Three‐dimensional topography of rat trigeminal ganglion neurons using a combination of retrograde labeling and tissue‐clearing techniques

The trigeminal nerve is the sensory afferent of the orofacial regions and divided into three major branches. Cell bodies of the trigeminal nerve lie in the trigeminal ganglion and are surrounded by satellite cells. There is a close interaction between ganglion cells via satellite cells, but the function is not fully understood. In the present study, we clarified the ganglion cells’ three‐dimensional (3D) localization, which is essential to understand the functions of cell–cell interactions in the trigeminal ganglion. Fast blue was injected into 12 sites of the rat orofacial regions, and ganglion cells were retrogradely labeled. The labeled trigeminal ganglia were cleared by modified 3DISCO, imaged with confocal laser‐scanning microscopy, and reconstructed in 3D. Histograms of the major axes of the fast blue‐positive somata revealed that the peak major axes of the cells innervating the skin/mucosa were smaller than those of cells innervating the deep structures. Ganglion cells innervating the ophthalmic, maxillary, and mandibular divisions were distributed in the anterodorsal, central, and posterolateral portions of the trigeminal ganglion, respectively, with considerable overlap in the border region. The intermingling in the distribution of ganglion cells within each division was also high, in particular, within the mandibular division. Specifically, intermingling was observed in combinations of tongue and masseter/temporal muscles, maxillary/mandibular molars and masseter/temporal muscles, and tongue and mandibular molars. Double retrograde labeling confirmed that some ganglion cells innervating these combinations were closely apposed. Our data provide essential information for understanding the function of ganglion cell–cell interactions via satellite cells.


INTRODUCTION
The trigeminal nerve is divided into three branches, each innervating a different orofacial region (Waite, 2003); the ophthalmic (first branch), maxillary (second branch), and mandibular divisions (third branch).
Depending on their location, cranial tissues are innervated by one of the three branches.In rats, as in other animals, the first branch supplies the dorsal part of the head, upper eyelid, supraorbital vibrissae, the cornea and conjunctiva, the glabrous and hairy skin over the dorsal and tip of the nose, and the intranasal mucosa.The second branch supplies the postorbital skin, the upper lip, mystacial vibrissae, the lateral nose, the intraoral upper jaw mucosa, and the upper teeth.The third branch supplies the temporomandibular joint, the external auditory meatus, the jaw muscles, the skin over the mandible and lower lip, the intraoral lower jaw mucosa, the lower teeth, and the anterior tongue.Trigeminal afferents from all three branches supply the dura mater and cranial blood vessels (Andres et al., 1987).Cell bodies of trigeminal afferents lie in the trigeminal ganglion (except for the mesencephalic trigeminal nucleus).Neurons within the trigeminal ganglion are organized roughly and somatotopically, with the cells innervating the ophthalmic division lying anteromedially and those supplying the mandibular division lying posterolaterally.Previous studies agree that somatotopy is only approximate with a considerable intermingling of somata from adjacent regions (Kawabata, 1981;Klein et al., 1986;Martin & Dolivo, 1983).However, no three-dimensional (3D) and quantitative data exist on which afferent neurons of the orofacial regions are intermingled and to what extent.
Trigeminal ganglion cell bodies are pseudounipolar and are surrounded by satellite cells.Within the trigeminal ganglion, interactions between ganglion cells via satellite cells are considered to be involved in related pain and ectopic allodynia that feel pain in areas distant from the causative site, such as inflammation and nerve injury (Goto et al., 2016;Gungikake et al., 2009;Messlinger & Russo, 2019;Shinoda et al., 2019).Mediators, such as nitric oxide, adenosine triphosphate (ATP), and nerve growth factors, are released from neuronal cell bodies due to inflammation and/or nerve injury and activate surrounding satellite cells and adjacent ganglion cells (Hanani & Spray, 2020;Kunkler et al., 2014;Liverman et al., 2009;Ohara et al., 2013;Rozanski et al., 2012;Shinoda et al., 2011;Sugiyama et al., 2013).Interactions between ganglion cells and satellite cells via gap junctions have also been observed (Garrett & Durham, 2008;Hanani & Spray, 2020;Kaji et al., 2016;Spray et al., 2019;Thalakoti et al., 2007;Vit et al., 2006).Animal models must be created to verify whether the crosstalk between ganglion cells and satellite cells via such mediators and gap junctions is the mechanism of allodynia.To this end, it is essential to obtain detailed 3D localization data of trigeminal ganglion cells.
In most previous studies, after injecting a retrograde tracer into the orofacial regions or performing partial axotomies of the trigeminal nerve, the harvested trigeminal ganglion was sliced into thin sections approximately 30 µm thick, and the localization of tracer-positive or degenerated neurons were observed in these sections (Gregg & Dixon, 1973;Kawabata, 1981;Klein et al., 1986;Mazza & Dixon, 1972).
However, reconstructing thin sections of the trigeminal ganglia is time-consuming and risks overestimating the number of ganglion cells.
Therefore, in the present study, we quantitatively elucidated the 3D localization of ganglion cells by combining a retrograde tracing method and a tissue-clearing method to the rat trigeminal ganglia to reduce the number of slices as much as possible.

Animals
All procedures involving animals were performed in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals.Rats were maintained under a 12 h light/dark cycle and provided access to food and water ad libitum.The experiments were approved by the Committees for Animal Experiment (D19005) and the Recombinant DNA Study (S29017) at Kagoshima University (Kagoshima, Japan).We used 58 Sprague-Dawley rats (males n = 31, females n = 27; 250-350 g; Shizuoka Laboratory Animal Center, Shizuoka, Japan).Because previous studies have reported that there are no differences in the number and distribution of retrogradely labeled ganglion cells between male and female rats (Ambalavanar et al., 2003), both sexes were used (but see Ibraham et al., 2023).
There was no apparent difference in the number of fast blue-labeled cells between male and female rats (Supporting Information Figure S1).Therefore, both male and female trigeminal ganglion samples were mixed and used for the analysis.All efforts were made to minimize animal suffering and reduce the number of animals used in the present study.

