Multiscale analysis of the adult human superior hypogastric plexus and hypogastric nerve.

KS contributed to concept/design, acquisition of data, data analysis/interpretation, drafting of the manuscript, critical revision of the manuscript, and approval of the article. JPFJ contributed to the concept/design, critical revision of the manuscript, and approval of the article. PO contributed to concept/design, critical revision of the manuscript, and approval of the article. MB contributed to concept/design, acquisition of data, drafting of the manuscript, critical revision of the manuscript, and approval of the article. JK contributed to concept/design, data analysis/interpretation, drafting of the manuscript, critical revision of the manuscript, and approval of the article.

Approval for the use of human tissue in this research project was granted from Human Research Ethics Committees at both the Centre Hospitalier Universitaire de Nîmes (CHU de Nîmes), France and the University of Melbourne, Australia. Specimens used for this project were obtained from the organ donation program based at the Centre Hospitalier Universitaire de Nîmes (CHU de Nîmes) and the Body Donor Program at the University of Melbourne (Table  1 ). All dissections were performed by Professor Martin Bertrand. The overall purpose of these dissections was to investigate the IHP and its neural connections, the first area of focus being the HN. To ensure completeness of the HN, the most caudal region of SHP was retained in the dissection. The present study included both tissues. Studies on the IHP will be reported separately. Donor and specimen information. Body donors at the University of Melbourne were embalmed with Genelyn, a commercial embalming fluid (Genelyn Pty Ltd., Australia) perfused via either the femoral or carotid artery. These samples will be referred to as embalmed . After dissection of the IHP and associated nerves, these samples were secured to the base of a box lined with silicon polymer to maintain their original structure during further fixation (4% buffered paraformaldehyde [PFA], 48 h, 4°C). Samples were then washed extensively with 0.1 M phosphate‐buffered saline, pH 7.2 (PBS; 6 × 1 h), to reduce the amount of embalming fluid remaining in the sample. Samples were stored in PBS containing the preservative 0.1% sodium azide until use. Specimens from CHU de Nîmes were obtained from organ donors. During dissection, sutures were used to tag selected structures to facilitate registration to major anatomical landmarks during later subdissection. After dissection (1–4 h), samples were immersed in freshly made 4% buffered paraformaldehyde (PFA) for 48–72 h, then washed in PBS and placed in PBS containing 0.1% sodium azide (pH 7.2) for storage and transportation. These samples will be referred to as unembalmed . Each specimen was photographed before being shipped to the University of Melbourne. Once received, the specimens were washed again in PBS, oriented by pinning out flat in a silicon‐lined dish, photographed and primary structures identified and labeled. Correspondence across the team ensured that the orientation and labels were correct and that they matched the original images. All samples were subdissected to carefully remove regions of excess adipose and connective tissue that were not closely adhered to the neural structures. The SHP and HN were cut into blocks of approximately 1 cm 3 for further study. An example of a specimen and subdissection can be seen in Figure  S1 . The boundaries of the HN were defined rostrally as the bifurcation of the SHP and caudally where each HN typically became broader before merging into the IHP. Therefore, only samples containing caudal SHP and ipsilateral IHP were considered to contain the entire HN and were included in all elements of this study. A partial HN identified in one donor was excluded from entire‐nerve analyses (Table  1 ). The complete workflow for the primary elements of the study can be seen in Figure  S2 and is described further below. To provide an initial assessment of the cellular and tissue integrity of samples used in this study, we performed conventional histology on tissue blocks selected from each of four samples that had been preserved and processed in different ways (Table  1 , Figure  S1 ). This allowed us to compare embalmed and unembalmed tissues and to determine the impact of processing tissues for tissue clearing. We also included blocks from samples that were rehydrated after tissue clearing (“reverse clearing”), a step we applied to some blocks to extend data collection to sections after prior analysis as whole mounts. The blocks were embedded in paraffin, sectioned (5 μm) using a microtome, then stained with hematoxylin and eosin (H&E). Slides were imaged using a Zeiss Axioscan7 Slidescanner (Carl Zeiss, Germany). All other tissue blocks were designated for tissue clearing or cryosectioning (Figures  S1 and S2 ). The large‐volume clearing and immunolabeling methodology used in this project was adapted from the iDISCO protocol developed for whole mounts by Renier et al. ( 2014 ). Due to the prevalence of adipose tissue in these samples, the Adipo‐Clear protocol designed by Chi et al. ( 2018 ) was incorporated. Subdissected blocks were viewed under a stereomicroscope (Olympus) and any visible fibers or contaminants carefully removed using fine forceps. All blocks were initially processed with Adipo‐Clear delipidation (Section “ Adipo‐Clear Protocol ”). While all blocks also eventually proceeded to iDISCO clearing (Section “ iDISCO Protocol ”), a subset of blocks underwent immunohistochemical processing (Section “ Immunohistochemistry on Whole Mounts ”) prior to iDISCO. Blocks were washed in B1n buffer (pH 7, 0.3 M glycine, 0.1% Triton X‐100, 0.01% sodium azide in H 2 O) then dehydrated with increasing concentrations of methanol (MeOH) in B1n buffer (20%, 40%, 60%, 80%, and 100%; 1 h each), then incubated in 100% dichloromethane (DCM; 1 h) before being transferred to fresh DCM and incubated overnight at 4°C with rotation at 12 rpm. For depigmentation, blocks were incubated in DCM for 1 h. This was repeated with fresh DCM until the block sunk to the bottom of the tube. Blocks were transferred to new tubes and washed in 100% MeOH for 2 × 1 h. All delipidation and depigmentation steps were performed at 4°C on rotation. Blocks were bleached in 6% H 2 O 2 in MeOH and incubated overnight at 4°C protected from light. Rehydration of the blocks involved increasing concentrations of B1n buffer (100%, 80%, 60%, 40%, 20%, 0%) with MeOH for 1 h each on ice on an orbital shaker. The blocks were then placed in DPBSG‐T (gelatine 0.2% w/v, Triton X‐100 (0.5%), thimerosal (0.01% w/v), Dulbecco's PBS) for immunohistochemistry as described in Section “ Immunohistochemistry on Whole Mounts ” or storage (no more than 4 days) prior to the final iDISCO clearing stages described in Section “ iDISCO Protocol ”. Blocks were incubated in DPBSG‐T for 4 days at room temperature with rotation at 12 rpm. They were then incubated in a working dilution of primary antibody in DPBSG‐T for 14 days in a hybridization oven at 37°C with orbital horizontal mixing (70 rpm), then washed in DPBSG‐T (6 × 1 h at room temperature at 12 rpm) and incubated in a working dilution of secondary antibody in DPBSG‐T for 2 days in a hybridization oven at 37°C with orbital horizontal mixing (70 rpm). Blocks were again washed in DPBSG‐T (6 × 1 h at room temperature, 12 rpm). Information on antibodies used in this study can be found in Table  2 . Primary and Secondary Antibodies. All blocks were again observed under a dissecting microscope to remove any contaminants previously missed or that had stuck to the sample during the clearing process. A final delipidation step involved dehydrating blocks in increasing concentrations of MeOH in DPBS (20%, 40%, 60%, 80%, 100%, 100%) before incubating in 2/3 DCM and 1/3 MeOH overnight in a fume hood with rotation at 12 rpm. Blocks were incubated in DCM for 30–60 min with rotation at 12 rpm, then repeated with fresh DCM until each block sank to the bottom of the tube. Blocks were then incubated in 100% dibenzyl ether (DBE) for 2 h with rotation at 12 rpm and stored in glass vials containing fresh DBE. DBE was replaced with ethyl cinnamate at least 3 h prior to imaging. Blocks selected for cryosections were cryoprotected in PBS containing 30% sucrose then embedded in an inert mounting medium (O.C.T., Tissue‐Tek, Sakura, Torrance, CA) before being sectioned (40 or 50 μm) using a cryostat (−20°C). Sections were mounted directly on glass slides, and an advanced PAP pen (Merck, Darmstadt, Germany) was used to draw a hydrophobic boundary around the sections. Slides were washed in PBS (3 × 10 min), then incubated in blocking solution (PBS containing 10% non‐immune horse serum and 0.1% Triton X‐100) for 2–3 h at room temperature. Sections were then incubated with combinations of primary antibodies overnight at room temperature. The sections were again washed in PBS (3 × 10 min) and incubated with combinations of species‐specific secondary antibodies for 2–3 h at room temperature while protected from light, then washed in PBS (3 × 10 min), mounted on glass slides and cover‐slipped using Vectashield‐Plus mounting medium (Vector Laboratories, Burlingame, CA). Antibodies were diluted with PBS containing 0.1% Triton X‐100 and 2% non‐immune horse serum. Primary and secondary antibodies are listed in Table  2 . The rationale behind the selection of antibody combinations is outlined in the relevant results section. For some primary antibodies, more than one host species was available, which allowed additional antibody combinations to be performed in multi‐channel labeling studies. Across all replicate samples (two SHP and three unilateral HN), at least two blocks from each sample were sectioned. From each block, two non‐sequential sections (40 or 50 μm thick; minimum interval of 100 μm) were collected and immunolabeled for each antibody combination. If sections contained no or a small number of neurons, additional sections were collected. Selected