SONIC HEDGEHOG (Shh) is a secreted glycoprotein

Role of Sonic Hedgehog in the Development of the Trachea and Oesophagus By Adonis S. Ioannides, Deborah J. Henderson, Lewis Spitz, and Andrew J. Copp London, England and Newcastle, England Backround/Purpose: The secreted glycoprotein, Sonic hedgehog (Shh) plays an important patterning role in the development of many organ systems. The authors aimed to study the temporal and spatial pattern of expression of Shh and its receptor Ptc1 during the development of the anterior foregut and to test the hypothesis that the Shh expression pattern is disturbed during the development of oesophageal atresia (OA) and tracheo-oesophageal fistula (TOF) in Adriamycin-treated mouse embryos. Methods: Saline and Adriamycin-treated (4 mg/kg) CBA/Ca embryos were harvested between embryonic days (E) 10.5 and 12.5, and Shh and Ptc1 expression was studied by whole-mount and section in situ hybridisation using digoxygenin-labelled riboprobes. Results: At E10.5, saline-treated embryos had an undivided foregut in which the ventrally placed prospective tracheal epithelium was positive for Shh, whereas the dorsal part was negative. At E11.5, this pattern had reversed with the separated trachea becoming negative and the oesophagus gaining expression of Shh. Ptc1 was expressed in the mesoderm adjacent to Shh expressing endoderm at both stages. Affected Adriamycin-treated embryos had an undivided foregut at E11.5, the epithelium of which showed diffuse Shh staining that lacked the dorso-ventral patterning seen in controls. Conclusions: The reversal in the dorso-ventral pattern of Shh expression during the narrow embryologic window in which tracheo-oesophageal separation is initiated suggests that Shh may play an important role in this process. Transient disturbance of this pattern may underlie the abnormal organogenesis in the Adriamycin model. J Pediatr Surg 38:29-36. Copyright 2003, Elsevier Science (USA). All rights reserved. INDEX WORDS: Tracheo-esophageal fistula, oesophageal atresia, Adriamycin, gene expression, Ptc1, Shh, foregut development, malformation, mouse. SONIC HEDGEHOG (Shh) is a secreted glycoprotein with multiple patterning roles in the developing embryo. Its signal is transduced via the transmembrane receptor Ptc1, which also is a target gene for Shh. 1 Shh is expressed in various organising centres and is involved in anteroposterior patterning of the limb bud and in dorso-ventral neural patterning. 2,3 Expression of Shh also suggests that it plays a role in the morphogenesis of various epithelial appendages such as the hair, teeth, tongue papillae, and respiratory system. 4 It is expressed in the respiratory primordium (E9.5 in the mouse) and later has an important role in branching morphogenesis of the developing lung. 5,6 Homozygous Shh / mutant mouse embryos are characterised by a number of development defects including failure of the trachea to develop as a separate structure from the oesophagus and severe disruption of distal lung structures. 7,8 We aimed to study in detail the temporal and spatial pattern of Shh expression in the anterior foregut at the time of normal tracheo-oesophageal development and the role it might play in this process. We showed previously that failure of tracheo-oesophageal separation is the fundamental embryonic defect preceding development of oesophageal atresia (OA) and tracheo-oesophageal fistula (TOF) in the Adriamycintreated mouse. 9 The specification of respiratory and gastrointestinal components of the foregut appears undisturbed, as indicated by the expression of the respiratory marker Nkx2.1, but the physical separation of the 2 components fails in these embryos. We aimed to test the hypothesis that the temporospatial pattern of Shh expression is disturbed in the foregut of Adriamycin-treated embryos that exhibit an undivided oesophagotrachea. We reasoned that this would provide a further test of the idea that Shh is involved in tracheo-oesophageal development and, specifically, the process of separation of these structures. MATERIALS AND METHODS Time-mated pregnant CBA/Ca mice were injected intraperitoneally with 4 mg/kg body weight of Adriamycin (doxorubicin hydrochloride; From the Neural Development Unit and the Surgery Unit, Institute of Child Health, University College London, and the Institute of Human Genetics, University of Newcastle, England. Presented at the 49th Annual Congress of the British Association of Paediatric Surgeons, Cambridge, England, July 23-26, 2002. Supported by grants from the Wellcome Trust and the Medical Research Council. Address reprint requests to Dr Adonis Ioannides, Neural Development Unit, Institute of Child Health, 30 Guilford St, London, WCIN IEH, England. Copyright 2003, Elsevier Science (USA). All rights reserved. 