Injection of retrograde tracers and tissue preparation
Information Figure S2).To label afferent neurons of the bilateral corneas, we inserted a Hamilton syringe needle into the corneal epithelium and injected 0.2 µL of 2% fast blue solution at five different locations (1.0 µL/eye in total).After the needle was removed, the overflow solution was removed with a cotton swab to prevent the uptake of fast blue into the surrounding tissue.To label afferent neurons of the vibrissa follicles, the skin posterior to the whisker pad was incised to expose the infraorbital nerve fiber bundles projecting to the vibrissa follicles.To prevent uptake of the tracer by the surrounding tissue, the nerve fiber bundle was wrapped in Parafilm, a drop of 1.0 µL of 2% fast blue solution was applied, and the tracer was imprinted onto the nerve fiber bundle by pinching the nerve fiber bundle several times with tweezers.After 20 min of exposure, any tracer not incorporated into the nerve fibers was swabbed off, and the skin was sutured.To label afferent neurons of the mucosa of the palate and tongue, a Hamilton syringe needle was inserted intramucosally, and 0.2 µL 2% fast blue solution was injected at each site.The palate was injected bilaterally at two sites (0.4 µL in total) and the tongue bilaterally at four sites (0.8 µL in total).To prevent overflow as much as possible, the needle was left in place for 5 min after injection and then removed.If overflow was observed, it was immediately swabbed off to prevent uptake by the surrounding tissue.We drilled down the center of each molar crown and exposed the pulp cavity to label afferent neurons of the papilla of the maxillary/mandibular first and second molars.A drop of 4% fast blue solution, 0.2 µL, was placed in each cavity, and these cavities were restored with glass ionomer cement (Fuji II LC, GC).To label afferent neurons of the temporal or masseter muscles, we shaved the skin covering these muscles and made a 3-mm-wide incision.We exposed each muscle, inserted a Hamilton syringe needle into it, and injected 1.0 µL of fast blue solution per location into four locations (4.0 µL per muscle).To prevent overflow as much as possible, the needle was left in place for 5 min after injection and then removed.If overflow was observed, it was immediately swabbed off to prevent uptake by the surrounding tissue, and the skin was sutured.To label afferent neurons of the temporomandibular joint, we shaved the skin covering it and made a 5-mm-wide incision.The temporal muscle attached to the temporal bone was partially detached to expose the articular capsule of the joint.A Hamilton syringe needle was inserted into the joint capsule, and 1.0 µL 2% fast blue solution was injected.Five minutes after the injection, the needle was removed and checked for drawbacks; any drawback present was swabbed off, and the skin was sutured.To label afferent neurons of the dura mater, we shaved the scalp hair and made a 2 cm longitudinal incision in the skin.A hole about 3 mm wide and 7 mm long was drilled into each side of the skull to expose the dura mater.A piece of Spongel (LTL Pharma, Tokyo) soaked with 3.0 µL 2% fast blue solution was placed over the dura mater and then covered with parafilm.To prevent the tracer from being taken up by surrounding tissue, the top of the parafilm was completely covered with glass ionomer cement (Fuji II LC).These rats survived for 1 week after the fast blue injection.
We used an adeno-associated virus vector (AAV), AAV PHP.S-CAG-tdTomato, which targets peripheral nerves (Chen et al., 2017), and a 2% fast blue solution for double-fluorescent retrograde labeling.Eight 4-week-old rats (males n = 2; females n = 6) were anesthetized as described above, and 2.0 µL of AAV PHP.S-CAG-tdTomato (1.0 × 10 13 GC/mL) was injected into the lingual mucosa (n = 1), scalp (n = 2), masseter muscle (n = 2), and temporal muscle (n = 3).Three weeks after injection, the rats were anesthetized again.The rats first injected in the lingual mucosa were subsequently injected with fast blue into the mandibular molar pulp; those first injected in the scalp were subsequently injected in the upper eyelid and dura mater; those first injected in the masseter muscle were injected in the lingual mucosa and mandibular molar pulp; and those first injected in the temporal muscle were injected in the lingual mucosa and maxillary/mandibular molar pulp.These eight rats survived for 1 week after the fast blue injection.
Fouty-eight fast blue-injected rats, eight fast blue-and AAV-injected rats, and two intact rats were deeply anesthetized with a mixture of medetomidine (0.3 mg/kg), midazolam (4.0 mg/kg), and butorphanol (5.0 mg/kg), and transcardially perfused with 50 mL 5 mM sodium phosphate-buffered 0.9% saline (PBS; pH 7.4), followed by 200 mL of 3% formaldehyde in 0.1 M PB.The bilateral trigeminal ganglions were harvested and then washed in the PBS.After cryoprotection with 30% sucrose in PBS, the retrogradely labeled trigeminal ganglia were cut into 500-µm-thick or 50-µm-thick horizontal sections on a freezing microtome, and these were serially collected in PBS.
The two left-side trigeminal ganglia from the intact male and female rats (two in total) were used to create a standardized trigeminal ganglion contour.After perfusion, the trigeminal ganglia were embedded in 10% (g/v) gelatin solution in PBS and postfixed overnight at 4 • C in the same fixative.They were then cut into 100-µm-thick horizontal sections on a freezing microtome.The sections were incubated overnight with 0.1 µg/mL 4' , 6-diamidino-2-phenylindole (DAPI) in PBS containing 0.3% Triton X-100 at 24 • C.

Tissue clearing with modified 3DISCO
We tried several types of tissue-clearing methods and found that the rat trigeminal ganglia were rapidly cleared using 3DISCO (Ertürk et al., 2012(Ertürk et al., , 2014)), and the fast blue fluorescent signal did not fade.We modified the original 3DISCO protocol (Ertürk et al., 2012(Ertürk et al., , 2014) ) by pretreatment with iDISCO (Renier et al., 2014) to make the rat trigeminal ganglion more transparent.The protocol details are listed in Table 1.

Imaging of rat trigeminal ganglion neurons
The trigeminal ganglia of the tracer-injected rats were observed under a confocal laser scanning microscope by illuminating the specimens with a single laser beam and collecting the images using appropriate emission filters for fast blue and DAPI (excitation 405 nm, emission bandpass filter 420-480 nm), and for tdTomato (excitation 543 nm, emission > 630 nm).At each wavelength, confocal images (1024 × 1024 pixels in size with an 8-bit pixel depth) were obtained twice and averaged to reduce noise.
The cleared horizontal slices containing fast blue-positive trigeminal ganglion neurons were imaged using an LSM900 confocal laserscanning microscope (Zeiss) for 3D reconstruction of the rat trigeminal ganglia under the following two conditions; "Tile Scan mode" with TA B L E 1 Modified 3DISCO.Incubation steps for clearing of the rat trigeminal ganglia.

DAY 1
(1) Wash 500 µm thick slices of the rat trigeminal ganglia for 30 min with PBS (5 mL/sections in a plastic six-well plate).
(4) Incubate for at least 12 h with 5% (w/v) H 2 O 2 in methanol (5 mL/sections in a plastic six-well plate).
(7) Incubate for 2 h in PBS-X (5 mL/sections in a plastic six-well plate).
(8) Incubate for at least 12 h with 50% (v/v) THF in DW (5 mL/sections in a glass petri dish; within the gap between two glass slides with a 500 µm spacer between them a ).

