Yasuhiro Kon, Laboratory of Anatomy, Department of Biomedical Sciences, Graduate School of Veterinary Medicine, Hokkaido University, Kita 18-Nishi 9, Kita-ku, Sapporo 060-0818, Japan. T: + 81 11 7065187; F: + 81 11 7065189; E: firstname.lastname@example.org
The mammalian gut undergoes morphological changes during development. We studied the developing mouse duodenojejunal flexure (DJF) to elucidate the mechanism of formation. During embryonic days 10.75–13.75, DJF formation was morphologically classified into three stages: the expansion stage, flexure formation stage, and flexure elongation stage. From the expansion to the flexure formation stages, the DJF wall showed asymmetric morphology and proliferation along the left-right intestinal axis. From the flexure formation to the flexure elongation stage, the DJF started to bend dorsally with counterclockwise rotation along the antero-caudal intestinal axis, indicating that the original right side of the duodenum was rotated towards the dorsal body wall during development of the DJF. The direction of attachment of the dorsal mesentery to the DJF did not correspond to the bending direction of the DJF during flexure formation, and this finding indicated that the dorsal mesentery contributed very little to DJF formation. During DJF formation, Aldh1a2 and hedgehog mRNAs were detected at the DJF, and their expression levels differed along the bending axis. In conclusion, DJF formation might be triggered by asymmetric morphology and proliferation along the left-right intestinal axis under the control of retinoic acid and hedgehog signaling.
The development of the murine gastrointestinal tract starts at around embryonic day (E) 7.0. Subsequently, the gut tube, comprising the foregut, midgut, and hindgut, is formed between E9.0 and E9.5. Simultaneously, the midgut begins to elongate and descend into the umbilical cord as a primary gut loop at around E10.5, and this structure, called the physiological umbilical herniation, exists until around E16.0. During the formation of the primary gut loop and physiological herniation, the gut tube assumes species-specific morphology by rotation, elongation, and looping (Noah et al. 2011; Spence et al. 2011).
During the development of the gastrointestinal tract, biological processes such as planar cell polarity (PCP), convergent extension (CE), or cell proliferation contribute to dynamic morphological changes (Reed et al. 2009; Mao et al. 2010; Yamada et al. 2010). PCP is the planar polarity in the alignment of a collection of cells within a cell sheet, in addition to the apico-basal polarity (Jones & Chen, 2007). PCP is mainly regulated by wingless-related MMTV integration site (Wnt) signaling, particularly non-canonical signaling (Schlessinger et al. 2009). Importantly, Wnt5a is reported to be essential for midgut development in mice (Cervantes et al. 2009). PCP is needed for the CE that occurs during gastrointestinal morphogenesis (Wallingford et al. 2002; Reed et al. 2009; Yamada et al. 2010). Briefly, CE is a morphogenetic process in which a cell sheet converges (narrows) along one axis and extends (elongates) along the perpendicular axis by cellular movements, such as polarized cell migration, cell shape change, and cell rearrangement (Wallingford et al. 2002). In Xenopus, endoderm CE mediated by the Ras homolog (Rho) gene family was reported to be an important process in gut tube elongation (Reed et al. 2009). In addition to these morphological changes in the cells, cell proliferation is also needed for gastrointestinal morphogenesis, which involves the elongation and thickening of the gut tube. In particular, hedgehog signaling promotes mesenchymal cell proliferation in the gut tube, and conditional removal of Indian hedgehog (Ihh) and sonic hedgehog (Shh) activities caused gastrointestinal abnormalities such as a reduction in the length and diameter of the gut tube (Mao et al. 2010).
The contributions of the mesentery and body axes formation are also proposed as other factors determining the morphological features of the gastrointestinal tract. Recently, it was reported that the chirality of midgut rotation was determined by left-right asymmetries of the cellular architecture in the dorsal mesentery, regulated by the expression of left-right determining genes such as paired-like homeodomain transcription factor 2 (Pitx2; Davis et al. 2008; Kurpios et al. 2008; Plageman et al. 2011). Furthermore, the looping morphogenesis of the jejunum was mediated by homogeneous and isotropic forces derived from different growth rates between the gut tube and the anchoring dorsal mesenteric sheet (Savin et al. 2011). Moreover, embryonic signaling mediated by retinoic acid (RA) and hedgehog have been suggested to influence body axis formation and to play important roles in gut tube morphogenesis, and the expression of aldehyde dehydrogenase (Aldh), which regulates RA synthesis, has been demonstrated during gut looping in Xenopus (Pitera et al. 2001; Lipscomb et al. 2006).
