Extrinsic innervation of the pelvic organs in the lesser pelvis of human embryos

Abstract Realistic models to understand the developmental appearance of the pelvic nervous system in mammals are scarce. We visualized the development of the inferior hypogastric plexus and its preganglionic connections in human embryos at 4–8 weeks post‐fertilization, using Amira 3D reconstruction and Cinema 4D‐remodelling software. We defined the embryonic lesser pelvis as the pelvic area caudal to both umbilical arteries and containing the hindgut. Neural crest cells (NCCs) appeared dorsolateral to the median sacral artery near vertebra S1 at ~5 weeks and had extended to vertebra S5 1 day later. Once para‐arterial, NCCs either formed sympathetic ganglia or continued to migrate ventrally to the pre‐arterial region, where they formed large bilateral inferior hypogastric ganglionic cell clusters (IHGCs). Unlike more cranial pre‐aortic plexuses, both IHGCs did not merge because the 'pelvic pouch', a temporary caudal extension of the peritoneal cavity, interposed. Although NCCs in the sacral area started to migrate later, they reached their pre‐arterial position simultaneously with the NCCs in the thoracolumbar regions. Accordingly, the superior hypogastric nerve, a caudal extension of the lumbar splanchnic nerves along the superior rectal artery, contacted the IHGCs only 1 day later than the lumbar splanchnic nerves contacted the inferior mesenteric ganglion. The superior hypogastric nerve subsequently splits to become the superior hypogastric plexus. The IHGCs had two additional sources of preganglionic innervation, of which the pelvic splanchnic nerves arrived at ~6.5 weeks and the sacral splanchnic nerves only at ~8 weeks. After all preganglionic connections had formed, separate parts of the inferior hypogastric plexus formed at the bladder neck and distal hindgut.


| INTRODUC TI ON
The pelvic organs, which occupy the lesser or 'true' pelvis, are innervated by the autonomic nervous system. The caudal portion of the vagal neural crest cells is the major source of neural crest-derived cells (NCCs) of the intrinsic enteric nervous system (ENS; Durbec et al. 1996;Anderson et al. 2006;Simkin et al. 2013;Espinosa-Medina et al. 2017). However, studies in chicken (Le Douarin and Teillet, 1973;Burns and Le Douarin, 1998), and subsequently in rodents (Serbedzija et al. 1991;Anderson et al. 2006), have shown that NCCs that originate distal to vertebral level L1-2, which corresponds to somite 28 in chicken (Le Douarin and Teillet, 1973) and somite 24 in mice (Dong et al. 2006), contribute to the mature ENS of the colon.
The timeline of the development and distribution of these sacral NCCs in mammals is described in most detail for rodents. In mice, these cells emigrate from the neural tube at embryonic day (ED) 9.0-9.5, aggregate in the para-aortic region at ED10.5-11.0, and form the pelvic ganglia ventrolateral to the hindgut at ED11.5-12.5. From here they enter the wall of the gut or base of bladder at ED13.5-14.0 and have colonized the entire postcoecal gut at ED14-14.5. Soon thereafter, differentiation into intramural ganglia and the formation of smooth muscle layers begins (Serbedzija et al.1991;Kapur, 2000;Dong et al. 2006;Wang et al. 2011;Erickson et al. 2012). Vagal NCCs emigrate from the neural tube at ED8.0-8.5, enter the gut at ED9.5, arrive at the coecum at ED10.5, bypass the coecum via the dorsal mesentery to move into the postcoecal gut at ED12.5, reach the midpoint of the colon by ED13.5 and the hindgut at ED14.5 (Kapur et al. 1992;Durbec et al. 1996;Young et al. 1998;McKeown et al. 2001;Druckenbrod and Epstein, 2005;Nishiyama et al. 2012). These data imply that the sacral NCCs take two to three times longer to move from the neural tube to the wall of the gut, but migrate approximately twofold faster through the postcoecal gut than the vagal NCCs.
From these studies, the general picture emerges that the development of the pelvic nervous system in humans is similar to that in rodents. Unfortunately, relatively few embryos were studied and the staging was rather imprecise. Perhaps even more limiting is the fact that these studies did not provide spatial models [apart from a single, elegant wax model of an 8-week-old embryo in Arango-Toro and Domenech-Mateu (1993)]. It is, therefore, difficult to appreciate topographic relations between the gut, nerves and pelvic organs.
In the present study, we investigated the development of the inferior hypogastric plexus and its preganglionic connections with the central nervous system: the hypogastric, pelvic and sacral splanchnic nerves. Recently, the pelvic splanchnic nerves were shown to be developmentally and phenotypically sympathetic (Espinosa-Medina et al. 2016), defining all three inputs as sympathetic. For this reason, we carefully mapped the timeline of their appearance and contact with the bilateral pelvic ganglionic cell clusters. Another reason to map the development of the hypogastric plexus and nerves was the difference in the timelines of the vagal (cranial) and sacral (caudal) contributions to colonic innervation. We further defined the topographical boundaries of the lesser pelvis in the embryo.