0022-3468/03/3801-0006$35.00/0 doi:10.1053/jpsu.2003.50005 Journal of Pediatric Surgery, Vol 38, No 1 (January), 2003: pp 29-36 29

30 IOANNIDES ET AL Pharmacia & Upjohn, Milton Keynes, England) in normal saline on embryonic days (E) 7.5 and E8.5. Control mice received intraperitoneal injections of an equivalent volume of normal saline on the same gestational days. Embryos were harvested between E10.5 and E12.5 and fixed in 4% paraformaldehyde solution. Whole mount in situ hybridisation 10 and in situ hybridisation on paraffin-embedded sections 11 were performed using digoxygenin-labelled riboprobes for Shh and Ptc1. Both techniques involved embryo/section pretreatment, followed by overnight hybridisation with the probe, application of the antidigoxygenin antibody (Roche, Mannheim, Germany) and signal development using the NBT/BCIP solution (Roche). Sections were mounted with Vectra mounting medium, and embryos were stored in PBT (Phosphate buffered saline with 0.1% Tween-20; Sigma, St Louis, MO) containing thimerosal (0.02% wt/vol). Whole-mount embryos were refixed after developing and sectioned with a vibratome (30- to 50- m thickness) for further analysis. RESULTS In E10.5 saline-treated control embryos, Shh is expressed in the brain, floor plate of the neural tube, notochord, limb buds, foregut, and hindgut (Figs 1a, c-h). At this stage, it also is expressed in both the dorsal and ventral endoderm of the pharynx but is absent from the lateral endoderm of the pharyngeal pouches (Fig 1c). This arrangement persists at E11.5 (Fig 2a). Caudal to the pharynx, at E10.5, Shh is expressed strongly in the ventral endoderm of the prospective trachea with the dorsal, prospective-oesophageal endoderm lacking expression. At that level, there is a sharp boundary between expressing and nonexpressing cells (Figs 1d-g). This boundary is maintained throughout the foregut at E10.5, although, at the level of the lung buds, Shh is expressed in all but the dorsal-most endoderm (Fig 1h). At E11.5, after tracheo-oesophageal separation has begun, Shh is expressed in the oesophagus but not in the trachea (Fig 2e). Sections taken just above the level of separation show that this ventral-to-dorsal switch already has occurred in the undivided foregut (Fig 2d). Even more cranial sections still exhibit the E10.5 pattern, in which Shh is expressed solely in the ventral foregut (Figs 2b,c). Hence, the ventral-to-dorsal switch in Shh expression appears to occur in a wave, spreading from caudal to cranial along the foregut just ahead of the zone of tracheo-oesophageal separation. At E12.5, the oesophagus remains strongly positive and the trachea completely negative for Shh (Fig 3e). In Adriamycin-treated embryos there is a disturbance in this precise pattern of Shh expression. At E10.5, 2 of 5 embryos examined lacked the dorso-ventral boundary seen in controls and exhibited a more uniform expression of Shh along the dorso-ventral axis in the caudal part of the undivided foregut (Fig 4b). Similar perturbations were seen in embryos with an undivided oesophagotrachea at E11.5. The dorso-ventral boundary was absent with a relatively uniform signal along the dorso-ventral axis of the undivided structure in 3 of 3 embryos examined (Fig 4d). Remarkably, at E12.5, the undivided oesophagotrachea of Adriamycin-treated embryos reestablishes the dorso-ventral pattern of Shh expression seen in controls. The dorsal (oesophageal) part of the structure was strongly positive in contrast to the ventral (tracheal) part that was completely negative in the one embryo examined (Figs 3d and f). About 60% of Adriamycintreated embryos have a separate trachea and oesophagus. All these embryos display dorso-ventral patterning