The Trpm8Flp mouse line captures Trpm8-expressing primary afferent neurons.

(A) Immunohistochemical staining for GFP (yellow) and NeuN (blue) in a dorsal root ganglion from a Trpm8Flp;RCE:FRT mouse. GFP is present in some of the small neurons. (B) Fluorescence in situ hybridisation with a probe for Trpm8 mRNA (magenta) superimposed on GFP native fluorescence (green). The great majority of GFP cells contain Trpm8 message, and vice versa. Some of the double-labelled cells are indicated with arrows, while the arrowhead points to a Trpm8-positive cell that lacks GFP. (C) Frequency distribution of soma diameter for cells that were GFP-/Trpm8+ (red bars), GFP+/Trpm8+ (blue bars) or GFP+/Trpm8- (green bars). Scale bars: (A) = 25 μm, (B) = 50 μm.

Distribution of Trpm8 afferents in the spinal dorsal horn.

Immunohistochemical staining for GFP at different segmental levels through the spinal cord in Trpm8Flp;RCE:FRT mice. GFP-labelled axons are largely restricted to lamina I, although occasional fibres penetrate into deeper parts of the dorsal horn (arrow). Within lamina I the distribution of GFP axons is patchy, and does not occupy the entire mediolateral extent. Scale bar = 200 μm.

Combined confocal and electron microscopic examination of a lamina I projection neuron with numerous contacts from Trpm8-expressing afferents.

(A) A confocal image (single optical section) from a horizontal section through lamina I, showing the cell body and part of a proximal dendrite of one of the tdTomato-expressing neurons that were analysed. The tissue was obtained from a Phox2a::Cre;Ai9;Trpm8Flp;RCE:FRT mouse. TdTomato expressed by the projection neuron is shown in magenta and GFP immunoreactivity in Trpm8 afferents in green. Several Trpm8 boutons contact the cell, and five of these are indicated with arrowheads. (B) A low magnification electron micrograph through the cell at approximately the same z-level as that shown in A. Although the cell does not contain an electron dense label, it can be recognised by its shape, and because it is surrounded by GFP-expressing axonal profiles that were labelled with an immunoperoxidase method. The locations of those corresponding to the boutons shown in A are again marked with arrowheads. The cell has been pseudocoloured magenta to show its location. (C,D) Progressively higher magnification EM images showing the five GFP- labelled boutons marked with arrowheads in A and B. In D the boutons are marked with asterisks and the locations of membrane darkening and vesicle clusters that presumably correspond to synapses on the cell body and dendrite of the projection neuron are indicated with arrowheads. Scale bars: (A,B) = 10 μm, (C) = 1 μm (D) = 500 nm.

Electrophysiological characterisation of retrogradely labelled lamina I projection neurons recorded in Trpm8Flp;RCE:FRT mice

(A) Fluorescence microscopy image of a retrogradely labelled (mCherry-positive) lamina I ALS neuron (soma marked with asterisk) that is densely coated by Trpm8-expressing (GFP-positive) axons. Note that bundles of GFP axons are closely associated with the dendrites of the retrogradely labelled neuron, which are marked with arrowheads. (B) Whole-cell recording from this cell reveals that only cold stimulation evoked action potentials (APs). (C) Pie chart showing the response properties of the 6 lamina I ALS neurons that were densely coated with Trpm8 afferents to mechanical, cold and heat stimulation of the skin. All of these cells responded only to cold stimulation: 3 cells showed APs, 1 cell showed EPSPs but no APs (subthreshold), while 2 cells were recorded in voltage clamp and showed EPSCs. (D) Fluorescence microscopy image of a lamina I ALS neuron that was not densely coated with Trpm8 axons. Again, the soma is marked with an asterisk and dendrites with arrowheads. Note that although the dendrites of this cell pass through bundles of GFP axons these bundles are not aligned with the dendrites. (E) Whole-cell recording from the cell shown in (D). AP firing was evoked by mechanical stimulation (von Frey filament 4.0g), but not by application of hot or cold saline to the skin. (F) Pie chart of response properties of 5 lamina I ALS neurons that lacked dense Trpm8 input to mechanical, cold and heat stimulation to the skin. Two cells were mechano-selective, while three were polymodal (2 cells: mechano, heat, 1 cell: mechano, heat and cold). Scale bar (A,D) = 50 µm. Scale bars in B apply to E.