DAY 3
(9) Incubate for 1 h with 80% (v/v) THF in DW (5 mL/sections in a glass petri dish; within the gap between two glass slides with a 500 µm spacer between them a ).
(10) Incubate for 1 h in 100% THF twice (5 mL/sections in a glass petri dish; within the gap between two glass slides with a 500 µm spacer between them a ).
(11) Incubate for 4 h in 100% DCM (5 mL/sections in a glass petri dish; within the gap between two glass slides with a 500 µm spacer between them a ).
(12) Incubate for more than 1 h in 100% DBE (5 mL/sections in a glass petri dish; within the gap between two glass slides with a 500 µm spacer between them a ).
(13) The cleared trigeminal ganglia were placed in a glass-bottomed dish filled with DBE for image acquisition.
a Tissues shrink and harden in THF solution.Therefore, tissues are lightly pressed by two glass slides with a 500 µm spacer between them to be flattened.a 10× objective lens (Plan Apochromat, numerical aperture = 0.45; Zeiss), the pinhole set at maximum, and a combination of "Tile Scan mode" and "z-stack mode," with a 20× objective lens (Plan Apochromat, numerical aperture = 0.8, working distance = 0.55 mm; Zeiss), the pinhole set at 2.34 Airy Unit (pinhole 55 µm), optical slice thickness of 2.0 µm, and a zoom factor of 0.45.Using the 20× objective lens, the LSM900 took a maximum of 120 images with a z-interval of 3.0 µm per stack (360 µm thick) in the Tile Scan and z-stack modes and fused the partially overlapping maximum of 21 tiles into a large z-stack image.
The 100-µm-thick trigeminal ganglion sections from the intact rats were stained with DAPI and imaged using an LSM900 confocal laserscanning microscope to develop a standardized rat trigeminal ganglion contour under the following condition; "Tile Scan mode" with a 10× objective lens (Plan Apochromat, numerical aperture = 0.45; Zeiss), the pinhole set at maximum.The captured images were traced in 3D using a Neurolucida 10 software (MBF bioscience).The procedure for 3D reconstruction is described in detail in the Section 3.

Measurement of the size of fast blue-positive cells and the volume in which the cells are distributed in the rat trigeminal ganglion
To measure the size of fast blue-positive ganglion cells, we used z-stack images, acquired using a confocal laser scanning microscope (Zeiss) with a 20× objective lens (Plan Apochromat, numerical aperture = 0.8, working distance = 0.55 mm; Zeiss) and a combination of "Tile Scan mode" and "z-stack mode."We measured the major axes of the fast blue-positive ganglion cells at the z-stack image with the largest area of the ganglion cells and corrected for shrinkage due to tissue clearing (see the Section 3 for details).The major axes were measured for all fast blue-positive cells within the trigeminal ganglia.
The number of trigeminal ganglion samples used for the measurements was as follows: for the upper eyelid, maxillary molar, temporal muscle, tongue, mandibular molar, and masseter muscle, n = 3; for the infraorbital nerve, n = 4; for the Scalp, cornea, palate, dura mater, and temporomandibular joint, n = 5.
Using 3D plot data of the location of fast blue-positive ganglion cells, the volume encompassing all the fast blue-positive cells was measured in each trigeminal ganglion.We used MATLAB built-in function "alphaShape" (MATLAB, version R2022b, MathWorks, Massachusetts, USA) to create polyhedra that encompassed all fast blue-positive cells within each trigeminal ganglion and measured their volume.For each injection site, polyhedra were created, and their volumes were measured for each of the five trigeminal ganglion samples (12 injection sites × 5 samples = 60 polyhedra).In addition, the volumes of the overlapping portions of these polyhedra were measured using MATLAB.

Statistical analysis
Sample sizes for the morphological analyses were deduced from previously published studies (Aker & Reith, 1981;Emrick et al., 2020;Launay et al., 2015).We used GraphPad Prism 9 (GraphPad Software, SanDiego, CA) or Excel (Microsoft, Redmond, WA) for statistical analysis, such as Tukey's post hoc multiple comparison tests following a oneor two-way analysis of variance (ANOVA) and a Student's t-test.p < 0.05 was considered statistically significant.

Combination of tissue clearing technique and retrograde tracing methods for the rat trigeminal ganglion
The rat trigeminal ganglia were treated using the modified 3DISCO method (Table 1) to examine clearing capabilities and were comparatively imaged for light transmission before and after (Figure 1a).The process of modified 3DISCO took 3 days, and the treated trigeminal ganglia showed high-quality transparency.In the present study, we used fast blue for retrograde labeling of rat trigeminal ganglion cells.
Thus, we checked whether the fluorescence signal of fast blue was retained after tissue clearing using the modified 3DISCO.Concerning the trigeminal ganglion in which a number of cell bodies were labeled by injecting fast blue into the infraorbital nerve, we compared whether the fluorescent signal changed before and after clearing by imaging the same cells with a confocal laser-scanning microscope.The fluorescence signals of the same fast blue-positive cells were retained in the modified 3DISCO treated trigeminal ganglion (see the XY-planes of "Before" and "After" in Figure 1b).Furthermore, tissue clearing with the modified 3DISCO enabled us to observe fluorescence signals from deep tissue (see the XZ-and YZ-planes of "Before" and "After" in Figure 1b).
After tissue clearing, fast blue-positive cells could be identified even in tissue deeper than 300 µm.
It has been reported that tissue size shrinks by clearing with 3DISCO (Ertürk et al., 2012;Richardson et al., 2015).We did observed shrinkage by modified 3DISCO at both the levels of the whole trigeminal ganglion (Figure 1a) as well as at the cellular level (Figure 1b).
In Figure 1b, the white dotted rectangle in the XY plane of "After" is the area shown in the XY plane of "Before."Because the tissue has shrunk, a wider area was observed in "After" than in "Before," even though the images were taken with the same objective lens and displayed on the same scale before and after clearing.We further noticed that the tissue contracted at different rates in the X-(mediolateral axis), Y-(anteroposterior axis), and Z-directions (dorsoventral axis).Thus, we measured the proportion of shrinkage in 3D and checked whether the shrinkage distorted the arrangement of cells in the rat trigeminal ganglion.After treatment with the modified 3DISCO, the size of the whole trigeminal ganglion shrank to 58±0.6% (mean ± SD) of its original size along the mediolateral axis, 80±1.9%along the anteroposterior axis, and 53±0.9%along the dorsoventral axis (Figure 1c; statistical analysis, one-way ANOVA and Tukey).Similarly, the size of ganglion cell bodies shrank to 57±6.6% of their original size, in the mediolateral axis, 80±7.7% in the anteroposterior axis, and 53±7.4% in the dorsoventral axis (Figure 1d; statistical analysis, one-way ANOVA and Tukey).Thus, the rat trigeminal ganglion was revealed to shrink strongly in the mediolateral and dorsoventral axis but moderately in the anteroposterior axis.Based on these results, we tried to compensate for the shrinkage; specifically, the size of z-stack images of the cleared trigeminal ganglion (Figure 1b, "After") was divided by 0.58, 0.80, and 0.53 in mediolateral, anteroposterior, and dorsoventral axis, respectively (Figure 1e, "After").Since the compensated image (Figure 1e, "After") was similar to the original image (Figure 1e, "Before"), it was suggested that the arrangement of cell bodies was not distorted by tissue clearing with modified 3DISCO.