of Shh expression closely similar to saline-treated controls (data not shown). Whole-mount in situ hybridisation studies in E10.5 saline-treated embryos show Ptc1 expression in various tissues throughout the embryo including the brain, neural tube, somites, and limb buds (Fig 1b). As a general rule, Ptc1 is expressed in areas immediately adjacent to Shh expressing tissues. It is found flanking Shh expression in the floor plate and notochord, in cells flanking the ventral midline in the brain and in the posterior part of the limb bud (Figs 1b,i-k). In the early foregut, Ptc1 is expressed in the mesoderm adjacent to Shh-expressing endoderm: at E10.5, there is expression in the mesoderm surrounding the ventral endoderm of the prospective trachea (arrowheads in Figs 1j and k). Expression of Ptc1 continues to be complementary to that of Shh at E11.5 when it follows the ventral-to-dorsal switch that characterises Shh expression and is found in the dorsal perioesophageal mesoderm and the mesoderm surrounding the lung buds (Figs 2g-l). DISCUSSION The pattern of Shh expression in the mouse embryo gives important clues to the possible molecular mechanisms that underlie foregut development. In the pharynx, Shh is weakly expressed in both the ventral and dorsal endoderm but is completely absent from the proliferating epithelium of the pharyngeal pouches, which give rise to structures such as the thymic primordium. This suggests that Shh expression boundaries may play a role in regulating organ budding from the endoderm of the gut tube. In the postpharyngeal foregut, such expression boundaries may play a role in the separation of the trachea from the oesophagus. A dorso-ventral boundary is established with strong ventral Shh expression in the epithelium of the prospective trachea and lack of expression in the epithelium of the prospective oesophagus. After tracheooesophageal separation, the boundary still is present even though the pattern has switched with the epithelium of the definitive trachea becoming negative and that of the definitive oesophagus positive for Shh. Remarkably, this complete and specific ventral-to-dorsal switch is itself propagated along the anteroposterior axis of the foregut in a caudal to cranial direction ahead of tracheo-oesophageal separation. The temporal and spatial relationship

Fig 1. Patterns of Shh expression and complementary Ptc1 patterns, in saline-treated E10.5 mouse embryos. In situ hybridisation for Shh (a,c-h) and Ptc1 (b,i,j,k): whole embryos and vibratome sections from whole-mount hybridisation experiments on E10.5 saline-treated embryos. (c-h, i-k) form 2 series of progressively more caudal sections. (a) Shh expression is seen in ventral areas of the brain, the floor plate (Fp), the notochord (arrowhead in a), the foregut (Fo) and hindgut (Hg) endoderm, including the lung buds (arrow in a) and the limb buds (Lb). (b) Ptc1 expression flanks the ventral neural tube (arrow in b) and ventral midline in the brain and also is found in the somites (S), the limb buds, and the mesoderm surrounding the foregut and hindgut. (c-k) Shh is expressed strongly in the floor plate of the neural tube (Fp) and the notochord (Nt), whereas Ptc1 is expressed in the mesoderm adjacent to the ventral neural tube and around the notochord (arrowhead in i). (c,i) At the level of the pharyngeal foregut (Ph), Shh is expressed in both the dorsal and ventral endoderm (arrows in c) but is specifically absent from the endoderm of the pharyngeal pouches (arrowheads in c). Ptc1 is expressed next to the Shh-expressing endoderm (arrows in i). (d,e,j,k) Shh is expressed in the ventral endoderm of the laryngotracheal groove (Lt) with a sharp boundary between ventral expressing and dorsal nonexpressing cells (arrowheads in d,e). Mesodermal expression of Ptc1 also is restricted to the ventral foregut (arrowheads in j,k) although the boundary is not as distinct because of adjacent perinotochordal expression. (f,g) Further caudally, Shh expression is restricted to the endoderm of the prospective trachea (Tr) (arrowheads). (h) At the level of the lung buds (Lb), the Shh dorso-ventral boundary still exists (arrow) but appears less distinct. D, diencephalon; T, telencephalon. Scale bar: a,b, 500 m; c-k, 100 m.