Characterisation of Calb1-expressing projection neurons.

(A) The distribution of retrogradely labelled cells in 10 transverse sections from the L5 segment of one of the Calb1Cre mice that had received an injection of AAV11.CreON.tdTomato into the LPB. The right side of the plot is the side contralateral to the brain injection. (B) Bar chart showing the proportion of retrogradely labelled cells in the L2, L3 and L4 segments of the Calb1Cre mice injected with AAV11.CreON.tdTomato into the LPB that received dense Trpm8 input. Repeated measures 1-way ANOVA showed a significant difference (p = 0.048), and post-hoc Holm-Šídák’s multiple comparisons test revealed a difference only between L2 and L4 (p = 0.0056). (C) Part of a horizontal section through the L2 segment of one of the mice, showing 4 lamina I neurons (numbered 1-4) that were retrogradely labelled with tdTomato (tdTom, magenta). (D) The same field scanned to reveal both tdTomato and GFP (green). (E-H) detailed views of the 4 cells shown in C,D. Three of these cells (1-3) have numerous contacts from GFP-labelled axons on their dendrites (marked with arrowheads) and cell bodies. The fourth cell (4) has very few such contacts. (I) Recordings from one of the cold-selective cells (upper traces) and the polymodal cell (lower traces) in experiments carried out in Calb1Cre mice that had received injections of AAV11.CreON.tdTomato into the CVLM. The cell in the upper traces shows action potential firing in response to application of cold saline to the skin, but no response to application of a 10g von Frey (VF) hair or hot saline. The other cell responds to brushing of the skin with a paint brush, as well as to application of hot and cold saline. (J) Pie chart showing response characteristics of the 6 cells recorded in these experiments. Scale bars: (C,D) = 50 µm, (E-H) = 25 µm. **P<0.01.

Experimental details for Calb1Cre mice used in anterograde tracing experiments.

Anterograde tracing following injections of AAV1.CreON.tdTomato into the medial dorsal horn of Calb1Cre mice

(A-C) Injection sites in the medial parts of the L3, L4 and L5 segments of one of the mice used in this part of the study. Immunostaining for tdTomato is shown in yellow and for NeuN in blue. Labelling of cell bodies is largely restricted to the medial part of the dorsal horn. There is some labelling in the lateral spinal nucleus (LSN), marked by asterisks, but this is likely to result from transport in axons of local interneurons. (D-F) Sections through selected regions of the brain immunostained to reveal tdTomato (yellow), with the corresponding dark-field images shown in blue. (D) At the level of the caudal medulla, the main ascending fibre bundle (arrowhead) lies in the ventrolateral region on the side contralateral to the spinal injections. Collateral axons innervate the nucleus of the solitary tract (NTS). (E) There is a dense projection of axons to the rostralmost part of the lateral parabrachial area (PBrel) and the caudal part of the PAG (cPAG) on the side contralateral to the injections. (F) Within the diencephalon there is sparse labelling in the contralateral ventral posterolateral nucleus (VPL) and posterior nucleus (Po) as well as within the medial thalamus. (G,H) The region indicated in the box in E is shown at higher magnification with immunostaining for Foxp2 shown in magenta. The overlap between arborisation of tdTomato and Foxp2-positive nuclei indicates that this region is indeed PBrel. (I, J) show higher magnification views of tdTomato-positive axons in the VPL and Po, respectively, corresponding to the boxes in (F). Numbers in D-F show approximate rostrocaudal locations in relation to the interaural line. Images in A-E, G-H are from animal #8, and those in F, I, J are from animal #7 in Table 1. Scale bars: (A-F) = 1 mm, (G-J) = 200 µm.

Proposed circuits involving cold-selective lamina I ALS neurons

Our results suggest that cold-selective lamina I projection neurons, which belong to the ALS3 cluster9 project, among other sites, to PBrel, the caudal part of the PAG (cPAG) and the ventral posterolateral nucleus of the thalamus (VPL). Cold-responsive cells in PBrel innervate the preoptic area and dorsomedial hypothalamus, which are integral components of circuits that underlie cold defence. The cPAG can indirectly activate brown adipose tissue (BAT) metabolism, also contributing to cold defence. The projection to VPL, through its connection to the primary somatosensory cortex (S1), presumably underlies conscious perception of cold stimuli applied to the skin.

