DEVELOPMENT OF THE ENTERIC NERVOUS SYSTEM
Overview of the development of the ENS
The ENS ganglia are formed, just like sympathetic and parasympathetic ones, by neural crest cells (NCC) (Le Douarin et al., 1981; Nagy & Goldstein, 2017). Even if most of modern research is made on mice and zebrafish (Ganz, 2018; Obata & Pachnis, 2016), the original studies on the development of this system were made in chicken, originally by ablation experiments (Yntema & Hammond, 1954) and then by quail-chicken xenografts (Douarin & Teillet, 1973) to discover that the cells of the enteric nervous system are derived from neural crest cells at two rostro caudal levels: so called “vagal neural crest”, adjacent to somites 1-7, and the sacral crest ((derived from the neural tube posterior to somite 28). The concept of “vagal neural crest” was recently revisited (Isabel Espinosa-Medina et al., 2017) and it appeared that only NCC at the level of somites 1-2 deserve that denomination, because they are in register with the roots of the vagal nerve and migrate along the nerve itself, like precursors of the parasympathetic ganglia were previously shown to do ((I Espinosa-Medina et al., 2014), while NCC facing somites 3-7 have no relation to the vagal nerve, participate in the formation of the superior cervical ganglion (the rostral-most ganglion of the paravertebral sympathetic chain) and continue their ventral migration pathway beyond the dorsal aorta to enter the foregut: they are better viewed as post-vagal (i.e. cervical) or “sympatho-enteric” crest. The vagal crest proper invades only the esophagus and oral stomach, while the cervical or sympatho-enteric crest invades the entire length of the digestive tube.
The process of ENS formation starts at E8.5 in mouse, when the neural crest cells delaminate from the neural tube, and invasion of the digestive tube starts at E9.5 (Anderson et al., 2006). These cells migrate along the tube caudally (Allan & Newgreen, 1980; Young et al., 2004) and this migration, while the gut itself is still growing, is thought to be fueled by proliferation (Landman et al., 2007), at a speed which reaches 35µm/h (Young et al., 2004). At the level of the cecum, where the gut forms a loop, the mode of migration shifts from following the length of the gut to a trans-mesenteric short-cut that bypasses the cecum, and cells directly start to colonize the hindgut – from cecum to the rectum (Nishiyama et al., 2012). At least to the cecum itself, the migration of the NCC is guided by a gradient of Glia Cell Line-Derived Neurotrophic Factor (GDNF) with a maximum of expression in the cecum (Natarajan et al., 2002; Young et al., 2001). Its expression starts as soon as E9.5 in the stomach and at E10.5 in the cecum before establishing a gradient between both regions. In addition, GDNF has a role in increasing the proliferation rate of NCC, as shown in vitro where GDNF can double the proliferation rate (Hearn et al., 1998). This factor also help to maintain cells in an undifferentiated condition: in the absence of GDNF the pool of progenitors is depleted by precocious differentiation and as a consequence the hindgut is not properly colonized (Gianino et al., 2003). GDNF acts through its receptor GFRα and the co-receptor Ret, a tyrosine kinase (Robertson & Mason, 1997), expressed at the surface of enteric NCC (Enomoto et al., 1998; Pachnis et al., 1993), while the ligand GDNF is expressed in the mesenchyme of the gut (Trupp et al., 1995). Inactivation of GDNF, GFRα or Ret leads to massive depletion of enteric ganglia, in most or all the length of the gut (Uesaka et al., 2008; Enomoto et al., 1998; Moore et al., 1996).
Other signaling pathways involved in ENS formation include G-protein coupled receptors (GPCR), and among them the endothelin receptor B (EDNRB) with its ligand endothelin-3 (EDN3), revealed by the study of knockout mice (Baynash et al., 1994; Hosoda et al., 1994) :these two mutants show an aganglionosis with an upstream formation of a megacolon. An interaction between the EDNRB and Ret pathways lead to a more severe phenotype, at least in a mouse model (McCallion et al., 2003). The receptor EDNRB is expressed by the NCC (Nataf et al., 1996), and its ligand, EDN3, is expressed in the mesenchyme (Leibl et al., 1999). EDN3 has a spatiotemporal specific expression appropriate to guide the NCC to their intermediate target organs one after the other, firstly the skin, then in the branchial arches and finally in the digestive tract (Nataf et al., 1998). This signaling pathways is also involved in survival, proliferation, migration and differentiation (Bondurand et al., 2018; Lahav et al., 1998; Wu et al., 1999).