As described above, several morphological and molecular mechanisms involved in gastrointestinal tract development have been proposed. On the other hand, morphological and molecular mechanisms controlling gut flexure formation are understudied. Importantly, different animal species have common intestinal flexures in the same positions, such as the cranial and caudal flexure of the duodenum or the duodenojejunal flexure (DJF). Therefore, we hypothesized the involvement of certain programmed controls in gastrointestinal morphogenesis.
In the present study, we morphologically investigated the development of the mouse gastrointestinal tract, with particular focus on the formation of the DJF. Gastrointestinal morphology affects digestive function, and malformations such as malrotation are sometimes fatal. Therefore, elucidating how the gastrointestinal tract develops morphologically is useful for understanding gastrointestinal morphology, as well as gastrointestinal physiology and pathology.
Materials and methods
C57BL/6 mice aged 0–50 days postnatal were maintained in our specific pathogen-free facility and used in this study. In addition, C57BL/6 pregnant mice carrying embryos of ages E10.75–13.75 were purchased from Japan SLC, Inc. (Shizuoka, Japan). Noon on the day when a vaginal plug was found was considered E0.5. Some pregnant mice were injected intraperitoneally with bromodeoxyuridine (BrdU; Wako, Osaka, Japan) at 1 mg g−1 body weight 2 h before being sacrificed. These mice were maintained and sacrificed according to the Guide for the Care and Use of Laboratory Animals of Hokkaido University, Graduate School of Veterinary Medicine (approved by the Association for Assessment and Accreditation of Laboratory Animal Care International).
Anatomical analysis and definition of anatomical terms
The postnatal and prenatal gastrointestinal tracts were fixed with 10% neutral buffered formalin and 4% paraformaldehyde (PFA) at 4 °C overnight, respectively. Their morphological pattern was observed with the point of attachment of the dorsal mesentery by the naked eye and through a dissection microscope.
Based on gross anatomical observation, the fetal ages of the mice were classified into three stages according to the existence and position of the DJF: the expansion stage (E10.75), the flexure formation stage (E11.25–11.75), and the flexure elongation stage (E12.75–13.75). Furthermore, Fig. 1 shows the anatomical terms based on the body or intestinal axes used in the present study: (i) the dorso-ventral and antero-caudal direction based on the body axis (D-V and A-C body axes); (ii) the dorso-ventral direction based on the intestinal axis relating to the attachment portion of the dorsal and ventral mesentery (D′-V′ intestinal axis, Fig. 1a); (iii) the left-right direction, meaning the axis perpendicular to the D′-V′ intestinal axis at right angles at the center of the gut lumen (L′-R′ intestinal axis, Fig. 1a); and (iv) the outer and inner direction based on the axis crossing the apices of the extensor and flexor sides of the DJF (bending axis, Fig. 1b). These terms have been applied to both anatomical and histological examinations.
Histological and histoplanimetric analyses
The fetal gastrointestinal tracts were fixed with 4% PFA and embedded in agarose gel before embedding in paraffin to confirm the exact direction of slicing. Semi-serial sagittal or cross-sections (3–4 μm thick) of the DJF were used for hematoxylin-eosin (HE) staining. The histological features of the gut were observed with the direction of attachment of the dorsal mesentery using these HE-stained semi-serial sagittal or cross-sections of the DJF at the three stages. At the expansion stage, the DJF was defined as the region located at the caudal position of the pancreas buds and attached to the dorsal and ventral mesenteries.
At the expansion stage, the total number and average area of the mesenchymal cell nuclei were evaluated at the definite area in the external layer of the gut wall shown in Fig. 1a, the gray area. This histoplanimetric analysis was performed in both the left and right definite areas based on the intestinal axis. Three HE-stained semi-serial sections were observed for each of five embryos, and the average of the three sections was expressed as the sample value.
At the flexure formation stage, the section in which the outer and inner sides of the mucosal epithelium were clearly separated from each other was selected for analysis. Based on the bending axis, the gut wall thickness in the DJF was measured with ImageJ (NIH, Bethesda, MD, USA), divided by the gut diameter, and expressed as a percentage, and the average was compared between the outer and inner sides (Fig. 1b). To measure the nuclear height/width ratios of the mesothelial cells of the DJF, a tangential line passing the outline of the DJF curvature was drawn, and the height and width of cells were defined as the cell length vertical and parallel to the tangential line, respectively. Five adjacent mesothelial cells were measured at the apices of DJF by using ImageJ, and their average nuclear height/width values were expressed as percentages and compared between the outer and inner sides of the flexure. For these analyses, one HE-stained sagittal section was used for each of 5 embryos.