cleared tissue blocks were rehydrated so that structures observed in whole mounts could be further examined in paraffin sections or in immunolabeled cryosections of the same blocks (see workflow in Figure  S2 ). Cleared blocks of embalmed samples were rehydrated for conventional histology only, whereas cleared blocks of unembalmed samples were rehydrated for conventional histology or immunofluorescence. This rehydration workflow is sometimes referred to as “reverse clearing” and was based on a protocol used by Blain et al. ( 2023 ). The selected blocks were rehydrated in decreasing concentrations of MeOH in DPBS (100%, 100%, 80%, 60%, 40%, 20%) for 1 h each. They were then washed in 100% DPBSG‐T for 2 × 1 h with rotation at 12 rpm, then processed for histology or immunofluorescence as described above. Cleared tissue blocks were visualized using an UltraMicroscope Blaze™ light sheet microscope (Miltenyi Biotec, Germany) equipped with an Andor Zyla 4.2 Mpx sCMOS camera (2048 × 2048 pixels) controlled by Inspector Pro 7.6.3 acquisition software (Miltenyi Biotec, Germany). The light sheet (4 mm thickness) was generated by lasers at separate wavelengths of 488 nm, 561 nm, and 640 nm. Three objectives (1×, 4× or 12×) were used with different magnification lenses (0.6×, 1×, 1.66× and 2.5×). Tissue blocks were supported by a sample holder placed within a container of ethyl cinnamate, illuminated by the laser light sheet from one side. When mosaics of image tiles were obtained, they were captured with an overlap of 10% between tiles. Images were acquired in a 16‐bit ome‐TIFF format. Acquisition parameters varied depending on the imaging requirements. Exposure times ranged from 50 to 200 ms. The light sheet width was set to 100% for overview images and reduced to 50% when using higher magnification objectives. Z‐step intervals ranged from 0.2 to 10 μm. Overview images were acquired with 1× or 4× objectives to capture mesoscopic structures, including vascular networks, nerve tracts, and neuron cluster distributions. For neuron counts and morphological analyses, images were acquired at 4× or 12× magnification, with z‐steps matched to the voxel pixel size for accurate 3D measurements. Image stacks were converted to Imaris (.ims) format using Imaris File Converter software (Oxford Instruments Andor, UK). For mosaic reconstruction, individual .ims tiles were imported into Imaris Stitcher (Oxford Instruments Andor, UK) where automatic alignment was selected. Alignments were then inspected and manually adjusted as needed. Visualization and descriptions of structures were obtained from images viewed on Imaris and Imaris Viewer (v9.6.1 to v10.1.1) (Oxford Instruments Andor, UK) software. Measurements of structures (e.g., blood vessel diameters) were obtained either using the line tool in Slice View or the Measurement Points tool in 3D View on Imaris. To better visualize specific anatomical structures, the virtual reality software syGlass (v2.4.0) (IstoVisio, Inc., USA) coupled to an Oculus Quest2 (Meta, USA) headset was used to manually segment objects in 3D. A custom Python code ( https://github.com/alicia‐ziying‐yang/syglass‐multichannel‐io‐utils ) was then used to extract the segmentation completed from the raw data, with the underlying fluorescence of each mask created as an independent file. These TIFF stacks were then converted back into Imaris files and imported as additional images in the original Imaris file. For neuron counts, all neuronal clusters containing 10 or more neurons within a sample block were imaged. Neurons were manually counted in Imaris using: (1) Presence of large autofluorescence aggregates in neural somata or (2) Immunostained somata. Autofluorescence‐based counting was possible due to the presence of highly fluorescent perinuclear lipofuscin granules, visible in the 488 nm and 561 nm channels. A pilot analysis was conducted to firstly determine the prevalence and distribution of lipofuscin. All neurons appeared to contain lipofuscin granules, but these were not uniformly distributed throughout the cytoplasm, being typically clustered in one or two regions of the cell (Figure  S3 ). We concluded that the appearance and distribution of autofluorescent lipofuscin would allow neurons to be quantified and mapped in unstained sample blocks (e.g., from embalmed specimens). To verify this, counts were compared using a sample of images containing both autofluorescence (488 nm) and a pan‐neuronal marker (neurofilament medium protein [NFM] or class III beta‐tubulin [Tuj1], 561 nm). Manual counts were conducted using the spot creation tool in Imaris, first in the autofluorescence channel and followed by counts in the pan‐neuronal marker channel. This was done on 12 images of neuron clusters across six HN blocks in one donor (Donor 3). We found that manual counts of NFM+ neurons ( n  = 1640) were 7% lower than manually counting neurons identified by lipofuscin autofluorescence ( n  = 1754). We did not explore the basis of this minor discrepancy but consider this may be due to a small population of neurons that do not express detectable levels of NFM‐immunoreactivity and/or autofluorescent aggregates present in a small number of non‐neuronal cells. Irrespective, this analysis encouraged us to collect data using autofluorescence as a surrogate marker of neuronal somata, providing the opportunity to include data from samples prepared by embalming. Cryosections were analyzed manually while viewing sections or images taken with Zeiss Axio‐Imager M2 wide‐field and LSM900 confocal microscopes (Carl Zeiss, Germany). Confocal images of representative single neurons were captured using the 63× objective and Airyscan2 function. Images were either captured as single optical slices or Z‐stacks (1–2 μm) depending on the structure visualized. The projection is specified for each panel in the figure legends. Colocalization patterns of neural markers were investigated to determine subpopulations of noradrenergic neurons and to make comparisons to homologous rodent structures and expression patterns. Colocalization was determined by manual observation on the microscope where possible and on digital images when using far red channels. Colocalization counts were recorded on images in FIJI ImageJ using the object‐based colocalization analysis (OBCA) tool plug‐in (Lunde & Glover,  2020 ). A minimum of 75 neurons per sample were analyzed for colocalization; neurons were only counted if the nucleus was visible. To provide additional visualization of synaptic boutons (varicosities) immunolabeled for calcitonin gene‐related peptide (CGRP) and substance P (SP), image stacks (1 μm) were converted to Imaris (.ims) format using Imaris File Converter software. The Surfaces creation tool on Imaris was used to manually segment the neuronal cell body. Varicosity segmentation was conducted using the semi‐automated Surface creation tool. The entire image was processed with the smoothing filter selected (Surface Grain Size = 0.2 μm). Background noise was eliminated by setting the diameter of the largest sphere as 2 μm. Thresholding was adjusted manually to ensure the surfaces matched the stained varicosities of the channel selected (either CGRP or SP). Overlapping surfaces were resolved by enabling the morphological split function and setting the region growing estimated diameter as 2 μm. Finally, the surfaces were then manually filtered based on several features including the sphericity and number of voxels. Varicosities positioned more than 2 μm from the cell surface were further removed using the filter function for better visualization of the segmented boutons positioned closer to the surface of the neuronal cell body. Primary antibodies used in this study have been described in previous immunofluorescence studies on the human peripheral nervous system. Examples of these studies are provided with the antibodies in Table  2 . No signal (apart from the lipofuscin autofluorescence) was evident when primary antibodies were omitted from the sections, indicating that the structures identified by the primary antibodies were not due to non‐specific binding of secondary antibodies. Each of the antibodies revealed types of structures expected from prior studies on the nervous system, for example, TH demonstrating neuronal cell bodies and bundles of axons but very few synaptic boutons, whereas synaptophysin demonstrated synaptic boutons but not neuronal cell bodies or axon tracts. We did not perform specificity tests on the primary antibody (e.g., isotype‐matched antibody controls or absorption tests) so we have referred to our results as immunoreactivity (‐IR) rather than the specific molecule of interest that was not directly identified. All 2D and 3D imaging is representative of all samples of the same type (i.e., SHP or HN), unless stated otherwise. The source of each image (sample, block) is indicated in Table  3 , with block locations shown in Figure  S1 Figures and movies were created using Adobe Creative Suite (Adobe InDesign, Illustrator, Photoshop and Premiere Pro; Adobe Systems, USA). Graphs were initially created using GraphPad Prism 10 (GraphPad Software, MA, USA) and edited using Adobe Illustrator. In multi‐channel images, individual channels are provided as monochrome (for individual channels) and digitally colorized to best demonstrate and compare immunolabeled structures (merged channels). Adjustments to contrast and brightness were made as necessary to accurately represent the immunostaining observed in cryosection images under the microscope using Adobe Photoshop. Occasionally, the brightness and contrast were adjusted to improve visibility and validate properties of less intensely stained structures. Adjustments of gamma for visualization of autofluorescent structures in cleared tissue were carried out using Imaris. The relevant figure legends indicate where these modifications have been performed. Donor and sample block information for representative images provided in figures and movies. See Table  1 for list of samples and donors, and Figure  S1 for information on Block location.

