The twists and turns of left-right asymmetric gut morphogenesis

Left-right (LR) asymmetry is a fundamental characteristic of vertebrate anatomy. In humans, this is evident in the left- or right-sided positioning of multiple organs in the thoracic and abdominal cavities (Fig. 1). Moreover, individual organs exhibit morphological laterality independent of their location, as evident in the unequally sized left and right lobes of the liver, or the disparate lobation patterns of the left versus right lungs. In many cases, these asymmetries are evolutionarily conserved and necessary for physiological function, indicating that the formation of LR asymmetry is an essential phase of vertebrate organogenesis. Indeed, abnormal LR axis formation – a condition known as heterotaxy – is associated with some of the most common and severe structural birth defects, including complex congenital heart defects (CHDs), intestinal malrotation, extrahepatic biliary atresia, asplenia/polysplenia and other anomalies (Bartram et al., 2005; Chinya et al., 2019; Desgrange et al., 2018; Gabriel and Lo, 2020; Kothari, 2014; Ticho et al., 2000). Therefore, understanding the mechanisms that shape individual organ lateralities is not only necessary for illuminating the morphogenesis of numerous organs but could also be crucial for explaining the etiology of a wide variety of birth defects.

Decades of studies have revealed the developmental processes that initially orient the LR body axis in vertebrate embryos (reviewed by Almirantis, 1995; Blum et al., 2014; Brueckner et al., 1991; Capdevila et al., 2000; Grimes and Burdine, 2017; Levin, 2005; Norris, 2012). In brief, a crucial phase of the symmetry-breaking process that occurs in many species involves cilia that are found within specialized LR organizer (LRO) regions (Dasgupta and Amack, 2016; Schweickert et al., 2017). The action of these cilia generates an asymmetrical fluid flow that establishes local asymmetries in gene expression (Cartwright et al., 2004; Okada et al., 2005; Schweickert et al., 2017; Smith et al., 2008), ultimately leading to the expression of Nodal, a transforming growth factor β superfamily ligand, in the left lateral plate mesoderm (LPM). Nodal then activates expression of the transcription factor Pitx2 in the left LPM and, ultimately, on the left side of developing organs, where it is required for the proper morphogenesis of various anatomical asymmetries (Burn and Hill, 2009; Campione et al., 1999; Davis et al., 2017; Liu et al., 2001; Mahadevan et al., 2014; Muller et al., 2003; Plageman et al., 2011; Ryan et al., 1998; Welsh et al., 2013; Womble et al., 2018). In some species, cilia are not involved and the early symmetry-breaking events remain unclear (Gros et al., 2009; Hamada and Tam, 2020; Kajikawa et al., 2020); nonetheless, the process culminates in LR asymmetrical gene expression in the LPM.

Although we know that asymmetrical Nodal/Pitx2 expression is required to confer ‘leftness’ to tissues left of the midline, what ‘leftness’ (or ‘rightness’) means, in terms of the morphogenesis of each individual organ, has been unclear and is only just beginning to be defined. Progress in understanding this last crucial phase of organogenesis has been slower than progress in elucidating the initial symmetry-breaking events. This is partly because, for most asymmetrical organs, the putative differences between the morphogenesis of their left and right sides is understudied or entirely unknown. Furthermore, what we think we know about the most basic developmental events that generate asymmetry in each organ is dominated by hypothetical descriptions proffered over one century ago by classical human embryologists, despite such ideas being supported by little quantitative or experimental data, or even being refuted.

In this Review, our goal is to reconcile historical models with contemporary studies that have provided new insights into asymmetric morphogenesis of the digestive organs, focusing on the stomach and intestine, but also touching on other gut-derived/associated organs such as the liver, pancreas, spleen and lung. We focus on tissue level changes, rather than molecular details, and aim to re-contextualize common dogma, highlight unanswered questions regarding the mechanisms that generate organ-specific LR asymmetries, and illuminate fruitful paths for future study.

The earliest LR asymmetrical morphology to appear during the development of the vertebrate digestive tract is the curvature of the stomach. This organ originates as a straight segment of the foregut that gradually acquires a leftward curved, J-shaped morphology with a convex (outward bending) curvature on the left side of the body and a concave (inward curling) curvature on the right (Fig. 1). The prevailing classical view of stomach development asserts that this curved shape arises as a result of repositioning the original dorsal wall of the stomach to the left side of the body via active ‘rotation’ of the primitive organ on its own longitudinal axis. In this model, the dorsal wall will become the convex side of the stomach (classically termed the ‘greater’ curvature), while the ventral wall, which is simultaneously rotated to the right side, becomes the concave side (or ‘lesser’ curvature; Fig. 2A).

This classical ‘rotation’ model arose in the late 19th and early 20th century, based largely on two observations made by human embryologists. The first observation was that the dorsal mesogastrium, which anchors the stomach to the dorsal body wall, is ultimately found attached on the left side of the organ after curvature (Fig. 2A). Interestingly, the right side of the dorsal mesogastrium is known to undergo vacuolization and breakdown as the structure lengthens to form the greater omentum (Fig. 2A; Liebermann-Meffert, 2000). This remodeling process was presumed to enable the mobility of the attached stomach during its rotation, although the morphogenetic and/or biomechanical mechanisms capable of driving such large-scale organ movement remained unexplained. The second observation was that the left and right vagus nerves arise along the left and right side of the stomach early in development, but are later branched along the ventral and dorsal faces of the curved organ, respectively (Larsen, 2014). Together, these two observations led to the conclusion that, in order to give rise to the observed repositioning of associated mesogastria and nerves during curvature formation, the stomach must rotate around its dorsoventral axis (Fig. 2A).

