The splanchnic mesodermal plate directs spleen and pancreatic laterality,and is regulated by Bapx1/Nkx3.2

From the initial bilaterally symmetrical mammalian embryo, left-right (LR)asymmetries appear throughout affecting the structure and location of internal organs (Capdevila et al.,2000; Hamada et al.,2002; Bisgrove et al.,2003). Such visceral asymmetries exist in all vertebrates,suggesting that many of the mechanisms that generate sidedness are conserved during evolution (Boorman and Shimeld,2002). The mechanism that establishes LR positional information has been intensively investigated. LR positional information initiates near the embryonic organiser by a process of nodal flow and is subsequently relayed to the lateral plate mesoderm (LPM)(Nonaka et al., 1998; Capdevila and Belmonte, 2000; Nonaka et al., 2002). Several genes have been identified which are necessary for establishing LR asymmetries many of which show remarkable side-specific expression patterns in the LPM. However, one question that has received little attention is how this information is ultimately conveyed to the organ primordia, resulting in the implementation of programs of asymmetric organ morphogenesis(Capdevila and Belmonte,2000).

The normal disposition of organs is called situs solitus. Disruptions of organ situs are common and heterotaxia conditions cover a broad range of gastrointestinal and cardiac defects(Burn, 1991; Aylsworth, 2001). In situs inversus complete mirror-image reversal of situs occurs. More commonly individuals have partial situs malformations owing to randomisation of patterning information and only a few organs are affected (situs ambiguus or heterotaxy). The mesenchymally derived spleen is normally situated on the left side of the abdominal cavity and as such is a readily identifiable landmark for detecting extensive laterality defects(Aylsworth, 2001). Conditions such as asplenia, and polysplenia syndromes are often associated with situs ambiguous, indicating an underlying, fundamental defect of asymmetric organogenesis (Bowers et al.,1996). Loss of splenic tissue (asplenia or rudimentary spleen)relates to bilateral right sidedness called right isomerism, whereas the contrasting phenotype of additional splenic tissue (polysplenia, either as additional splenules or multilobulated single spleen) relates to bilateral left sidedness (left isomerism). Hence, spleen abnormalities are frequently associated with situs malformations of other organ systems dependent on LR signalling (Ivermark's syndrome) (Ivemark,1955).

In this report, we have examined two asplenic mouse mutations, dominant hemimelia (Dh) and the Bapx1-targeted disruption, to investigate the relationship of spleen development and LR asymmetry. Dh is an established, spontaneously derived mouse mutation that disrupts visceral and limb development(Green, 1967). The Bapx1 gene (also referred to as Nkx3.2) is a member of the NK family of homeobox-containing genes(Tribioli et al., 1997) first described in Drosophila and is most closely related to the Drosophila bagpipe (bap) gene(Azpiazu and Frasch, 1993). Targeted mutations of Bapx1 results in loss of the spleen and vertebral defects (Lettice et al.,1999; Tribioli and Lufkin,1999; Akazawa et al.,2000). Examination of visceral development in Dh and Bapx1 mutants led to the identification of a mechanism whereby the primordial spleno-pancreatic mesoderm that surrounds the gut endoderm controls localised asymmetric growth.

Postimplantation mouse embryos were collected at the desired stages,considering the day of the vaginal plug as E0.5, and were genotyped as described (Lettice et al.,1999). Whole-mount immunohistochemistry was carried out as described (Sharpe et al.,2002), while combined immunohistochemistry and in situ hybridisation was performed as detailed by Hecksher-Sørensen and Sharpe(Hecksher-Sørensen and Sharpe,2001). Following staining, tissue was embedded, sectioned and counterstained as described by Hecksher-Sørensen and Sharpe(Hecksher-Sørensen and Sharpe,2001). Alternatively, samples were embedded in 1% LMP agarose and analysed by Optical Projection Tomography as described(Sharpe et al., 2002). The PDX1 and ISL1 antibodies have been described previously(Ahlgren et al., 1997).