Cellular and tissue structure in sample blocks of the SHP and HN from two of the five available donors were initially assessed using conventional histology in paraffin sections. Histological assessment was performed on samples that had been preserved and processed in different ways (Figure  S1 ). Examples of each are provided in Figure  2 (SHP) and Figure  3 (HN). Unless stated otherwise, the structures described below were consistent across tissues prepared three different ways prior to paraffin embedding: (i) embalmed, (ii) unembalmed, and (iii) unembalmed, cleared then reverse cleared. In all cases and for both tissues, histological sections demonstrated neural tissues (axon tracts and aggregates of neuronal cell bodies) and blood vessels embedded within adipose and fibrous connective tissue. Histology of the superior hypogastric plexus (SHP). Longitudinal sections (5 μm) of paraffin‐embedded SHP from (a–d) unembalmed and (e–h) embalmed tissue. (a) The SHP contains axon tracts embedded in adipose tissue (AT), with clusters of neuronal cell bodies (outlined, black arrowheads) embedded within the fascicles. Higher magnifications of neuron clusters indicated by boxed regions are provided in b and c. Further magnifications of region shown in c are shown in panel d and its inset, showing that all neuronal cell bodies are surrounded by glial cell nuclei (d, inset; 5× magnification). Blood vessels are indicated (examples shown by black asterisks). (e) A similar range of tissues can be identified in embalmed tissue, including clusters of neuronal cell bodies (outlined, black arrowheads), blood vessels (examples shown by black asterisks) and multi‐axonal fascicles of axons (examples shown by white asterisks), most of which have been cut obliquely. Images show axon fascicles with a diverse range of diameters and several blood vessels located between the fascicles. Higher magnifications of boxed regions are provided in (f–h). Comparison of preservation methods shows that in embalmed tissue most neuronal cell bodies are surrounded by a space (“halo”), whereas these spaces are smaller or rare in unembalmed tissue. Scale bars: 500 μm (a, e). All other scale bars are 100 μm. Tissue shown in (a–d) is from Donor 3 (male, 78 years) and (e–h) from Donor 2 (male, 75 years). Donor and specimen details are provided in Table  1 , block number for each image panel in Table  3 , and block location shown in Figure  S1 . Histology of the hypogastric nerve (HN). Paraffin sections (5 μm) of HN from unembalmed tissue (Donor 3, male, 78 years) that underwent tissue clearing then reverse clearing prior to paraffin embedding. (a) A cross‐section through the HN highlights large blood vessels (examples shown by black asterisks) and axon tracts of variable diameters (examples shown by white asterisks). (b) A section through axon tracts embedded in adipose tissue (AT) include clusters of neuronal cell bodies embedded within tracts (outlined, black arrowheads). One of these neuron‐dense regions is shown at higher magnification (c), with a further magnification of a region in c in (d). The inset (4.7× magnification) in d displays a single neuronal soma and its closely surrounding glial cells. Scale bars: 500 μm (a) and 100 μm (b–d). Donor and specimen details are provided in Table  1 , block number for each image panel in Table  3 , and block location shown in Figure  S1 . Within the caudal SHP, neural structures were present within all blocks, with most of each block being comprised of adipose and fibrous connective tissue. Longitudinal sections through nerve tracts showed the presence of converging fascicles (Figure  2a ), whereas transverse and oblique sections through these tracts revealed a diversity of fascicle diameters and blood vessels located between nerve fascicles (Figure  2e,g,h ). Neuronal cell bodies (somata) identified within the SHP were typically arranged in clusters embedded within neural tracts rather than as discrete, encapsulated ganglia (Figure  2a–f ). All SHP neurons were surrounded by satellite glial cells (Figure  2d,f ). In unembalmed samples, nuclei of satellite glial cells closely surrounded neuronal somata (Figure  2d , inset), whereas in embalmed tissue, a gap often separated a neuronal soma from its nearest glial cells (Figure  2f ). We deduce this ’halo’ effect was likely due to neuronal shrinkage prior to paraffin embedding. Similarly to the SHP, the HN contained axon tracts and blood vessels of diverse diameters (Figure  3a ). Therefore, from this approach alone it would be challenging to distinguish between these tissues and obtain a thorough understanding of their anatomical organization. Specifically, multi‐fascicular axon tracts with interconnecting fascicles were noted in longitudinal sections of HN (Figure  3b ). Neuronal cell bodies were present in the HN, appearing in clusters embedded within neural tracts (Figure  3b–d ). As identified in the SHP, all neurons in the HN were surrounded by satellite cells (Figure  3d , inset). These satellite cells closely surrounded neural somata and did not show the perisomatic halo observed in embalmed SHP tissue. This indicates that the improved preservation of cellular structure in unembalmed, freshly fixed samples was not later impaired by the subsequent processes (tissue clearing, LSFM and then reverse clearing). A limitation of applying fluorescence microscopy approaches to human tissues is the high level of autofluorescence in many cell types. This is particularly the case for neurons, which contain cytoplasmic aggregates of the protein lipofuscin that are highly fluorescent at the wavelengths used for viewing many tagged antibodies (Terman & Brunk,  2004 ). Though these lipopigments have been shown to accumulate with age, studies have identified lipofuscin within sympathetic ganglia of donors across younger ages, including in children (Helén,  1983 ; Hervonen et al.,  1986 , Hervonen et al.,  1978 ). This limits the range of fluorophores that can be used in immunohistochemical experiments. We therefore investigated if autofluorescence alone could be used to visualize the 3D distribution of blood vessels, axon tracts, and neuronal cell bodies within the SHP and HN. If successful, this would provide the opportunity to analyze these structures in embalmed tissues that are generally unsuited to immunofluorescence. Several tissue blocks of SHP (Figure  4 ; Movie  1 https://doi.org/10.26188/30736406 ) and HN (Figure  5 ; Movie  2 https://doi.org/10.26188/30736397 ) were cleared and viewed with LSFM, without prior processing for immunohistochemistry. This was performed with both embalmed and unembalmed specimens. Autofluorescence allowed identification of numerous individual neuronal cell bodies (due to their brightly fluorescent granules, presumed to be lipofuscin), and less intense, more uniform autofluorescence of axon tracts and blood vessels in each of these samples. Structure of the caudal superior hypogastric plexus (SHP) of an embalmed specimen (Donor 2, male, 75 years) processed for tissue clearing and viewed by light sheet fluorescence microscopy. (a) Neural tracts and blood vessels can be seen with autofluorescence but here have been highlighted using segmentation (syGlass); this shows aggregation of nerve tracts into distinct planes, each colorized differently here (yellow, green, and blue). A blood vessel has also been segmented (red). Fascicular connections between these neural planes are indicated by arrowheads. Individual segmentations are shown in (d–g). The yellow area in the schematic indicates the location of this tissue block in the context of the entire sample. (b) Raw image showing strong vascular autofluorescence due to fixative effects primarily due to residual blood. (c) Optical cross‐section through the block showing neural tracts corresponding to the segmented structures shown in a. Higher magnification of regions with interconnecting fascicles are shown in (h) and (i), the location indicated by boxes in d and g, respectively. Both insets in h (12× magnification) and i (24× magnification) show examples of neuronal cell bodies embedded in the tracts. Scale bars: 1000 μm (a–g) and 500 μm (h, i). Donor and specimen details are provided in Table  1 , block number for each image panel in Table  3 , and block location shown in Figure  S1 . Structures associated with the hypogastric nerve (HN) in embalmed (a–e) or unembalmed (f–j) tissue, followed by tissue clearing and light sheet fluorescence microscopy imaged using 561 nm autofluorescence (no antibody staining). The yellow area in each schematic (a, f) indicates the location of this tissue block in the context of the entire sample. In (a), nerve tracts are tightly grouped with a prominent vessel running parallel and giving off perpendicular branches (b). Here, selected vasculature has been segmented (syGlass) to demonstrate the structure and trajectory of the primary vessel (red), and distinct secondary branches (magenta, green, yellow, blue). (c) An optical section (Imaris Ortho Slicer tool) from the location indicated in a demonstrates how the primary vessel and a smaller branch, identifiable by their lumen (indicated by asterisks), run parallel to the fascicles (examples shown by arrowheads). The boxed region (d) shows a neuron cluster embedded within a nerve tract, with (e) providing a higher magnification view of individual autofluorescent neurons. (f) This HN showed a wider fascicular distribution and more complex vascular branching. (g) A horizontal optical slice from the location indicated in f, showing neural tracts (arrowheads) and a blood vessel (asterisk), which can be identified by its lumen and