Tcof1�Treacle is required for neural crest cellformation and proliferation deficienciesthat cause craniofacial abnormalitiesJill Dixon*†, Natalie C. Jones†‡, Lisa L. Sandell‡, Sachintha M. Jayasinghe‡, Jennifer Crane‡, Jean-Philippe Rey‡,Michael J. Dixon*§¶, and Paul A. Trainor‡¶

*School of Dentistry and §Faculty of Life Sciences, University of Manchester, Oxford Road, Manchester M13 9PT, United Kingdom; and‡Stowers Institute for Medical Research, 1000 East 50th Street, Kansas City, MO 64110

Edited by Kathryn V. Anderson, Sloan–Kettering Institute, New York, NY, and approved July 10, 2006 (received for review May 16, 2006)

Neural crest cells are a migratory cell population that give rise tothe majority of the cartilage, bone, connective tissue, and sensoryganglia in the head. Abnormalities in the formation, proliferation,migration, and differentiation phases of the neural crest cell lifecycle can lead to craniofacial malformations, which constituteone-third of all congenital birth defects. Treacher Collins syndrome(TCS) is characterized by hypoplasia of the facial bones, cleft palate,and middle and external ear defects. Although TCS results fromautosomal dominant mutations of the gene TCOF1, the mechanisticorigins of the abnormalities observed in this condition are un-known, and the function of Treacle, the protein encoded by TCOF1,remains poorly understood. To investigate the developmentalbasis of TCS we generated a mouse model through germ-linemutation of Tcof1. Haploinsufficiency of Tcof1 leads to a deficiencyin migrating neural crest cells, which results in severe craniofacialmalformations. We demonstrate that Tcof1�Treacle is requiredcell-autonomously for the formation and proliferation of neuralcrest cells. Tcof1�Treacle regulates proliferation by controlling theproduction of mature ribosomes. Therefore, Tcof1�Treacle is aunique spatiotemporal regulator of ribosome biogenesis, a defi-ciency that disrupts neural crest cell formation and proliferation,causing the hypoplasia characteristic of TCS craniofacial anomalies.

craniofacial development � embryo � mouse � Treacher Collins syndrome

Neural crest cells are a multipotent, migratory cell populationthat generate a diverse array of cell types during vertebrate

development. These include bones, tendons, neurons, glia, mela-nocytes, and connective, endocrine, and adipose tissue. With alimited capacity for self-renewal and a wide range of differentiationfates, neural crest cells bear many of the hallmarks of stem cells (1).Neural crest cells are born at the interface between the surfaceectoderm and the dorsal margin of the neural plate, a region termedthe neural plate border (2, 3). This induction process requires aprecise threshold concentration gradient of bone morphogeneticprotein (BMP) signaling in the neural plate border, which isdetermined by contact-mediated interactions across the neuralplate–surface ectoderm interface (4–7). The delamination of neu-ral crest cells involves an epithelial-to-mesenchymal transforma-tion, which is driven by the repression of cell adhesion moleculessuch as E-cadherin by Snail1 (6).

In the head, cranial neural crest cells exhibit a unique ability todifferentiate into cartilage and bone, underpinning their funda-mental importance to vertebrate craniofacial evolution (7). Defectsin neural crest cell formation, proliferation, migration, and�ordifferentiation are considered responsible for craniofacial abnor-malities, which constitute up to one-third of all congenital birthdefects (8). Therefore, it is critical to understand the mechanismsregulating each phase of neural crest cell development to compre-hend the precise etiology of specific craniofacial anomalies, andcurrently there is a dearth of knowledge describing the regulationof neural crest cell proliferation.

Treacher Collins syndrome (TCS) (Online Mendelian Inheri-tance in Man database accession no. 154500) is one example of acongenital craniofacial disorder, this condition being characterizedby numerous anomalies that are restricted primarily to the head andneck. The phenotype of TCS includes hypoplasia of the facial bones,particularly the zygomatic complex and mandible, cleft palate, andmiddle and external ear defects that result in conductive deafness.TCS is caused by autosomal dominant mutations in the TCOF1gene; however, the in vivo functions of TCOF1 and the nucleolarprotein Treacle that it encodes (9) remain poorly understood. Anumber of hypotheses have been proposed to account for theorigins of TCS craniofacial malformations. These include abnormalneural crest cell migration (10), improper cellular differentiation(11), and extracellular matrix abnormalities (12). However, to datethere have been no molecular or cellular data to support any ofthese hypotheses.