Interestingly, an alternative model of stomach curvature morphogenesis was put forward by embryologists in the mid-20th century. This alternative view posited that the asymmetrical shape of the organ forms independently of its rotation. In this model, the development of curvature is attributed to predominant growth of the left side of the stomach over the right, such that the original left wall expands to form the convex side (Macarulla-Sanz et al., 1996; Miete, 1960). The concave side, by contrast, is derived from the original right side. This ‘LR asymmetrical growth’ model (Fig. 2B) originated from skepticism about the practicability of whole-organ rotation, as well as observed differences in epithelial architecture on the left versus right sides of the human stomach during curvature (Dankmejer and Miette, 1961).

Perhaps owing to the simplicity of the classical rotation model, the LR asymmetrical growth model is less prevalent in the literature and often overlooked in medical textbooks. This is likely because the non-rotation-based mechanism is thought to be at odds with the apparent rotation of the mesogastria and vagus nerves. However, proponents of the asymmetrical growth hypothesis argued that the connection of the dorsal mesogastrium to the dorsal midline of the stomach is not an immutable tether but a dynamic linkage that shifts as the stomach grows. They suggested that breakdown of the right side of the dorsal mesogastrium allows it to be ‘pulled’ to the left with expansion of the left stomach wall (Kanagasuntheram, 1957). In this model, the final position of the dorsal mesogastrium on the left side of the curved stomach is not indicative of a rotation event, but merely the result of the original midline attachment point shifting to the left as the organ grows preferentially leftward (Fig. 2B). But what about the left and right vagus nerves becoming located on the ventral/anterior and dorsal/posterior surfaces of the curved stomach? Non-rotation proponents point out that the left and right vagus nerves also assume ‘rotated’, i.e. ventral and dorsal, positions in the esophagus (Kanagasuntheram, 1957), a region of the gut that does not curve or rotate (Borghi et al., 2002). Therefore, the vagus nerves may pathfind by mechanisms independent of organ orientation per se, such that their final positions are potentially irrelevant to the argument of stomach curvature.

It should be noted that both the rotation and non-rotation (i.e. LR asymmetrical growth) models were proposed prior to the advent of modern developmental biology and were based largely on qualitative, retrospective observations of a series of preserved human embryos. However, several groups have recently addressed the origin of stomach curvature using quantitative methodologies and in vivo experiments in animal models. Interestingly, these studies largely support the non-rotation, LR asymmetrical growth hypothesis. For example, highly quantitative, 3D morphometric analyses of human embryonic stomachs revealed predominant growth of the left gastric wall over the right, finding little evidence for rotation around the dorsoventral axis that is proposed by the classical rotation model; only a gradual deflection of the stomach caudally and to the left was observed, thus reflecting differential growth (Kaigai et al., 2014; Nebot-Cegarra et al., 1999). More recently, animal model studies have directly addressed the cellular and molecular basis for stomach laterality (Davis et al., 2017). These revealed that, in both mouse and frog embryos, differences in the morphogenesis of the left versus right sides of the stomach facilitate preferential thinning and expansion of the left stomach wall, consistent with predominant ‘growth’ of the left side of the organ (Fig. 2B). Surprisingly, this expansion was found to be independent of LR asymmetries in cell numbers; instead, it appears to be driven by asymmetrical cell rearrangements, facilitated by precocious polarization and radial intercalary rearrangement of the endoderm cells that differentiate into the gastric epithelium in the left stomach wall (Davis et al., 2017; see Fig. 4B). This early morphogenetic difference is thought to leave the primitive stomach tube with unequal left and right side dimensions, manifested as whole organ-scale curvature. In both species, the changes in LR asymmetric cell polarization and rearrangement were found to be dependent on earlier LR patterning events, including cilia function (in mouse), Nodal activity (in frog), and left side-specific activity of Pitx2c (in frog) (Davis et al., 2017). Thus, accumulating evidence suggests that, despite the widespread propagation of the rotation theory, the early curvature of the stomach is likely generated by asymmetrical growth of its contralateral walls.

Caudal/posterior to the curved stomach, discrete segments of the small and large intestine also become positioned to the left or right of the midline. In humans, for example, the descending duodenum is on the right, the duodenojejunal flexure is on the left, the ascending colon and cecum are on the right, and the descending colon is on the left (Fig. 1). These asymmetries are derived from the development of the embryonic midgut, which forms a prominent ‘primary’ loop during gut morphogenesis (Fig. 3A). The cranial limb of this loop is composed of the future duodenum, jejunum and proximal ileum, while the caudal limb will become the distal ileum, cecum and colon [the cranial and caudal limbs were traditionally designated based on their relative position with respect to the superior mesenteric artery (SMA); Fig. 3A].