autofluorescent internal elastic lamina. (h) A higher magnification of embedded neurons from the location indicated in f, with the inset (2× magnification) showing individual neurons visualized by their autofluorescent lipofuscin. The boxed region (i) shows a putative paraganglion, validated by immunostaining for tyrosine hydroxylase (TH) in (j), showing numerous SIF cells. Immunostaining was performed on the block after reverse clearing and cryosectioning. Scale bars: 1000 μm (a–c, f–g) and 100 μm (d, e, h, i). Tissue shown in (a–e) is from Donor 2 (male, 75 years) and (f–j) from Donor 3 (male, 78 years). Donor and specimen details are provided in Table  1 , block number for each image panel in Table  3 , and block location shown in Figure  S1 . The caudal SHP exhibited a complex arrangement of nerve fascicles interwoven with an extensive vascular network that was highly variable between blocks (Figure  4 ). In embalmed samples, nerve tracts were identified by their uniform signal intensity, visible in 3D View, Slice and Section modes on Imaris. Blood vessels were readily identified, as confirmed in optical sections by their lumen (Figure  4c ), or higher levels of autofluorescence due to the signal from blood cells present at the time of initial preservation (Figure  4b,c ). Segmentation with syGlass was performed to more clearly distinguish in 3D the primary neural tracts and vascular elements of SHP tissue (Figure  4a,c–g ; Movie  1 https://doi.org/10.26188/30736406 ). The fascicles themselves varied greatly in diameter, ranging from fine fibers to larger, more prominent bundles. While most fascicles travelled in generally parallel trajectories, they frequently branched and were dispersed across multiple planes. Within this network, fascicles regularly converged, diverged, and interconnected, forming diverse branching patterns (Figure  4d,e,g ). Closer inspection of the whole mount SHP sample blocks showed nerve tracts that were surrounded by rounded small spaces (presumed to be adipose cells) and they occasionally contained clusters of autofluorescent neuronal cell bodies (Figure  4h,i ). These were aggregated into groups of diverse numbers and as single neurons embedded in axon tracts. Their location could not be deduced from the size or shape of the tracts, that is, there was no enlargement of the tracts at the location of neuronal cell bodies and the neuronal cell bodies were not specifically aggregated at intersections of axon tracts. The arrangement of neuronal cell bodies was highly variable between SHP samples and blocks. The HN comprised numerous neural tracts (Figure  5a,c,f ). Most of these ran in parallel but were not fully isolated from one another; instead, they frequently branched and connected with neighboring fascicles, occasionally forming interconnecting patterns like those observed in the SHP. Blood vessels coursed alongside the neural tracts, and in most blocks, a prominent, tortuous vessel was present (Figure  5a–c ). Segmentation with syGlass was performed to more clearly demonstrate this vascular trajectory. This large vessel typically gave off perpendicular branches, which either extended smaller parallel branches that followed the nerve's path or continued outward to supply surrounding adipose tissue. This prominent blood vessel varied in diameter across specimens (diameter 317 ± 16 μm; mean ± SEM, range 152–594, n  = 5 nerves). In some HN blocks, the vasculature was more complex, where the prominent vessel was observed bifurcating numerous times to form smaller vessels running alongside the nerve tracts (Figure  5f ). Blood vessels were identified as described above for the embalmed SHP and confirmed using the brighter autofluorescence of their elastic laminae as seen in Slice mode in Imaris (Figure  5g ). Within many of the HN nerve tracts, clusters of neuronal cell bodies and individual neurons were clearly visible by their intense lipofuscin autofluorescence (Figure  5d,e,h ), identifiable even at low magnification (Figure  5a,f ). Putative paraganglia were occasionally identified due to their unique features, including their nodular shape, connection to axon tracts, and scarcity of neurons (Figure  5i ). This included an unembalmed sample, where we then undertook reverse clearing and cryosectioning to then immunolabel the sections for the catecholamine synthetic enzyme, tyrosine hydroxylase (TH). This identified dense aggregates of small TH‐IR cells; these were not considered neurons as they were too small (~5–20 μm) and lacked typical autofluorescent lipofuscin. Instead, they were identified as small intensely fluorescent (SIF) chromaffin‐like cells as these synthesize and store catecholamines (Figure  5j ). These structures will be investigated further in a separate study. Our initial 3D visualization of cleared (i.e., unsectioned) blocks of HN revealed numerous embedded neuronal cell bodies (see Section  3.2 ). To determine the prevalence and distribution of neurons throughout the entire length of the HN, we then allocated the majority of remaining HN tissue blocks to tissue clearing. This enabled mapping and quantitation of neurons along the full length of each nerve. By utilizing the autofluorescent features of neuronal cell bodies, we were able to include embalmed tissues in our analysis. For many of the unembalmed samples, we performed immunohistochemistry prior to clearing, using immunolabeling for a pan‐neural protein (NFM [neurofilament medium protein] or Tuj1 [class III beta‐tubulin]). In one of these nerves, co‐labeling for TH was performed to determine the proportion of catecholamine‐synthesizing (presumed noradrenergic) neurons. In each HN, neurons were embedded directly within the neural tracts. Neuronal clusters within the HN displayed considerable variation in their arrangement (Figure  6a–c ). Neuronal clusters varied widely in size, ranging from large aggregations comprising hundreds of neurons to small groups of fewer than 10 cells. Isolated single neurons could also be identified throughout the HN (Figure  6a , white arrow). Many neuron clusters were embedded within the unbranching neural tracts (Figure  6a ), whereas others were positioned close to or within intersections of fascicles (Figure  6b ). For larger clusters, neurons appeared spread along the length of individual fascicles and their interconnecting branches. A small number of neurons were occasionally identified within paraganglia, either near their fascicular connection or located deeper within the non‐neural tissue (Figure  6c ). Neuron distribution within the hypogastric nerve (HN). (a–c) Cleared blocks of HN were immunolabelled for neurofilament medium chain (NFM) and visualized by light sheet fluorescence microscopy. This showed the diversity of patterning of neuronal cell bodies within the HN, including large aggregates embedded in tracts (a), isolated neurons (a, arrow), clusters positioned close to intersections of multiple neural tracts (b), or neurons embedded near the surface of a paraganglion (c). Neural tracts in a and b and the paraganglion in c are outlined by the dotted line. In the paraganglion the only NFM‐positive structures are neurons, as SIF cells do not express neurofilament proteins. (d) Quantification of neurons within all blocks available for each HN (5–7 blocks per nerve). Blocks are numbered from rostral (SHP end) to caudal (IHP end) of each HN which corresponds to the subdissection illustrations in Figure  S1 . The total number of neurons present in the nerve is shown in bold after the last block. The size of each block is plotted against its length, and the heat map represents the number of neurons per mm of length of HN, with 370 per mm being the highest density found across the specimens. For three specimens, HN was available from both sides and the rest were counted using immunolabelling for NFM or Tuj1. In one nerve (Donor 3, right), immunostaining for TH was performed and expressed here as the percentage (%) of all neurons for each block. (e) A group of neurons (NFM), most of which are TH+, from block 4 of donor 3, right HN. Examples of TH‐negative neurons are indicated (arrowheads). Scale bars: 100 μm. Unembalmed tissue shown in panels (a–c, e) was from Donor 3 (male, 78 years). Donor and specimen details are provided in Table  1 , block number for each image panel in Table  3 , and block location shown in Figure  S1 . Thousands of neurons were identified in cleared samples from the entire length of each HN (Figure  6d ). Where HN from each side were available, the number of neurons was similar on each side (right 6714 ± 1077, left 6321 ± 1358, mean ± SEM; n  = 4 samples, 3 bilateral, 1 unilateral). Neurons were distributed along the whole length of each HN, rather than being concentrated close to the SHP or IHP. In one HN where cleared blocks were also immunostained for both NFM and TH, in 5 of the 6 blocks more than 90% of the neurons were TH‐IR (Figure  6d,e ). In this nerve, the block closest to the IHP had a lower proportion of TH‐IR neurons (66.9%). A selection of cleared, immunolabeled blocks ( n  = 8) of SHP from one donor showed that 97.2% (10333/10635) of neurons were TH‐IR, showing their similarity to HN neurons. The SHP is a component of the prevertebral sympathetic ganglia and the HN a major trajectory for sympathetic postganglionic axons; therefore, we first determined the features of neural structures immunolabeled for tyrosine hydroxylase (TH; presumed noradrenergic). This was performed in cryosections of three unembalmed specimens, viewed with confocal microscopy. In all cases, the cellular features were quite similar, showing numerous bundles of TH‐IR axons (Figure  