Cell proliferation analysis
Proliferating cells were detected by immunostaining for BrdU. Briefly, antigen retrieval was performed by citrate buffer (pH 6.0) for 20 min at 105 °C. Rat monoclonal anti-BrdU antibodies (1 : 200; Abcam, Cambridge, UK) and biotin-conjugated goat anti-rat IgG antibodies for BrdU (1 : 100; Caltag, London, UK) were used as primary and secondary antibodies, respectively. BrdU-positive reactions were developed by the streptavidin-biotin complex (SABPO® kit; Nichirei, Tokyo, Japan) and 3,3′-diaminobenzidine tetrahydrochloride-H2O2 solution. Cell proliferation was analyzed using BrdU-immunostained DJF sections. At the expansion stage, the number of total nuclei and BrdU-positive nuclei in the mesenchymal cells were counted at the areas in the dorsal, ventral, left, and right gut walls based on the intestinal axis as shown in the grid area in Fig. 1a. The number of BrdU-positive nuclei was expressed as a percentage of the total number of nuclei, and the averages for each side were compared.
At the flexure formation stage, the number of total nuclei and BrdU-positive nuclei of the mesenchymal cells were counted at the areas in the outer and inner bend of the gut wall as shown in the slash area in Fig. 1b. The percentage of BrdU-positive nuclei was calculated as in the expansion stage, and the averages from bend were compared. In addition, the number of BrdU-positive nuclei in the gut mucosal epithelium was counted in the outer and inner bends (Fig. 1b). These values were divided by the area of the measured mucosal layer and expressed as BrdU-positive cell density. The averages in each bend were compared.
For the analyses of mesenchymal cells in both the expansion and flexure formation stages, 3 semi-serial sections were observed for each of 4 embryos, and the sample value reflects the average value of the three sections. For analysis of the mucosal epithelium, 1 section was used for each of 3 embryos at the flexure formation stage.
Reverse transcription and quantitative real-time polymerase chain reaction (qPCR) analysis
To identify candidate genes associated with DJF formation, mRNA expression in the DJF at the expansion stage was investigated. The DJF was isolated and divided into the dorsal and ventral parts with tweezers and needle according to the body axis. Three pairs of dorsal and ventral parts were obtained. The total RNA of each DJF part was extracted, and complementary DNAs (cDNA) were synthesized by the ReverTra Ace reverse transcriptase enzyme (Toyobo, Osaka, Japan) and random primers (Promega, Madison, WI, USA). The qPCR analysis was performed using the Brilliant II SYBR Green qPCR Master Mix (Agilent Technology, Tokyo, Japan) and a real-time thermal cycler (MX 3000P; Agilent Technology) according to the manufacturer's instructions. Gene expression levels were normalized to the expression of the actin, beta (Actb) gene. Details of the specific primer pairs used for each gene are provided in Table 1.
Table 1. Summary of specific gene primer pairs
Genes (Accession No)
Primer sequence (5′-3′) F: Forward, R: Reverse
Product size (bp)
Actb, actin, beta; Aldh1a2, aldehyde dehydrogenase family 1, subfamily A2; Ccnd1, cyclin D1; Ihh, Indian hedgehog; Lef1, lymphoid enhancer binding factor 1; Lrp5/6, low density lipoprotein receptor-related protein 5/6; Rhoa/c: ras homolog gene family, member A/C; Shh, sonic hedgehog; Smo, smoothened homolog; Wnt5a, wingless-related MMTV integration site 5A.
Complementary RNA (cRNA) probes were synthesized in the presence of digoxigenin (DIG)-labeled UTP using a DIG RNA Labeling Kit (Roche Diagnostics, Mannheim, Germany). The primer pairs for the generation of each probe are shown in Table 1. The embryos were treated with ethanol containing 6% H2O2 and then digested with 10 μg mL−1 proteinase K for 15 min at RT. The embryos were incubated with prehybridization and then hybridization buffers containing 50% deionized formamide, 2% Blocking Reagent (Nucleic Acid Detection Kit; Roche Diagnostics), 50 μg mL−1 yeast rRNA gene (Sigma-Aldrich, St. Louis, MO, USA), 50 μg mL−1 heparin sodium salt (Sigma-Aldrich), 0.1% N,N-dimethyl formamide, Triton X-100, 5× SSC (pH7.0), 5 mm EDTA, and sense or antisense RNA probe (final concentration 300–400 ng mL−1) overnight at 70 °C. After washing and blocking, positive signals were detected using 0.2% polyclonal sheep anti-DIG Fab fragments conjugated to alkaline phosphatase (Nucleic Acid Detection Kit; Roche Diagnostics), nitroblue tetrazolium, and X-phosphate.