DIG-in situ hybridisation was performed essentially as described by Wilkinson (Wilkinson, 1992). Prior to staining, the embryos were washed twice in 0.1 M Tris (pH 8.2) for 30 minutes and the signal was visualized in 2 ml 0.1 M Tris (pH 8.2) containing one Fast Red tablet (Roche). The probes for Bapx1, Nkx2.5, Hox11 (Tlx1 – Mouse Genome Informatics), Wt1,capsulin (Tcf2l – Mouse Genome Informatics), Barx1, Pitx2,Fgf9, Fgf10 and Fgfr3 have been described previously(Mundlos et al., 1993; Peters et al., 1993; Roberts et al., 1994; Tissier-Seta et al., 1995; Semina et al., 1996; Bellusci et al., 1997; Lu et al., 1998; Colvin et al., 1999; Lettice et al., 2001).

E10.5 embryos were labelled with BrdU by intraperitoneal injection of 200μl BrdU (10 mg/ml) into pregnant females. After a 30 minute labelling period, the embryos were dissected out and fixed in 4% PFA. Embryos were then embedded in paraffin wax and sectioned. After rehydration, sections were trypsinised for 10 minutes at 37°C, washed in PBS and then incubated for 10 minutes in 1 N HCl. Incorporation of BrdU was detected using anti-BrdU monoclonal antibodies and Alexa Fluor® 488 goat anti-mouse secondary antibody (1:200 dilution; Mol. Probes). Finally, sections were counterstained with Propidium Iodide (1:5000).

The spleen is a wholly left-sided organ in the wild-type mouse, closely associated with the pancreas, and located on the left lateral side of the stomach. Here, we examine the location of the pre-splenic rudiment relative to the stomach and pancreatic primordia at the initiation of organogenesis. Histological analysis of late gestation mammalian embryos indicates that the splenic rudiment is recognisable as a single condensation of mesenchyme along the left side of the mesogastrium dorsal to the stomach(Thiel and Downey, 1921). Analysis of genes expressed in the mouse spleen, such as Hox11(Roberts et al., 1994; Dear et al., 1995), enable the tracing of spleen development to earlier embryonic stages. By E11.5, Hox11 expression is established as left-handed and in the dorsal mesogastrium lateral to the stomach(Kanzler and Dear, 2001).

We have identified the initial lateral position of the pre-spleen rudiment at a stage before the spleen is associated with the stomach. At E10.5,cross-sections taken through the embryo in the region of the developing forelimb reveal the spatial relationship between the stomach(Fig. 1A,B) and dorsal pancreas(Fig. 1A,C) during the early stages of leftward growth. At this stage, Hox11 expression is detected in the dorsal mesenchyme positioned on the left-lateral side, but lying posterior to the stomach primordium(Fig. 1D). The pancreas-specific anti-PDX1 antibody shows that this mesenchyme also supports the dorsal pancreatic bud (Fig. 1D) lying ventral to the splenic rudiment.

Expression of other genes implicated in spleen development confirmed that this region of the embryo participates in spleen formation. In Xenopus,Nkx2.5 is expressed during early stages of spleen development(Patterson et al., 2000). In mouse embryos, Nkx2.5 expression was detected at E10.5 in two distinct domains around the dorsal pancreatic bud; a ventral domain and a dorsal domain that overlaps Hox11(Fig. 1E). Targeted mutations in capsulin/Pod1 (Lu et al.,2000) and Wt1 (Herzer et al., 1999) show that both genes are required in spleen formation. The expression patterns of Wt1 and capsulinoverlap with those of Hox11 and Nkx2.5 in the dorsal region of the mesenchyme (Fig. 1F,G),indicating that this region will ultimately give rise to the spleen. By E11.5,the splenic rudiment, which continues to express Wt1(Herzer et al., 1999),capsulin (Lu et al., 2000), Nkx2.5 and Hox11 (Dear et al., 1995; Kanzler and Dear, 2001; Roberts et al.,1994) (data not shown), is located in the dorsal mesogastrium,which is now lateral to the stomach. While subsequent splenic development occurs in close association with the morphogenesis of the stomach, we show that the pre-spleen mesenchyme and endodermal-derived dorsal pancreatic bud are closely associated posterior to the stomach at an early developmental stage.