Finally, a third signaling pathway has emerges as involved in the ENS development: the ErbB3/Neuregulin1 (Nrg1) pathway. Nrg1 belongs to the EGF-like signaling molecules. It was already known to be essential for the development of parasympathetic and sympathetic ganglia (Britsch et al., 1998; Dyachuk et al., 2014). The receptor ErbB3 is expressed by the NCC, while its ligand Nrg1 is expressed at the surface of the axons (Britsch et al., 2001; Perlin et al., 2011), as well has another source, likely to be the mesenchyme, which provide secreted isoforms of Nrg1 (Birchmeier, 2009; Birchmeier & Nave, 2008). Its absence leads to a gradient of atrophy along the ENS from to duodenum to the rectum with a loss from 75 to 50% (Isabel Espinosa-Medina et al., 2017b).
The second major source of cells of the ENS is the sacral neural crest, delaminating at the level of somite 28 and below, mostly responsible for colonizing the hindgut (Burns & Douarin, 1998). These sacral NCC first form the pelvic ganglia (Serbedzija et al., 1991), and then a sub-population continue to migrate along the pelvic nerve to reach the rectum at E13.5 (Wang et al., 2011) and start a rostral migration through the hindgut where they encounter and mix with the caudally migrating cervical (formally vagal) neural crest cells. These sacral neural crest cells never go further than the umbilicus (Burns & Douarin, 1998), even if in the absence of the caudally migrating cervical population (Burns et al., 2000). With the redefinition of the pelvic ganglia as sympathetic (I. Espinosa-Medina et al., 2016) (which means that there is no longer any reason to call the sacral crest “parasympathetic”), the distinction of the “sympatho-enteric” neural crest (Isabel Espinosa-Medina et al., 2017) from the vagal, and the newly recognized contribution of trunk Schwann cell precursors to the ENS (Uesaka et al., 2015), it appears that from the cervical level to the sacral, the neural crest has a dual fate, sympathetic and enteric.
This whole process initially form the myenteric plexus of the ENS; later a portion of these cells start a second wave of migration, this time in the radial direction, to go deeper in the wall of the gut, toward the mucosa, and form the submucosal plexus (McKeown et al., 2001). Because of this delayed process, the differentiation of the neurons of this plexus is shifted in time: whereas the first differentiated neurons are visible at E10 in the myenteric plexus, they appear only at E14 in the submucosal plexus (Pham et al., 1991). An intriguing evolutionary complication is that in the hindgut of the chick, the neurons from a sacral origin firstly form the submucosal plexus and then effectuate a radial migration to form the myenteric one (Burns & Douarin, 1998). In zebrafish the submucosal plexus is missing altogether (Holmberg et al., 2003).
Regulation of neuron numbers in the ENS II.A. Embryonic proliferation and death
The enteric neural crest probably represents one of the most proliferative cell populations during embryogenesis, since a small number of crest cells (albeit not precisely quantified to my knowledge) at vagal, cervical and sacral level gives rise to a number of neurons equivalent to that in the spinal cord (Furness et al., 2014). However, the exact extent and pattern of cell division during ENS formation is incompletely understood.
Initial size of the pre-enteric population is essential for ENCC migration as demonstrated after partial ablation of the neural crest that leads to reduced speed of the wavefront (Young et al., 2001) and distal agangliosis of the GI tract (Barlow et al., 2008; Burns et al., 2000; Druckenbrod & Epstein, 2005), which could be relevant to the common developmental defect observed in Hirschsprug disease (see below). An even more compelling demonstration that the number of pre-enteric neural crest cells is essential to generate a pool big enough to colonize the whole gut was obtained by back transplanting one-somite length of neural crest in chicken embryos where the crest facing somites 3-6 had been removed. Unexpectedly, this operation restored colonization of the entire gut even in cases where the graft itself could not participate in the colonization per se (i.e. if it was taken at thoracic level), but presumably restored an adequate population pressure in migrating neural crest prior to its entrance in the gut (Barlow et al., 2008).