32 IOANNIDES ET AL Fig 2. Changing patterns of Shh expression and complementary Ptc1 patterns in E11.5 saline-treated embryos. Slide in situ hybridisation for Shh (a-f) and vibratome sections from whole-mount hybridisation experiments for Ptc1 (g-l). (a-f and g-l) represent corresponding levels in progressively more caudal sections through 2 comparable embryos. (a,g) In the pharyngeal foregut (Ph), Shh is expressed ventrally and dorsally but is absent laterally from the endoderm of the pharyngeal pouches (arrowheads in a mark boundary). Ptc1 is expressed adjacent to the ventral (but not dorsal) Shh expressing endoderm (arrow in g). (b,h) At the level of the laryngotracheal groove (Lt), Shh is expressed in a well-defined ventral domain (arrows in b) with Ptc1 expression in the adjacent mesoderm (arrow in h). (c,i) At a slightly more caudal level, Shh is still ventrally expressed (arrowhead in c), but the expression appears to be shifting dorsally. This is reflected by a dorsalward shift in mesodermal Ptc1 expression (arrow in i). (d,j) At an even more caudal level, the pattern of Shh expression has reversed with a well-defined dorsal Shh expression domain (arrowheads in d). Ptc1 is now expressed in the mesoderm adjacent to the dorsal foregut (Fo) (arrow in j). (e,k) Shh is expressed in the oesophagus (Oe), and Ptc1 is expressed in the perioesophageal mesoderm. (f,l) Shh expression in the oesophagus and lung buds (Lb). Ptc1 expression around the lung buds is stronger than perioesophageal expression at this level (asterisk in l). Fp, floor plate; Nt, notochord; Tr, trachea. Scale bar: 100 m.

SONIC HEDGEHOG IN THE TRACHEA AND OESOPHAGUS 33 Fig 3. Dorso-ventral pattern of Shh expression in the oesophagotrachea of E12.5 Adriamycin-treated embryos reestablishes a similar pattern to saline-treated control embryos. In situ hybridisation for Shh on transverse sections from saline- (a,c,e) and Adriamycin-treated (b,d,f) E12.5 embryos. (a,b) At E12.5, the cranial-most part of the laryngotracheal groove (Lt) has yet to separate from the oesophageal foregut. Both uniformly express Shh in saline- and Adriamycin-treated embryos. (c,d) At a slightly more caudal level, the separated trachea (Tr) is completely negative for Shh, whereas the oesophagus (Oe) retains a strong signal (c). In the Adriamycin-treated embryo (d), the undivided oesophagotrachea (Oet) reflects this dorso-ventral pattern in expression with a sharp boundary between dorsal expressing and ventral nonexpressing cells (arrows). (e,f) This pattern persists further caudally and is unchanged as far as the level of the trifurcation to form lung buds and fistula to the stomach. Scale bar: 100 m.