An outer layer of epithelium surrounding the splenopancreatic mesenchyme was noted in the E10.5 embryos (broken lines in Fig. 1D,E). Earlier stages were examined to establish when this tissue originates. The splanchnic mesoderm is the region of the lateral plate mesoderm (LPM) that surrounds the gut. In the region of the dorsal pancreatic bud, we found that fluorescently tagged phalloidin detects the accumulation of f-actin at the boundary between the endoderm and splanchnic mesoderm and at the outer margin of the splanchnic mesoderm, thus outlining a thickened epithelial structure(Fig. 2A,B). The splanchnic mesoderm surrounds the foregut and appears as a bilateral, highly organised layer of elongated cells ∼50-60 μm thick that is easily distinguishable from the unorganised mesenchymal cells that compose a large proportion of the embryo. The position of the dorsal pancreatic bud was detected with the PDX1 antibody (Fig. 2A,C,D,O,P). The pancreatic bud is situated at the embryonic midline(Fig. 2A) and extends dorsally flanked tightly on both sides by the splanchnic mesoderm.

We have analysed development of this region in the asplenic dominant hemimelia (Dh) mouse mutant. The Dh/Dh mutant was of particular interest based on the original description by Green(Green, 1967), which suggested that the structural integrity of the splanchnic mesoderm was disrupted. At E9.5 in Dh/Dh mutant embryos, the organised splanchnic mesoderm is not apparent but is replaced by unorganised mesenchyme(Fig. 2D,E). The Dh/Dhmutant embryos highlight the distinct structure of the wild-type mesoderm,which, similar to the term used by Green(1967), we propose to name the splanchnic mesodermal plate (SMP).

By E10 the spleno-pancreatic region in wild-type embryos acquires a characteristic triangular shape in cross-section(Fig. 2C). The SMP is prominent and located on the left side (white arrowheads in Fig. 2C). The mesenchyme on the right side of the gut tube remains positioned near the embryonic midline and the thick epithelial plate-like structure is lost. TUNEL staining for apoptotic nuclei reveals negligible cell death in this region between the stages E9.5 and E10.5 (data not shown). This argues against apoptosis as the driving force behind the observed loss of the right-sided SMP. Cells of the right SMP persist after E9.5, most probably contributing to the underlying mesenchyme and the thin layer of cuboidal cells (indistinguishable from the mesothelial layer that lines the mesenchyme in the remainder of the coelomic cavity and shown by yellow arrowheads in Fig. 2C) that is found in place of the right SMP from E10.0 onwards. Analysis of nuclear morphology in the spleno-pancreatic region between E9.5 and E10.5 reveals that the elongated nuclear structure that characterises the left SMP is progressively lost from the right-hand side during this period(Fig. 2F-N). By E10.0, nuclei positioned ventrally along the right margin resemble those present in the underlying mesenchyme or in adjacent mesothelia(Fig. 2I,K). At this stage,nuclei positioned dorsally along the right margin retain their elongated morphology (Fig. 2I,J). By E10.5, nuclei along the length of the right-hand margin are overtly indistinguishable from those in the underlying mesenchyme or in adjacent mesothelia (Fig. 2L-N),suggesting that the observed loss of the right-hand SMP is actually the result of a spatiotemporal change in cell morphology.

Dorsal pancreatic bud growth accompanies the asymmetric expansion of the mesenchyme such that it becomes displaced to the left of the embryonic midline. The SMP, which by bright-field microscopy appears as a translucent cell layer (indicated by broken lines in Fig. 6A), entirely encompasses the left, lateral spleno-pancreatic mesenchyme extending from the posterior half of the stomach to the pancreatic buds. During the subsequent day of development, the shape and position of the region changes dramatically and by E11.5 the thick epithelial structure is no longer detected (data not shown).

