) cells were analysed in sections from 5 mice. For the nNOSCreERT2 mice, Sst2a immunoreactivity was used to identify inhibitory mCherry+ neurons (mean 86 cells, range 72-120), because both Pax2 and nNOS antibodies were raised in rabbit. The proportion of mCherry neurons that were Pax2-immunoreactive in PdynCre mice was 72.7% (62.9-90.2%), while the proportion of mCherry neurons that were Sst2a-immunoreactive in the nNOSCreERT2 mice was 57% (45-67.1%). Similar results were obtained in 5 mice of each genotype. B. In both PdynCre and nNOSCreERT2 mice the great majority of mCherry neurons were Fos+ after CNO treatment, irrespective of their neurotransmitter phenotype (95.9% [188/196] and 92.1% [93/101], respectively, for inhibitory and excitatory mCherry cells in PdynCre mice and 98.2% [108/110] and 86.1% [142/165], respectively, for inhibitory and excitatory mCherry cells in nNOSCreERT2 mice). Data were pooled from 3 CNO-treated mice of each genotype and absolute cell numbers are shown in square brackets. Very few Fos+ cells were seen in vehicle-treated mice: 1.3% [2/154] and 5.7% [2/35], respectively for inhibitory and excitatory mCherry cells in PdynCre mice and 0% [0/63] and 1.1% [1/93], respectively, for inhibitory and excitatory mCherry cells in nNOSCreERT2 mice. Data were pooled from 2 vehicle-treated mice of each genotype and absolute cell numbers are shown in square brackets. Similarly, the majority of Fos+ cells were also mCherry-positive in both mouse lines in the CNO-treated mice: 84.6% [281/332] and 63.5 [251/395], respectively for the PdynCre and nNOSCreERT2 lines. Few Fos cells were seen in the vehicle-treated mice, and some of these were mCherry-positive: 19% [4/21] and 3% [1/33], respectively for the PdynCre and nNOSCreERT2 lines. Experiments were performed on 5 mice of each genotype, which had received intraspinal injections of AAV2.flex.hM3Dq-mCherry into the L3 and L5 spinal segments. For each genotype, three of the mice were treated with CNO and two with vehicle two hours before perfusion fixation. Graphs show individual percentages for each of these mice.
Supplementary Figure 4 Laminar distribution of mCherry expression and lack of expression in dorsal root ganglia in PdynCre and nNOSCreERT2 mice that had received intraspinal injections of AAV2.flex.hM3Dq-mCherry.
A-D show transverse sections through the rostral part of the L3 segment of a PdynCre mouse (A,B) and through the caudal part of L3 of a nNOSCreERT2 mouse (C,D). In both genotypes, mCherry-positive cells form a dense band in the superficial dorsal horn, and there are scattered neurons in deeper laminae of the dorsal horn, consistent with the distribution of dynorphin- and nNOS-expressing neurons. Images to the right (B, D) show mCherry-immunoreactivity superimposed on a dark-field image, which indicates the location of the gray matter. Similar results were obtained from 5 animals of each genotype. E-J show scans of L4 dorsal root ganglia from the right side (ipsilateral to the intraspinal injection) of PdynCre (E-G) and nNOSCreERT2 (H-J) mice. In each case, immunostaining for mCherry is shown on the left, a dark-field image in the middle and a merged image on the right. No mCherry-positive cells were observed in the DRGs in either strain. Scale bar = 200 μm. Similar results were obtained from 4 nNOSCreERT2 animals and from 2 PdynCre animals. The time spent licking the calf in response to intradermal injection of chloroquine (100 μg) was reduced following chemogenetic activation (CNO) in PdynCre mice (K), whereas there was no effect on responses in nNOSCreERT2 animals (L). Significant differences were assessed using two-sided unpaired Student’s t-tests (t21 = 2.82, *p = 0.0103; ns not significant, t23 = 1.2, p = 0.2422). Data represent means ± SEM (n=11, 12, 12, and 13 animals, for PdynCre treated with CNO and vehicle and for nNOSCreERT2 treated with CNO and vehicle, respectively).
Supplementary Figure 5 Dynorphin excitatory neurons are chemogenetically activated; these include vertical cells, and there is a spatial segregation of inhibitory and excitatory preprodynorphin cells in glabrous and hairy skin territories in the dorsal horn.