7a,b ) and aggregates of TH‐IR neuronal cell bodies (Figure  7c,d ). Within larger aggregates of axons, fortuitously cut obliquely or transversely, the TH‐IR axons were distributed throughout the tract rather than showing a specific pattern of aggregation (Figure  7a ). In sections cut through the major axis of smaller tracts (Figure  7b ), merging of axon fascicles was identified, resembling observations made with LSFM; occasionally, individual TH‐IR axons could be followed travelling from one branch to another (Figure  7b ). Tyrosine hydroxylase‐immunoreactivity (TH) distribution in axons and neurons of the superior hypogastric plexus (SHP) and hypogastric nerve (HN). (a) A large tract of TH‐IR axons within the SHP, cut obliquely and showing that TH‐IR axons are distributed throughout the tract. (b) Small bundles of TH‐IR axons associated with the HN, cut along their major axis and showing a region of intersecting fascicles. (c, d) TH‐IR neurons in the SHP (c) and HN (d) have a variable soma size (examples of small neurons indicated by arrowheads and large neurons by arrows). (e, f) Cholinergic varicosities (synaptic boutons), identified by vesicular acetylcholine transporter (VAChT) immunoreactivity, are primarily located in the neuropil between somata, with only a minority closely associated with the soma surface of some neurons (examples indicated by asterisks). Confocal microscopy images are maximum projections of a 12 μm Z‐Stack at 2 μm steps (a, b) and single optical sections (c–f). Scale bars: 100 μm (a, b) and 50 μm (c–f). Unembalmed tissue shown in panels (a–c) was from Donor 3 (male, 78 years), (d) from Donor 5 (female, 42 years) and (e, f) from Donor 4 (male, 22 years). Donor and specimen details are provided in Table  1 , block number for each image panel in Table  3 , and block location shown in Figure  S1 . TH‐IR neurons were embedded within neural tracts in both the SHP and HN. These neurons exhibited considerable variability in size and shape (Figure  7c,d ). To visualize synaptic boutons of spinal preganglionic neurons that innervate autonomic ganglion neurons, we performed immunolabeling for vesicular acetylcholine transporter (VAChT). This revealed VAChT‐IR varicosities (putative cholinergic synaptic boutons) in close proximity to TH‐IR neuronal somata of the SHP (Figure  7e ) and HN (Figure  7f ). However, the majority were located in the neuropil between TH‐IR somata, where they are presumed to be making synaptic connections with dendrites of these somata. VAChT‐IR was not found in smooth (non‐varicose) axons. To investigate non‐noradrenergic axons and somata, we immunolabeled for two substances previously identified in cholinergic peripheral nerves (Elfvin et al.,  1997 ; Lundberg et al.,  1979 ; Persson et al.,  1998 ), vasoactive intestinal peptide (VIP) (Figure  8 ) and neuronal nitric oxide synthase (nNOS) (Figure  9 ). Vasoactive intestinal peptide (VIP) distribution in axons and neurons in the superior hypogastric plexus (SHP) and hypogastric nerve (HN). (a) Section through axon tracts showing separate tyrosine hydroxylase‐immunoreactive (TH‐IR) and VIP‐IR axon populations in the SHP, where VIP‐IR axons comprise the minority. (b) No VIP‐IR neuronal somata were identified in the SHP, but many VIP‐IR axons were located near TH‐IR somata, traversing regions close to aggregates of neurons or forming perisomatic varicosities surrounding specific TH‐IR neurons (b, arrowhead). The asterisk indicates a TH‐/VIP‐negative neuron, visible with autofluorescence (lipofuscin) but not immunostaining. (c) Separate TH‐IR and VIP‐IR axon populations in the HN have a similar distribution to what was seen in the SHP. (d) No VIP‐IR neuronal somata were identified in the HN; however, occasional perisomatic varicosities surround specific TH‐IR neurons as seen in the SHP (d, arrowhead). Confocal microscopy images are single optical sections. Scale bars: 50 μm. Unembalmed tissue shown in panels (a, b) was from Donor 3 (male, 78 years) and (c, d) from Donor 5 (female, 42 years). Donor and specimen details are provided in Table  1 , block number for each image panel in Table  3 , and block location shown in Figure  S1 . Neuronal nitric oxide synthase (nNOS) distribution in axons and neurons in the superior hypogastric plexus (SHP) and hypogastric nerve (HN). (a) Section through axon tracts showing separate tyrosine hydroxylase‐positive (TH‐IR) and nNOS‐IR axon populations in the SHP, where nNOS‐IR axons comprise the minority and are interspersed between TH‐IR axons. (b) No nNOS‐IR neuronal somata were identified in the SHP, but nNOS‐IR axons traverse regions close to aggregates of neurons. (c) TH‐IR and nNOS‐IR axon populations in the HN have a similar distribution to what was seen in the SHP. (d) In the HN, nNOS‐IR neurons were occasionally identified but these were TH‐negative (arrowheads). Confocal microscopy images are single optical sections. Scale bars: 50 μm. Unembalmed tissue shown in panels (a–c) was from Donor 3 (male, 78 years) and (d) from Donor 5 (female, 42 years). Donor and specimen details are provided in Table  1 , block number for each image panel in Table  3 , and block location shown in Figure  S1 . VIP‐IR axons were present in many neural tracts of the SHP and HN. VIP‐IR axons were not TH‐IR and overall comprised the minority of axons within tracts (Figure  8a,c ). Typically, the VIP‐IR axons were in small clusters distributed amongst the TH‐IR axons (Figure  8a,c ). For both SHP and HN, the prevalence of VIP‐IR axons in nerve tracts varied between individual tracts. In both the SHP and HN, VIP‐IR axons were also located close to TH‐IR somata, where they appeared to be traversing neuron‐dense regions, or as VIP‐IR varicosities (putative synaptic boutons) closely surrounding individual TH‐IR neurons (Figure  8b,d ). No VIP‐IR neurons (somata) were found in either the SHP or HN across the three specimens. In both the SHP and HN, the distribution of nNOS‐IR axons was similar to that of VIP‐IR axons, with nNOS‐IR axons arranged in bundles dispersed among the TH‐IR axons, and no co‐localization with TH (Figure  9a,c ). Similarly to VIP‐IR axons, the number of nNOS‐IR axons varied across nerve tracts. While no nNOS‐IR neurons were identified in the SHP (Figure  9b ), four TH‐negative/nNOS‐IR neurons were identified in a single middle sample block of the left HN. These accounted for a small number of the 188 total neurons identified across that specimen's HN cryosections (Figure  9d ). In autonomic ganglion neurons, VIP is frequently co‐expressed with nNOS (Ceccatelli et al.,  1994 ; Elfvin et al.,  1997 ; Parr & Sharkey,  1996 ), but in the current study, we were unable to investigate this directly due to limited availability of samples. The neuropeptides calcitonin gene‐related peptide (CGRP) and substance P (SP) are expressed by many visceral sensory neurons innervating the pelvic organs (Dail & Dziurzynski,  1985 ; Papka & McNeill,  1992 ) and in rodents have a high level of co‐localization in sensory axons that traverse pelvic ganglia (Bertrand et al.,  2020 ). We immunolabeled for both peptides as surrogate markers of visceral afferents traversing the SHP and HN. Numerous axons co‐labeled for both peptides were present in the nerve tracts of each sample, with many of these axons showing punctate staining (Figure  10a,c ). Some axons labeled for only CGRP or SP were also observed, but these were in the minority. The density of CGRP‐IR and SP‐IR axons varied between different nerve tracts in the SHP and HN. Calcitonin gene‐related peptide (CGRP) and substance P (SP) distribution in axons in the superior hypogastric plexus (SHP) and hypogastric nerve (HN). (a) CGRP and SP‐immunoreactive axons strongly colocalize within fascicles of the SHP. (b) Two neurons within the SHP visualized by their lipofuscin autofluorescence in the 488 nm channel show distinct patterns of innervation, with only one (left) encircled by a dense network of CGRP/SP‐immunoreactive varicosities and the other (right) devoid of CGRP/SP innervation. (c) Nerve tracts in the HN had a similar distribution of CGRP and SP‐IR axons as seen in the SHP (a), although a small minority of axons are labelled for only CGRP or SP (an example of a CGRP‐negative SP‐IR axon indicated by the arrow). (d) Cluster of neurons in the HN that have been visualized by their autofluorescence, showing that some but not all of these are in close proximity of perisomatic CGRP/SP‐IR varicosities (arrowheads). Confocal microscopy images are maximum projections of a 19 μm (a, d), 22 μm (b), and 27 μm (c) Z‐Stack at 1 μm steps. Scale bars: 50 μm. Unembalmed tissue shown in panels (a) was from Donor 3 (male, 78 years), (b) from Donor 4 (male, 22 years) and (c, d) from Donor 5 (female, 42 years). Donor and specimen details are provided in Table  1 , block number for each image panel in Table  3 , and block location shown in Figure  S1 . We also utilized the lipofuscin autofluorescence of neuronal somata to assess the relationship between varicose CGRP and SP axons and neurons embedded within the SHP and HN. In both structures, we found that a minority (approximately 1 in 5 neurons observed) of neuronal somata were surrounded by CGRP/SP‐IR axons (Figure  10b ). When the entire soma was present within a section, three‐dimensional visualization of these boutons demonstrated the dense perisomatic network of putative sensory axons (Movie  3 https://doi.org/10.26188/30736439 ). Less dense perisomatic putative sensory boutons can also be seen (Figure  10d ). The majority of perisomatic boutons showed co‐localization of both CGRP and SP, but occasional boutons contained only CGRP‐ or SP‐IR.