Immunohistochemical characterisation of GFP- expressing somata in dorsal root ganglia of Trpm8Flp;RCE:FRT mice.

In each set of images GFP immunoreactivity is shown in green and other types of immunoreactivity in magenta, red or blue. (A-C) GFP and Vglut3 are present in largely separate populations, although some cells contain both proteins (one cell in this field, marked with an arrow). Two cells containing only GFP are indicated with arrowheads, and two cells containing only Vglut3 with double arrowheads. (D-F) There is a greater degree of overlap between GFP and Trpv1, and three double-labelled cells are shown with arrows. Some cells with only GFP or Trpv1 are indicated with single and double arrowheads, respectively. (G-I) GFP and somatostatin (Som) are found in separate populations, and some labelled cells are marked with single and double arrowheads, respectively. Note that somatostatin is seen as a partial ring around the nucleus in cells that express the peptide. (J-M) GFP cells often contain substance P (SP) but lack calcitonin gene-related peptide (CGRP), and 2 of these are marked with arrows. A single and double arrowhead mark cells with CGRP only and with both SP and CGRP, respectively. Scale bar: (A-M) = 100 μm.

Expression of Vglut3 in the central terminals of Trpm8 afferents.

(A-C) A low magnification confocal image from a transverse section of the L2 segment in a Trpm8Flp;RCE:FRT mouse immunostained for GFP (green) and Vglut3 (magenta). GFP-labelled axons are restricted to lamina I, while a dense band of Vglut3-immunoreactive terminals in the inner part of lamina II (marked with arrowheads) corresponds to the central terminals of C low-threshold mechanoreceptors. The dashed line in C shows the outline of the dorsal horn and the box indicates the area shown in (D- F). (D-F) At higher magnification a few Vglut3 profiles are also seen in lamina I (arrows) and these co-express GFP. (G-J) A horizontal section through lamina I from a Phox2a::Cre;Ai9;Trpm8Flp;RCE:FRT mouse immunostained to reveal tdTomato (tdTom, red), GFP (green) and Vglut3 (blue). This field includes a tdTomato labelled projection neuron that receives numerous contacts from GFP-labelled axons. Some of these are also immunoreactive for Vglut3 and these are seen more clearly (some marked with arrows) when only tdTomato and Vglut3 are revealed (J). The insets in J show two of these Vglut3 profiles (those marked with arrows numbered 1 and 2) contacting dendrites of the projection neuron in limited z-projections (2 optical sections at 0.5 μm z-spacing). In these images the coexpression of Vglut3 (blue) and GFP (green) is clearly visible. Scale bars: (A-C) = 100 µm, (D-F) = 50 µm, (G-J) = 50 µm, (insets in J) = 2 µm.

The relationship between tdTomato and GFP labelling for the cell illustrated in

Figure 3. The top pane shows immunostaining for tdTomato (tdTom, magenta) in a horizontal section through lamina I from a Phox2a::Cre;Ai9;Trpm8Flp;RCE:FRT mouse. The tdTomato-labelled projection neuron illustrated in Figure 3 is visible. The soma is marked with an asterisk and parts of the dendritic tree with arrowheads. The cell body and dendrites are associated with dense bundles of GFP-labelled axons. Scale bar = 50 μm.

Confocal and electron microscope images of two Trpm8-innervated cells examined with the combined method in spinal cord sections from a Phox2a::Cre;Ai9;Trpm8Flp;RCE:FRT mouse.

(A) Projected confocal z-stack showing parts of the two cells (numbered 1, 2) labelled with tdTomato (magenta) in a horizontal section. (B) a single optical section from the same confocal z-stack showing the association between axons labelled with GFP (green) and the cell body of cell 1 and dendrites of both cells. (C-J) Examples of synapses between the GFP-containing boutons, which are labelled with an immunoperoxidase method, and (unlabelled) dendrites (D) or soma (S) of the tdTomato-positive projection neurons. C-F show synapses on cell 1, and G-J those on cell 2. Presumed synaptic active zones (marked by arrowheads) can be recognised by the presence of vesicles on the presynaptic side and darkening of the membrane. However, these synapses do not show prominent postsynaptic densities. Scale bars: (A,B) = 50 μm, (C-J) = 500 nm.