| Embryos
This study was undertaken in accordance with the Dutch regula- were studied (Table 1). In addition, digital images of carefully staged human embryos from the Carnegie collection (Washington, DC, USA) were included from the Digitally Reproduced Embryonic Morphology project (http://virtu alhum anemb ryo.lsuhsc.edu).

| Image acquisition, 3D reconstruction and visualization
Human embryos of between 4 and 8 weeks post-fertilization development were investigated. The modified O'Rahilly's criteria were used to define the Carnegie stage (CS) of development and postfertilization age (O'Rahilly and Muller, 2010; Table 1). A graph relating the CSs of human embryos to days of development in mice or Hamilton-Hamburger stages (Hamburger and Hamilton, 1951) in chicken are found in Figure S1. Serial sections from the named historical collections were digitized with an Olympus BX51 or BX61 microscope and the Dotslide program (Olympus) to provide highresolution digital images. Serial sections of the Blechschmidt collection were digitized with a Zeiss Axio Scan.Z1 (Carl Zeiss Microscopy).
All digital images were transformed into greyscale 'JPEG' format and imported into Amira3D (version 6.5; FEI Visualization Sciences Group Europe). The imported images were aligned automatically with the least-squares function and then manually optimized by correction for the embryonic curvature with the aid of photographs and magnetic resonance images of the same stages of human embryos (Pooh et al. 2011). Structures of interest were segmented manually and used to generate 3D shapes with the Amira3D program.
To eliminate the distracting noise in the Amira3D output attributable to section processing and stacking, Amira3D polygon meshes were exported via 'vrml export' to Cinema 4D (version R21; MAXON Computer GmbH) and remodelled using the Amira3D model as a template. Concurrent visualization of the Amira3D template and the remodelled Cinema4D model in Cinema 4D was used to verify the accuracy of the Cinema4D models ( Figure S2). The Cinema4D models were transferred via 'wrl export' to Adobe Acrobat version 9 (http://www.adobe.com) to generate interactive 3D PDF files, which are a user-friendly format for 3D visualization ( Figures S3 and   S4). We mostly refer in the text to the figures to relate histology to TA B L E 1 Metadata of human embryos and foetuses that were studied