In this multiscale imaging study on the human sympathetic nervous system, we provide new insights into the organizational features and cell types of two major components that regulate the pelvic viscera, the HN and SHP. These advances were based on the unique availability of specimens from an organ donor program, facilitating multi‐channel immunofluorescence and application of tissue clearing of whole mount tissue blocks, LSFM, and high‐resolution confocal microscopy. We also found that the intrinsic autofluorescence of neural and vascular tissue (Hieda et al.,  2013 ; Hosaka et al.,  2014 ) was maintained in cleared samples dissected from embalmed specimens. This enabled additional 3D mapping of neuronal cell bodies, nerve tracts, and vasculature across unsectioned samples of HN and SHP. This overcomes a common limitation in human anatomical research where unembalmed tissue is not always available and opens further opportunities for quantitative, large‐volume neural mapping in the periphery. An additional novel aspect of our study was the use of tissue clearing as an aid to search for sparse clusters of neurons that could then be studied further and at higher resolution in sections. This approach will be valuable for many future studies on sparsely distributed structures within large samples. This study has several limitations. Full‐length analysis of the HN could only be performed where we were confident that the complete length of the nerve was present. We also note that the SHP was only sampled from the caudal region of this plexus so should not be considered representative across the entire SHP. The limited number of donors and their differences in age and sex precluded statistical analysis to investigate potential variation based on these or other biological features. The utilization of neuronal autofluorescence to visualize neurons in all samples (embalmed and unembalmed) enabled us to quantify the total number of autofluorescent neurons and their distribution; however, the small number of donors precluded statistical correlation of autofluorescence with donor age. Our visualization of lipofuscin in whole mounts clearly indicated the clumped nature of these autofluorescent granules within the neuronal cytoplasm, highlighting the possibility of false negatives if using autofluorescence alone to quantify neurons in thin sections. Our study identified many structural similarities between the SHP and HN, both comprising complex branching axon tracts with embedded aggregates of neuronal cell bodies. The locations of these neuronal aggregates could not be inferred at the macroscopic level but were clearly identifiable in cleared, unsectioned blocks of embalmed or PFA‐fixed specimens. Our specimens of caudal SHP were known to be incomplete, so we focused our quantitative study of neuronal cell bodies on the HN. Although ganglion cells have also been observed in histological sections of the human HN (Imai et al.,  2006 ; Jang et al.,  2015 ; Kraima et al.,  2015 ; Takenaka et al.,  2005 ) and rodents (Costa & Furness,  1973 ; Elfvin et al.,  1997 ), our study builds on upon these findings by providing 3D visualization and quantification of neurons within unsectioned samples of the adult human HN. Comparable numbers of neurons (2000–10,000 per nerve) were identified in semi‐serial sections of the HN (i.e., 10 μm sections taken every 20 μm) (Imai et al.,  2006 ). The HN is typically regarded as a simple nerve tract that is a major conduit for autonomic and sensory axons; however, the prevalence of embedded neuronal cell bodies raises questions about the origin of its neurons. Developmental studies have shown that the SHP and IHP ganglion cell clusters form before the HN (Arango‐Toro & Domenech‐Mateu,  1993 ; Kruepunga et al.,  2020 ); therefore, small ganglia and solitary neurons that are present in the fetal HN (Bruska,  2005 ) could potentially originate from the SHP or IHP, migrating along the HN during development. We dedicated one complete HN specimen to quantifying the proportion of TH‐IR (presumed noradrenergic) neurons within different HN regions and found that >90% of HN neurons were TH‐IR, except for the block closest to the IHP (67% TH‐IR). This may indicate that most of the HN neurons migrate from the SHP, with only the most caudal HN neurons migrating from the mixed sympathetic‐parasympathetic IHP. Reported proportions of TH+ neurons within the HN range from 99% observed by Imai et al. ( 2006 ) to a more variable 20–83% found by Takenaka et al. ( 2005 ). While both studies provide overall proportions, they did not describe any variability along the length of the nerve as seen in our study. Sexual dimorphism was not investigated in this study due to the small sample size; however, Imai et al. ( 2006 ) found higher numbers of neurons within the male unilateral HN compared to the female. Animal studies also predict a greater number of axons in male than female HN (McLachlan,  1985 ); this may reflect a larger number of sympathetic preganglionic axons in males, where there are a greater number of sympathetic pelvic ganglion neurons (Greenwood et al.,  1985 ). In this study, we visualized the nerve tracts and vasculature of the HN in 3D and in conventional histological sections. To our knowledge, these tracts and vessels have not previously been demonstrated microscopically in their intact form, revealing an unprecedented view of their patterning and relationships. The complex branching and merging of nerve tracts precluded definitive counts; however, it would be feasible in future studies to investigate specific axon classes and their trajectories within this network. Likewise, mapping vascular structures in the context of the entire nerve could provide valuable insights into neural function and dysfunction associated with ischemic conditions or surgical manipulation. The presence of numerous neuronal cell bodies within the HN has significant clinical implications. Recent studies (Cho et al.,  2024 ; Ye et al.,  2023 ) emphasized the importance of identifying and preserving the HN during surgical interventions. The multi‐functionality of the HN underpins the diverse impact of injury to the HN, which may result in urinary retention, ejaculatory disorders, or altered rectal motility. It is appreciated that due to their positional variability, individualized dissection strategies are required to minimize the damage to axon tracts and optimize functional outcomes. The presence of many neurons embedded within the entire length of the HN raises several further considerations. Physical or ischemic injury to these neuronal cell bodies during surgery would have additional impact on pelvic organ regulation beyond that caused by axon damage. Development of targeted therapies to promote neural regeneration or neuroprotection should aim to replace the HN‐embedded neurons in addition to restoring axon connections from the many nerve tracts traversing the HN. The presence of so many neuronal cell bodies within the HN is also relevant to the development of neuromodulation therapies for conditions related to pelvic organs. Neuromodulation has been previously applied to the SHP to control visceral pain and improve symptoms of the bladder (Kim et al.,  2016 ; Possover & Chiantera,  2009 ). To our knowledge, the presence of neurons within the HN has not been previously considered in refinement of this approach or interpretation of its outcomes. The SHP and HN also showed a striking similarity at the cellular and molecular levels, having comparable molecular classes of neurons and axons, and with similar structural relationships. Our functional interpretation of these observations, as outlined below, is largely based on animal studies of the autonomic and visceral sensory systems, as there has been comparatively little neuronal subclassification by direct multi‐label immunostaining or transcriptome analysis in the human peripheral nervous system. Extensive studies in experimental animals have characterized the pre‐ and postganglionic sympathetic axons in the HN, in addition to identifying visceral afferents from thoraco‐lumbar dorsal root ganglia (Baron & Jänig,  1988 ; Guo et al.,  2020 ; Harji et al.,  1998 ; McLachlan,  1985 ; Nadelhaft & Vera,  1991 ; Nance et al.,  1988 ; Neuhuber,  1982 ; Papka et al.,  1996 ; Steinman et al.,  1992 ). We also indicate below where comparable studies have been performed on prevertebral ganglia of rodents, noting that in these species the caudal prevertebral ganglia are consolidated into the inferior mesenteric ganglion (IMG), and the term SHP is not used. In one HN sample dedicated entirely to TH‐immunostaining, we found that more than 90% of the neurons we identified in the SHP and HN were immunoreactive for TH. We presumed these TH‐IR neurons were noradrenergic, although co‐expression of dopamine β‐hydroxylase would confirm this. Based on rodent studies, it is predicted that these neurons would be controlled by thoraco‐lumbar preganglionic neurons, so considered part of the sympathetic autonomic nervous system (Anderson et al.,  2009 ; Gibbins,  2013 ; Jänig & McLachlan,  1987 ). Our results concur with a previous study showing that 99% of neurons in the HN were TH+ (Imai et al.,  2006 ), but contrast with another study that found a much broader range of TH expression (20–83%) (Takenaka et al.,  2005 ). Both studies discussed significant interindividual variations, but from the data provided it was not possible to determine if rostrocaudal location in the HN contributed to this variability. TH‐IR neurons of the SHP and HN showed a wide range of sizes and shapes, raising the possibility that these represent functionally distinct subpopulations, as identified in rodent sympathetic ganglia (Boyd et al.,  1996 ; Gibbins,  1991 ). Based on rodent data, we would further predict distinct molecular profiles for functionally distinct populations of neurons (Gibbins,  1992 ; Wang et al.,  2025 ). The presence of a small but consistent population of TH‐negative neurons concurs with analyses of sympathetic ganglia in other species, where a small population of cholinergic neurons has been consistently described. For example, cholinergic neurons in the guinea pig IMG, identified by lack of TH or presence of choline acetyltransferase (ChAT), are typically clustered near the origin of the HN (Sann et al.,  1995 ). We did not investigate the cholinergic nature of the TH‐negative neurons in our study, due to limitations of antibodies required to demonstrate these directly. VIP and nNOS are consistently reported in small populations of non‐noradrenergic (presumed cholinergic) neurons in rodent prevertebral ganglia (Elfvin et al.,  1997 ; Heym et al.,  1984 ; Hökfelt, Elfvin, Schultzberg, Fuxe, et al.,  1977 ; Masliukov et al.,  2015 ; Parr & Sharkey,  1996 ; Sann et al.,  1995 ). We