Thus, the SMP, a derivative of lateral plate mesoderm, is bilateral at initial stages. This columnar epithelium lines the gut endoderm with few intervening mesenchymal cells. Within ∼12 hours the SMP is preferentially maintained on the left-hand side accompanied by lateral growth and appearance of underlying mesenchymal cells. The gut tube remains at the embryonic midline, while the dorsal pancreatic bud moves leftward accompanying the outgrowth of the SMP. Expression of Hox11, Nkx2.5 and capsulin/Pod1 in E10.5 embryos is located in the mesenchyme directly underlying the SMP, which itself does not express these genes at a high level(Fig. 1D,E,G), suggesting that the SMP is distinct from, but closely associated with, the developing spleen.

To determine whether SMP asymmetry is under control of the LR signalling pathway, we analysed expression of Pitx2, a gene known to be expressed in the left LPM during early embryogenesis(Logan et al., 1998; Piedra et al., 1998; Shiratori et al., 2001). Pitx2 has a role in lung isomerism, heart development and rotation of the duodenum (Gage et al.,1999; Kitamura et al.,1999; Lin et al.,1999; Lu et al.,1999; Liu et al.,2001). Expression of the Pitx2 gene persists later than many other genes in the LR signalling cascade, with expression still detectable at E9.5. At this stage, before asymmetric organogenesis is apparent, Pitx2 is expressed predominantly in the left-sided SMP in the region of the dorsal pancreatic bud(Fig. 3A). Thus, Pitx2expression is an indicator of left/right differences and shows that the left SMP expresses LR-specific genes. The homeobox-containing Barx1(Tissier-Seta et al., 1995),which has not previously been associated with LR asymmetry, also shows left-specific mesenchymal expression, substantiating the phenotypic differences between left and right mesenchyme at these early stages(Fig. 3B).

A previous report predicted that Bapx1 has an early role in LR asymmetry as the gene is expressed in a side-specific manner(Schneider et al., 1999). Paradoxically, Bapx1 mesodermal expression is highest in the left LPM in chick and the right LPM in mouse. The asymmetric expression in mouse is initially detected at E8.0 and continues up to ∼E9.5. We suggest that by E9.5, however, expression of Bapx1 in the region of the pancreas is in the process of converting to left-sided expression. In the SMP, the expression is bilateral but the level is detectably higher on the left side(Fig. 3D). By E10.5, Pitx2 is no longer detected (data not shown) while both Barx1 (Fig. 3C) and Bapx1 (Fig. 3E) are expressed at the extreme lateral domain of the SMP and in the underlying mesenchyme. Thus, there is a conversion in the sidedness of the Bapx1expression in the mouse and by E9.5 is in accordance with the expression pattern of the chick (Schneider et al.,1999).

To further investigate the link between SMP and the genetic LR asymmetry program, the spleno-pancreatic region in mice carrying the inversion of body turning (inv) mutation was analysed(Fig. 3F,G). Homozygous inv mutant embryos all show situs inversus(Yokoyama et al., 1993; Morgan et al., 1998; Watanabe et al., 2003) with both the spleen and pancreas developing on the right hand side. At E10.5 in the inv/inv embryos, the SMP is positioned on the opposite,right-hand side (Fig. 3G). The underlying spleno-pancreatic mesenchyme has grown to the right and the region is the mirror image of that in the wild type. These findings show that the maintenance of the SMP, and subsequent growth of the spleno-pancreatic mesenchyme, is a downstream consequence of the LR signalling pathway.

The Dh mutant embryo enables the examination of left lateral asymmetry in the absence of the SMP. In wild-type embryos at E10 the SMP and associated mesenchyme grows laterally (Fig. 4A). By contrast, in the Dh/Dh embryos the SMP remains undetectable and specific leftward growth of the mesenchyme is impaired(Fig. 4C). The mesenchyme is symmetrical and surrounds the dorsal pancreatic bud that remains situated along the embryonic midline. This arrangement persists at later stages and at E10.5 the pancreas is still observed at the midline (data not shown).