In situ hybridization (ISH) analysis
Semi-serial cross-sections of the DJF at the expansion stage and cRNA probes for Shh were used (Table 1). The sections were deparaffinized and digested with 10 μg mL−1 proteinase K for 1 h, and incubated with a prehybridization buffer followed by incubation with a hybridization buffer containing 40% deionized formamide, 1× Denhardt's solution (Sigma-Aldrich), 20 μg mL−1 yeast rRNA gene (Sigma-Aldrich), 10 μg mL−1 salmon testis DNA (Sigma-Aldrich), 10% dextran sulfate, 10 mm Tris-HCl (pH7.4), 600 mm NaCl, 1 mm EDTA, 1 × TE, and sense or antisense RNA probe (final concentration 350 ng mL−1) overnight at 58 °C. After washing and blocking, the signal was detected by the same method as used in WISH. The integrated densities of the epithelium in the left, right, dorsal, and ventral areas relative to the intestinal axis (Fig. 1c) were measured by ImageJ and compared with each other. For this analysis, 3 sections from a single embryo were used.
The results are expressed as the mean ± standard deviation (SD) and analyzed by the Wilcoxon signed-rank test (P <0.05) or the paired t-test (P <0.05).
Definition of the body axis and intestinal axis
To observe the DJF of fetal mouse, these two axes are defined as follows:
‘Body axis' based on the perspective of the entire mouse body.
‘Dorsal’ is defined as the side of the spinal column, and ‘ventral’ is defined as the direction opposite to ‘dorsal’.
‘Left’ is defined as being to the left of the spinal column from the perspective of the fetus body, and ‘right’ is defined as the direction opposite to ‘left’.
The body axis does not shift throughout the development of fetus.
‘Intestinal axis' based on the perspective of an intestine.
‘Dorsal’ is defined as the direction of the portion attached to the dorsal mesentery, and ‘ventral’ is defined as the direction opposite to ‘dorsal’.
‘Left’ is defined as the left side of the dorsal mesentery from the perspective of the intestine, and ‘right’ is defined as the direction opposite to ‘left’.
Importantly, during E10.75–11.75 (the expansion to flexure formation stage), the right of the DJF based on the intestinal axis gradually shifts from the right to the dorsal based on the body axis due to gut rotation.
Morphological and flexure patterns in the postnatal and prenatal gastrointestinal tracts
To confirm the presence of programmed controls in gastrointestinal morphogenesis, the morphology of the gut tubes was observed in postnatal and prenatal mice. In postnatal mice, although the intestines had different morphologies in the jejunum, ileum, and cecum at the same age, common morphological patterns and flexures were observed in some regions, such as the descending and ascending parts of the duodenum, the colon, the cranial and caudal flexures of the duodenum, and the DJF (Supporting Information Fig. S1a,b). In prenatal mice, all gut tubes had the same pattern among fetal mice on the same embryonic days (Fig. 2a,b,d,e,g,h). In particular, the DJF was clearly observed as an independent flexure through the period from E10.75 to E13.75. DJF formation was classified into three stages according to the existence and position of the flexure: (i) the expansion stage, characterized by increased gut diameter without clear flexure (Fig. 2c); (ii) the flexure formation stage, characterized by flexure along the D-V body axis (Fig. 2f); and (iii) the flexure elongation stage, characterized by elongated flexure around the stomach with counterclockwise rotation along the antero-caudal intestinal axis (A′-C′ intestinal axis; Fig. 2i). In this study, we focus on each stage of DJF formation to elucidate how intestinal flexures develop.
Location of the dorsal mesentery during DJF formation
To assess the contribution of the dorsal mesentery to DJF formation, the attached portions of the dorsal mesenteries in the fetal gastrointestinal tracts were analyzed at each DJF stage (Fig. 2b,c,e,f,h–k). At the expansion stage, the dorsal mesentery was attached to the dorsal side of the DJF based on the body axis (Fig. 2b,c). The cross-section of the fetus at the expansion stage also revealed that the gut tube was held by the dorsal and ventral mesenteries, and tilted toward the left side of the body axis (Fig. 2j). At the flexure formation stage based on the body axis, the right side of the DJF was not covered by the mesentery, whereas the left side of the DJF was covered (Fig. 2e,f). This finding implied that the position of dorsal mesentery in fetus was altered from the dorsal to the left side of the DJF based on the body axis during development. However, importantly, the attaching position of dorsal mesentery to the DJF did not change and was still located to the dorsal side of the DJF based on the intestinal axis during development. Taken together, these findings indicate that the dorsal mesentery was attached to the dorsal side of the DJF based on the intestinal axis during DJF formation, and the DJF was formed along the L′-R′ intestinal axis with a 90° counterclockwise rotation along the A′-C′ intestinal axis (Fig. 2f). At the flexure elongation stage, the dorsal mesentery was attached across the two gut tube parts forming the ascending and descending portions of the flexure (Fig. 2h,i,k). The position of attachment of the dorsal mesentery was the lateral side of the DJF, marking the caudal position of the stomach (Fig. 2i). These results suggested that the direction of attachment of the dorsal mesentery to the DJF (the left-right body axis; L–R body axis) did not correspond to the bending direction of the DJF (the D-V body axis) during flexure formation.