Three-dimensional reconstruction of the retrogradely labeled trigeminal ganglion
For each injection site, fast blue was injected into both sides of four rats, and eight trigeminal ganglion samples were obtained.Of the eight samples, five samples per injection site were used for 3D reconstruction, excluding those with weak fast blue labeling and those with extremely low and high numbers of labeled cells.All of the procedures of 3D reconstruction are shown in Figure 2. The maximum imaging depth was 550 µm because of limitations of the 20× objective lens (Plan Apochromat, numerical aperture = 0.8, working distance = 0.55 mm; Zeiss).Therefore, each whole rat trigeminal ganglion, with a thickness of around 1.5 mm, was horizontally sectioned into three 500-µm-thick sections.These were cleared and imaged using an LSM900 confocal laser-scanning microscope (Zeiss).
The locations of the retrogradely labeled cell bodies were manually mapped in 3D using the Neurolucida 10 software.Briefly, contours of the trigeminal ganglia were traced using images of the entire trigeminal ganglion taken under "tile scan mode" with a 10× objective lens.Images of fast blue-positive cell bodies taken with a 20× objective lens using a combination of "Tile Scan mode" and "z-stack mode" were aligned on the images of the entire trigeminal ganglion.The 3D localization of fast blue-positive cell bodies was then marked using the aligned z-stack images (such that each cell body was marked as a single point in the 3D space; yellow dots in Figure 2).Data from all sections were aligned and integrated using the bifurcation points of the first/second branch and third branch, and the protuberance of the third branch.Finally, the size of the reconstructed trigeminal ganglion was divided by 0.58, 0.80, and 0.53 in the mediolateral, anteroposterior, and dorsoventral axes, respectively, for compensation of shrinkage.
To analyze the distribution of all fast blue-positive cells in a single rectangular coordinate system, the 60 trigeminal ganglia reconstructed in 3D were aligned to a single standardized trigeminal ganglion.The standardized trigeminal ganglion is the left side ganglion.The reconstructed right trigeminal ganglia were inverted and superimposed on the standardized left trigeminal ganglion.

F I G U R E 1
Clearing of retrogradely labeled rat trigeminal ganglion using the modified 3DISCO method.(a) Comparison of the rat trigeminal ganglion before and after clearing.The rat trigeminal ganglion on the left side is shown.The modified 3DISCO made the rat trigeminal ganglion transparent.A, anterior; L, lateral; M, medial; P, posterior.(b) Z-stack images of the rat trigeminal ganglion labeled with fast blue before and after clearing.Injection of fast blue into the infraorbital nerve retrogradely labeled numerous cell bodies in the trigeminal ganglion (b; Before).The same trigeminal ganglion was treated by modified 3DISCO (b; After).The fluorescence of fast blue was not attenuated by the clearing procedure.Before clearing, fast blue-positive cells could only be observed 20 µm deep from the surface of the trigeminal ganglion.However, after clearing, fast blue-positive cells were visible even at 300 µm deep from the surface.The X-direction corresponds to the mediolateral axis, the Y-direction corresponds to the anteroposterior axis, and the Z-direction corresponds to the dorsoventral axis.D, dorsal; V, ventral.We measured the shrinkage rate in the mediolateral axis (M-L), anteroposterior axis (A-P), and dorsoventral axis (D-V), for the whole trigeminal ganglia (c) and for the individual fast blue-positive cell bodies (d).The whole trigeminal ganglia and individual ganglion cells showed a similar shrinkage ratio; A-P shrinkage was less than M-L and D-V shrinkage (c, d; the original size is 100%).When the shrinkage was adjusted (e; After), the size and arrangement of fast blue-positive cells were similar to those of the original one (e; Before).In (e), the "20 µm" indicates that the image is 20 µm deep from the surface, and "Merge" means that the image is projected z-stack image from the surface to a depth of 300 µm.Scale bar in a = 2 mm, in b = 100 µm (applies to b, e).

Three-dimensional localization of retrogradely labeled cell bodies in the rat trigeminal ganglion; the first division
To clarify the localization of neurons in the rat trigeminal ganglion, a retrograde tracer, fast blue, was injected into the scalp, upper eyelid, and corneal mucosa, innervated by the first division of the trigeminal nerve.Five trigeminal ganglia per injection site were reconstructed in 3D. Figure 3a-c Overview of the three-dimensional reconstruction of retrogradely labeled cells in the rat trigeminal ganglia.The rat trigeminal ganglion (approximately 1.5 mm thick) was cut horizontally into three 500 µm thick slices, which were cleared using the modified 3DISCO method and imaged with a confocal laser-scanning microscope.We used a 10× objective lens to capture an image of the entire trigeminal ganglion slice and a 20× lens to capture z-stack images of areas where fast blue-positive cell bodies existed.Based on the confocal laser-scanning microscopy images, the contour of the trigeminal ganglion and the distribution of fast blue-positive cells were reconstructed in 3D using the Neurolucida 10 software.After correction for shrinkage, the reconstructed rat trigeminal ganglion was aligned to a standardized left-side trigeminal ganglion contour to compare the 3D topography.
brightness.Trigeminal ganglion cells projecting their axons to the first branch were concentrated in the dorsal part within the trigeminal ganglion.Ganglion neurons that project to the scalp were evenly distributed in the mediolateral axis, whereas those that project to the upper eyelid and cornea were more medially distributed.

3.4
Three-dimensional distribution of retrogradely labeled cell bodies in the rat trigeminal ganglion; the second division

3.5
Three-dimensional distribution of retrogradely labeled cell bodies in the rat trigeminal ganglion; the third division  S3.The five reconstructed trigeminal ganglia were superimposed, with areas of higher cell body density indicated by brighter colors; the lightest color is the area where the distance between cells is within 50 µm, the intermediate color is the area where the distance between cells of 50-100 µm, and the darkest color is the area that contains all of the cells (d-f).The labeled cells projecting to the ophthalmic division were densely distributed in the dorsal part of the trigeminal ganglion.The scale bar in a = 1 mm (applies to a-f).