Historically, the process of establishing small and large intestinal laterality was deemed to occur via ‘rotation’. This was because the final anatomical positions of the various midgut derivatives and the twisted appearance of the attached mesentery suggested that the cranial and caudal limbs of the primary loop must rotate around each other in a counter-clockwise (CCW) direction during development (Blechschmidt and Kircheiss, 1973; Gasser, 1975; Kim et al., 2003; Mall, 1898; Snyder and Chaffin, 1954). In this classical view (the ‘Rotation’ model, Fig. 3B), the entire midgut loop, as a complete entity, is thought to undergo a series of twisting turns, ultimately completing a 270° course of rotation to place the various segments of the intestine and colon in their final left- or right-sided locations. The first 90° rotation occurs prior to formation of the primary loop itself, as the cranial end of the primitive midgut bends to the right and the caudal end to the left, forming an ‘S’-shape when viewed ventrally (Fig. 3A; Snyder and Chaffin, 1954). In amniotes, rapid lengthening of the midgut tube compared with the embryonic body cavity then leads to extension of the partially rotated primary loop into the umbilical space (known as the physiological umbilical hernia; see Fig. 3A); later, the gut must re-enter the abdominal cavity. According to the rotation model, an additional 180° CCW rotation occurs while the midgut is inside the umbilicus (Kim et al., 2003) and/or as it undergoes re-entry (Frazer and Robbins, 1915; Snyder and Chaffin, 1954); the cranial limb shifts caudally and then to the left, and the caudal limb moves cranially and then to the right (Fig. 3B). In this classical view, the cranial and caudal limbs of the primary loop cross over each other as they rotate, twisting around the SMA at the root of the mesentery, to place the segments of the mature duodenum, jejunum, ileum, cecum and colon in their final left- or right-sided positions (Fig. 3D; Kim et al., 2003; Mall, 1898; Snyder and Chaffin, 1954). In humans with congenital intestinal defects, the abnormal relative locations of these segments were interpreted to be the result of the midgut failing to properly complete the rotation process, leading to such anatomical configurations being termed intestinal ‘malrotations’ (Andrew, 1961; Deitch and Engel, 1980; Dott, 1923; Freitas and Ventura, 1980; Glowniak, 1988; Price and Kane, 1955; Schwalbe, 1906; Torres and Ziegler, 1993; Valioulis et al., 1997).

Like stomach rotation, the concept of midgut rotation distills the convoluted topological transformations of intestine morphogenesis into a straightforward succession of rotating turns; this likely underlies the persistence of the century-old rotation model in modern literature and textbooks. However, over the years, several authors have questioned the assumption that intestine laterality is derived from such a user-friendly operation, suggesting that the rotation model is oversimplified and misleading. Below, we discuss classical and recent observations that challenge ‘simple’ rotation as the mechanism that generates intestine laterality, and outline an alternative interpretation of the process as a hierarchical series of spatiotemporally coordinated looping and lengthening events that occur in discrete regions of the midgut (Fig. 3C).

As mentioned above, a key symmetry-breaking event in midgut morphogenesis is the formation of the initial S-shape in which the cranial portion of the primitive midgut curves to the right, while the caudal region curves to the left (Fig. 3A). Although the classical rotation model interprets this configuration as the initial 90° turn in an active 270° rotation process (Snyder and Chaffin, 1954), recent 3D imaging analyses of human embryos (Hikspoors et al., 2018; Soffers et al., 2015) suggest that the contour of the early midgut loop merely follows the helical shape of the body axis, which is created by the normal ‘turning’ of the mammalian embryo – a process that happens to coincide with the early stages of gut development. Moreover, it has been noted that the formation of the S-shape coincides with the normal caudalward descent of the curved stomach, which would logically be expected to shift the cranial end/limb of the midgut (which is contiguous with the right-sided, pyloric end of the stomach) caudally and rightward (Hikspoors et al., 2018; Soffers et al., 2015). Thus, although the first break in midgut symmetry may be described (geometrically) as a 90° rotation of the primitive midgut, this configuration alone is not necessarily evidence of an active organ-scale rotation process, as the S-shape could arise as a passive response to the pre-existing helical asymmetry of the body axis and/or the influence of apposed organs or attached tissues (discussed later).

After the establishment of the S-shape, continued lengthening of the midgut causes it to form the primary loop, the apex of which protrudes into the umbilicus, with cranial and caudal limbs still in their partially ‘rotated’ right- and left-sided positions (Fig. 3A). While herniated, the cranial limb then forms smaller loops along its length, resembling ‘sine wave’ undulations, likely a result of the rapidly lengthening gut tube buckling under the mechanical strain of its attachment to the mesentery (Fig. 3B; Savin et al., 2011); these smaller loops have recently been termed ‘tertiary loops’ (discussed in detail later). In contrast, the caudal limb does not undergo such convolutions, allowing it to be visually distinguished from the cranial limb. Multiple authors have used this morphological distinction to observe that, while herniated, the segments of the cranial and caudal limbs leading to (and contained within) the umbilicus do not alter their positions or twist around each other but remain on the right and left sides of the SMA, respectively. Thus, in contrast to depictions in classical rotation models, the midgut loop does not actually rotate prior to re-entry (compare Fig. 3B with C; Frazer and Robbins, 1915; Hikspoors et al., 2018; Soffers et al., 2015). Moreover, both classical and modern studies have noted that the cecum – a morphological landmark at the apex of the midgut loop, often used to track rotation – moves only ‘passively’ while herniated, i.e. it adopts different orientations as the mass of intestinal loops forms in the cranial limb but nonetheless remains to the left of the midline until re-entry (Fig. 3C; Frazer and Robbins, 1915; Metzger et al., 2011; Ueda et al., 2016). Consequently, the position of the cecum, although often depicted as adopting successively more CCW orientations during intestinal rotation (as in Fig. 3B; Sivakumar et al., 2018), is likely not an accurate reflection of gut situs. Indeed, in several species (e.g. rat and chick), the tip of the midgut loop even appears to (at least initially) rotate clockwise (Metzger et al., 2011; Southwell, 2006).