Histological differences in the DJF along each intestinal axis at the expansion stage
To investigate histological differences in the DJF along the intestinal axis, cross-sections of the DJF were observed at the expansion stage (Fig. 3a). The mucosal layer of the DJF was lined by columnar-shaped epithelium, and the outer layer of the mucosa was composed of mesenchymal cells with a narrow cytoplasm. The outermost layer of the DJF was covered by mesothelium extending from the dorsal and ventral mesenteries. In the morphological features of these cells, no constant difference was observed along the D′-V′ intestinal axes. Because the DJF was formed along the L′-R′ intestinal axis (Fig. 2f), the nuclear number and nuclear area of mesenchymal cells in the external layer of the gut wall were compared between the left and right sides of the intestinal axis (Fig. 1a). Because the cytoplasm of mesenchymal cells was quite narrow, the nuclear number and nuclear area indicate the number of assembly cells and cell area, respectively. We observed no difference in the nuclear number of the mesenchymal cells between the left and right measured areas (Fig. 3b), but the left area showed higher values for the nuclear area than did the right area (Fig. 3c).
Furthermore, the contribution of cell proliferation to DJF formation was assessed at the expansion stage. As shown in Fig. 3d-f, the number of BrdU-positive mesenchymal cells in the gut wall was examined as the percentage of positive nuclei in the dorsal, ventral, left, and right areas based on the intestinal axis (Fig. 1a). The left area showed the highest value and had significantly higher values than did the dorsal or ventral areas (Fig. 3e). The summation value of the left and right areas was higher than that of the dorsal and ventral areas (Fig. 3f). In addition, although we performed ssDNA immunohistochemistry to detect apoptosis, there were no positive cells in the mesenchyme of the gut wall (data not shown).
Histological differences in the DJF along the bending axis at the flexure formation stage
The histological features of the DJF at the flexure formation stage are shown in Fig. 4. The gut wall was thinner on the outer side based on the bending axis than on the inner side (Figs 1b and 4a), and the results of histoplanimetric analysis revealed a significant difference (Fig. 4b). Furthermore, the shape of the mesothelium lining the outermost DJF was squamous on the outer bend (Fig. 4c), but cuboidal or round on the inner bend (Fig. 4d). According to this observation, the nuclear height/width ratio of the mesothelium was significantly lower on the outer than on the inner bend (Fig. 4e). In addition, the difference of the cell shape between the outer and inner bending sides tended to be observed in the mesenchyme as well as in the mesothelium (Fig. 4d,e). For the cell proliferation assay, the outer gut wall based on the bending axis tended to have more abundant BrdU-positive mesenchymal cells than did the inner wall (Figs 1b and 4f) and a significant difference was observed in the measurement of BrdU-positive nuclei (Fig. 4g). The mucosal epithelium on the outer bend also tended to have higher values than on the inner bend in the evaluation of BrdU-positive cells (Figs 1b and 4h).
The mRNA expression levels of candidate genes associated with DJF formation
At the expansion stage, to identify candidate genes associated with DJF formation, the mRNA expression levels of several genes were compared between the dorsal and ventral parts of the DJF based on the body axis (Fig. 5). We examined genes associated with gastrointestinal morphogenesis based on previous reports; these genes are involved in RA signaling (Aldh1a2), canonical Wnt signaling (Ccdn1, Lef1, Lrp5, Lrp6), non-canonical Wnt signaling (Rhoa, Rhoc, Wnt5a), and hedgehog signaling (Ihh, Shh, Smo; Table 1). qPCR analysis revealed that the expression levels of Aldh1a2, Ccdn1, Lrp5, Lrp6, Rhoa, Rhoc, Ihh, Shh, and Smo tended to be higher in the ventral part of the DJF than in the dorsal part based on the body axis, and significant statistical differences were observed in the levels of Ccdn1, Lrp6, and Ihh. Importantly, because the gut tube was tilted toward the left side based on the body axis at the expansion stage (Figs 2f and 3a), the dorsal and ventral parts of the DJF based on the body axis included mainly the left and right parts based on the intestinal axis, respectively. Therefore, these results also indicate that the right side of the DJF tends to express higher Aldh1a2, Ccdn1, Lrp6, Ihh, and Shh than the left side does based on the intestinal axis. Interestingly, the dorsal expression levels of Ihh and Shh were undetectable in two of three DJF samples, and Aldh1a2, which is associated with axis formation in several organogenesis processes, showed the third largest difference (approximately twice) of expression between two parts in the DJF compared with the other examined genes. To more accurately evaluate expression differences between the ventral and dorsal parts of the DJF, we analyzed mRNA expression levels in the DJF by subsequent hybridization methods, with a particular focus on Aldh1a2 and hedgehog.