To investigate the possibility that the dorsal pancreas itself drives the leftward growth of the spleno-pancreatic mesenchyme, we analysed development of this region in Fgf10 mutant mice. Loss of Fgf10expression has been reported previously to result in a severe disruption of pancreatic growth (Bhushan et al.,2001). Consistent with these findings, no dorsal pancreatic tissue was detected at E10.5 in Fgf10^–/– embryos following staining with the PDX1 antibody(Fig. 4G). In addition, no other endodermal-derived structures were found in the spleno-pancreatic mesenchyme and the duodenum remained at the embryonic midline. Despite this,the leftward growth of the spleno-pancreatic mesenchyme surrounded by the SMP occurred to approximately the same extent as the wild type(Fig. 4F,G).

Thus the Dh and Fgf10^–/– mouse mutations provide two contrasting developmental conditions in which to examine leftward asymmetry. In the absence of the SMP, no leftward pancreatic growth was observed, whereas in the absence of the pancreatic bud, the SMP and underlying mesenchyme expand laterally. This suggests that asymmetry of the spleno-pancreatic mesenchyme and SMP is autonomous. The pancreatic endoderm in wild-type embryos is embedded in the left-sided mesenchyme and develops in close association with the SMP, but has no role, either physically or inductively, in the asymmetric growth of these structures.

As endodermal development is not required for asymmetric growth in the spleno-pancreatic region, we hypothesised that regional variations in cellular proliferation might be a key factor in the morphogenetic changes that characterise the development of the SMP and the underlying tissue. At E9.5,prior to overt LR asymmetry in this region, BrdU incorporation can be detected at high levels in both the left and right sides of the SMP(Fig. 5A). This pattern is reiterated at E9.75, during the initial stages of asymmetric morphogenesis. However, by this stage, regional differences in the BrdU-uptake pattern exist. Comparable levels of BrdU incorporation are observed in the left and right-sided SMP anterior to the dorsal pancreatic bud, with positive nuclei distributed evenly along the dorsoventral axis(Fig. 5B). However, in the region of the dorsal pancreatic bud, BrdU is detected in clusters of nuclei located around the apex of the outgrowing SMP(Fig. 5C).

By E10.5, distinct populations of cells can be readily discerned, enabling comparative analysis of BrdU incorporation throughout the stomach and pancreatic region. Wild-type embryos at ∼ E10.5 were labelled with BrdU for 30 minutes and transverse sections through different regions of the gut(represented in Fig. 6B) were stained for BrdU incorporation (Fig. 6C-F). Mesothelial layers contiguous with the SMP in two embryonic regions were analysed; sections were taken anterior to the SMP in the region of the developing stomach (D in Fig. 6B) and in the region of the dorsal pancreatic bud that includes the SMP (E in Fig. 6B). Embryonic incorporation of BrdU was high throughout the embryo(Fig. 6C); ∼50-55% of the cells labelled in the gut mesenchyme while in the gut endoderm about 35% of the cells labelled. Analysis of sections taken through the stomach show an associated thin mesothelial layer (white arrowheads in Fig. 6D) in which about 60%(±2%) of the cells incorporated BrdU(Fig. 6G). By comparison, the thickened SMP associated with the spleno-pancreatic region shows ∼77%(±3%) incorporation (Fig. 6G). The mesothelium that lines the lateral coelomic walls (blue arrowheads in Fig. 6D-F)incorporates BrdU at a rate of about 50%(Fig. 6G) or less. Posterior to the pancreas (F in Fig. 6B) the gut mesoderm is no longer flanked by the SMP and the BrdU uptake in the mesoderm and associated mesothelium are appreciably lower than the SMP(<50% incorporation) (Fig. 6F). Thus, the SMP shows a higher rate of BrdU incorporation when compared with the mesothelia in other regions of the embryo and the mesenchyme underlying the SMP. We suggest that the SMP has a role in the left lateral disposition of the mesenchyme that is crucial for both the formation of the spleen and the asymmetric morphogenesis of the pancreas. The asymmetric growth is driven by the high rate of cellular proliferation in the SMP.