While the umbilical region of the cranial limb is forming its mass of smaller loops, the segment of this limb still inside the abdominal cavity, which includes the duodenum and proximal jejunum, forms only a single large loop (labelled as ‘1’ in Fig. 3C). This intra-abdominal loop (i.e. as opposed to the herniated, intra-umbilical loops) originates on the right side but elongates extensively, crossing the midline caudal to the SMA, ultimately becoming the left-sided duodenojejunal flexure (DJF; see Figs 1 and 3C; Kluth et al., 1995; Kluth et al., 2003; Metzger et al., 2011). The final asymmetrical position of the DJF shifts the cranial end of the midgut leftward, thereby influencing the motility and orientation of the contiguous jejunal and ileal loops (Long et al., 1996). Indeed, striking correlations between intestinal malrotation/volvulus and abnormal orientations of the DJF in humans suggest proper morphogenesis of this unique intra-abdominal loop is essential for normal intestine laterality (Long et al., 1996). Although classical embryologists concluded that the DJF must be brought into its critical left-sided position by the coordinated rotation of the entire midgut, it actually achieves its asymmetric location early in gut development, well before the hypothetical rotation and re-entry of the herniated segments of the primary loop (Bardeen, 1914; Kluth et al., 2003). Thus, at least one essential intestinal laterality must form independently of the hypothetical rotation process (see also ‘Hierarchical looping’ model below).

Proponents of the rotation model suggested that growth of the caudal limb during and/or after re-entry into the body cavity must provide the force to propel rotational movement of the primary loop (Frazer and Robbins, 1915). In this view, the caudal limb would push the cranial limb leftward as it extends further rightward, each limb presumably re-orienting the other in a yin-yang duality (Fig. 3B). Yet multiple studies (Bardeen, 1914; Mall, 1898; Ueda et al., 2016) have shown that the caudal limb does not grow rapidly or extensively enough to impel the hypothesized rotation. Moreover, although the mechanisms that underlie re-entry remain poorly understood, it has been known since the early 20th century (Frazer and Robbins, 1915) that the herniated intestine does not return to the body cavity all at once. Instead, the mass of jejunal and ileal coils (derived from the cranial limb) returns to the body cavity first and moves to the left while the cecum and colon (derived from the caudal limb) are still herniated (Fig. 3C; Frazer and Robbins, 1915). Thus, the derivatives of the cranial limb assume their respective LR positions independently of the growth trajectory of the caudal limb.

The discrepancies discussed above, especially the spatially compartmentalized growth of the intra-abdominal (i.e. the DJF) versus herniated segments of the cranial limb, and the temporally independent re-entry of the cranial versus caudal limbs, suggest that coordinated rotation of the entire midgut during development is an implausible mechanism for generating digestive laterality. Without an actively orchestrated rotation process how, then, is the final LR positioning of the various derivatives of the midgut accomplished? Interestingly, recent 3D reconstructions of discrete stages of human intestine development (Soffers et al., 2015) have led to the rediscovery and further characterization of a second generation of four invariant ‘secondary’ loops that form along the length of the midgut. These anatomical features, first recognized over one century ago (Mall, 1898), are demarcated by areas of shorter mesentery between them, an embryological artifact that remains identifiable even in the adult intestine (Mall, 1898). These four stereotypical loops form in a predictable, hierarchical order along the cranial-caudal axis of the midgut, corresponding to the order of their re-entry into the body cavity (Fig. 3C; Soffers et al., 2015).

The identification of these loops led to the proposition of the ‘Hierarchical looping’ model (Fig. 3C; Soffers et al., 2015). In this model, the first secondary loop forms within the intra-abdominal region of the cranial limb (labelled ‘1’ in Fig. 3C; Soffers et al., 2015). As discussed above, this first secondary loop becomes the DJF, elongating extensively to assume a left-sided position in the body prior to re-entry of the herniated loops. The second and third secondary loops (labelled ‘2’ and ‘3’ in Fig. 3C) form in the intra-umbilical segments of the cranial limb, within the distal jejunum and proximal ileum, respectively. During re-entry, the coils of these second and third secondary loops pass sequentially under the SMA, spreading to the left, likely guided by their connection with the left-sided DJF. The fourth secondary loop (labelled ‘4’ in Fig. 3C) forms last, in the caudal limb, between the tip of the primary loop and the cecum, within the distal ileum (Soffers et al., 2015). This fourth loop is thought to be the last to re-enter the body cavity, filling the remaining available space on the right (Metzger et al., 2011; Soffers et al., 2015). Following re-entry, the distal ileum and cecum then slide dorso-caudally (rather than rotate) into their final position in the right lower abdomen (Fig. 3D).

In this contemporized view, individual segments of the midgut assume their left- or right-sided anatomical positions according to the order of looping and/or re-entry, the space available in the abdominal cavity, and/or the relative orientation and forces provided by neighboring intra-abdominal segments of the gut tube. Thus, according to the hierarchical looping model, the final orientation of the intestine segments, flexures and twisted mesentery merely creates the illusion that the gut rotates 270° around the SMA; the end result is the same but the developmental process generating the final laterality is distinctly different. Although more challenging to comprehend than a coordinated rotation process, hierarchical looping has the advantage of being entirely compatible with the spatially compartmentalized and temporally distinct development of the cranial versus caudal and intra-abdominal versus intra-umbilical segments of the primary midgut loop discussed above. This re-interpretation may have important implications for future studies of the morphogenetic mechanisms that underlie normal and abnormal intestinal laterality (discussed in detail by Soffers et al., 2015).