mRNA expression levels of Aldh1a2 and Shh in the DJF at the expansion and flexure formation stages
To confirm the results of qPCR, the mRNA expression levels of Aldh1a2 and Shh were examined at the expansion and flexure formation stages by the WISH method. At the expansion stage, Aldh1a2-positive signals were observed in the thoracic and tail spinal cords, lateral plate mesoderm, otocyst, and gastrointestinal tract (Fig. 6a); in particular, the mesenchymal layer of the DJF showed strong signals (Fig. 6b). In the DJF, the ventral side based on the body axis tended to show a stronger positive signal than did the dorsal side (Fig. 6b). At the flexure formation stage, Aldh1a2-positive signals were observed in positions similar to those seen in the expansion stage (Fig. 6c). In particular, in the mesenchymal layer of the DJF, the positive signal was slightly higher in the dorsal side based on the body axis (i.e. the outer side based on the bending axis) than in the ventral side (Fig. 6d). No positive signal was observed in the sense-negative control (Fig. 6c, inset).
At the expansion stage, positive signals for Shh expression were observed in the cerebral ventricles, neural canals, otocyst, caudal region of the limbs, and gastrointestinal tract (Fig. 7a); in particular, the mucosal layer of the gut tube including DJF showed a strong positive reaction (Fig. 7b). At the flexure formation stage, the localization of Shh expression was similar to that seen in the expansion stage (Fig. 7c), and Shh expression was observed along the gut tube including the DJF (Fig. 7d). At both the expansion and flexure formation stages, no apparent difference was observed in the intensity of Shh-positive signals between the dorsal and ventral parts of the DJF (Fig. 7b,d). No positive signal was observed in the sense-negative control (Fig. 7c, inset).
To observe Shh expression in detail, ISH was performed on cross-sections of the DJF at the expansion stage. The results of ISH indicate that Shh was expressed in the mucosal epithelium of the DJF (Fig. 7e). Furthermore, the integrated density of Shh-positive signals was compared among the left, right, dorsal, and ventral mucosal areas based on the intestinal axis on the ISH section (Figs 1c and 7f). The values in the left and right areas tended to be higher than those in the other two areas (Fig. 7f). In the adult intestine, clear Shh-positive signals were observed in the epithelium, particularly at the crypts (Fig. 7g), and no positive signal was observed in the sense-negative control (Fig. 7h).
Genetic programs are likely to control gastro-intestinal morphology
Individual postnatal mice exhibited both common and unique morphological features in the intestines. The common features, including the DJF, are believed to be developed under programmed control. In the postnatal stage, peristalsis causes the intestines to change position in the abdominal cavity. During peristalsis, the intestines contract and stretch, forming irregular flexures covering the regular flexures conserved among the species. Furthermore, some regions of the intestine, especially the jejunum, ileum, and colon, are less tightly attached via the mesentery to the abdominal wall. Therefore, environmental factors rather than genetic factors are responsible for the formation of intestinal morphology in postnatal mice.
In prenatal mice, every gut tube had the same pattern among different individuals during E10.75–13.75. At earlier prenatal stages, the gastrointestinal tract is still undergoing developmental processes such as morphological changes and cell differentiation in the mucosa or mesenchyme. Therefore, only genetic factors would likely affect intestinal morphology, not peristalsis derived from the contraction of smooth muscles. In fact, the process of gastrointestinal formation had common features among individual fetal mice and, in particular, common features in the development of the DJF were independently and clearly observed. From these results, we propose that DJF formation is strongly affected by genetic factors and is appropriate as a representative model for the analysis of gut flexure formation.