Consequently, we investigated the function of Tcof1�Treacle inneural crest cell development and craniofacial morphogenesis.Through germ-line mutation of Tcof1 we generated a mouse modelof TCS, and we demonstrate that Treacle is a novel spatiotemporalregulator of ribosome biogenesis that is cell-autonomously requiredfor neural crest cell generation and proliferation.

ResultsTcof1 Is Expressed in a Dynamic Spatiotemporal Pattern. Individualsaffected by TCS exhibit distinctive craniofacial anomalies, and wehypothesized that these abnormalities result from defects in neuralcrest cell patterning. As a first step to determining the mechanisticbasis of TCS and any role for Tcof1 in neural crest cells, wecharacterized in detail the spatiotemporal pattern of Tcof1 expres-sion during early embryogenesis (Fig. 6, which is published assupporting information on the PNAS web site). In whole embryoswe observed strong, spatiotemporally specific expression of Tcof1 inthe neuroepithelium at embryonic day (E) 8.5 (Fig. 7A) and in thefrontonasal and branchial arch mesenchyme at E9.5 (Fig. 7D).

Section in situ hybridization for Tcof1 demonstrated the speci-ficity of staining throughout the neuroepithelium at E8.5, whichcorresponds with the generation of neural crest cells. Interestingly,staining was most intense at the lateral edge of the neuroepithelium,which correlates with cells in the G1�S phases of the cell cycle.Strong Tcof1 expression was also observed in migrating neural crestcells in the craniofacial mesenchyme but was noticeably absent in

Conflict of interest statement: No conflicts declared.

This paper was submitted directly (Track II) to the PNAS office.

Freely available online through the PNAS open access option.

Abbreviations: En, embryonic day n; BMP, bone morphogenic protein; TCS, Treacher Collinssyndrome.

†J.D. and N.C.J. contributed equally to this work.

¶To whom correspondence may be addressed. E-mail: pat@stowers-institute.org or mike.dixon@manchester.ac.uk.

© 2006 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0603730103 PNAS � September 5, 2006 � vol. 103 � no. 36 � 13403–13408

DEV

ELO

PMEN

TAL

BIO

LOG

Y

the more medially located cranial mesoderm (Fig. 7 B and C). AtE9.5 Tcof1 was strongly expressed throughout the neural tube andcould also be observed in craniofacial tissues heavily populated withneural crest cells such as the frontonasal, maxillary, and mandibularmesenchyme and the sensory ganglia (Fig. 7 E and F). By E10.5 thelevels of Tcof1 expression had significantly weakened, becomingvery diffuse, and by E11.5–E12.5 Tcof1 expression was largelyundetectable. This demonstration of the dynamically regulatedspatiotemporal pattern of expression of Tcof1 during neural crestcell formation and migration provided evidence of a tangible linkbetween neural crest cells and TCS craniofacial abnormalities.

Tcof1 Haploinsufficient Embryos Display Severe Cranioskeletal De-fects. To investigate the role of Tcof1 in neural crest cell patterning,our laboratory generated chimeric mice carrying a germ-line mu-tation in one allele of the murine orthologue of TCOF1. Breedingof male chimeras with female C57BL�6 mice generated heterozy-gous offspring that exhibited a number of features reminiscent ofthe human disorder, including abnormalities of the maxilla andmandible. However, additional anomalies not observed in TCSpatients, including severe developmental delay, anophthalmia, andexencephaly, were also found (13). The extreme nature of thisphenotype, which resulted in premature neonatal death, precludedbreeding and expansion of the mutant line and furthermore pre-vented a detailed analysis of neural crest cell patterning defects inthese animals. To circumvent this lethality we crossed the Tcof1mutation onto several different backgrounds and determined that,on the DBA genetic background, Tcof1 heterozygosity was com-patible with postnatal life, enabling congenic lines to be generated(14). Although we maintained the Tcof1�/� line on a pure DBAbackground, we outcrossed the Tcof1�/� DBA line to C57BL�6 togenerate embryos with the characteristic features of TCS. Impor-tantly, although the embryos analyzed were of mixed DBA �C57BL�6 background, the characteristic mutant phenotype wasconsistently reproducible with minimal interembryo variability ateach developmental stage. This permitted a detailed investigationof the function of Tcof1 and its relationship to neural crest cellpatterning.