Conversely, if apoptosis is blocked early on, by electroporation of an inhibitor, thus presumably in pre-enteric crest, the mature ENS is hypertrophic (Wallace et al., 2009), with an increase in the number of ganglia rather than of their size, in the midgut – from duodenum to the caecum – and hindgut. Incidentally, this study revealed an unexpected level of apoptosis in the early migrating crest.
The GDNF gradient also has a function in the regulation of neuronal numbers: it increases the proliferative capacities of cells, and in absence of GDNF the pool of precursors differentiate before reaching the caecum, the point of highest GDNF concentration in the gut (Gianino et al., 2003). The wavefront of migration is composed of progenitors that are dividing at the same time as migrating, but once a segment of the gut is colonized the proliferation occurs only in 4% of the cells, while the other cells are differentiating into neurons – expressing HuC/D – or even into mature neurons – expressing such specific neurotransmitter as Nitric Oxyde Synthase (NOS) (Young et al., 2005). While the highly proliferative ENCC wavefront migrates into unpopulated gut regions, less proliferative rearguard cells populate already colonized regions by migrating non-directionally (Theveneau & Mayor, 2011). The previous studies mostly focused on the colonization of the distal bowel, but the early migration from the neural tube to the foregut is still poorly understood.
The first extensive study of the pattern of cell division in the ENS is very recent (Lasrado et al., 2017). It makes use of lineage analysis with the confetti reporter or the MADM complementation system. Recombination of these reporters were triggered at E12.5 to show the following: i) cells did not migrate extensively once they were behind the wave front; ii) they formed a mosaic of overlapping clones; iii) their degree of dispersion was proportional to the number of divisions and best explained by a model where neuronal progenitors intermingle with unrelated dividing cells of the mesenchyme; iv) cell fate decisions occurred at the last or penultimate division; v) formation of ganglia in the submucosal plexus was by descendants of myenteric precursors situated immediately radial to them; vi) Clonal analysis of Ret— cells showed that decrease in Ret signaling cell-autonomously favors proliferation over neuronal differentiation (i.e. leads to larger clones containing fewer neurons and more glia).
A limitation of this study is that it was performed at a time when most of the gut is colonized, and a lot of proliferation has already occurred, and might thus explain the favored proliferation in absence of Ret. It would be interesting to repeat this analysis of clonal architecture of the ENS at earlier time points, to better understand the lineage of the ENS. However, image analysis of colored clones might be more challenging, if too many clones become intermingled.
Post-natal proliferation and death
The development of the ENS continues during post-natal life, including neurogenesis and neuronal differentiation, so that the birth of the animal might be considered as a somewhat arbitrary time point during ENS development. This is a debated topic however, and rather confusing in its details. I use the occasion of this report to try and clarify the literature on the subject. I will distinguish evidence concerning i) the number of neurons at different post-natal ages, ii) cell division and iii) cell death.
Evolution of the number of enteric neurons after birth
Gabella (1971), referring to the papers by Altman & Das (1965) and collaborators which famously put an end the “no new neuron” dogma in the central nervous system, sought evidence for the same phenomenon in the ENS. The author counted the density of neurons in the myenteric plexus of the small intestine of rats (i.e. the number of neurons per mm2 of intestinal wall) in the newborn rat and the adult (6 months old) rat. After correction of this number by the enlargement of the gut wall during the same time window, an increase in neuronal numbers was estimated, from 420,000 to 1, 850,000, i.e. 4.4 fold. This would mean that around 75% of the enteric neurons found in the adult are born postnatally (during an unspecified time window between P0 and 6 months). This spectacular level of neurogenesis seems to have been neither contradicted nor confirmed by later studies and, to the best of my knowledge, the paper has not been cited for this piece of evidence in several decades.