The dorsal mesentery is unlikely to be involved in DJF formation
Recently, the dorsal mesentery was reported to contribute to processes that determine gastrointestinal morphology, such as gut rotation and looping of the future jejunum (Davis et al. 2008; Kurpios et al. 2008; Savin et al. 2011). In this study, the dorsal mesentery was attached to the dorsal side of the DJF based on the body axis at the expansion stage. However, at the flexure formation stage, the dorsal mesentery was attached to the left side of the DJF based on the body axis. Furthermore, at the flexure elongation stage, the dorsal mesentery was attached to the lateral side of the DJF, which faced the stomach. These observations indicate that the dorsal mesentery is located not on the dorsal side but on the left lateral side of the DJF based on the body axis during DJF formation. Importantly, the DJF was bent toward the dorsal direction based on the body axis at the flexure formation stage. Therefore, we propose that the dorsal mesentery is less likely to contribute to the formation of the DJF, because the direction of its attachment (left side based on the body axis) does not correspond with the bending direction of the DJF (dorsal side based on the body axis). This result strongly supports the existence of genetic factors affecting DJF bending.
Histological differences based on the bending axis contribute to DJF formation
As described above, the DJF was formed by a 90° counterclockwise rotation along the A′-C′ intestinal axis during the expansion stage to the flexure formation stage (Fig. 2c,f). Therefore, the D′-V′ intestinal axis of the DJF at the expansion stage is parallel to the L–R body axis at the flexure formation stage (Fig. 2c,f). Furthermore, the left side based on the intestinal axis at the expansion stage became the inner bend at the flexure formation stage. Therefore, it is likely that the asymmetric development of the DJF along the L′-R′ intestinal axis contributed to DJF formation.
The left-right asymmetry in morphological development is well known to be concerned with various internal organs, including the gastrointestinal tract (Levin, 2005). In this study, we have clarified that the left and right sides based on the intestinal axis become the future inner and outer bend, respectively. Furthermore, at the expansion stage, the nuclear area of the mesenchymal cells in the external layer of the DJF gut wall were larger on the left side based on the intestinal axis than on the right side, despite the presence of the same cell density in the both sides. These results demonstrate the morphological asymmetry in the cell size of the mesenchymal cells between the left and right sides based on the intestinal axis (Fig. 8a, left scheme). Another possible interpretation is that of asymmetric cell polarity (Fig. 8a, right scheme). Briefly, it is possible that the cells on the left side based on the intestinal axis tend to be large and long along the apico-basal intestinal axis, and the cells on the right side tend to be small and long along the A′-C′ intestinal axis. In addition, more active cell proliferation is observed along the L′-R′ intestinal axis than along the D′-V′ intestinal axis at the expansion stage. These results indicate that mesenchymal proliferation with asymmetric cell morphology along the L′-R′ intestinal axis at the expansion stage would be a force for DJF formation.
Furthermore, at the flexure formation stage, the outer side had a thinner gut wall than did the inner side based on the bending axis. From this finding, we propose that CE in the bend occurred more frequently in the outer bend than in the inner bend of the DJF. Moreover, the mesothelium was more squamous on the outer side than on the inner side in the DJF. This result confirmed the asymmetrical cell morphology of the DJF along the bending axis.
In summary, we propose a process for DJF formation as shown in Fig. 8b. Briefly, from the expansion stage to the flexure formation stage, the mesenchymal cells in the gut wall proliferate and accumulate to expand the DJF along the L′-R′ intestinal axis, and these accumulated cells contribute to the force of DJF bending (Fig. 8b, upper schemes). At the same time, differences in mesenchymal cell morphology and CE rate along the L′-R′ intestinal axis also contribute to the asymmetric elongation of the gut wall as another force involved in DJF bending (Fig. 8b, lower schemes). DJF bending is accomplished by the combined effect of the asymmetric mesenchymal cell morphology and the CE rate (Fig. 8b, right-most scheme).