Recording strategy and characteristics of spontaneous EPSCs in cold-selective and other lamina I ALS cells.

(A) Cells were identified by expression of mCherry following injection of AAV9.mCherry into the CVLM or LPB of Trpm8Flp;RCE:FRT mice, or by expression of tdTomato following injection of AAV9.Cre_GFP into the CVLM of Trpm8Flp;RCE:FRT;Ai9 mice. Note that the level of GFP expression resulting from injection of AAV9.Cre_GFP was extremely low, and would have been restricted to the nucleus of labelled neurons, due to the presence of a nuclear localisation signal. In both cases, Trpm8-expressing afferents were labelled with GFP. (B) The semi-intact somatosensory preparation retained skin attached to lumbar spinal cord through intact peripheral nerves and was used for recording responses to natural skin stimuli. (C) Representative traces from an ALS neuron with dense Trpm8 input (left) and an ALS neuron without dense Trpm8 input (right). (D, E) Quantification of sEPSC frequency (D) and amplitude (E) of ALS neurons. Box plots are median and interquartile range with symbols representing data points from individual ALS neurons (with dense Trpm8 input: n=6, without dense Trpm8 input: n=5; * indicates p < 0.05, Mann-Whitney U test). (F) Overlay of the firing time course of 3 cold-selective cells in response to cold, heat and mechanical stimulation. Individual traces shown in grey and averages in black.

Further characterisation of Calb1-expressing projection neurons.

(A) UMAP plot showing the distribution of cells in the ALS1-5 clusters, together with a plot for expression of Calb1. Note the presence of many Calb1- positive cells in the ALS3 cluster. (B) The experimental approach used to generate data presented in this figure and in Figure 5. (C) An example of a section through the L5 segment from a Calb1Cre mouse injected with AAV11.CreON.tdTomato in the LPB. Expression of tdTomato (red) is superimposed on a dark field image. (D-F) The distribution of retrogradely labelled cells in 9 (D,F) or 8 (E) transverse sections from the L5 segments of the remaining 3 Calb1Cre mice injected with AAV11.CreON.tdTomato in the LPB. The labelling for the other mouse is shown in Figure 5A. (G, H) sEPSC frequencies and amplitudes for the 5 cold-selective cells and the 1 polymodal cell recorded in Calb1Cre mice injected with AAV11.CreON.tdTomato in the CVLM. Note that although 2 of the cold- selective cells in this sample had relatively high sEPSC frequencies, when data from these cells were pooled with those reported for other sample of cold-selective cells (shown in Figure 4–Figure supplement 1) and compared with the polymodal cells shown in that Figure supplement, there was still a significant difference for both sEPSC frequency (p = 0.037) and amplitude (p = 0.015) (Mann-Whitney U test). (I) Overlay of the firing time course of 5 cold-selective cells in response to cold, heat and mechanical stimulation. Individual traces shown in grey and averages in black. Scale bar: (C) = 500 µm.

Anterograde labelling following injections of AAV1.CreON.tdTomato into the central part of the spinal dorsal horn of Calb1Cre mice.

(A) The experimental approach used to generate data for Figure 6 and for Figure 6– figure supplements 1-3. Intraspinal injections were made into either the L3 or the L3, L4 and L5 segments, and were targeted on either the central or medial part of the dorsal horn. (B) Example of an injection into the central part of the L4 segment of a Calb1Cre mouse. Immunostaining for tdTomato is shown in yellow and for NeuN in blue. There is extensive tdTomato labelling throughout most of the dorsal horn, and this extends into the lateral spinal nucleus (arrowhead), which contains tdTomato-labelled cell bodies. (C-E) Sections through selected regions of the brain have been immunostained to reveal tdTomato (yellow), and these are shown on corresponding dark-field images (blue). (C) There is strong labelling in the rostralmost part of the contralateral LPB, corresponding to the PBrel, as well as in the caudal PAG (cPAG). Unlike the pattern seen with injections restricted to the medial dorsal horn, there is also strong labelling in the contralateral superior colliculus (SC). (D) At a slightly more caudal level there is axonal labelling in several subnuclei of the contralateral LPB, including external lateral (el), dorsolateral (dl), central lateral (cl) and internal lateral (il), but not ventrolateral (vl). The location of the superior cerebellar peduncle (SCP) is indicated. (E) Within the diencephalon there is sparse axonal labelling in contralateral VPL and posterior (Po) thalamic nuclei, as seen following medial injections. However, there is also dense labelling in the medial thalamus, which was not seen with medial injections. Numbers in C-E show approximate rostrocaudal locations in relation to the interaural line. All images are from animal #2 in Table 1. Scale bars: (B,C,E) = 1 mm, (D) = 500 µm.