| Terminology
Intestinal development in avian (Southwell, 2006) and mammalian embryos (Soffers et al. 2015)  in both birds and mammals (Le Douarin and Teillet, 1973;Dong et al. 2006) and only colonize the 'post-umbilical' gut in a caudocranial gradient that becomes progressively steep during development (Le Douarin and Teillet, 1973;Burns and Le Douarin, 1998;Anderson et al. 2006). Whereas the avian vitelline duct remains present and patent until hatching (Esteban et al. 1991), the mammalian vitelline duct already disappears at ~CS15 in rodent (Lamers et al. 1987) and human (Soffers et al. 2015) embryos. Due to the disappearance of the vitelline duct in mammalian embryos, the cranial boundary of the post-umbilical gut can no longer be delineated accurately.
Accordingly, the part of the gut colonized by sacral NCCs is variously referred to as the hindgut (Young and Newgreen, 2001;Wang et al. 2011), post-coecal hindgut (McKeown et al. 2001, or colorectum (Young and Newgreen, 2001). In agreement with our accompanying study, we will identify the mammalian equivalent of the avian postumbilical gut as the distal loop of the midgut and hindgut.
Some terminology is confusing because embryonic and definitive structures differ markedly in appearance. Relevant for the present study are the superior hypogastric nerve and plexus, the inferior hypogastric (ganglionic) cluster and plexus, and the hindgut and rectum. The superior hypogastric plexus acquires its definitive configuration upon the division of the single large splanchnic nerve into many smaller nerve strands during CS18-20. Similarly, we describe the inferior hypogastric plexus as a cluster of ganglionic cells until it becomes populated by nerves at CS20.
Our detailed study of the extrinsic innervation of the caudal part of the gut showed that the junction of the mid-and hindgut corresponded with the location of the stem of the inferior mesenteric artery (IMA) rather than the cranial end of its left colic branch. The stem of the IMA branches from the aorta at vertebral level T12-L1 during CS15 and has descended to L2-3 at CS20 (Evans, 1912). This level corresponds, in turn, with the cranial boundary of the sacral neural crest (Le Douarin and Teillet, 1973;Dong et al. 2006) and, as we will show, with the rectosigmoidal junction rather than the transverse colon in the adult. We will describe the development of the junction between the mid-and hindgut trunk in more detail in a separate study.

| RE SULTS
In Kruepunga et al., in press), we identified the neural crest-derived ganglionic cells in the thoraco-lumbar region by their topography and intense staining properties. In this study, we investigated the appearance and migration of NCCs in the caudal-most portion of the body, the lesser pelvis.

| Boundaries of the lesser pelvis in the embryo
We define the lesser pelvis in the embryo as the pelvic area caudal to