A shows part of a transverse section of spinal cord taken from a PdynCre mouse that had been injected with AAV2-flex-hM3Dq-mCherry and treated with CNO 2 hours prior to perfusion fixation. A neuron in the outer part of lamina II, which is marked with an asterisk (*), expresses hM3Dq-mCherry (mCh, red), and is also immunoreactive for Fos (green), showing that it was chemogenetically activated. This is a projection of 6 confocal optical sections at 2 μm z-spacing, to show details of the dendritic tree. Arrowheads indicate prominent ventrally-directed dendrites, which are characteristic of vertical cells (a type of excitatory interneuron). The images on the right show the same cell in a projected image from only 2 optical sections. The nucleus of this cell is negative for Pax2 (blue) confirming that this is an excitatory neuron (see Figure S3 for numbers). Scale = 20 μm. Similar results were obtained from 3 animals. We identified 17 mCherry-positive lamina II neurons with ventral dendrites that entered lamina III in the sections analysed (n = 5 mice) and found that 16 of these (94%) were Pax2-negative (i.e. excitatory neurons). B. Excitatory dynorphin cells are largely restricted to regions of the superficial dorsal horn that are innervated from glabrous skin. Plots of inhibitory (Pax2-positive, blue circles) and excitatory (Pax2-negative, red circles) neurons that were immunoreactive for preprodynorphin (PPD) in three segments from the cervical enlargement: C7, C8 and T1 (data pooled from 2 mice) are shown. For C7 and C8, the border between regions innervated by hairy and glabrous skin, determined by immunostaining for VGLUT3 (not shown), is indicated by a dashed line. The regions lateral and medial to this line are innervated from hairy and glabrous skin, respectively. In the T1 segment, the VGLUT3 plexus extended right across the mediolateral extent of the dorsal horn, indicating that the whole region was innervated from hairy skin. C shows quantification of the of inhibitory and excitatory PdynCre neurons in the lumbar (L3, L4, L5) and cervical (C7, C8, T1) enlargements. All cells illustrated in the plates in panel A and in Fig 5 are included, and these have been divided into those in hairy and glabrous skin territories. Note that the proportions of excitatory PPD cells differ significantly between hairy skin territory (11.1% for lumbar, 10.6% for cervical) and glabrous skin territory (52.9% for lumbar, 62.8% for cervical). These differences were highly significant (Fisher exact probability test, two-sided, p = 7.5 × 10−11 for cervical and 5.4 × 10−9 for lumbar). Lumbar and cervical enlargements therefore display similar differential distribution of excitatory and inhibitory PdynCre neurons in the medial and lateral dorsal spinal cord. Data for the lumbar and cervical enlargements were obtained from 3 and 2 animals, respectively.
A, Sstf/f; Trpv1cre, and B, Sstf/f; Lbx1cre mice do not display significant differences in their scratch responses to intra-dermal injection of a variety of compounds, compared to littermate controls. Data are means ±SEM (n= 7, 8, 6, 6, 6, 6, 8, 8, 6, 6, 5, 6, 6, 6, 8, 8, 8, 9, 6, and 7); no significant (ns) difference in response were observed between genotypes assessed using two-sided unpaired Student t-tests (t13 = 0.5919, t10 = 0.3334, t10 = 1.703, t14 = 0.7476, t10 = 0.011, t9 = 1.565, t10 = 0.3265, t14 = 0.7429, t15 = 0.2209, t11 = 1.273, *p = 0.5641, 0.7457, 0.1194, 0.4671, 0.9913, 0.1520, 0.7508, 0.4698, 0.8282, and 0.2292). C, testing with von Frey filaments revealed significantly lower thresholds for reflex paw withdrawal responses for Sstf/f;Trpv1cre, Sstf/f;Lbx1cre and Sstf/f;Wnt1cre mice compared to control littermates, differences assessed using two-sided unpaired Student t-tests (t10 = 2.607, 2.507, 1.581, *p = 0.0262, 0.0311, and ns p = 0.1449). Data are means ±SEM (n= 6 animals).
Supplementary Figure 7 Response of dynorphin cells to somatostatin. Whole-cell patch-clamp recordings were made from eight eGFP+ neurons in spinal cord slices from PdynCre mice that had received intraspinal injections of AAV.flex.eGFP.