It has been hypothesized that, during vertebrate evolution, the primitive gut tube is the first structure to acquire LR asymmetry (i.e. as opposed to the heart; Blum et al., 2014), allowing longer, compartmentalized digestive tracts to be consistently packaged within the body cavity. However, LR asymmetries are not only formed by segments of the gut tube itself; many other lateralities in digestive and non-digestive organs originate from gut tube-derived or -associated structures (Fig. 1). Below, we touch briefly on a few examples for which recent studies have begun to shed light on the morphogenesis of anatomical LR asymmetries (Fig. 4).

The liver diverticulum buds off from the ventral surface of the primitive gut tube at the foregut-midgut boundary, posterior to the stomach. Notwithstanding some variation in accessory lobes between species, the right side of the liver usually becomes larger than the left side (Abdel-Misih and Bloomston, 2010). The classical explanation for this asymmetry is that the displacement of the cardiac sinus venosus to the right during heart looping enables greater blood flow (i.e. a growth advantage) to the right side of the embryonic liver (Grover and Moore, 1988). However, even prior to vascularization (i.e. shortly after the initial budding of the liver diverticulum from the primitive gut tube), the majority of the volume of the organ was found to lie on the right, suggesting the early bud is already inherently LR asymmetric (Grover and Moore, 1988; Heisler, 1907). Indeed, recent lineage tracing investigations identified the presence of left- and right-side clones in the embryonic liver, suggesting that cells contributing to each lobe become segregated early, and that the left and right halves of the organ develop largely independently (Weiss et al., 2016).

Consistent with this idea, a recent study in Xenopus showed that, as the early liver diverticulum buds off from the gut tube, hepatic endoderm cells on the right side become more apically constricted, with a greater length-to-width ratio, whereas left-sided cells appear rounder and more compact, indicating that distinct left- and right-sided morphogenetic programs are distinguished very early in liver morphogenesis (Fig. 4A; Womble et al., 2018). In this study, asymmetric expression of pitx2c in the mesoderm surrounding the left side of the early hepatic diverticulum (Fig. 4A) was found to be necessary and sufficient to elicit left-sided epithelial morphogenesis and lobe formation in the underlying endoderm. The nature of the molecular events mediating this tissue interaction, and the relationship between early asymmetries in epithelial architecture and later differences in the gross size and shape of the left versus right lobes, remain to be determined. Pitx2c is also expressed in the left side of the septum transversum mesenchyme surrounding the budding liver in mammalian embryos (Shiratori et al., 2006), but whether similar asymmetries in the cellular morphogenesis of the hepatic endoderm exist in other vertebrates, and how such differences might affect the later morphogenesis of the mammalian liver, have not been investigated.

In zebrafish, the liver bud does not exhibit (obvious) asymmetrical lobation but the organ does become positioned asymmetrically, albeit on the left (not right) side of the gut tube, as driven by asymmetrical leftward migration of early hepatoblasts from the hepatic endoderm. This process is dependent on epithelial-mesenchymal interactions mediated via ephrin B1/EphB3b signaling (Cayuso et al., 2016). Although hepatic expression of ephrin B1 has also been reported in other species (Costa et al., 2003; Fletcher et al., 1994), it is unknown if this pathway is involved in asymmetrical liver lobation or anatomical positioning in other vertebrates.

Both the pancreas and the spleen arise at the foregut-midgut boundary (Brendolan et al., 2007; Jennings et al., 2013) and eventually become situated on the left side of the body near the convex side of the stomach (Fig. 1). These tissues arise within the dorsal mesentery (DM), an embryonic structure derived from the splanchnic layer of the LPM that connects the primitive gut tube to the body wall (Fig. 4C). A thickened columnar epithelium known as the splanchnic mesoderm plate (SMP; Green, 1967; Hecksher-Sørensen et al., 2004) covers the DM from the posterior stomach to the anterior duodenum. Although the SMP is initially bilaterally symmetrical, the SMP on the right recedes while the left SMP, which expresses Pitx2, maintains its columnar morphology, undergoes rapid proliferation and bulges outward to the left of the gut tube. The leftward outgrowth of the SMP and underlying splenopancreatic mesenchyme forms a dorsal, left-sided protrusion posterior to the stomach that contains both the spleen anlage and the dorsal pancreas bud (Fig. 4C). In embryos with no or abnormal SMP development, the spleen is absent and the pancreas adopts abnormal orientations (Hecksher-Sørensen et al., 2004). Moreover, in embryos with a reversed LR axis (i.e. inv/inv mice; Watanabe et al., 2003), the SMP bulges towards the right side, confirming its asymmetry is determined by global LR patterning cues (Hecksher-Sørensen et al., 2004).

It should be noted that, although the asymmetrical influence of the SMP may compel the dorsal pancreas to grow to the left of the midline, the morphogenesis of the mature pancreas involves the fusion of both dorsal and ventral pancreas rudiments. Although the fusion process itself remains virtually unstudied, it is often attributed to the rotation of the gut tube (Gittes, 2009; Jennings et al., 2013; Pan and Brissova, 2014), with most textbooks illustrating the ventral pancreas bud(s) moving rightward around to the dorsal side of a rotating duodenum to meet the dorsal bud (Larsen, 2014). However, the means by which such a translocation might occur are unknown.