| Formation of the inferior hypogastric ganglionic-cell cluster
In the pelvic area, NCCs were first found dorsolaterally to the median sacral artery near level S1 in CS15-early embryos (~36 days of development; blue dots in Figure 2A and blue arrows in Figure 2C).
Topographically, these ganglionic cells were a caudal continuation of similarly located ganglionic cells in the lumbar region and, accordingly, passed the bifurcating umbilical arteries dorsally ( Figure S2A, CS14early). At CS15-late (~37 days), the dorsolateral NCCs had extended The course of the umbilical arteries defines the upper boundary of the lesser pelvis. Panels a-f show side views of the dorsal aorta and its major ventral branches between Carnegie Stage (CS)14 and CS22, with the notochord representing the embryonic curvature and the grey spheres marking segments L1, S1 and S5. At CS14 (33 days; panel a) the plane through the bifurcation of the umbilical arteries was orthogonal to a frontal plane through the aorta. As the caudal body axis unfolded (Kruepunga et al. 2018), the angle between both planes straightened to 160-170° at CS22. Concomitantly with the unfolding of the embryonic axis, the line through the subpubic arch and vertebra S5 unfolded from CS16 onwards (panels c-f). Note that the unfolding process plateaus in the lumbar region at CS17, but continues in the sacral region. Bars = 500 µm. NC, notochord; UA, umbilical arteries; UR, umbilical ring. [Colour figure can be viewed at wileyonlinelibrary.com] their presence to level S5 (blue dots in Figure 2F and blue arrows in Figure 2H,J). The most cranial portion of these cells began to consolidate as ganglia of the sympathetic trunk along the median sacral artery (blue arrows in Figure 2H). Furthermore, many scattered cells had now accumulated ventrally to the median sacral artery and laterally to the hindgut, where they occupied on both sides a triangular area with its base dorsally between S1 and S5, and its apex ventrally near the entrance of the common nephric portion of the Wolffian ducts into the urogenital sinus. These ganglionic cells were found caudal to the umbilical arteries and therefore did not extend into the abdominal cavity ( Figure 2F). Their ventral extension was situated laterally to the pelvic pouch of the coelomic cavity. This pelvic pouch located in the urorectal septum and surrounded the hindgut ventrally and laterally down to the cloaca (Kruepunga et al. 2018; Figure S2B), so that the ventrolateral ganglionic cells were not in direct contact with the wall of the hindgut. At CS16 (~39 days of development; Figure 3A-C), ganglionic cells had advanced further ventrally along the lateral side of the pelvic pouch to form the left-and right-sided, sagittally oriented inferior hypogastric ganglionic cell clusters (IHGCs; Figure 3B). The presence of the pelvic pouch explains why both IHGCs persisted as separate entities throughout subsequent development. More cranially, in a plane perpendicular to the notochord at level S1 and through the junction of mid-and hindgut, cells of the IHGCs that remained located mediodorsally to the hindgut connected to the well-developed superior hypogastric nerve (SHN in Figure 3A; asterisk in Figure 4A), which is the common caudal continuation of the left-and right-sided lumbar splanchnic nerves along the superior rectal branch of the IMA.
Strikingly, no caudal extensions of ganglionic cells toward the cloaca were seen. Ganglionic cells dorsolaterally to the median sacral artery had now aggregated to such an extent that they could be identified as the sacral sympathetic trunk ( Figure 3A).
At CS18 (~44 days of development), the cell density in the IHGCs had further increased ( Figure 3D) and came to resemble the preaortic ganglia seen more cranially (Kruepunga et al., in press) Figure 3D). At CS20 (~49 days of development), the developmental events described for CS18 had continued to advance, with a major feature being a quantitative increase in the density of ganglionic cells in both IHGCs ( Figure 5A-C).
In CS22 embryos (~53 days of development; Figure 5D-G), the left and right IHGCs were still separate, but each cluster had fragmented (blue dots in panels 5D and E). Its middle portion, which occupied the niche between the urogenital sinus ventrally and pelvic coelomic pouch at the level of the entrance of the Wolffian and Müllerian ducts into the urogenital sinus dorsally, had evolved as the biggest and densest (blue dots in panel 5E, blue arrows in panels 5F and G).
Because each IHGC now also incorporated nerve fibres, it could be labelled the inferior hypogastric or pelvic plexus (blue dots and beige network in Figure 5D). The more cranial and caudal portions of the IHGCs formed small clusters near the ureteric entrance into the bladder and around the lower part of the rectum, respectively ( Figure 5D). Interestingly, the caudal extension of the IHGC reached the rectal wall caudal to the pelvic coelomic cavity, which at this stage had started to regress in a cranial direction (Hikspoors et al. 2019).

| Formation of the nerve fibre network of the inferior hypogastric plexus
Nerve fibres were first observed in the lesser pelvis at CS16, when the superior hypogastric nerve formed as a median continuation of both lumbar splanchnic nerves along the superior rectal artery.
This nerve formed cranial to the bifurcation of the umbilical arteries, but began to extend caudally across the bifurcation towards the cranial part of both IHGCs described in the previous section (asterisks in Figure 4A,E). In CS18 embryos, the superior hypogastric nerve had increased in diameter and length, so that it now passed the aortic bifurcation. This single nerve trunk (asterisks in Figure 4B,F) bifurcated just caudal to the umbilical arteries to into a bundle of smaller nerves that began to resemble the superior hypogastric plexus (asterisks in Figure 4C,G). Fragmentation of the nerve into a plexus continued through CS22 (asterisks in Figure 4D,H).
Concomitantly, medial branches of sacral spinal nerves S2-4, known as the pelvic splanchnic nerves, extended medially towards the IHGCs at CS16 (yellow arrows in Figure 6A,C). The pelvic splanchnic nerves reached the IHGCs at CS18 (yellow arrows in Figure 6D,F) and lined up with nerves in the IHGCs in CS20 embryos ( Figure 7A). The resulting network of fibres inside the left and right IHGCs had, therefore, transformed these clusters into the left and right inferior hypogastric plexuses ( Figure 5A-C).
Furthermore, nerve fibres from the sympathetic trunk had started to extend ventrally as the sacral splanchnic nerves and connected to the inferior hypogastric plexus at CS22 (orange arrowheads in Figure 8D,F,J).  Figure 7C,E) and ventral extensions that formed the pudendal nerve ( Figure S3C).
In addition, extrinsic enteric nerve fibres (beige arrows) within the inferior hypogastric plexus had connected intrinsic enteric nerve fibres (light green arrows) in the urogenital and hindgut mesenchyme ( Figure 9). While the network of nerve fibres in the wall of the hindgut was well defined (Figure 9C), such a network of nerve fibres was not present in the wall of, for example, the duodenum ( Figure 9H). At CS22 and CS23, no differences in pelvic innervation between male and female embryos were apparent.