Similar to the Dh mutant, the Bapx1^–/– embryos are asplenic. Mice homozygous for the Bapx1 mutation were examined at the early stages of spleno-pancreatic outgrowth, E10 and E10.5(Fig. 6B,E, respectively). In contrast to Dh embryos, the Bapx1 mutants retain the left SMP throughout these stages. The spleno-pancreatic mesenchyme of the Bapx1 mutant expands laterally but lags behind that of wild type and the characteristic triangular shape of the wild-type SMP is compromised. Most significantly, the dorsal pancreatic bud of the Bapx1 mutant does not grow laterally but remains positioned along the embryonic midline.

To investigate the effect that the early malposition of the pancreas has on later development, we examined both the Dh and Bapx1 mutants at a later stage in organogenesis. Wholemount, three-dimensional analysis of organ position by optical projection tomography (OPT)(Sharpe et al., 2002) revealed a significant change in the orientation(Fig. 7A-C). At E13.5 in both Dh and Bapx1 mutants, the pancreas is growing along a different embryonic axis. In wild-type embryos, the dorsal pancreas grows laterally in a plane that is nearly perpendicular to the stomach(Fig. 7A). In both mutants, the dorsal pancreas grows along embryonic axes that are close to parallel with the stomach. In the Bapx1 embryos, the pancreas grows along the lateral wall of the stomach (Fig. 7C),whereas in the Dh embryo the pancreas is oriented ventrally to the stomach (Fig. 7B).

A number of genes predicted to be relevant to embryonic growth were assayed for expression in the spleno-pancreatic region. Using RT-PCR, 60 genes,including eight Fgfs were examined. Fgf 9, Fgf10, Fgf11 and Fgf13 were found to be expressed in the spleno-pancreatic region at E10.5. Expression of these genes was analysed by in situ hybridisation in transverse sections. Whereas Fgf11 and Fgf13 were expressed at detectable levels throughout the mesenchyme (data not shown), both Fgf9 and Fgf10 were restricted to domains on the left-side encompassing the SMP. Fgf9 is expressed most highly in the dorsal rim, including the lateral tip of the SMP and extends into the underlying mesenchyme (Fig. 8B). Fgf10 was previously reported to be expressed in the pancreatic mesenchyme (Bhushan et al.,2001). Here, we show that, more specifically, Fgf10 is expressed in the ventral region of the SMP and in the associated underlying mesenchyme (Fig. 8A). In addition, we detected expression of Fgfr3 in the extreme tip of the SMP (Fig. 8C). FGFR3 is a high affinity receptor for FGF9 and its expression in this region therefore suggests a model of Fgf9 induced outgrowth of the SMP.

In the Dh mutant, in the absence of the SMP, expression of Fgf9 and Fgf10 were both substantially reduced (data not shown). In the Bapx1 mutant, Fgf10 expression in the SMP was dramatically downregulated (Fig. 8D) although there was persistent expression in the mesenchyme surrounding the pancreatic bud. Fgf9 expression was reduced but still detectable (data not shown). The SMP therefore appears to be a source of the secreted factors Fgf9 and Fgf10. In the absence of the SMP,or following perturbation in its development, expression of Fgfs is compromised. Our findings suggest that Bapx1 is upstream of, and required for, Fgf10 expression in the SMP.