In many vertebrates, the left lung exhibits fewer lobes than its right counterpart (Matthew et al., 2009). For example, in humans, the left lung has two lobes whereas the right lung is trilobed (Fig. 1; Larsen, 2014). In individuals with aberrant LR asymmetry, the laterality of lung lobation may be reversed or even eliminated (e.g. both lungs develop a bilobed left lung or a trilobed right lung morphology; Aylsworth, 2001; Casey, 1998; Lin et al., 2014; Yim et al., 2018). Early in foregut development, the left and right lung buds emerge from the foregut just anterior to the stomach and undergo distinct branching events; the bifurcating patterns are reversed in embryos with reversed LR asymmetry, suggesting that left and right lung morphologies result from early initiation of disparate side-specific morphogenetic programs (Metzger et al., 2008). The left side program is likely specified by Pitx2c, which is expressed only in the left lung in mice, as loss of Pitx2c leads to both lungs developing right lung morphology (Cardoso and Lü, 2006; Hogan, 1999; Liu et al., 2001). However, the downstream cellular events and molecular targets of Pitx2c in the lung remain unknown. Interestingly, a recent investigation suggested that, in snakes, the left lung often develops as a shortened or even vestigial organ (van Soldt et al., 2015). This LR asymmetry is hypothesized to be achieved by the left lung bud undergoing slowed or arrested growth compared to the right (van Soldt et al., 2015). It remains to be seen whether such mechanisms contribute to the Pitx2-mediated asymmetries in lung morphology observed in other vertebrates.

Although our knowledge of asymmetric gut morphogenesis is still limited, some commonalities between organs may be noted. For example, in every organ, the cell populations involved originate from the splanchnic layer of the LPM and/or from endoderm immediately juxtaposed to splanchnic mesoderm-derived tissues (Fig. 4). Thus, the executors of asymmetric organ morphogenesis are directly derived from, or influenced by, the LPM, i.e. the tissue in which LR asymmetric gene expression patterns are broadly established earlier in development (although the exact, organ-specific fates of Nodal/Pitx2-expressing cell populations from the LPM have not been explicitly delineated for most species). Another common theme is that the emergence of laterality occurs early in organogenesis and often involves LR asymmetries in the architecture of mesoderm- and/or endoderm-derived epithelial tissues within or associated with each organ. Finally, these early tissue asymmetries are all directly or indirectly dependent on late-stage, organ-specific (left-sided) expression of Pitx2 within the developing gut tube and/or its associated mesenteries (Fig. 4).

Despite these similarities, there are many differences in the morphogenetic programs that drive the formation of asymmetric morphology in each discrete region of the gut tube (Fig. 4). For example, it is obvious that there is no universal tissue-level shape change that signifies ‘leftness’ or ‘rightness’. The left stomach wall thins and expands while the left side of the liver bud remains compact. Likewise, while tissue expansion characterizes the left side of the dorsal mesentery-derived SMP, tissue condensation contracts the left side of the midgut dorsal mesentery (see below). Furthermore, while the endoderm exhibits cellular-level asymmetries in the developing stomach and liver, no such lateralities have been identified in this tissue layer in the early midgut (Davis et al., 2008). The underlying cellular mechanisms at play in each tissue also vary widely, from LR differences in proliferation to asymmetries in cell shape, polarity, rearrangement and/or ECM remodeling (Davis et al., 2008, 2017; Hecksher-Sørensen et al., 2004; Kurpios et al., 2008; Sivakumar et al., 2018; Welsh et al., 2013; Womble et al., 2018; Fig. 4). Finally, at the molecular level, there appears to be little consensus regarding the effectors and networks that function in parallel with or downstream of Pitx2, although only a small number of genes are currently known to be associated with asymmetric organ morphogenesis (Table 1). Further elucidation of the relevant tissue-, cellular- and molecular-level asymmetries operating in both the left and right sides of multiple organs is necessary to determine whether this apparent divergence might in fact be underlain by common themes.

With new insights into the development of stomach and intestine laterality, the field is now poised to address key unanswered questions in the context of contemporized models of asymmetrical gut morphogenesis.