In apparent contradiction Marese et al. (2007) found a stable number of neurons from 21 days-old rats to 60 days-old rats, albeit redistributed on a larger surface. However, this study leaves unexplored a time-window from P0 to P21, which is when the massive increase reported by Gabella could conceivably have taken place.
Liu et al. (2009) report an increase in neuronal density in the first 4 months after birth in wild type mice, by 20% (contrary to Gabella, who found that the surface of the gut increased more than the number of neurons, thus that there was a decrease in density). The total size of the intestine was not taken into account in that study and the Methods section is clear that the total number of neurons was not calculated. Thus, if the size of the gut wall increases between P0 and P120, which is likely, it is possible that the increase in total cell numbers after P0 is much larger than 20% increase in density. In line with this possibility, the authors also report direct detection of neurogenesis (discussed below).
I could not find published evidence for the evolution of neuronal numbers during most of the adult life. Kulkarni et al. (2017) mention that the “numbers of enteric neurons in the healthy adult rat remain remarkably constant for most of adult life” with a reference to Gabella, which contains in fact only one sentence on that topic in the discussion: “the number of nerve cells seems only slightly reduced, if at all, in rats more than 1 year old”, not illustrated by any data or reference.
On the other hand, a large literature has been devoted to enteric neuron loss in the aging animal by up to 50%. (Marese et al., 2007; Thrasivoulou et al., 2006) provide references for studies in the myenteric plexus of rats, mice, guinea pigs and humans, with evidence that cholinergic neurons are more affected than nitrergic ones. Thrasivoulou et al. (2006) show that caloric restriction, of the type that prolongs the life of rats by 40%, completely prevents this neuronal loss.
Evidence for ongoing neurogenesis
Early postnatal stages and juveniles
Gabella is probably the first to discuss this question, without providing original data, however. The author deduces a requirement for neurogenesis after birth from considering i) the need to generate new neurons in the young adult (i.e. to account for the massive increase in neuronal numbers during the first 6 months of life he reported, see above), ii) electron microscopy evidence of neuronal death (from the authors, but unpublished), despite iii) stability of the number of neurons in adults (with no data or no reference on the subject, see above). Speculations on the mechanism of neurogenesis are accompanied by ancient references (back to the 1920’s and 1930’s), with the underlying idea that a persisting pool of progenitors would explain the increase in neuronal numbers in juveniles and their stability in adults, while its exhaustion would explain neuronal loss in aged animals.
Pham et al. (1991), might be the first to report direct evidence for significant levels of neurogenesis until P30, mostly in the submucosal plexus, by administration of tritiated thymidine during 24 hours, at ages ranging from E8 to P21, and detection of the label at P30. Only the cells that are about to withdraw from the cell cycle at the time of labeling can retain the label by this method.
Detection of incorporated H3 was combined with that of the neurotransmitter phenotype: serotonergic, cholinergic, and peptidergic (including Vip, Enk, Npy and Cgrp). Different dynamics of cell birth (i.e. withdrawal from the cell cycle) were detected for different phenotypes. Cholinergic and serotonergic neurons were all born before birth, as well as Vip, Enk and Npy neurons of the myenteric plexus. In contrast, a large fraction of Vip, Npy and Cgrp neurons of the submucosal plexus were born after birth. (These data were interpreted as arguing for an influence of early differentiating neurons on later ones, although they do not rule out the existence of separate lineages each with its intrinsic dynamic).
Two decades later, the same lab, (H. Wang et al., 2010), could detect BrdU incorporation (i.e. neuronal birth) a few hours after treatment, at stages ranging from P0 to P8, in different part of the digestive tube, accounting for a proportion of total PGP9.5+ cells (i.e. neurons) ranging from 16% in submucosal plexus of the colon at P8 to 2% for the myenteric plexus of the small intestine at P8.