3.6
The number of fast blue positive cells and the volume in which the cells are distributed in the rat trigeminal ganglion The numbers of fast blue-positive cells in the trigeminal ganglia are shown in Figure 6a.Injection of fast blue into the infraorbital nerve, which mainly projects to the vibrissa follicles, labeled a remarkably large number of cell bodies (∼2000, 400 in other areas).Rats obtain information about their surroundings utilizing vibrissa, and therefore, the number of ganglion cells innervating the vibrissa is considered to be high.
Polyhedra containing all fast blue-positive cells were created for individual trigeminal ganglion (Figures 3a-c, 4a-c, 5a-c, and 5g-i, Supporting Information Figures 3-5) and their volumes were measured F I G U R E 4 Three-dimensional distribution of retrogradely labeled cells in the trigeminal ganglion produced by fast blue application to the infraorbital nerve, palate, and maxillary molar teeth-maxillary division.A representative of the reconstructed trigeminal ganglion of the rats that received a fast blue injection into the infraorbital nerve (a), palate (b), or maxillary molar teeth (c).Each dot represents one fast blue positive cell body in a-c.Five trigeminal ganglia were reconstructed for each injection site.The remaining four trigeminal ganglia are shown in Supporting Information Figure S4.The five reconstructed trigeminal ganglia were superimposed, with areas of higher cell body density indicated by brighter colors; the lightest color indicates a distance between cells within 50 µm, the middle color indicates a distance between cells of 50-100 µm, and the darkest color contains all cells (d-f).Cells projecting axons to the infraorbital nerve were densely distributed in the center of the trigeminal ganglion (a, d).Cells projecting axons to the palate were localized ventrally (b, e).Cells projecting axons to the maxillary molars were distributed laterally (c, f).The scale bar in a = 1 mm (applies to a-f).

F I G U R E 5
Three-dimensional distribution of retrogradely labeled cells in the trigeminal ganglion produced by fast blue application to the dura mater, temporal muscle, temporomandibular joint, masseter muscle, mandibular molars, and tongue-mandibular division.A representative of the reconstructed trigeminal ganglion of the rats that received a fast blue injection into the dura mater (a), temporal muscle (b), temporomandibular joint (c), tongue (g), mandibular molar teeth (h), or masseter muscle (i).Each dot represents one fast blue-positive cell body in a-c and g-i.Five trigeminal ganglia were reconstructed for each injection site.The remaining four trigeminal ganglia are shown in Supporting Information Figure 5.The five reconstructed trigeminal ganglia were superimposed, with areas of higher cell body density indicated by brighter colors; the lightest color indicates a distance between cells within 50 µm, the middle color indicates a distance between cells of 50-100 µm, and the darkest color contains all of the cells (d-f, j-l).Cells projecting to the mandibular division are mainly distributed in the posterolateral part of the trigeminal ganglion.Although few, the cells innervating the dura mater were not limited to the posterior part, but were also widely distributed anterodorsally (d).The scale bar in a = 1 mm (applies to a-l).

F I G U R E 5 Continued
(Figure 6b).Note that Figures 3d-f

Size distribution of fast blue positive cells in the rat trigeminal ganglia
The soma size of the ganglion cells is related to their chemical marker classification and to the axon projection pathway (Ambalavanar et al., 2003;Kummer & Heym, 1986;Tashiro et al., 1984).Therefore, we measured the major axes of fast blue-positive cell bodies in the rat trigeminal ganglia in each orofacial region, corrected for shrinkage due to tissue clearing, and created histograms (Figure 7).The peaks of the distributions of major axes were calculated using a nonlinear curve-fitting method ("Lognormal," GraphPad Prism 9) and divided into two major groups: those with peaks < 30 µm and those with peaks > 30 µm.Peaks < 30 µm were in trigeminal ganglia injected with fast blue into the scalp, upper eyelid, cornea, infraorbital nerve (vibrissae follicles), and palatal and lingual mucosa; the > 30 µm peaks were found in cases injected into the dura mater, temporomandibular joint, masseter muscle, temporal muscle, and upper and lower molars.Thus, the populations of trigeminal ganglion cells innervating the skin and mucosa have smaller cell bodies than those innervating the deep tissues, suggesting that the components of each population are different.

Somatotopy of the distribution of retrogradely labeled cell bodies in the rat trigeminal ganglion
We superimposed all the fast blue-positive cells in 60 trigeminal ganglia on a single standardized trigeminal ganglion (Figure 8a previous study showing that all three branches of the trigeminal nerve project to the dura mater (Andres et al., 1987).A tendency for somatotopy of ganglion cells was also observed within the divisions.For example, the ganglion cells innervating the upper eyelid and cornea within the first division tended to be distributed more anteromedially than those innervating the scalp.Within the second division, cell bodies innervating the palatal mucosa were located more ventral than those innervating the infraorbital nerve projecting to the vibrissa follicles, and cell bodies innervating the pulp of the maxillary molars tended to be distributed more laterally.In the third division, the distribution of the ganglion cell bodies innervating each region did not show clear somatotopy and were quite evidently overlapping each other.

Overlap in the distribution of trigeminal ganglion cells innervating each orofacial region
The present results revealed that in the trigeminal ganglion, the distribution of ganglion cells innervating each orofacial region was scattered and overlapped each other considerably.The ganglion cells have been known to communicate with adjacent ganglion cells via surrounding satellite cells (Hanani and Spray, 2020).When ganglion cells innervating different regions are present in close apposition, cytokines secreted by ganglion cells activated by tissue injury may affect adjacent ganglion cells innervating undamaged regions via satellite cells, causing hyperalgesia and ectopic pain.The greater the volume of overlapping distribution of ganglion cells, the more likely ganglion cells are to interact.Therefore, we measured their overlapping volumes.The overlap between the distribution regions of the labeled cells was measured and is shown as a heat map (Figure 9a).Since there are five reconstructed trigeminal ganglia for each orofacial region (Figures 3-5; Supporting Information Figures S3-S5), 5 × 5 = 25 measurements are shown for each combination.Figure 9b shows the combinations of orofacial regions arranged in order of increasing volume of ganglion cell distribution overlap.
There was a large overlap in the distributions of ganglion cells that project to the same branch of the trigeminal nerve.For example, the distribution of ganglion cells projecting to the scalp, upper eyelid, and cornea within the first division overlapped widely.Within the second division, the distribution of ganglion cells with axonal projections to the vibrissa follicles (infraorbital nerve), palatal mucosa, and maxillary molars significantly overlapped.Within the third division, the distribution of ganglion cells projecting to the temporal and masseter muscles, mandibular molars, and tongue showed a large overlap with each other.
There was also a large overlap in the distributions of ganglion cells that project to the different branch of the trigeminal nerve.For example, the distribution of ganglion cells innervating the infraorbital nerve (vibrissa follicles) highly overlapped with the distribution of ganglion cells projecting to the scalp, upper eyelid, and cornea in the first division and with the distribution of ganglion cells projecting to the temporal and masseter muscles, and dura mater in the third division.In addition, the distribution of ganglion cells projecting to the maxillary molars in the second division overlapped with the distribution of ganglion cells projecting to the masseter and temporal muscles, and mandibular molars in the third division.Ganglion cells projecting to the areas with combinations of large overlap volumes may interact.