did not identify VIP‐IR somata in any sample and only found a small number of nNOS‐IR somata in two samples of HN, consistent with the prior rodent studies; these were TH‐negative. Immunohistochemistry revealed two structural classes of axons in the SHP and HN: smooth, presumed non‐terminal axons and varicose axons, where the small swellings (varicosities, also referred to as boutons en passant) indicate sites of transmitter storage and release. Smooth non‐terminal axons were primarily TH‐IR, with smaller groups being nNOS‐IR or VIP‐IR (but TH‐negative). We propose that these TH‐IR axons were postganglionic axons originating from neurons in the SHP, more rostral prevertebral ganglia, or the HN. The dominance of TH‐IR axons in the HN is consistent with the HN being the typical trajectory of pelvic organ‐projecting prevertebral sympathetic ganglion axons, which reach the organs via the IHP. Smooth axons immunoreactive for nNOS or VIP potentially originate from cholinergic neurons located in more rostral regions of SHP than sectioned here. In guinea pigs, accumulation of VIP on the caudal side of a HN crush, but nNOS on both sides of the crush, provides evidence of caudally projecting VIP axons in the HN but two populations (rostrally and caudally projecting) nNOS axons (Elfvin et al.,  1997 ). Enteric viscerofugal neurons provide another possible source of nNOS axons in the SHP and HN, but very few of these neurons express VIP (Chen et al.,  2024 ). We did not perform simultaneous immunolabeling with VIP and nNOS antibodies due to tissue availability, so we could not determine their level of co‐expression in our samples. No smooth axons were VAChT‐IR, consistent with this transporter being typically concentrated near the synaptic terminals of cholinergic axons. Varicose (presumed terminal) axons immunoreactive for VAChT were associated with all groups of neuronal cell bodies observed in the SHP and HN. We interpreted this as representing cholinergic synaptic inputs from spinal preganglionic neurons, although some may originate from enteric viscerofugal neurons (most of which are cholinergic; (Chen et al.,  2024 )). The proximity of VAChT‐IR varicosities to SHP and HN neuronal cell bodies varied widely between neurons, likely representing differences in synapse location on the soma or on dendrites. It is also important to recognize that in thin tissue sections, only a fraction of each neuron is visible, so the complete synaptic connectivity of each neuron cannot be deduced. VIP‐IR varicose axons were closely associated with a minority of neurons in the SHP and HN, where they surrounded TH‐IR somata and potentially indicated sites of synaptic contact. We did not identify VIP‐IR somata in these samples of HN and caudal SHP, so deduce that these VIP terminals arise from more rostral cholinergic prevertebral neurons. Our observations partly align with a prior study of human SHP (Kraima et al.,  2015 ), reporting VIP boutons associated with most of the SHP ganglion neurons; however, in contrast to our study, they did not identify ganglion neurons in the HN so did not report on VIP boutons in that site. Our results contrast with studies of the guinea pig IMG, where VIP varicose axons formed a much denser network, associated with most but not all neurons (Elfvin et al.,  1997 ; Hökfelt, Elfvin, Schultzberg, Fuxe, et al.,  1977 ). However, in this species, VIP is expressed by many viscerofugal neurons (Brookes,  2001 ), so cannot directly inform interpretation of human data. We also cannot discount the possibility that VIP terminals may have a sensory or preganglionic origin, although VIP is expressed in only a small proportion of sacral dorsal root ganglion neurons in rats (Keast & de Groat,  1992 ) and has very low expression in human sensory neurons (Yu et al.,  2024 ). Several immunohistochemical studies have been performed on sections of human HN or SHP (Imai et al.,  2006 ; Jang et al.,  2015 ; Kraima et al.,  2015 ). These all reported TH axons, interpreted as sympathetic, and some observed VIP axons that the authors categorized as parasympathetic. From the data presented, it is not clear why these were considered parasympathetic, as the anatomical source of their preganglionic innervation was not defined (i.e., as required for the conventional distinction of sympathetic and parasympathetic). It is possible that the term parasympathetic was used by the authors as a synonym for cholinergic, which does not necessarily infer that their functional classification (sympathetic ganglia also contain cholinergic neurons; (Ernsberger & Rohrer,  1999 )). Both studies also identified axons immunolabeled for myelin basic protein, which could represent preganglionic axons (as interpreted by the authors) or myelinated visceral sensory axons (Hulsebosch & Coggeshall,  1982 ). In the present study, we used immunoreactivity for the neuropeptides SP and CGRP as commonly co‐expressed surrogate markers for a major group of visceral sensory axons that include nociceptors (Alvarez et al.,  1988 ; Dail & Dziurzynski,  1985 ; Dalsgaard et al.,  1982 ; Gibson et al.,  1985 ; Maggi,  1995 ). We observed a highly localized association of SP/CGRP‐IR varicose axons with a small subpopulation of SHP and HN neurons. This relationship was seen as a dense basket‐like covering of varicose boutons, raising the possibility of synaptic communication between sensory and autonomic systems in discrete neural pathways rather than broadly across the sympathetic system. Sensory modulation of autonomic signaling in prevertebral ganglia has been characterized in animal studies (Dalsgaard et al.,  1982 ; Hökfelt, Elfvin, Schultzberg, Goldstein, & Nilsson,  1977 ; Matthews et al.,  1987 ), but to our knowledge has not been investigated in human prevertebral ganglia. Pericellular SP and CGRP boutons have been reported in the infant male human IHP (Jen et al.,  1996 ) and adult rat pelvic ganglia (Bertrand et al.,  2020 ), supporting this functional role.

In this multiscale microscopy study, we characterized in 3D the anatomical and immunohistochemical features of the SHP and HN in adult humans. This new data demonstrate thousands of neuronal cell bodies within the HN (indicating that this nerve does not function solely as a conduit for axons) and provides new insights into the anatomical organization, neurochemical classes, and spatial relationship of cell types within the SHP and HN. These findings have important implications for both anatomical research and clinical practice, particularly in the context of nerve‐sparing surgery and neuromodulation therapies, where precise knowledge of neural structures is essential.

Neural control of the pelvic organs is mediated by a complex network of peripheral nerves, sympathetic and parasympathetic autonomic ganglia (motor) and dorsal root ganglia (sensory). The autonomic ganglia are controlled by specific regions of the brain and spinal cord that integrate the interoceptive and nociceptive sensory signals from pelvic organs to activate and coordinate specific motor outputs. Together, these neural circuits maintain homeostasis and coordinate with volitional activity to perform voiding, continence, and sexual function (de Groat et al.,  2015 ; Giovannetti et al.,  2023 ; Jänig,  2022 ; Keast & Osborne,  2018 ). Selective activation of specific populations of sympathetic nerves is required for many pelvic organ functions, including smooth muscle contraction in the bladder neck, prostate gland, and vas deferens, reduced motility of the lower bowel, penile detumescence, and vasoconstriction. Axons innervating these organs and their vascular supply are located within several groups of sympathetic ganglia (Figure  1 ), defined as sympathetic because of their thoraco‐lumbar (T10‐L2) source of spinal preganglionic innervation (Gibbins,  2013 ; Jänig & McLachlan,  1987 ). The paired lumbosacral sympathetic chain (paravertebral ganglia) mediates vasoconstriction in all organs, whereas sympathetic regulation of other pelvic organ functions is mediated by the sympathetic ganglion neurons embedded in the superior hypogastric plexus (SHP) (a caudal continuation of the inferior mesenteric plexus (IMP)), and sympathetic neurons of the inferior hypogastric plexus (IHP); here they are intermingled with parasympathetic ganglion neurons innervated by sacral preganglionic neurons (Anderson et al.,  2009 ; Hancock & Peveto,  1979 ). The paired hypogastric nerves (HN) macroscopically connect the SHP and IHP. Autonomic and sensory pathways innervating the pelvic viscera. (a) The position of peripheral nerves and autonomic ganglionated plexuses in an adult human, showing the caudal prevertebral ganglia (inferior mesenteric plexus [IMP], superior hypogastric plexus [SHP]), hypogastric nerves (HN), pelvic and sacral splanchnic nerves, and the inferior hypogastric plexus (IHP). “Mixed” refers to the intermingling of sympathetic and parasympathetic neurons and pathways within the IHP. All of these structures other than the prevertebral ganglia are bilateral, although here for simplicity the IHP and its inputs from the pelvic splanchnic and sacral splanchnic nerves are shown on only one side. (b) A schematic representation of the peripheral neural circuits that are required for pelvic organ regulation. Where there are paired structures, only one side is shown here. The initial aim of our study was to define organizational features and cellular components of the adult human HN, comprising a major source and trajectory of sympathetic nerves regulating the pelvic organs. As well as aiming to extend the foundational understanding of the human autonomic nervous system, this study is also driven by surgical goals to inform nerve‐sparing procedures. Recent surgical studies (Cho et al.,  2024 ; Kim et al.,  2021 ; Ye et al.,  2023 ) emphasized the importance of identifying and preserving the HN during radical hysterectomy, total mesorectal excision, and para‐aortic lymphadenectomy, as injury to the HN may result in urinary retention, ejaculatory disorders, or altered rectal motility. To ensure that the complete length of HN was analyzed, our samples also included the most caudal region of SHP, and therefore our aim was extended to include characterization of this SHP tissue. The SHP has very close proximity to the aortic bifurcation, inferior mesenteric artery, ureters, and iliac vessels, making it highly vulnerable during rectal, vascular, and gynecologic surgery. Surgical mapping of the SHP (Kostov et al.,  2024 ) has clarified the relationships between the plexus, major vessels, and peritoneal reflections, emphasizing the importance of identifying this structure during pelvic dissection. Building foundational knowledge on the organization and cellular structure of these neural pathways may also provide new insights into mechanisms of pelvic