Although the early curvature of the stomach is initially generated by asymmetrical expansion of its contralateral walls, the curved organ does rotate to varying degrees in different directions as it grows larger and descends into the abdominal cavity (Kaigai et al., 2014), and other factors contribute to its final size, shape and anatomical orientation. For example, Wnt/PCP signaling contributes to the lengthening of the anterior stomach via oriented cell division (Matsuyama et al., 2009). However, the relationship between this process and the radial intercalation of endoderm cells (Davis et al., 2017) is unknown. Likewise, the mesoderm layer of the developing stomach also exhibits LR asymmetries in thickness and tissue architecture (Davis et al., 2017), but the cellular morphogenetic events that underlie these LR differences, and their influence on the nascent gastric musculature or vagus innervation remain to be elucidated. There is some evidence for interaction between the mesoderm and endoderm tissue layers (Davis et al., 2017) and, although Pitx2 targets are likely involved, the molecules that mediate this interplay have not been identified. Whether instructive morphogenetic changes also occur on the right side of the stomach, and whether there is coordination and crosstalk between the contralateral sides, are intriguing unresolved issues with important implications for the development of other asymmetric organs. Finally, it will be interesting to ascertain whether and/or how tissues outside the stomach tube itself (e.g. the splanchnic mesoderm-derived, asymmetrically remodeling mesogastrium) influence the process of curvature, and how the various cellular- and tissue-level asymmetries in the developing stomach influence, and are influenced by, the dynamic mechanical properties of the entire organ or neighboring regions of the gut tube. Indeed, it is intriguing that the stomach lies in the midst of a concentrated zone of asymmetric morphogenesis at the foregut-midgut boundary that also gives rise to several other asymmetric structures, including the lungs, pancreas, spleen and liver (Fig. 4). Thus, defining the spatiotemporal and mechanistic inter-relationships between the morphogenesis of the stomach walls, mesogastrium, lungs, SMP and hepatic diverticulum has the potential to yield tremendous insight into how multiple organogenesis events are coordinated to generate consistent anatomical complexity. In the context of experimental models of abnormal LR asymmetry (e.g. Hummel and Chapman, 1959; Ryan et al., 1998; Sempou and Khokha, 2019; Yokoyama et al., 1993), such studies may help explain the convoluted, seemingly random anatomical patterns and combinations of laterality-related organ defects seen in humans (Casey, 1998; Shiraishi and Ichikawa, 2012; Sutherland and Ware, 2009).

The formation of the initial S-shape of the midgut places the cranial limb of the prospective primary loop on the right and the caudal limb on the left, making it arguably the most important symmetry-breaking event in intestine morphogenesis. However, the mechanisms that underlie the development of this key morphological feature remain unclear. As discussed above, it is hypothesized that the formation of the S-shape may be a passive response to the constraints of a helical body axis and/or the descent of more cranial organs. However, the midgut also forms an initial S-shape in non-amniote vertebrates that do not have helical body axes, e.g. Xenopus (Muller et al., 2003), suggesting other factors must contribute to the initial break in midgut symmetry, at least in lower vertebrates. Moreover, active symmetry-breaking events linked to LR patterning pathways may influence the formation of this crucial asymmetry in amniote species (discussed below).

Despite the enigmatic expression of Pitx2 on the left side of the primitive midgut (Fig. 4D; Logan et al., 1998; Piedra et al., 1998), cellular- or tissue-level asymmetries have yet to be found in the early midgut tube itself (Kurpios et al., 2008). However, prior to the formation of the S-shape, striking differences in cell/tissue morphology, extracellular matrix (ECM) composition and gene expression exist between the left and right sides of the DM that suspends the midgut, i.e. posterior and/or caudal to the SMP (see Fig. 4D and Table 1; Davis et al., 2008; Kurpios et al., 2008; Welsh et al., 2013). Tissue expansion on the right side of the DM, followed by Pitx2-dependent compression on the left side (Sivakumar et al., 2018), creates an asymmetrical architecture that is essential for properly lateralized gut vasculature (Mahadevan et al., 2014; Sivakumar et al., 2018). In addition, because the asymmetrical morphogenesis of this region distorts the shape of the mesenteric stalk itself, tilting the attached gut tube to the left, it has been hypothesized that asymmetrical remodeling of the DM is also the first step in establishing the intestine laterality that is observed later (Davis et al., 2008). However, it is not obvious exactly how the early DM asymmetry might affect the later morphogenesis of the intestine, as the tilting occurs early, during closure of the open midgut into a tube and prior to primary loop formation (Davis et al., 2008; Kurpios et al., 2008; Sivakumar et al., 2018; Southwell, 2006; Welsh et al., 2013). It has been hypothesized that the early DM tilt could somehow influence the initial S-shape of the midgut (Davis et al., 2008), although the cranial region of the ‘S’ curves rightward, in the opposite direction to the mesentery-mediated leftward tilt. It is also possible that the leftward tilt of the DM somehow influences the laterality of a mechanical buckling event in the early midgut, resulting in the S-shape (Davis et al., 2008), although this remains to be demonstrated. Experiments that perturb DM tilting in the chick embryo do result in abnormal gut laterality, as revealed by reversals in the curvature of the stomach (Davis et al., 2008). Unfortunately, however, the experimentally manipulated embryos do not survive to the point of midgut looping (Davis et al., 2008), and the intestine of the chicken does not develop the same ‘rotated’ LR asymmetries as the mammalian tract (Southwell, 2006), so the relationship between DM tilting and specific intestine lateralities may only be inferred in the chick model. However, in mice, knockout of an ECM-modifying gene normally restricted to the right side of the cranial DM, Tsg6, was recently found to cause striking intestinal malrotation and volvulus phenotypes (Sivakumar et al., 2018). Investigating the formation of the initial S-shape, DJF or later secondary loops in this model has great potential to illuminate our understanding of the development of mammalian intestinal laterality and the underlying pathogenesis of human malrotation and volvulus.