The spleen originates in the mesenchyme that is located posterior to the stomach and adjacent to the dorsal pancreas. The leftward expansion of the mesenchyme supplies the tissue in which the spleen forms and in which the leftward growing dorsal pancreas resides. Thus, initiation of splenogenesis and leftward pancreatic growth are closely linked. The close proximity in early development is consistent with studies by Patterson et al.(Patterson et al., 2000) that showed a close association between spleen and pancreatic laterality in experiments designed to randomise the embryonic LR axis. These indicated that the location of the spleen strongly correlated with the position of the pancreas and not other organs, such as the heart. This idea is further supported by the observation that some individuals with heterotaxy that includes spleen malformations have no detectable heart defects(Debich et al., 1990), and by analyses of the iv mutant mice, where no consistent relationship of heart and spleen laterality defects was shown(Seo et al., 1992).

Therefore, while determination of LR asymmetry is a general developmental mechanism, it appears that LR information is interpreted differently by each organ primordia. We suggest that Bapx1 and Dh mutants are good models for heterotaxy syndromes that include asplenia (double right isomerism). Neither mouse mutant shows cardiac, lung or liver malformations(Green, 1967; Herbrand et al., 2002) (data not shown) and, therefore, they provide insights into laterality disorders in which only a restricted number of organ systems are affected. We propose that the developmental mechanism that drives asymmetric organ morphogenesis in spleen and pancreas differs from that responsible for lobation of the lung and morphogenesis of the heart tube and that it is dependent on a mesodermal-derived structure, the SMP.

The SMP is central to our model for left lateral specific morphogenesis of the spleen and the pancreas (Fig. 9). The process of LR asymmetry is divided into four steps(Hamada et al., 2002). The first three steps are responsible for transferring an initial breaking of symmetry from near the node (step 1), to the LPM (step 2) with the subsequent asymmetric expression of TGFβ-related molecules (step 3). Less clearly understood is the fourth step, which is the relay of information from the LMP to the organ primordia. The SMP appears to be an important element in this final step and in the spleno-pancreatic region may be the primary target for the LR positional information.

At the earliest developmental stages examined, E9.5, the SMP is bilateral and flanks the midline-positioned dorsal pancreatic bud. There is no intervening space and few mesenchymal cells are found in between the SMP and the pancreatic endoderm. The left side of the SMP is specified at this early stage, as shown by the side-specific expression of both Pitx2 and Barx1. The LR positional information is subsequently interpreted as maintenance of the left lateral SMP and loss of the right. The invmutant mouse confirms that the unilateral maintenance of the SMP is under the genetic control of the LR asymmetry cascade.

The Dh mutation underscores the distinctive structural nature of the SMP. The mutation operates early in embryogenesis to disrupt the bilateral organised, columnar structure. We suggest that the highly organised structure of the SMP plays a significant role in morphogenesis. In several cases,disruption of the structural integrity of embryonic epithelia has resulted in profound perturbations in subsequent morphogenesis. A similar highly organised epithelia, which is derived from the lateral plate mesoderm, is key to the process of gut looping in zebrafish(Horne-Badovinac et al.,2003). Genetic disruption of the cellular organisation results in the lack of morphogenesis. Thus, the structural context of the SMP may be crucial to lateral outgrowth and we speculate that the structure may provide a rigid tissue layer for guidance. Accordingly, the SMP is characterised by the accumulation of f-actin at the apical surfaces as detected by phalloidin staining. Localisation of actin filaments to the apical surface of columnar epithelial cells has been described in a number of organisms (reviewed by Jacinto and Baum, 2003) and networks of actin filaments have recently been shown to play a crucial role during the morphogenesis of the pharyngeal pouches by directing expansion along specific axes (Quinlan et al.,2004).

A second characteristic of the SMP important for our model of lateral outgrowth is the high rate of cellular proliferation in this tissue layer. Outgrowth of the spleno-pancreatic mesenchyme and SMP is autonomous and entails no discernible input from the pancreatic endoderm. Proliferation within this cellular layer appears to be the motive force for driving lateral growth. We suggest that within a highly organised cellular framework proliferation within the SMP drives rapid, disproportionate expansion of this tissue layer behind which the mesenchyme, perhaps passively, populates.