pain (including endometriosis) and neuromodulation strategies targeting pelvic organ dysfunction. To address our aim, specimens of HN and caudal SHP were obtained from an established organ donor program and a body donor program. In both cases, methodologies were developed to visualize neural structures across scale—from macroscopic to microscopic. Current understanding of pelvic organ innervation is largely macroscopic, drawn from nerve‐sparing surgical literature (Aurore et al.,  2020 ; Baader & Herrmann,  2003 ; Beveridge et al.,  2018 ; Maas et al.,  2005 ; Mauroy et al.,  2007 ; Paraskevas et al.,  2008 ; Ripperda et al.,  2017 ) or from embryonic (Arango‐Toro & Domenech‐Mateu,  1993 ; Kruepunga et al.,  2020 ) and fetal tissue studies (Alsaid et al.,  2009 ; Bruska,  2005 ; Zaitouna et al.,  2013 ). Some of these studies have created or inferred three‐dimensional (3D) visualizations of pelvic autonomic structures; however, to date these studies have predominantly relied on thin (4–5 μm) sections of paraffin‐embedded samples from selected regions of formalin‐fixed cadaveric tissue (Hinata et al.,  2014 ; Imai et al.,  2006 ; Jang et al.,  2015 ; Kraima et al.,  2015 ; Takenaka et al.,  2005 ). While some of these studies have employed immunohistochemistry to distinguish neural structures or individual neural classes, observations from sections are difficult to interpret in a macroscopic context, and specific structures such as axon terminals or complex branching patterns are difficult to validate in 2D approaches. Many aspects of biology and biomedical science have been transformed by the development of new technologies of tissue clearing (making tissues fully transparent) and large‐volume microscopy (light sheet fluorescence microscopy; LSFM) (Stelzer et al.,  2021 ; Tian et al.,  2021 ; Ueda et al.,  2020 ). In combination with multi‐label immunofluorescence or other visible tags, these approaches have enabled 3D visualization of large, complex structures from their macroscopic context to their detailed microscopic and molecular features. We have adapted and applied these methods to define the HN and caudal SHP features in 3D in the adult human. Specifically, we have performed tissue clearing (a modification of the iDISCO clearing method (Renier et al.,  2014 )) and immunofluorescence of HN and associated caudal SHP specimens obtained from an established organ donor program. Availability of these specimens enabled use of a fixation method suitable for optimal molecular preservation of neural markers and multi‐channel immunofluorescence analysis. We selected markers for various neuronal classes, based on animal studies of visceral nerves, viewed structures in 3D using LSFM and segmented specific components within the HN using the virtual reality technology, syGlass (v2.4.0) (IstoVisio, Inc., USA). We also repurposed some of these specimens by applying a “reverse clearing” (i.e., rehydration) method (Blain et al.,  2023 ). This enabled us to re‐use the large‐volume cleared tissues for a second round of higher resolution analyses in sections (paraffin or cryosections). This was performed to further examine specific cellular and subcellular structures and locations of interest identified during LSFM. We also analyzed the HN and caudal SHP obtained from a body donor program where tissue preservation and treatment (embalming) can be unsuitable for immunofluorescence (Fritschy,  2008 ). Here we built upon our initial observations from organ donor specimens, identifying high levels of autofluorescence specifically within autonomic ganglion cells of the SHP. This led us to investigate if this specific neuronal autofluorescence was retained and neuronal structures sufficiently well preserved within the embalmed SHP and HN obtained from an anatomical body donor program.

Supplementary Figure 1. Subdissection and experimental allocation of donor samples. (a) Representative image of a complete sample prior to subdissection, showing the location of the superior hypogastric plexus (SHP), hypogastric nerve (HN), and inferior hypogastric plexus (IHP). Scale bar: 20 mm. (b–f) Schematic representations of the SHP (labelled s) and right (r) and left (l) HN subdissected into numbered blocks for specific experimental protocols: Grey blocks represent whole mount samples cleared using an Adipo‐clear iDISCO tissue clearing protocol prior to light sheet fluorescence microscopy (LSFM) for either autofluorescence only (AF; plain grey) or immunohistochemistry (IHC; grey with green dots); purple–pink gradient blocks indicated paraffin‐embedded samples for hematoxylin and eosin (H&E) staining; and green blocks are samples that underwent cryosectioning with IHC and subsequent confocal microscopy. Faded black horizontal lines signify whole mount samples that were originally processed for AF LSFM before being rehydrated and reprocessed for the indicated sectioning methodology. Note that Donor 5 (f) was an incomplete sample where the HN was identified during subdissection and confirmed via LSFM. Supplementary Figure 2. Workflow for tissue processing through steps of clearing and, for selected blocks, reverse clearing (rehydration), paraffin sections, and cryosectioning. The time (days) and protocol steps are associated with the adapted Adipo‐Clear (blue boxes) iDISCO (orange boxes) with immunohistochemistry (IHC; green boxes) and without immunolabeling, imaged with autofluorescence only (AF). Subdissected tissue could also be used for paraffin‐embedded sections with subsequent H&E staining or cryosections with IHC. Select whole mount blocks which underwent clearing without immunolabeling were rehydrated (i.e., “reverse cleared”) and reprocessed. Unembalmed tissue block shown here is from Donor 3 (male, 78 years). Donor and specimen details are provided in Table  1 , block number in Table  3 , and block location shown in Figure  S2 . Supplementary Figure 3. Examples of autofluorescent and immunolabeled neurons in whole mount cleared sample blocks imaged using light sheet fluorescence microscopy. The Imaris spot tool neuron counting method has been used on identified neuron clusters (green or cyan spheres). (a) A sample block of HN from an embalmed donor with the spot tool used on unstained autofluorescent neurons (a2). (b) A higher magnification (3×) image of the boxed region in a. The white arrowheads indicated two examples of individual neurons, outlined by the white dotted line. The red dashed line outlines the lipofuscin granule clusters in each neuron. (c) A sample block of HN from an unembalmed donor with the spot tool used on unstained autofluorescent neurons. (d) A sample block of HN from an unembalmed donor stained for neurofilament medium chain (NFM). Lipofuscin can be seen in the autofluorescence channel (d1, d2), cyan spheres have been used to count neurons using lipofuscin only (d2). Neurons are immunoreactive for NFM (d3, d4), green spheres have been used to count NFM+ neurons (d4) (e) A higher magnification (2×) optical section (1.62 μm thick) using the Ortho Slicer tool in Imaris of the boxed region in d. The white asterisks indicate the neurons identified in the maximum projection image (d1 and d3) and the optical section image (e). Scale bars: 100 μm. Embalmed tissue shown in panels (a, b) was from Donor 1 (male, 101 years) and (c, d) from Donor 3 (male, 78 years). Donor and specimen details are provided in Table  1 , block number in Table  3 , and block location shown in Figure  S1 . Movie 1: 3D visualization of the adult human caudal superior hypogastric plexus (SHP). DOI: https://doi.org/10.26188/30736406 . A sample block of the adult human (male, 75 years) caudal superior hypogastric plexus (SHP) of an embalmed specimen processed for tissue clearing and viewed by light sheet fluorescence microscopy using 561 nm autofluorescence (no antibody staining). Neural tracts and blood vessels can be clearly observed using autofluorescence without additional staining. Some strong vascular autofluorescence is present, primarily due to residual blood cells from the fixation method. Neural tracts in distinct planes were segmented using syGlass and are shown in yellow, green, and blue. A blood vessel is segmented in red. The arrangement of these structures can clearly be seen in a transverse plane using the Imaris Ortho Slicer tool. In a coronal plane, regions of interconnecting fascicles, neuronal cell bodies embedded in the tracts and vascular branching are evident. Donor and specimen details are provided in Table  1 , block number for this image in Table  3 , and block location shown in Figure  S1 . Movie 2: 3D visualization of the adult human hypogastric nerve (HN). DOI: https://doi.org/10.26188/30736397 . A sample block of the adult human (male, 78 years) hypogastric nerve (HN) preserved by paraformaldehyde (PFA) fixation (unembalmed), followed by tissue clearing and light sheet fluorescence microscopy imaged using 561 nm autofluorescence (no antibody staining). Neural tracts and blood vessels can be clearly observed using autofluorescence without additional staining. The neural tracts run parallel to one another, and some contain embedded neuron clusters. A prominent vessel overlies these tracts and demonstrates a complex vascular branching pattern. Transverse optical slices using the Imaris Ortho Slicer tool show these structures running parallel to one another for a region of the block. A putative paraganglion, which was later validated by immunostaining for tyrosine hydroxylase (TH) (not shown), has anatomical fascicular connections to the HN, is highly vascularized, and contains a small number of neurons embedded in it. Donor and specimen details are provided in Table  1 , block number for this image in Table  3 , and block location shown in Figure  S1 . Movie 3: 3D visualization of calcitonin gene‐related peptide (CGRP)‐ and substance P (SP)‐immunoreactive varicosities around a neuron within the superior hypogastric plexus (SHP). DOI: https://doi.org/10.26188/30736439 . A sample block of adult human (male, 22 years) SHP was preserved by paraformaldehyde (PFA) fixation (unembalmed) and prepared for immunofluorescence. Cryosections (50 μm) of the SHP were stained with CGRP and SP to investigate putative visceral afferent fibers. A Z‐stack of the neuron and associated CGRP‐ and SP‐immunoreactive varicosities was captured with confocal microscopy. Manual segmentation of the neuronal cell body (white) using autofluorescence (captured in the 488 nm channel) was completed using the Surfaces tool in Imaris. Semi‐automated segmentation of CGRP (green) and SP (magenta) varicosities show their density, arrangement, and colocalization around the neuron in a three‐dimensional view. Donor and specimen details are provided in Table  1 , block number for this image in Table  3 , and block location shown in Figure  S1 .