Aldh1a2 and hedgehog are candidate genes associated with DJF formation
Aldh1a2 mRNA was expressed only at the mesenchyme of the DJF in the gastrointestinal tract and showed a gradation along the D-V body axis at the expansion and flexure formation stages. Aldh1a2, an RA generating enzyme, participates in RA metabolism by oxidizing retinol to retinal and then to RA (Marlétaz et al. 2006). The RA signal regulates the expression levels of genes including homeobox genes and several transcription factors (Marlétaz et al. 2006). RA signaling was found to be critical for both embryonic body shaping and organogenesis via tissue homeostasis and cell proliferation, differentiation, and apoptosis (Mark et al. 2009). RA signaling is implicated in embryonic axis formation including the D-V body axis, the antero-caudal body axis, and the L–R body axis (Kam et al. 2012). During axis formation, the gradation of RA concentration along the axis is maintained by Aldh and degradative enzymes (Kam et al. 2012). Disruption of RA signaling perturbs gut rotation, elongation, and looping in the Xenopus gastrointestinal tract (Lipscomb et al. 2006) and is concerned with duodenal atresia via fibroblast growth factor 10 (Fgf10) in human and mouse (Nichol et al. 2012). Interestingly, RA signaling also plays an essential role in the regionalization of the endoderm to specify the dorsal pancreas, and it has been reported that the RA synthesized from the paraxial mesoderm acts on the endoderm (Kam et al. 2012). Based on these findings, we propose that the graded expression of Aldh1a2 in the DJF mesenchyme is associated with the induction of RA signaling, providing positional and directional information for flexure formation.
Ihh and Shh are members of the hedgehog family, and induce the hedgehog signaling pathway that is important for diverse aspects of animal development (van den Brink, 2007; Jiang & Hui, 2008). The hedgehog proteins are secreted proteins and their signals control hedgehog target gene expression (van den Brink, 2007). Hedgehog signaling is essential for the formation of the body axis, and Ihh and/or Shh mutant mice exhibited gut malrotation and L–R body axis disruption due to the collapse of the left side decision in the node at an early developmental stage (E7.25–9.5; van den Brink, 2007). In addition, hedgehogs act as paracrine mitogens to promote the cell proliferation of myofibroblasts, smooth muscle cells, and enteric neurons in the mesenchyme of the gut wall, thereby increasing the size and length of the gut tube (Jiang & Hui, 2008; Mao et al. 2010; Powell et al. 2011; Biau et al. 2012). Our observations in this study indicate that the cell proliferation rate along the L′-R′ intestinal axis is higher than that along the D′-V′ intestinal axis at the expansion stage. Furthermore, Shh mRNA expression in the epithelium along the L′-R′ intestinal axis tends to be higher than that along the D′-V′ intestinal axis; this result corresponds with that of cell proliferation analysis. Therefore, for DJF formation, the gradation in hedgehog expression might cause differences in the cell proliferation rate, especially that of the mesenchyme between the left-right and dorso-ventral areas based on the intestinal axis.
Cross-talk between candidate genes and other signaling molecules in DJF formation
In addition, it is important to discuss the cross-talk among the RA, hedgehog, and other signaling pathways. It has recently been proposed that RA signaling must be coordinated with other embryonic signaling pathways, including hedgehog, Fgf, Wnt, and Notch signaling (Niederreither & Dollé, 2008; Kam et al. 2012). RA signaling induces Shh at the proper position to establish antero-caudal digit patterning during forelimb development, and affects the expression of Shh downstream target genes (Niederreither & Dollé, 2008). Furthermore, G-protein-coupled receptors induced by RA were implicated in activating non-canonical Wnt signaling via frizzled receptors (Harada et al. 2007), and non-canonical Wnt signaling pathways including the PCP signaling pathway were associated with CE (Schlessinger et al. 2009). Indeed, the different cell morphologies and CE rates along the L′-R′ intestinal axis in this study might be associated with differences in PCP signaling effects, such as the intensity of the response to the signal. However, no differences were evident along the L′-R′ intestinal axis in the expression levels of related Wnt/PCP signaling factors such as Wnt5a and Rho in qPCR at the expansion stage. We assumed the existence of other Wnt ligands and downstream targets. Moreover, hedgehog signaling has also been reported to be associated with canonical Wnt signaling (Katoh & Katoh, 2006). In the intestine, hedgehog signaling induces an antagonist of canonical Wnt signaling and inhibits cell proliferation in the intestinal epithelium (Katoh & Katoh, 2006). Therefore, further investigation of non-canonical and canonical Wnt signaling pathways is needed to elucidate the cross-talk among these signaling pathways.
Taken together, we propose that the gradation in the expression of RA signaling molecules might regulate cell proliferation and morphology via other signaling pathways such as hedgehog and Wnt signaling, and subsequently cause the bending of the intestine to form the DJF.
In conclusion, gastrointestinal morphology, especially DJF formation, is likely to be controlled by multiple genes. We propose a model of DJF formation, in which Aldh1a2 and embryonic signals including RA and hedgehog are involved in DJF formation by regulating cell proliferation and morphology.
The research described in this paper was chosen for the Encouragement Award (undergraduate section) at the 154th Japanese Association of Veterinary Anatomists in Iwate (14–16 September 2012). This work was partially supported by a Grant-in-Aid for Scientific Research (B) (No. 24380156).