By a completely different approach, Laranjeira et al. (2011) evaluated the number of neurons in an adult mouse that were born from Sox10+ progenitors present at several embryonic and postnatal stages, using lineage tracing with a tamoxifen inducible Cre transgene driven by the Sox10 promoter (Sox10::CreERT2). The main finding was that, although large amounts of neurons in the adult ENS are born from progenitors present at E8.5 or E12.5, very few ((2.8%) are born from Sox10-positive progenitors present at P0, even fewer (1.6%) from progenitors present at P30 (those being presumably included in the former), and none from progenitors present at P84 (6 months). So that, for all practical purposes, no neurogenesis from Sox10+ cells present at post-natal stages would contribute to make-up the adult myenteric plexus. These findings are hard to reconcile with the massive increase in neuronal numbers proposed by Gabella, or even the increase in cell density described in Liu et al. (2009), and are even substantially lower than those reported by Wang et al. (2010), but the different experimental designs prevent direct confrontation of the data. Possible caveats of the Laranjeira study are: i) the possibility that neurons generated postnatally would undergo a massive turnover, while those born in the embryo would last for the life of the animal, but this is unlikely. ii) the possibility that postnatal progenitors do not express Sox10 (see below). iii) The possibility that the efficiency of recombination of the reporter transgene dramatically goes down with age (see below).
Finally, Uesaka et al. (2015), taking for proven the existence of a postnatal neurogenesis, with references to Pham et al. (1991), Wang et al. (2010) and Laranjeira et al. (2011) (despite the low level reported by the latter, see above), explore a possible mechanism for this neurogenesis. The authors propose that one of its sources are Schwann Cell Precusrors (SCPs) of the extrinsic nerves of the gut. By lineage tracing SCPs with a Cre driven by the immature SCP marker Desert Hedgehog (Dhh) and a reporter gene for enteric neurons (a conditional knock-in of GFP in the GFRα locus), they show that about 5% of the submucosal ENS of the small intestine, and 20% of both myenteric and submucosal plexi of the colon are made of descendant of SCPs. Conversely, the destruction of these cells by a conditional mutation of Ret depletes the distal colon by about 30% of its neurons. Importantly for this discussion, SCP-derived neuronal precursors start acquiring neuronal features (expression of tyrosine kinase Ret and the neuronal marker PGP9.5) at P1 and all of them are Ret+ and PGP9.5+ at P21, so that neuronal differentiation from SCPs is post-natal. However, this observation cannot ascertain the birth date of these cells, so does not quite answer the question of postnatal neurogenesis per se, only of neuronal differentiation.
Table of contents :
1 – General Introduction
2 – Development of the enteric nervous system
I – Overview of the development of the ENS
II – Regulation of the neuron numbers in the ENS
A – Embryonic proliferation and death
B – Post Natal proliferation and death
1 – Evolution of the number of enteric neurons after birth
2 – Evidence for ongoing neurogenesis
3 – Neuronal Diversity in the ENS
I – Introduction
III – Types of enteric neurons defined by classical studies
A – Myenteric plexus
2 – Sensory (afferent) neurons
3 – Interneurons.
4 – Intestinofugal neurons
B – Submucosal plexus
1 – Secretomor/Vasodilator neurons
2 – IPANs
3 – Other types
C – Conclusion on the detection of enteric neurons types.
D – Glia.
IV – Types of enteric neurons newly defined by single cell transcriptomics.
V – Differentiation of enteric neurons into their different types.
A – Timing of neuronal diversification.
B – Mechanisms of neuronal differentiation
VI – Transcriptional control of the ENS development
4 – The transcription factor TBX3
I – The T-Box family of TFs.
II – Tbx3 in stem cells
III – Tbx3 in cancer
IV – Developmental roles of Tbx3
V – Tbx3 and the enteric nervous system
5 – Hmx2 and Hmx3 in development
I – Hmx2 and Hmx3 during embryonic development
II – Expression of Hmx2/3 in the nervous system
1 – Article 1
2 – Article 23 – Unpublished results
I – Genetic interaction between ErbB3 and Ret
II – Hmx2 and Hmx3 knockouts
A – Construction
B – Phenotype of the Hmx knockouts
III – Role of Tbx3 in the development of the ENS
A – Expression of Tbx3 in the ENS
B – Gross phenotype of a Tbx3 conditional KO
C – Histological analysis of the ENS of Phox2b::Cre; Tbx3lox/lox mutants
Material and Methods