Offspring of congenic DBA Tcof1 heterozygous mice inter-crossed with wild-type C57BL�6 mice exhibit a distinctive pheno-

type that mimics TCS. Confirmed Tcof1�/� neonates were charac-terized by a reduction in the size of the head, which was domed inappearance and shortened in the anteroposterior direction withobvious severe frontonasal dysplasia (Fig. 1 A and B). Neonatesdisplayed gasping behavior and abdominal distension and diedwithin 24 h of birth.

Skeletal analyses at E17.5 highlighted the severity of thecranioskeletal abnormalities associated with heterozygosity ofTcof1. The cranial vault was domed and the nasal and frontalbones were dysmorphic and hypoplastic in Tcof1�/� mutants(Fig. 1 C–F). Ventral views of the viscero- and chondrocraniumrevealed that the premaxilla, maxillary, and palatine bones werealso abnormally shaped and truncated (Fig. 1 E and F). Amidline cleft was visible between the displaced palatal shelves ofthe maxilla and palatine bones, allowing partial visualization ofthe presphenoid skull bone elements (arrow in Fig. 1F). Man-dibles isolated from the same embryos were also shorter inmutant as compared with wild-type embryos (Fig. 1 G and H).In addition, the temporal bone was hypoplastic and dysmorphicin mutant embryos (Fig. 1 I and J). These cranioskeletal abnor-malities in Tcof1�/� mutants were evident as early as E15.5 in theform of frontonasal, premaxillary, and maxillary hypoplasia(Fig. 1 K and L). Transverse histological sections through themidfacial region of E14.5 Tcof1�/� embryos confirmed thepresence of cleft palate by demonstrating that the palatal shelveshad failed to fuse completely (Fig. 1 M and N). Furthermore, thenasal passages were poorly formed, with no evidence of a nasalseptum or conchae.

Collectively, these data indicated that Tcof1�/� neonates diedfrom respiratory arrest due to malformations of the nasal, premax-illa, maxilla, and palatine skeletal elements. These defects arecharacteristic of the midfacial abnormalities associated with severecases of TCS in humans. Interestingly, the cranioskeletal structuresthat are malformed in Tcof1�/� embryos are derived primarily fromneural crest cells. These combined observations strongly suggestedthat the craniofacial anomalies associated with TCS arise specifi-cally as the result of defects in neural crest cell patterning.

Tcof1 Haploinsufficient Embryos Exhibit Neural Crest Cell Hypoplasia.To establish whether a migration defect was responsible for thecraniofacial defects characteristic of TCS as originally proposed

Fig. 1. Analysis of DBA � C57BL�6Tcof1��� embryos. Comparison of wild-type (A) and mutant (B) newborn pups.Tcof1�/� mutant mice exhibit shortenedand domed-shaped heads with frontonasaldysplasia (arrow). Skeletal staining of E17.5wild-type (C, E, G, and I) and Tcof1�/� (D, F,H, and J) embryos reveals hypoplasia ofnumerous craniofacial bones including thenasal (n), frontal (f), premaxillary (pmx),maxillary (mx), mandibular (md), and tem-poral bones (t) together with cleft palate(arrow) in mutant embryos. In comparisonsof wild-type (K) and mutant (L) embryoskeletal preparations, cranioskeletal hyp-oplasia is evident at E15.5 in the frontal,premaxillary, and maxillary elements.Transverse histological sections of E14.5wild-type (M) and mutant (N) embryoshighlight the absence of complete midlinefusion (arrowhead) and the lack of a nasalseptum (ns) and conchae (c) in Tcof1�/�

embryos.

13404 � www.pnas.org�cgi�doi�10.1073�pnas.0603730103 Dixon et al.