Since the description of asymmetric Bapx1 expression during early mouse and chick embryogenesis, it has been postulated that Bapx1 may have a role to play in the establishment of LR asymmetry(Rodriguez Esteban et al.,1999; Schneider et al.,1999). The exact nature of involvement of Bapx1 in this process has, however, remained undetermined. Based on the reported expression patterns, it is tempting to speculate that Bapx1 could be playing a role in the early establishment of positional information. Paradoxically, the laterality of the expression is not conserved in evolution(Schneider et al., 1999). Our observations suggest that Bapx1 in mouse loses side-specific expression at a stage (E9.5) before any phenotype is observed in the Bapx1^–/– spleno-pancreatic region. Expression of the Bapx1 gene persists after that of Pitx2, a late stage gene in the LR cascade. The defects observed in the Bapx1^–/– embryos reported here relate specifically to the process of morphogenesis and are in agreement with a role for Bapx1 in the translation of positional information into asymmetric morphogenesis. Thus, we suggest that Bapx1 has a key role in the process of linking organ morphogenesis to events that define LR positional information.

Two phenotypic consequences result from the lack of Bapx1expression. The first is the lateral growth of the spleno-pancreatic mesenchyme is reduced. The second is the downregulation of Fgf10expression. Bapx1 is expressed in the ventral region of the SMP overlapping the Fgf10 domain, consistent with Bapx1 as a regulator of Fgf10 expression. FGF10 itself is responsible for early pancreatic growth (Bhushan et al.,2001), and in its absence the pancreas does not progress beyond the early bud stage. Fgf10 is expressed in the SMP and neighbouring mesenchyme at a source distal to the pancreatic bud. FGF10 is a known chemotactic factor in lung development(Weaver et al., 2000). We suggest that, in addition to its role in maintaining proliferation in the pancreatic endoderm, a role for FGF10 is to promote leftward pancreatic growth toward the source of high FGF10 production, thus resulting in the initial pancreatic asymmetry.

Recent analysis in the developing zebrafish has addressed the question of gut asymmetry (Horne-Badovinac et al.,2003). The digestive organs originate from a solid rod of endodermal cells and the first leftward bend arises from a morphogenetic process known as gut looping. The tissue layer that appears responsible for the initial mechanism in gut looping is the LPM. The relationship of the LPM in zebrafish and the SMP in mouse is not clear. The SMP is derived from the mesoderm of the lateral plate and both the LPM described in fish and SMP described here, are similar epithelial layers composed of columnar cells. However the mechanism of LR organogenesis in which these embryonic tissues participate is different. In the zebrafish, the lateral plate mesoderm (LPM)flanks the gut tube on the left and right sides, and by coordinated tissue migration drives the initial asymmetry of the gut tube by a pushing mechanism(Horne-Badovinac et al.,2003). In the mouse embryo, the mechanism relies on the gut mesoderm, in which leftward expansion of the mesenchyme and lateral morphogenesis of the pancreas appears to follow growth of the SMP.

Is it possible that organ morphogenesis shows appreciable species differences? Many of the genes that specify organ identity and cellular differentiation are highly conserved. However species have different requirements in the gut depending on food source and diet, and in vertebrates separated by such an evolutionary distance as mammals and fish some differences may be expected to occur. An important example is that the zebrafish has no stomach (Smith,1982). In addition, some endodermal components of the gut are generated differently. It is clear, for example, that fish and mammals generate the pancreatic rudiment in a different manner. In fish, the pancreas does not bud from the endodermal gut tube, but instead is derived from endoderm that is peripheral to the intestine(Wallace and Pack, 2003). The mechanisms for organ morphogenesis must therefore respond to these different requirements and may underlie species differences.

We thank James Sharpe for his help in all aspects of the Optical Projection Tomography and Judith Goodship for crucial reading of the manuscript. We would particularly like to thank Severio Bellusci for supplying the Fgf10mutant embryos.