34 IOANNIDES ET AL Fig 4. The ventral-to-dorsal switch in Shh expression is disturbed in Adriamycin-treated embryos. In situ hybridisation for Shh on saline- (a,c) and Adriamycin-treated (b,d) embryos. Vibratome sections, at the level of the atrial chamber of the heart (a,b) and the level of the atria (c,d), from whole-mount experiments. (a,c) In normal development, Shh expression shifts completely from the ventral, tracheal-prospective epithelium at E10.5 (arrowheads in a mark boundaries), to the oesophageal epithelium at E11.5 (after tracheo-oesophageal separation). (b) In 2 of 5 E10.5 Adriamycin-treated embryos, the undivided foregut (Fo) expresses Shh uniformly along its dorso-ventral axis. (d) All E11.5 Adriamycin-treated embryos with an undivided oesophagotrachea exhibit relatively diffuse staining along the dorso-ventral axis of the common structure. In this embryo, the signal appears stronger at the 2 poles and contrasts sharply with the pattern seen in saline-controls (c). Scale bar: 100 m. between the switch in expression and separation suggests that Shh may have a role in this process. At the caudal end of the prospective tracheal epithelium, the lung buds are strongly positive for Shh and retain their Shh-positive status even after the separation of the trachea from the oesophagus. This suggests that their development is a separate process from that of tracheo-oesophageal separation and is consistent with the role of Shh in lung proliferation and branching morphogenesis. 5,6 One can only speculate as to the factors that control Shh expression in the foregut. Signals from the notochord could play a role in this regulation, but their proposed effect on the E10.5 dorsal foregut endoderm would be in contrast to their effect on the adjacent neuroepithelium where they induce Shh expression. 12,13 However, notochord grafting and ablation experiments have shown that in the posterior foregut, dorsal endodermal expression of Shh is repressed by signals from the notochord (activin- B), thereby allowing development of the dorsal pancreas. 14 It is possible that similar mechanisms regulate Shh expression in the anterior foregut. Expression of Shh in the ventral foregut endoderm could be under the control of signals from the ventral meso-

SONIC HEDGEHOG IN THE TRACHEA AND OESOPHAGUS 35 derm (cardiac mesoderm). Coculture experiments have shown that ventral foregut endoderm expresses Shh under the influence of cardiac mesoderm, and fibroblast growth factor 2 (FGF2) has been shown to be a potential mediator of this induction. 15 The effect of Adriamycin on the pattern of Shh expression suggests a possible mechanism for abnormal organogenesis in Adriamycin-treated embryos. At E10.5, some embryos lack the well-defined ventral Shh expression and show a more uniform pattern of expression in the caudal part of the undivided foregut. The morphologic data suggest that although all Adriamycin-treated embryos appear normal at E10.5, 40% to 50% are destined to have an undivided foregut. 9 It seems likely that the E10.5 embryos that exhibit diffuse expression of Shh are the ones that are destined to be abnormal. This is supported by the finding at E11.5, that all Adriamycintreated embryos with an undivided oesophagotrachea lack the sharp dorso-ventral boundary in Shh expression seen in saline-treated controls, exhibiting a more diffuse signal along the dorso-ventral axis of the common tube. It is not possible at this stage to determine whether this disturbance in Shh expression is related causally to the faulty tracheo-oesophageal separation or, rather, whether it is secondary to another factor that causes failure of separation. However, given the potential role of Shh in the separation process, it is reasonable to suggest that disturbance of the tightly regulated Shh expression pattern may underlie the action of Adriamycin in treated embryos. This hypothesis also is consistent with the lack of lung malformations in mouse embryos with OA/TOF. Whereas failure of ventral Shh downregulation interferes with tracheo-oesophageal separation, it should not affect branching morphogenesis, because the tips of the lung buds normally are strongly Shh positive throughout pulmonary development. The abnormal tracheo-oesophageal development in Shh / null mutant embryos also supports the hypothesis that a disturbance in Shh expression patterns can lead to tracheo-oesophageal malformations. A recent study on the Adriamycin rat model of OA/ TOF has also suggested that the Shh expression pattern may be disturbed, but the described pattern differs from our experience. 