A.An example trace showing membrane hyperpolarization in response to bath-applied somatostatin (2 μM) in a dynorphin cell. All 8 cells tested showed clear hyperpolarization within a few minutes of the start of somatostatin application. B. Changes in the membrane potential of individual cells (circle, n = 8) are plotted, and the mean values (gray bar ± SD) are shown in control conditions (−58.0 ± 5.2 mV) and in the presence of somatostatin (−67.4 ± 6.5 mV). Somatostatin caused statistically significant hyperpolarization (t7 = 3.935, p = 0.0056, two-sided paired t-test, n = 8) in the dynorphin cells. C. Subthreshold currents measured in response to a voltage step protocol (−90 to −50 mV, 5 mV increments, 500 ms duration) while the cell was held at −60 mV under control conditions and in the presence of somatostatin. A reduction in the input resistance and an outward shift of the measured current during somatostatin application are apparent in this example trace. D. A plot of the values for each cell (circles, n = 8) before and during somatostatin application shows that all of the neurons have their input resistance reduced during somatostatin application. The mean values (gray bar ± SD) are shown for control (858.7 ± 380.7 MΩ) and somatostatin-treated conditions (516.9 ± 288.7 MΩ). A two-sided paired t-test indicates that this effect is significant (t7 = 6.243, p = 0.00043). E. The current-voltage (I-V) relationship was plotted using the voltage step protocol (as in C) in control conditions and in the presence of somatostatin. The I-V relationship in control conditions was then subtracted from that in the presence of somatostatin, which demonstrates the I-V relationship for the current responsible for the somatostatin-induced hyperpolarization. This current appears to reverse at −80.6 mV, which suggests that activation of somatostatin receptors was coupled to a downstream effect of opening G-protein coupled inwardly rectifying potassium (GIRK) channels. This intracellular coupling mechanism was previously suggested for another population of sst2a-expressing inhibitory interneurons in the mouse superficial dorsal horn3. Data were obtained from 8 cells and are shown as mean ± standard deviation. F. Patterns of action potential firing in control conditions (top) and during somatostatin application (bottom). Action potentials were evoked by injecting square current pulses (1 s). Of the 8 cells tested, 5 initially exhibited a tonic firing pattern while 3 cells showed an initial bursting pattern. However, after somatostatin application, only 1 cell continued to show tonic firing and the remaining 7 cells exhibited an initial bursting pattern. Note that the hyperpolarizing effect of somatostatin was counteracted by adding bias currents to the recorded cell to restore the membrane potential to around −60 mV. This conversion in firing patterns after somatostatin application indicates that not only the resting membrane potential, but also the ability of the cell to generate action potentials, was inhibited by somatostatin. G-I. Confocal optical section through the cell body of one of the recorded neurons, showing expression of GFP (green) and avidin-rhodamine (magenta), which was used to detect Neurobiotin. The recorded cell is GFP+. J. The morphology of the cell seen in a projection of 55 optical sections at 1 μm z-separation. Scale bars I = 20 μm, J = 50 μm. 3.Iwagaki, N., Garzillo, F., Polgar, E., Riddell, J.S. & Todd, A.J. Neurochemical characterisation of lamina II inhibitory interneurons that express GFP in the PrP-GFP mouse. Mol Pain 9, 56 (2013).
Supplementary Figure 8 Contacts from boutons that contain somatostatin and VGLUT2 onto dynorphin cells that express the Sst2a receptor.
A-C confocal image of part of a sagittal section from a PdynCre mouse that had received an intraspinal injection of AAV.flex.eGFP. eGFP is shown in green and sst2a receptor in magenta. Nine eGFP+ cells are present in this field, and 8 of these (marked with asterisks) are sst2a -immunoreactive, while 1 (arrowhead) lacks the receptor. Between 70 and 79 (mean 74.3) eGFP+ cells were identified in sections from the L3 segment of 3 mice, and 78.2% (68.9-78.5%) of these were sst 2a-immunoreactive. This is consistent with our finding that 90% of lamina I-II preprodynorphin neurons in hairy skin territory are inhibitory, and that 91% of inhibitory PPD cells are sst 2a-immunoreactive4. The box indicates the cell body of the neuron that is illustrated in D-I. D-I Contacts from boutons that co-express somatostatin (SST, red) and VGLUT2 (blue) onto the cell body (D-F) and dendrite (G-I) of a eGFP+ cell that was immunoreactive for sst 2a (magenta in D). Contacts are indicated with arrows. J. Plot of the contacts from SST/VGLUT2 boutons onto the cell bodies (yellow circles) or dendrites (blue circles) of 5 of the cells that were analysed. The top left cell is the one in illustrated in A (surrounded by box) and D-I. Confocal images in A-C are projected from 4 optical sections at 1 μm z-spacing. Those in D-F and G-I are projected from 3 and 5 (respectively) optical sections at 0.5 μm z-spacing. Scale bars: A-C = 50 μm, D-I = 5 μm, J = 50 μm. 4. Boyle, K.A. et al. A quantitative study of neurochemically defined populations of inhibitory interneurons in the superficial dorsal horn of the mouse spinal cord. Neuroscience 363, 120-133 (2017).