(10), we performed neural crest cell lineage tracing in combinationwith whole-embryo culture of wild-type and Tcof1�/� mouse em-bryos (Fig. 2). Regions of the anterior hindbrain, midbrain, andforebrain were focally labeled with the lipophilic dye DiI at E8.25to coincide with the emergence of the earliest waves of neural crestcells. Fluorescence analysis indicated that numerous DiI-labeledrhombomere 2- and rhombomere 4-derived neural crest cellsmigrated ventrolaterally in discrete streams into the first and secondbranchial arches, respectively, in wild-type embryos (Fig. 2 A, B, E,and F). An identical pattern of migration from the hindbrain to thebranchial arches was observed in Tcof1�/� mutant embryos (Fig. 2C, D, G, and H). Similarly, no migratory or path-finding differenceswere observed in midbrain and forebrain neural crest cell migrationto the maxillary and frontonasal regions of the face in comparisonsbetween wild-type and Tcof1�/� embryos (Fig. 3 A–D). Hence, thesegmental migration of cranial neural crest cells is normal andunaffected in Tcof1�/� embryos, which clearly demonstrates thataberrant neural crest cell migration is not the underlying cause ofTCS craniofacial abnormalities. Therefore, Tcof1 does not play arole in neural crest cell migration.

Interestingly, despite the absence of a migration defect, we didreproducibly observe fewer migrating neural crest cells in theTcof1�/� embryos compared with their wild-type littermates ateach axial level examined in our lineage-tracing experiments (Fig.2, compare B and F with D and H). The reduction in the numberof migrating cranial neural crest cells was further supported by insitu hybridization staining for the neural crest marker Sox10, inwhich the staining was visibly more sparse in E8–E9 Tcof1�/�

embryos compared with wild-type littermates (Fig. 3 A–D). Con-sequently, by E10.5 defects in cranial neural crest cell-derivedstructures were clearly evident. Although the cranial sensory gan-glia formed in the correct axial locations, they were significantlysmaller and more disorganized in Tcof1�/� embryos compared withwild-type littermates (Fig. 3 E and F). This, however, was a verysubjective view, and it was critical in establishing the mechanisticbasis of TCS to be able to consistently and reproducibly quantify thedegree of neural crest cell reduction. This quantification wasachieved by flow cytometric analyses of GFP-labeled neural crestcells. E9 embryos with GFP-labeled neural crest cells were obtainedthrough intercrossing DBA Tcof1�/� and homozygous C57BL�6Pax3GFP mice (Fig. 3G). Reproducible flow cytometry enumera-tion of GFP-positive neural crest cells demonstrated that, as aproportion of the total number of cells in the craniofacial mesen-

chyme, there were �22% fewer migrating cranial neural crest cellsin Tcof1�/� embryos compared with wild-type littermates (Fig. 3Hand Table 1, which is published as supporting information on thePNAS web site). Taken together with the dynamic expressionpattern exhibited by Tcof1, this finding suggested that Treacle wasrequired for neural crest cell generation, proliferation, and�orviability. Loss of Tcof1 therefore results in a neural crest cell deficit,which causes hypoplasia of the cranial sensory ganglia and skeletalelements.

Neural Crest Cell Hypoplasia Is a Direct Consequence of Neuroepithe-lial-Specific Apoptosis. Collectively, our analyses indicated that therewas a significant deficiency in the number of migrating cranialneural crest cells in Tcof1�/� mutant embryos, which underlies the

Fig. 3. Neural crest cell quantification in Tcof1��� embryos. Sox10 in situhybridization in E8.5 (A and B), E9.5 (C and D), and E10.5 (E and F) wild-type (A,C,andE)andTcof1�/� (B,D,andF)embryosreveals thedeficiency incranialneuralcrest cell contribution to the frontonasal prominence (fnp), first and secondbranchial arches (ba1 and ba2; arrows in A and B), cranial ganglia (cg; arrows inCandD) suchasthetrigeminalganglion(t inE),andtothehypoglossopharyngealand vagal ganglia posterior to the otic vesicle (double arrows in E and F). (G) E9.0wild-type embryo obtained by intercrossing C57BL�6 Pax3GFP and DBA Tcof1�/�

mice to quantify the number of migrating cranial neural crest cells in flowcytometry analyses. Neural crest cells (GFP-labeled) contribute to the frontonasalprocess (fnp), the mesenchyme surrounding the optic placode (op), and themaxillary (mx) and mandibular (md) prominences of the first branchial arch (ba1),as well as the more caudal branchial arches. fb, forebrain; mb, midbrain; hb,hindbrain. (H) Histogram demonstrating a 22.3% reduction in the relative pro-portion of GFP-labeled migrating neural crest cells to cranial mesenchyme cells inTcof1�/� mutant (green bar) as compared with wild-type littermate (yellow bar)andpositivecontrol (redbar)embryos.Thebluebar representsanegativecontrolfor GFP in wild-type DBA � C57BL�6 embryos.

Fig. 2. Neural crest cells at the level of rhombomere 2 (r2) and rhombomere4 (r4) in wild-type (A, B, E, and F) and Tcof1�/� (C, D, G, and H) embryos werelabeled with DiI (red) at E8.25 and cultured for 24 h. Fewer migrating neuralcrest cells from r2 and r4 populating the first branchial arch (ba1) (B and D) andsecond branchial arch (ba2) (F and H), respectively, in E9.5 Tcof1�/� (D and H)embryos were observed.

Dixon et al. PNAS � September 5, 2006 � vol. 103 � no. 36 � 13405

DEV

ELO

PMEN

TAL

BIO

LOG

Y

neurogenic and cranioskeletal hypoplasia observed in TCS indi-viduals. This deficiency could arise through compromised viabilityof migrating neural crest cells or anomalies in neural crest cellformation. To discriminate between these mechanisms we initiallyassayed for apoptosis using TUNEL staining in E8–E9.5 embryos,which corresponded with the formation and migration phases ofneural crest cell development. We detected a surprisingly high levelof apoptosis in Tcof1�/� embryos relative to their wild-type litter-mates, which was largely confined to the neural plate (Fig. 4 A andB). These observations suggested that Treacle was critically re-quired for neuroepithelial survival and neural crest cell formationbut was not essential for neural crest cell viability.

To confirm the viability of migrating cranial neural crest cells, webroadly labeled the midbrain and anterior hindbrain of E8–E8.5mutant embryos with DiI (red fluorescence) and cultured theembryos for at least 6 h. This time period was sufficient for thenewly formed cranial neural crest cells to migrate to the peripheryof the head. These labeled embryos were then assayed for apoptosisby TUNEL (green fluorescence). The absence of double-labeled(yellow fluorescence) migrating neural crest cells conclusively dem-onstrated that migrating neural crest cells in Tcof1�/� embryos wereviable (Fig. 4C). In contrast, colabeled DiI- and TUNEL-positivecells highlighted the spatiotemporal restriction of cell death to theneuroepithelium. The elevated levels of apoptosis were mediated bycaspase3 as evidenced by enhanced activated capsase3 immuno-staining in the neuroepithelium of mutant embryos (Fig. 4 D andE and data not shown). Consequently, during neural crest cellinduction the neural plate in Tcof1�/� embryos is significantlythinner than in wild-type littermates (Fig. 4 D and E). Thus,haploinsufficiency of Tcof1 leads to the apoptotic depletion ofneural stem cells and migrating neural crest cells, which stronglyargues that Treacle is critically required for neural crest cellformation.

To further refine the mechanistic origins of TCS craniofacialabnormalities and the role of Tcof1, it was important to investigatewhere Tcof1 functions in the neural crest cell formation process.Because much of the information for neural crest cell patterning isderived from the neural tube and the hindbrain in particular, wefirst assayed for the presence of any anteroposterior patterning orsegmentation defects in the neural tube during early embryogen-esis. Otx2, which is strongly expressed in the forebrain and midbrainneuroepithelium of wild-type embryos, was unaltered in Tcof1�/�

embryos (Fig. 7 A and B, which is published as supporting infor-mation on the PNAS web site). Similarly, Msx2 strongly labeled thedorsal folds of the neural plate in both wild-type and Tcof1�/�

embryos (Fig. 7 C and D). Furthermore, using genetic markers suchas Krox20, which distinguishes rhombomeres 3 and 5 in the hind-brain, we could discern no segmental patterning defects in theneural tubes of Tcof1�/� embryos (Fig. 7 E and F). Therefore, therewere no obvious anteroposterior or segmentation defects underly-ing the neuroepithelial apoptosis and corresponding reduction inthe number of cranial neural crest cells. Hence, we focused ouranalyses on the morphogenetic process of neural crest cell inductionand assayed for BMP and WNT signaling owing to their wellcharacterized roles in neural crest cell development (15, 16). Bmp2is normally strongly expressed in the surface ectoderm immediatelyadjacent to the neuroepithelium, and no differences in spatiotem-poral expression or intensity were observed in comparisons be-tween wild-type and mutant embryos (Fig. 7 G and H). Similarly,no differences were observed between wild-type and mutant em-bryos for Wnt1 activity, which is strongly expressed throughout thedorsal neuroepithelium during the induction phase of neural crestcell development (Fig. 7 I and J). Therefore, the conserved mor-phogenetic signals involved in vertebrate neural crest cell inductionremain unaltered in Tcof1�/� embryos, suggesting that the defectin neural crest cell formation lies downstream of BMP�WNTsignaling.

In contrast, we observed a significant reduction in Snail1 expres-sion in mutant compared with wild-type embryos, which wasconsistent with diminished epithelial-to-mesenchymal transforma-tion and generation of neural crest cells (Fig. 7 K and L). These dataindicate that the function of Tcof1 lies upstream of Snail1 in thegeneration of neural crest cells. Therefore, although the morpho-genetic machinery for neural crest stem cell induction is spatio-temporally intact in Tcof1�/� embryos, haploinsufficiency of Tcof1results in elevated neuroepithelial apoptosis and a diminishedcapacity to generate neural crest stem cells, leading directly tohypoplasia of migrating cranial neural crest cells and also of theirderived craniofacial structures.

Deficient Production of Mature Ribosomes Causes Apoptosis andReduced Proliferation. As a corollary to the elevated levels ofapoptosis observed specifically in the neuroepithelium of Tcof1�/�

embryos, it was important to assay for alterations to cell prolifer-ation in the neuroepithelium and neural crest. We therefore

Fig. 4. Neuroepithelial-specific apoptosis in Tcof1��� embryos. TUNEL stainingfor apoptosis (green) in E8.25 wild-type (A) and Tcof1�/� (B) embryos and in acultured Tcof1�/� embryo (C) in which forebrain-derived (fb), midbrain-derived(mb), and hindbrain-derived (hb) neural crest cells were labeled with DiI (red).Double-stained cells (yellow) in the neural plate (np) and not in migrating neuralcrest cells (red) highlight the specificity of apoptosis in the neuroepithelium ofTcof1�/� embryos. Caspase3 immunohistochemistry of parasagittal cranial sec-tions from E8.75 wild-type (D) and Tcof1�/� (E) embryos revealed enhancedcaspase3 activity in the neuroepithelium (arrows) of Tcof1�/� mutants.

Fig. 5. Cellular analysis of Tcof1��� embryos. (A–D) BrdU incorporation (pink inAandB) andY10B immunostaining (red inCandD)of transverse sectionsofE8.75wild-type (A and C) and Tcof1�/� (B and D) embryos. A reduction in cell prolifer-ation (arrows) in the neuroepithelium (ne) and cranial mesenchyme (cm) corre-sponds with diminished production of mature 28S rRNA (arrows in D) in Tcof1�/�

embryos. No proliferation anomalies were observed in the endoderm (arrow-heads in A and B) or surface ectoderm (se). (E–H) TUNEL-stained embryos afterhomotopic hindbrain neural plate (np) transplantations of DiI-labeled (red) wild-type neuroepithelium in cultured E8.25 wild-type (E) and Tcof1�/� (F) embryos.Transplanted cells exhibit minimal apoptosis and generate migrating neural crestcells. (G and H) E8.75–E9.0 TUNEL-stained embryos after homotopic transplanta-tions of DiI-labeled Tcof1�/� hindbrain neuroepithelium in cultured E8.25 wild-type(G)andTcof1�/� (H)embryos.Transplantedcellsexhibit significantapoptosis(arrows in G and H).