Close apposition of cell bodies that innervate different orofacial regions
Quantitative analysis of the overlap in the distribution of trigeminal ganglion cells revealed a pronounced overlap in the combination of ganglion cells projecting to the masseter muscle and mandibular molars and to the mandibular molars and tongue.These are areas where ectopic pain is frequently observed in clinical dentistry.However, a large overlap in the distribution of ganglion cells does not necessarily mean that these cells are actually in close apposition.Therefore, we injected two fluorescent tracers, fast blue and AAV PHP.S-CAG-tdTomato, which is targeted to peripheral nerves, into the two different orofacial sites and performed double retrograde labeling to determine whether the ganglion cells innervating each site were in close apposition.Closely apposed ganglion cell bodies were observed in the following combinations (Figure 10); maxillary/mandibular molars and masseter/temporal muscles, the lingual mucosa and mandibular molars, the lingual mucosa and masseter/temporal muscles, the scalp and upper eyelid, and the scalp and dura mater.Hence, we suggest that interactions might occur between the ganglion cells innervating these orofacial regions via satellite cells.

Summary
We created a detailed 3D map of the rat trigeminal ganglion.Fast blue was used as a tracer to efficiently retrograde label rat trigeminal ganglion cells.We developed a protocol combined with tissue

Technical considerations
There are many methods for tissue clearing have been published (Ertürk et al., 2012;Hama et al., 2015;Ke et al., 2013;Renier et al., 2014;Richardson & Lichtman, 2015;Vigouroux et al., 2017).We concluded that 3DISCO and/or iDISCO are the most suitable for clearing rat trigeminal ganglia in terms of the time required for clearing, quality of transparency, and the presence or absence of fading of fast blue.
In the original 3DISCO method (Ertürk et al., 2012(Ertürk et al., , 2014)), fixed tissues are immersed in 50% tetrahydrofuran (THF), and clearing begins immediately.However, because the rat trigeminal ganglion contains many myelinated axons and has a high content of lipids, sufficient transparency could not be obtained with the original protocol.Therefore, pretreatment with methanol, a part of the iDISCO protocol combining immunostaining and tissue clearing (Renier et al., 2014), was performed.This effectively cleared the rat trigeminal ganglion.
The 3DISCO and iDISCO methods shrink tissues (Bossolani et al., 2019).This phenomenon was also observed in our study but was not evenly; the mediolateral-axis and dorsoventral-axis shrinkage ratios were significantly greater than the anteroposterior-axis shrinkage ratio.In the trigeminal ganglion, axons run along the anteroposterior axis.The anteroposterior axis was less affected by shrinkage after clearing with 3DISCO/iDISCO because of the presence of oriented microtubules within the axon, structurally supporting the axonal morphology.The tissue shrinkage ratios in the mediolateral-, anteroposterior-, and dorsoventral-axis were stable each time, and a correction was applied to reproduce the tissue architecture before clearing.However, the tissue shrinkage had the advantage of shortening the imaging time using a confocal laser-scanning microscope.In general, the working distance of the objective lens decreases as the magnification increases; for example, the working distance of the 20× objective lens we used was 0.55 mm.The rat trigeminal ganglions in our studies were cut into 0.5-mm-thick slices and cleared using modified 3DISCO, so the thickness of the slices after clearing was approximately 0.3 mm, which was sufficiently thin to observe with the 20× objective lens.Some tissue-clearing methods do not cause tissue contraction, whereas others conversely cause expansion (Bossolani et al., 2019).When such clearing methods are used, the rat trigeminal ganglion must be sliced thinner or turned upside down and imaged twice, which doubles the imaging time and complicates the reconstruction procedure.By using a light-sheet microscope or an objective lens with a long working distance developed for cleared tissues, we are possible to photograph the trigeminal ganglion in its entirety without slicing it.However, such devices are not widely available in many research facilities.The protocol used in the present study is considered to be the best method for devices that are widely available in most research facilities.
The reason we used rats and not mice in the present study is that the smaller size of the mice made it difficult to inject localized tracers into the teeth pulp.Further, it is thought that the tracer solution injected with the Hamilton syringe is more likely to overflow through the needle hole than in rats, because the body size of mice is smaller than rats.
In the present study, efforts were made to prevent overflow as much as possible (see Section 2).As shown in the Figure 13B  Besides fast blue, we also tried other retrograde tracers such as Fluorogold, Fluororuby, tetramethylrhodamine-dextran amine, and fluorescence-conjugated cholera toxin B subunit.However, these retrograde tracers were less efficient in labeling than fast blue, and the red fluorescence signal was weakened by tissue clearing with modified 3DISCO.Therefore, in the present study, we used fast blue as a retrograde tracer in combination with tissue clearing using the modified 3DISCO method.Previous studies have reported that a single trigeminal ganglion cell innervates multiple regions.For example, single cells are known to project to both the nasal epithelium and olfactory bulb (Schaefer et al., 2002), to multiple masticatory muscles and temporomandibular joint (Hovhannisyan et al., 2023), and to both the tongue and pulp of a mandibular molar (Ohara et al., 2013).In the last experi-