Anterograde labelling following injections of AAV.CreON.tdTomato into the medial dorsal horn of Calb1Cre mice.

(A) A section through the C7 segment of a mouse that received injections into L3, L4 and L5 segments. Immunostaining for tdTomato is shown in yellow and for NeuN in blue. There is a bundle of axons (arrow) in the contralateral dorsolateral white matter, with a much more weakly stained bundle in the same location on the opposite side. A few collateral branches are present in the deep dorsal horn and the area around the central canal. (B,C) Sections through selected regions of the brain have been immunostained to reveal tdTomato (yellow), and this is shown on corresponding dark-field images (blue). (B) At this level of the contralateral LPB, there is axonal labelling in the external lateral (el) and dorsolateral (dl) subnuclei, with very little labelling in ventrolateral (vl), central lateral (cl) or internal lateral (il). The location of the superior cerebellar peduncle (SCP) is indicated. (C) At a level corresponding to ∼ 0.4 mm rostral to the interaural line there is very sparse axonal labelling in the posterior triangular nucleus of the thalamus (PoT) and in the rostral PAG (rPAG) on the side contralateral to the spinal injection. This is shown at higher magnification in (D) and (E), which correspond to the boxes in (C). Numbers in B and C show approximate rostrocaudal locations in relation to the interaural line. Images in A and C-E are from animal #7, and that in B from animal # 8 in Table 1. Scale bars: (A,C) = 1 mm, (B,D,E) = 500 µm.

Retrograde labelling of Calb1-expressing ALS neurons from PAG.

(A) TdTomato labelling at the site of injection of AAV11.CreON.tdTomato in one of the 3 Calb1Cre;Trpm8Flp;RCE:FRT mice used in this part of the study. Note that this will not reveal the precise location of the injection, because only nearby Calb1-expressing cells will be labelled. (B-D) Labelling for tdTomato (tdTom) and GFP in a horizontal section through lamina I from the case illustrated in (A). Several tdTom-positive neurons are visible in this field, and all of them are densely coated with GFP-labelled (Trpm8-positive) axons. (E) Histogram showing the percentages of retrogradely labelled neurons in the L2-L4 segments that were densely coated with Trpm8- positive axons in this set of experiments (PAG), compared with the corresponding values from the 4 cases in which AAV11.CreON.tdTomato was injected into the LPB. There was a highly significant difference between the percentages between the two injection sites (p < 0.0001, unpaired t-test with Welch’s correction). Scale bars: (A) = 1 mm, (B-D) = 100 µm.

Identification of spinothalamic lamina I neurons with dense Trpm8 input.

(A-C) CTB injection sites in the 3 experiments. In each case there is dense labelling in the ventral posterolateral (VPL) and ventral posteromedial (VPM) nuclei, together with some labelling in the posterior (Po) nucleus, and in two cases also in the ventrolateral (VL) nucleus. Note the lack of labelling in medial thalamic nuclei (centrolateral, CL; centromedial, CM; mediodorsal, MD). Numbers show approximate rostrocaudal locations in relation to the interaural line. (D-I) Examples of retrogradely labelled (CTB-positive) lamina I neurons (marked with asterisks) that are associated with numerous GFP-labelled (Trpm8-expressing) axons in horizontal sections through the C7 segments in the experiments illustrated in A and B. (J-L) In the experiment illustrated in C, we also identified 5 CTB-labelled cells in the L4 segment, and one of these, which is associated with numerous GFP-labelled boutons, is shown here (asterisk). Note that we also found CTB-labelled cells associated with numerous GFP-positive boutons in the C7 segment in this experiment. In each case, the images are shown below the injection site from the corresponding experiment. Scale bars: (A-C) = 1 mm, (D-L) = 50 µm.

UMAP plot showing the distribution of cells in the ALS1-5 clusters, together with plots for expression of Trhr and Calcrl.

Note the presence of both Trhr and Calcrl in several of the ALS clusters, including ALS3.