| D ISCUSS I ON
We studied the development of the extrinsic innervation in the lesser pelvis to illustrate and explain its spatiotemporal population with autonomic ganglionic cells and nerve fibres (for a pictorial summary, see Figures 10 and 11).

| Boundaries of the lesser pelvis
The

| Migratory timeline of neural crest-derived ganglionic cells
Cell tracing studies showed that the ganglionic cells in the lesser pelvis arise from sacral NCCs in mammalian embryos (Serbedzija et al. 1991 However, the NCCs that continue to migrate ventrally need just ~1 day to reach the pre-arterial region at CS15-late (~37 days) and form the IHGC (equivalent to pre-aortic ganglia) (Figure 11). This comparison shows that the timeline of the migration of ganglionic cells in the thoracolumbar and sacral region differs, but that the migratory pathway and time of arrival at their destination are very similar.

| Differences in innervation of the neural crest-derived ganglionic-cell aggregates
The pre-aortic plexuses in the thoracolumbar region are irregularly shaped median masses of neural crest-derived ganglionic cells that surround the roots of the ventral branches of the dorsal aorta. In addition, agglomerates of the SCP subgroup of NCCs are found in association with the coeliac and inferior mesenteric plexuses as the (developing) chromaffin cells of the adrenal medulla and para-aortic bodies, respectively (Furlan et al. 2017;Kastriti et al. 2019). These two agglomerates differ in that the chromaffin cells of the adrenal medulla are positioned dorsolaterally to the coeliac plexus, whereas those of the para-aortic bodies touch or merge in the midline ( Figure S5A,B), with very few para-aortic bodies found distal to the umbilical bifurcation (Coupland, 1952, Kruepunga et al., in press). The preganglionic nerve fibres that innervate the pre-aortic ganglionic-cell agglomerates arrive at CS15 late (~37 days of development; Figure 10). The IHGCs form at the same developmental stage as, and are contacted only slightly later (CS16; ~39 days) by the superior hypogastric nerve than the inferior mesenteric cluster by the lumbar splanchnic nerves (Kimmel and McCrea, 1958;Arango-Toro and Domenech-Mateu, 1993).
In fact, our reconstructions show that the superior hypogastric nerve develops as an unpaired mediocaudal extension of the lumbar splanchnic nerves along the superior rectal artery (CS16-20), which explains the common origin of the lumbar splanchnic and superior hypogastric nerves in segments L1 and L2. At CS20, the superior hypogastric nerve has split up into several and at CS22 into many separate nerve fibres. This remarkable splitting of a compact nerve into a distributed plexus may well explain its morphological variability and many alternate names in the adult (Davis, 1934).
Apart from the hypogastric nerves, the IHGC has two additional sources of preganglionic innervation, of which the pelvic splanchnic nerves arrive at CS18 (~44 days) and the sacral splanchnic nerves only at CS22 (~53 days). Phenotypically and developmentally, the prenatal pelvic splanchnic nerves were recently characterized as sympathetic rather than parasympathetic (Espinosa-Medina et al. 2016), so of comparable phenotype to the hypogastric and sacral splanchnic nerves. Our findings refine earlier timelines in human embryos (Kuntz, 1952;Kimmel and McCrea, 1958;Arango-Toro and Domenech-Mateu, 1993) and better allow comparison with data obtained in experimental animal models, such as mice. However, cause(s) and consequence(s) of this phased innervation by phenotypically similar preganglionic nerves remain to be clarified.