Soma size of ganglion cells innervating superficial and deep parts of the body
Previous studies have shown that the cell body size distribution of muscle afferents is significantly larger than that of cutaneous afferents in trigeminal ganglia and spinal dorsal root ganglia (Ambalavanar et al., 2003;Reid et al., 2011).In the present study, the cell body size distribution of primary afferents in the rat trigeminal ganglion was found to be larger in cells innervating deep tissues such as muscle and joints than in ganglion cells innervating skin and mucosa, and this result is in good agreement with that of previous studies.Other previous studies have reported that the soma size of oral thermosensitive primary afferent neurons is significantly smaller than the average size of trigeminal ganglion cells (Leijon et al., 2019).It has also been reported that significantly large numbers of thermosensitive trigeminal ganglion neurons have been distributed in the cornea, rather than in the dura mater (von Buchholtz et al., 2020).The greater distribution of small thermosensitive ganglion neurons in the skin/mucosa, rather than in deep tissues, may be important for protection against burns from hot and cold temperatures in the external environment.Recently, it has been reported that not only differences in cell body size but also the expression patterns of chemical markers are characteristics in ganglion cells innervating different regions (Ambalavanar et al., 2003;Buchholtz et al., 2020;Hovhannisyan et al., 2023;Ibrahim et al., 2023;Leijon et al., 2019;Lindquist et al., 2021;Reid et al., 2011;Wu et al., 2018;Yang et al., 2022).In the future, the integration of data on chemical markers expressed by ganglion cells, axonal projection targets, and cell body localization within the trigeminal ganglion should provide a complete picture of trigeminal ganglion function.
Unlike the trigeminal ganglion, the central nervous system exhibits precise somatotopy, for example, the barrellets of the trigeminal nucleus, the barreloids of the thalamic nucleus, and the barrels of the cerebral cortex, which correspond to the sensation of each vibrissa follicle (Adibi, 2019).The lack of precise somatotopy in the trigeminal ganglion suggests that the trigeminal ganglion processes sensory information using a different strategy than the central nervous system.
Previous animal model studies of ectopic pain have demonstrated interactions between the tongue and mandibular molar afferent neurons within the trigeminal ganglion (Kanno et al., 2020;Katagiri et al., 2012;Ohara et al., 2013).Crosstalk between ganglion cells innervating the scalp and dura mater has been suggested as a possible cause in a rat model of headache (O'Connor & van der Kooy, 1986).
Besides, there are human clinical cases in which inflammation in the muscles of the head and neck region resulted in glossodynia (Rhodus et al., 2003;Svensson & Kaaber, 1995).In clinical dentistry, pain in the masseter muscle often results in associated pain in another area, namely, the mandibular molars, and a source of pain in the temporal muscle often results in associated pain in the maxillary molars.
These are referred to as myogenic toothaches (Fukuda, 2016;Travell & Simons, 1983).Previous studies have shown that single ganglion cells can simultaneously project their axons to the multiple masticatory muscles and temporomandibular joints (Hovhannisyan et al., 2023), as well as to the pulp of molars and tongue (Ohara et al., 2013).This multiregional control by single neurons may be involved in ectopic pain.
Furthermore, crosstalk between ganglion cells via satellite cells may also be involved in the ectopic pain and myogenic toothache described above.Indeed, we have shown that in the trigeminal ganglion, afferent neurons of the tongue and mandibular molars, those of the scalp and dura mater, those of the temporal/masseter muscles and tongue, and those of the temporal/masseter muscles and molars are closely apposed and capable of interacting.Our 3D map of the trigeminal ganglion will provide essential anatomical data for the elucidation of ectopic pain and will contribute to the advancement of research in this field.
Figure3d-f.The areas of higher cell density are indicated by higher Figure S4.The five trigeminal ganglia with the same injection site are superimposed on one standardized trigeminal ganglion in Figure 4d-f.The areas of higher cell density are indicated by higher brightness.The cell bodies innervating the infraorbital nerve projecting to the vibrissa follicles were present as a dense population in the central portion of the trigeminal ganglion (Figure 4a, d).The cell bodies innervating the palatal mucosa were distributed in a ventral-biased manner (Figure 4b, e), and those innervating the dental pulp of the maxillary molar were distributed in a posterolateral part-biased manner (Figure 4c, f).

Finally
, we examined the somatotopic localization of ganglion cells projecting axons to the third branch of the trigeminal nerve.We retrogradely labeled ganglion cells by injecting fast blue into the following six locations: the dura mater, temporal muscle, temporomandibular joint, lingual mucosa, pulp of the first and second mandibular molars, and masseter muscle.For each tracer injection site, five trigeminal ganglia were reconstructed in 3D.Representative examples are shown in Figure 5a-c and g-i.The remaining reconstructed trigeminal ganglia are shown in Supporting Information Figure S5.In five of the trigeminal ganglia with the same injection site, retrogradely labeled cell bodies were observed in similar locations.In Figure 5d-f and j-l, five reconstructed trigeminal ganglia are superimposed on a single standardized trigeminal ganglion.Areas of higher cell density are shown at higher luminosity, that is, areas of higher luminance have a higher probability of distributing the ganglion cells that innervate the orofacial region.As shown in Figure 5 and Supporting Information Figure S5, the ganglion cells sending axons to the third branch of the trigeminal nerve were distributed mainly in the posterolateral portion of the trigeminal ganglion.Those innervating the third branch were distributed over a wide area and no clear somatotopy was observed.The ganglion cells innervating the dura mater differed from other cells projecting into the third branch in that the cell bodies were unevenly distributed dorsally and distributed medially as well as laterally.F I G U R E 3 Three-dimensional distribution of retrogradely labeled cells in the rat trigeminal ganglia produced by fast blue application to the scalp, upper eyelid, and cornea-ophthalmic division.A representative of the reconstructed trigeminal ganglion of the rats that received a fast blue injection into the scalp (a), upper eyelid (b), or cornea (c).Each dot represents one fast blue-positive cell body in a-c.Five trigeminal ganglia were reconstructed for each injection site.The remaining four trigeminal ganglia are shown in Supporting Information Figure , 4d-f, 5d-f, and 5j-l show superimposed fast blue-positive cells in five trigeminal ganglia and are not used to create Figure 6b.The volume encompassing fast blue-positive cells was not necessarily proportional to the number of fast blue-positive cells; for example, when fast blue was injected into the scalp, upper eyelid, and cornea, the volume in which cells were distributed was small compared to the number of labeled cells.Conversely, when fast blue was injected into the temporal muscle, masseter muscle, or mandibular molar teeth, the volume, including cells, was large compared to the small number of fast blue-positive cells.The density of fast blue-positive cells was calculated by dividing the number of fast blue-positive cells by the volume containing those cells (Figure 6c).The top six orofacial regions with the highest cell densities, in order, were the infraorbital nerve, upper eyelid, scalp, cornea, lingual mucosa, and palatal mucosa.All of these cell-dense groups are skin or mucosa and areas with a high level of discriminative sensation.The areas with a low density of fast blue-positive cells were deep tissues of the maxillary and mandibular molar teeth, dura mater, temporal and masseter muscles, and temporomandibular joint.Thus, ganglion cells innervating the skin and mucosa were densely distributed in the F I G U R E 6 Number, volume of distribution, and density of fast blue-positive cells in the rat trigeminal ganglion.The number of cells (a), cellular distribution volume (b), and cell density (c) were measured for each reconstructed trigeminal ganglion.Five trigeminal ganglia were analyzed for each injection site.Bars, error bars, and dots indicate the mean, SD, and individual measurements of the five data, respectively.The cellular distribution volume was measured as a polyhedron containing all the fast blue-positive cells for individual trigeminal ganglion (Figures 3a-c, 4a-c, 5a-c, and 5g-i, Supporting Information Figures S3-S5).trigeminal ganglion, whereas those innervating the deep tissues were scattered over a wide area.
, Supporting Information Movie 1) to compare the localization of the fast bluepositive ganglion cells innervating each orofacial region in 3D.The 3D distribution of fast blue-positive ganglion cells (Figure 8a) was projected onto the XZ, YZ, and XY planes and shown in the quadrant range in Figure 8b-d, respectively.In Figure 8b-d, the unfilled rectangles indicate the interquartile range, the crosshairs represent the range from minimum to maximum, the intersections of crosshairs indicate the median, and the filled squares indicate the mean.Ganglion cells projecting axons to each of the three branches of the trigeminal nerve tended to be distributed in different parts of the trigeminal ganglion with considerable overlap.Cell bodies innervating the first division, scalp, upper eyelid, and cornea were predominantly localized in the anterodorsal part of the trigeminal ganglion.Cell F I G U R E 7 Histograms of the major axes of the fast blue-positive cell bodies.The histograms show the occurrence percentage of the major axes of the fast blue-positive cell bodies in the rat trigeminal ganglia.The major axes of all fast blue-positive cells in the reconstructed trigeminal ganglia were measured for each injection site (for the upper eyelid, maxillary molar, temporal muscle, tongue, mandibular molar, and masseter muscle, n = 3; for the infraorbital nerve, n = 4; for the Scalp, cornea, palate, dura mater, and temporomandibular joint, n = 5) and corrected for shrinkage due to tissue clearing.The peaks of the distributions of major axes were calculated using a nonlinear curve-fitting method ("Lognormal," GraphPad Prism 9), and they are indicated by arrows and values.bodies innervating the second division, the infraorbital nerve, palatal mucosa, and pulp of the maxillary molars were distributed in the center of the trigeminal ganglion.Cell bodies innervating the third division, the temporal muscle, temporomandibular joint, lingual mucosa, masseter muscle, and pulp of mandibular molars were mainly distributed posterolateral of the trigeminal ganglion.The only exception was the dura mater, where cell bodies were distributed not only posterolateral but also anterodorsal of the trigeminal ganglion, where cell bodies innervating the first division are distributed.This result reflects a