The hierarchical looping model re-interprets the complex events of intestine development, providing a plausible alternative to rotation-based theories of asymmetric morphogenesis, but fundamental unanswered questions still exist. For example, what genetic patterning events or morphogenetic gradients underlie the stereotypical placement of discrete growth zones at defined positions along the cranio-caudal, dorsal-ventral or left-right axis of the gut tube? Recently, LR differences in tissue thickness and cellular orientation suggestive of asymmetrical cell rearrangement were reported to exist in the inner versus outer curvatures of the DJF, i.e. within the first secondary loop (Onouchi et al., 2013, 2015, 2016). Although the inner and outer curvatures of the DJF are derived from the original left and right sides of the primitive duodenum, respectively, the relationship of the observed cellular asymmetries to LR axis formation, DM tilting or the expression of LR genes such as Pitx2 is unknown. It will be interesting to determine whether similar cellular asymmetries accompany the formation of other secondary loops. Furthermore, determining whether and/or how the formation of secondary loops is altered in models with LR asymmetry defects (such as situs inversus or heterotaxy) versus isolated intestinal malrotation (e.g. the Tsg6 mouse) could reveal whether and/or how LR patterning affects specific features of intestine laterality, and/or whether secondary looping is executed independently of LR patterning per se. Another fascinating area to explore is the conservation of secondary loops between species. The existence of species-specific patterns of secondary loops could underlie the wide variety of vertebrate intestinal morphologies that are difficult to explain by the rotation of a single primary loop, e.g. the large hairpin bends observed in the horse colon or the multi-layered spirals seen in ruminants (Singh, 2017). Finally, the biomechanical factors that control the hierarchical formation of the secondary loops or provide the force for their sequential retraction and placement within the body cavity, also remain to be identified.

Regardless of whether rotation, differential growth and/or hierarchical looping generates the various twists and turns of vertebrate digestive anatomy, physical and mechanical forces must underlie the massive topological changes that occur. It is important to note that the entire process of intestine morphogenesis coincides with, and indeed is driven by, proper gut tube elongation. Therefore, the biomechanical mechanisms impelling and supporting the lengthening of the gut tube are likely to be integral to the development of proper laterality. Underscoring this point is a recent review of cases of congenital short bowel (from 1969 to present), which revealed that malrotation occurs in 98.4% of individuals with shortened gut tubes (Negri et al., 2020). Moreover, intestinal malrotation has been observed in multiple contexts in which short guts were induced by experimental perturbations of signaling pathways involved in gut elongation (Pitera et al., 2001; Lipscomb et al., 2006; Yamada et al., 2010).

The link between gut elongation and laterality is also supported by computational modelling indicating that gut lengthening is mechanically coupled with looping. As mentioned above, disproportionate lengthening of the gut tube versus the attached DM is thought to create a growth strain that drives compressive buckling of the tube, resulting in the formation of predictable ‘sine wave’ undulations (Fig. 3B; Savin et al., 2011). Soffers et al. (2015) named these buckling events 'tertiary' loops; they exist within the secondary loops discussed above, but are thought neither to arise in precise locations along the gut tube (Soffers et al., 2015) nor to represent LR asymmetries in and of themselves. Nonetheless, their existence could influence the formation, re-entry and/or retraction of the secondary loops, thus altering final laterality. Although the role of the attached DM in secondary loop formation is unknown, the shorter mesentery observed between secondary loops (Mall, 1898; Soffers et al., 2015) may serve to define and/or mechanically isolate these key segments.

As the number and curvature of gut loops is dependent not only on the rate of gut lengthening, but also on the radius of the gut and the length/thickness of the mesentery (Nerurkar et al., 2017), any perturbation that alters the geometric and/or physical properties of different segments of the gut tube and/or the attached DM might influence the mechanics of looping and ‘rotation’. For example, defects in the radial patterning and diameter of the gut tube itself may affect its mechanical properties and ‘loopability’; indeed, intestinal narrowing (stenosis and/or atresia) has been associated with malrotation in humans (Adams and Stanton, 2014; Chinya et al., 2019; Ishii et al., 2020; Morikawa et al., 2009). Interestingly, these narrowing defects are often attributed to ‘vascular accidents’ that inhibit blood flow during gut development (Adams and Stanton, 2014; Ishii et al., 2020; Martin and Shaw-Smith, 2010). Thus, the DM, which is crucial for development of the gut vasculature (Sivakumar et al., 2018), may play a key role in asymmetric intestine morphogenesis simply by supporting proper gut tube growth and/or elongation. Further questions arise: e.g. how does perturbing the early leftward tilting of the DM affect its dimensions and mechanical properties, and how does abnormal DM tilting affect the vascularization, length and/or mechanical properties of associated loops of intestine? Pursuing such integrative lines of inquiry could begin to define the relative contributions of early versus late, cell biological versus physiological, and molecular versus mechanical mechanisms in the complex multi-phasic process that sculpts the final configuration of the vertebrate gut tube.

The overall breadth and variety of vertebrate organ lateralities suggests that variation in the asymmetric morphogenesis of gut-related tissues may have been a rich source of novel functional morphology during evolution. Although we are just beginning to understand what makes individual organs develop as LR asymmetric entities, some trends are emerging. First, rotational translocations of large segments of the gut tube are likely not the primary events that shape the anatomical lateralities of the stomach or intestine, despite the dogmatic illustrations found throughout the literature and textbooks. Second, many of the events that shape anatomical asymmetries involve tissues directly derived from, or immediately influenced by, the splanchnic layer of the LPM. Third, within these tissues, a wide variety of asymmetric cellular morphogenetic processes, dependent on localized, tissue-specific, left-sided expression of Pitx2, are deployed early in organogenesis. Unraveling how all these levels and layers of LR asymmetric morphogenesis have been integrated into the development of the vertebrate gut tube could provide profound insight into not only the etiology of laterality-related birth defects, but also the fundamental nature of organogenesis itself.

We thank J. Soffers and members of the Nascone-Yoder lab for helpful comments and feedback.