| Differences in topography of the abdominal and pelvic plexuses
Compared to the midline pre-aortic abdominal ganglia, the bilateral presence of the inferior hypogastric plexuses is striking. This pronounced topographic difference can be ascribed to the pelvic pouch, a caudal extension of the peritoneal cavity between the hindgut and the urogenital sinus that extends down to the muscular pelvic floor. The temporary presence of this narrow pouch in human embryos was first described by Cunéo and Veau, 1899 and confirmed by Tobin andBenjamin, 1945, andUhlenhuth et al. 1948.  (Tobin and Benjamin, 1945;Uhlenhuth et al. 1948;Fritsch, 1988). Obliteration of the pouch may protect bipedal hominids from rectal prolapse, which is quite common in quadrupeds (Pettan-Brewer and Treuting, 2011). Accordingly, a straight course of the sigmoid colon with a correspondingly short mesentery (both embryonic features) often co-occur with a persisting pelvic coelomic pouch and enterocele (Baessler and Schuessler, 2006

| Limitations of the study
The present study provides detailed reconstructions of the autonomic nervous system in the developing pelvis in six human embryos at between 5 and 8 weeks of development. Although one can object that six models cannot visualize all of the ENS in the developing embryo, we were able to provide a continuous account of the developmental appearance of relevant structures. A valid question is, nevertheless, whether all variation is accounted for.
Although the answer is obviously 'no', differences between specimens could usually be explained as differences in degree of development rather than deviation from the expected morphology. The most important limitation of the present series is probably that the F I G U R E 9 Connection of extrinsic and intrinsic enteric nerve fibres in the lesser pelvis. Panels a, e show nerve fibres in the lesser pelvis and a magnified view (rectangle). Panels b-d and f-g show histological sections from the levels indicated by dotted lines in panel a and magnified views (rectangles). Extrinsic enteric nerve fibres (beige arrows) extend into the mesenchymal cuffs of the hindgut, where intrinsic nerve fibres (light green arrows) form an intrinsic nervous network (panels c, g In conclusion, the extrinsic innervation, both ganglionic cells and nerve fibres, in the lesser pelvis is organized in a similar fashion to that in its abdominal counterpart. Its three topographically separate preganglionic connections can be ascribed to differences in developmental timeline and its bilateral appearance to local peritoneal topography. Up to and including CS23 pelvic innervation is phenotypically indifferent with respect to sexual dimorphism.

| DATA AVAIL AB ILIT Y S TATEMENT:
The data that support the findings of this study are available in the supplementary material of this article.

CO N FLI C T S O F I NTE R E S T
None declared.

F I G U R E 11
Timeline of neural crest-cell migration to ganglionic cell clusters in, and preganglionic innervation of the inferior hypogastric plexus. Panel A shows schematic transverse sections at different developmental timepoints. The development of the inferior hypogastric plexus can be divided into (1) migration of neural crest cells (blue dots) towards their para-arterial [Carnegie stage (CS)14 and CS15] and pre-arterial positions (CS15 and CS16); and (2) association of nerve fibres with the inferior hypogastric ganglionic cell cluster. The first nerve to arrive is the superior hypogastric nerve at CS16, followed by the pelvic splanchnic nerves at CS18 and the sacral splanchnic nerves at CS22. By that time small nerve fibres had entered the gut wall via the dorsal mesentery of the hindgut and contacted intrinsic enteric nerves (asterisk in far right subpanel