F
Summary of 3D topography of the rat trigeminal ganglion.(a) All the reconstructed rat trigeminal ganglia (12 injection sites × 5 samples each = 60 samples) were superimposed on a standardized left trigeminal ganglion.For 3D data of (a), see the movie in Supporting Information Movie 1.The distributions of ganglion cells in the XZ plane (b), YZ plane (c), and XY plane (d) of the rat trigeminal ganglion are shown as mean (filled squares), interquartile range (unfilled squares), range (crosshair), and median (intersections of crosshairs).Small round dots indicate outliers.Eyelid, upper eyelid; ION, infraorbital nerve; Masseter, masseter muscle; Mandibular, mandibular molar teeth; Maxillary, maxillary molar teeth; Temporal, temporal muscle; TMJ, temporomandibular joint.The scale bar in a = 1 mm.
clearing and retrograde labeling to reconstruct the trigeminal ganglion accurately and efficiently.Fast blue was injected into 12 sites of the orofacial region: six into the skin/mucosa and six into the deep tissues.Histograms of the major axes of the fast blue-positive cell bodies revealed that the peak major axes of the cell bodies innervating the skin/mucosa were <30 µm, while those of cell bodies innervating the deep structures were >30 µm.The ganglion cells innervating each of the regions of the first, second, and third branches of the trigeminal nerve were localized to different portions of the trigeminal ganglion.The ganglion cells innervating the first branch were distributed in the anterodorsal portion, those innervating the second branch were distributed in the central portion, and those innervating the third branch were distributed in the posterolateral portion of the trigeminal ganglion.However, the distribution of ganglion cells projecting to each trigeminal nerve branch showed considerable overlap at their border regions.In particular, trigeminal ganglion cells projecting to the third branch region were sparsely distributed and considerably intermingled with each other.Finally, double-fluorescent labeling of trigeminal ganglion cells with fast blue and AAV PHP.S-tdTomato revealed the presence of ganglion cells in close apposition to each other in combinations of the orofacial regions that showed a large overlap of ganglion cell distribution.Specifically, in combinations of the lingual mucosa and masseter/temporal muscles, maxillary/mandibular molars and masseter/temporal muscles, lingual mucosa and mandibular molars, scalp and eyelid, scalp and dura mater, and in these combinations of orofacial regions, it was suggested that ganglion cell interacts with each other via satellite cells.

F
Overlaps of volumes in which cell bodies are distributed.The heat map (a) was produced using data from Figures3a-c, 4a-c, 5a-c, and 5g-i, and Supporting Information Figures S3-S5.The volume, which encloses all of the fast blue-positive cells, was measured, and the overlap was calculated using MATLAB software.(b) shows the combinations of orofacial regions arranged in order of increasing volume of ganglion cell distribution overlap.Bars, error bars, and dots indicate the mean, SD, and individual measurements of the data, respectively.Eyelid, upper eyelid; ION, infraorbital nerve; Masseter, masseter muscle; Mandibular, mandibular molar teeth; Maxillary, maxillary molar teeth; Temporal, temporal muscle; TMJ, temporomandibular joint.F I G U R E 1 0 Close apposition of trigeminal ganglion neurons.Combinations that showed a large overlap in the distribution of fast blue-positive ganglion cells were examined for cell bodies in close apposition using two kinds of fluorescent tracers.AAV PHP.S-CAG-tdTomato was injected into the lingual mucosa, scalp, masseter muscle, and temporal muscle (red), and fast blue was injected into the maxillary/mandibular molar pulp, upper eyelid, dura mater, and lingual mucosa (cyan).Some ganglion neurons showed close apposition to each other (arrowheads).The scale bars = 100 µm.
of Lindquist   et al., 2021, however, skin  is present in contact with the surface of the masseter and temporalis muscles.We should interpret the results with caution because fast blue could infiltrate the surface of the muscles, and the nerves innervating the muscle as well as the nerves innervating the skin covering the muscle could be labeled with fast blue.
ment of the current study, double fluorescent labeling was performed by injecting AAV PHP.S-CAG-tdTomato in combination with fast blue, but no double-labeled ganglion cells were observed.This could be due to the low efficiency of AAV PHP.S-CAG-tdTomato infection or insufficient injection volume of virus solutions.In the future, when improved retrograde fluorescent tracers with better labeling efficiency become available, and when improved tissue transparency methods that do not affect the fluorophores become available, retrograde multiple fluorescent labeling and tissue clearing would be combined.This technique is expected to allow the more precise location of ganglion cells innervating different regions to be analyzed within the same trigeminal ganglion.