16 It reports uniform expression in some normal embryos that becomes downregulated specifically at the point of tracheo-oesophageal separation. This downregulation is absent in the Adriamycin-treated rat embryos in which separation fails. This difference in apparent Shh expression patterns may be attributed to species differences In our study, the undivided oesophagotrachea of Adriamycin-treated embryos reestablishes correct dorsoventral patterning of Shh expression at E12.5 with the dorsal oesophageal part positive and the ventral tracheal part negative. This could indicate that the action of Adriamycin is transient and interferes with foregut expression patterns only around the time of tracheooesophageal separation. This would be consistent with the terminal half-life of the drug of up to 48 hours. Once the foregut has failed to divide, the gradual reestablishment of the dorso-ventral gradient may be of little consequence. 1. Tabin CJ, McMahon AP: Recent advances in Hedgehog signalling. Trends Cell Biol 7:442-446, 1997 2. Riddle RD, Johnson RL, Laufer E, et al: Sonic hedgehog mediates the polarizing activity of the ZPA. Cell 75:1401-1416, 1993 3. Roelink H, Porter JA, Chiang C, et al: Floor plate and motor neuron induction by different concentrations of the amino-terminal cleavage product of sonic hedgehog autoproteolysis. Cell 81:445-455, 1995 4. Chuong CM, Patel N, Lin J, et al: Sonic hedgehog signaling pathway in vertebrate epithelial appendage morphogenesis: Perspectives in development and evolution. Cell Mol Life Sci 57:1672-1681, 2000 5. Pepicelli C, Lewis PM, McMahon AP: Sonic hedgehog regulates branching morphogenesis in the mammalian lung. Curr Biol 8:1083-1086, 1998 6. Bellusci S, Furuta Y, Rush MG, et al: Involvement of Sonic hedgehog (Shh) in mouse embryonic lung growth and morphogenesis. Development 124:53-63, 1997 7. Chiang C, Litingtung Y, Lee E, et al: Cyclopia and defective axial patterning in mice lacking Sonic hedgehog gene function. Nature 383:407-413, 1996 8. Litingtung Y, Lei L, Westphal H, et al: Sonic hedgehog is essential for foregut development. Nat Genet 20:58-61, 1998 9. Ioannides AS, Chaudhry B, Henderson DJ, et al: Dorsoventral REFERENCES patterning in oesophageal atresia with tracheo-oesophageal fistula: Evidence from a new mouse model. J Pediatr Surg 37:185-191, 2002 10. Copp AJ, Cogram P, Fleming A, et al: Neurulation and neural tube closure defects, in Tuan RS, Lo CW (eds): Developmental Biology Protocols, Vol 1. New York, NY, Academic Press, 1999 11. Breitschopf H, Suchanek G, Gould RM, et al: In situ hybridisation with digoxygenin-labeled probes: Sensitive and reliable detection method applied to myelinating rat brain. Acta Neuropathol (Berl) 84:581-587, 1992 12. Ericson J, Morton S, Kawakami A, et al: Two critical periods of Sonic Hedgehog signalling required for the specification of motor neuron identity. Cell 87:661-673, 1996 13. Patten I, Placzek M: Opponent activities of Shh and BMP signalling during floor plate induction in vivo. Curr Biol 12:47-52, 2002 14. Hebrok M, Kim SK, Melton DA: Notochord repression of endodermal Sonic hedgehog permits pancreas development. Genes Dev 12:1705-1713, 1998 15. Deutch G, Jung J, Zheng M, et al: A bipotential precursor population for pancreas and liver within the embryonic endoderm. Development 128:871-881, 2001 16. Orford J, Manglick P, Cass DT, et al: Mechanisms for the development of esophageal atresia. J Pediatr Surg 36:985-994, 2001

36 IOANNIDES ET AL Discussion J. Langer (Toronto, Ontario): There is some evidence that sonic hedgehogs may be involved in anorectal malformations that are part of the same syndrome as tracheo-oesophageal problems. Is there any theory that unifies these things, and what are your strategies for taking this forward to identify what happens in humans? A.S. Ioannides (response): Yes, there is, but it is speculation at present. We believe that a generalised disturbance in Sonic hedgehog function is unlikely, because it would lead to a very severe phenotype in many systems. Tracheo-oesophageal and anorectal malformations often are isolated defects, and it is likely that we are looking at a transient disturbance in the Sonic hedgehog pathway in specific regions of the gut. To take this forward we will try to link gene expression to mechanisms that could be associated with tracheo-oesophageal separation. We currently are studying programmed cell death, cell adhesion, and changes in the cytoskeleton and already have shown that the pattern of programmed cell death closely follows the pattern of Sonic hedgehog expression.