13406 � www.pnas.org�cgi�doi�10.1073�pnas.0603730103 Dixon et al.

analyzed proliferation in cultured E8.5–E9.0 embryos that werepulsed with BrdU. Analyses of BrdU-labeled embryos clearlydemonstrated a significant reduction in neuroepithelial cell prolif-eration in Tcof1�/� embryos compared with wild-type littermates(Fig. 5 A and B). Although this finding was not surprising consid-ering the high level of apoptosis observed in the neural plate, wealso discovered that there was reduced proliferation in the neuralcrest-derived cranial mesenchyme. It is important to note, however,that there may be decreased proliferation in the paraxial mesoderm,which also contributes to the cranial mesenchyme. In contrast,proliferation in the cranial endoderm was not affected, whichsuggests that the proliferation anomaly was specific to the neuro-epithelium and neural crest in Tcof1�/� embryos (Fig. 5 A and B).These observations demonstrate that Treacle is dynamically re-quired for cell proliferation in the neuroepithelium and also in theneural crest, which correlates with the specific expression pattern ofTcof1 during embryogenesis.

Consistent with the nucleolar localization of Treacle, the phos-phoprotein encoded by Tcof1 (17), we hypothesized that Treaclemay regulate neural crest cell proliferation by playing key roles inribosome biogenesis. The final stage of ribosome maturation occursas ribosomal subunits such as 28S rRNA are processed from rRNAprecursors and transferred to the cytoplasm. Hence, we analyzedwild-type and mutant embryos by immunohistochemistry with themonoclonal anti-rRNA antibody Y10B (18, 19), which serves as amarker of mature ribosomal integrity by specifically labeling the 28Ssubunit of rRNA (18, 20) (Fig. 5C). We observed significantlyreduced Y10B immunoreactivity in the neuroepithelium and neuralcrest-derived craniofacial mesenchyme of E8.75–E9 Tcof1�/� em-bryos as compared with wild-type littermates (Fig. 5 C and D). Incontrast, no changes were observed in the surface ectoderm orendoderm. These data suggest that Treacle plays a critical spatio-temporally specific role in regulating the production of matureribosomes in the neuroepithelium and neural crest.

Treacle Functions Cell-Autonomously. Our data suggest a model inwhich haploinsufficiency of Tcof1 leads to insufficient matureribosome biogenesis specifically in neuroepithelial cells and neuralcrest cells. As a direct result, the proliferative capacity of neuro-epithelial and neural crest cells is compromised. The spatiotempo-ral correlation between these defects and the domain of activity ofTcof1 predicts that Treacle functions in a cell-autonomous manner.Therefore, to test our prediction we homotopically transplantedsmall focal regions of DiI-labeled midbrain and hindbrain recipro-cally between E8.5 wild-type and Tcof1�/� mutant embryos (Fig. 5E and F). The viability of the grafted cells was then assayed byTUNEL staining after 6- to 8-h periods of in vitro whole-embryoculture. Wild-type DiI-labeled midbrain or rhombomere 2 cellstransplanted homotopically into wild-type isochronic host embryosincorporated into the neural plate and generated neural crest cellswhile exhibiting minimal apoptosis as evidenced by the absence ofTUNEL staining (Fig. 5E). Similarly, wild-type midbrain or rhom-bomere 2 cells were also viable when grafted homotopically intoTcof1�/� isochronic host embryos (Fig. 5F). Transplanted wild-typecells incorporated into the neural plate and generated neural crestcells despite being surrounded by a sea of endogenous apoptosiswithin the host Tcof1�/� neural plate. These observations imply thatTreacle functions in a cell-autonomous manner.

We confirmed the cell-autonomous function of Treacle by trans-planting Tcof1�/� midbrain or rhombomere 2 cells homotopicallyinto wild-type and Tcof1�/� isochronic host embryos (Fig. 5).Tcof1�/� neuroepithelial cells incorporated into the neural plate ofwild-type and Tcof1�/� host embryos, and in both cases a significantproportion of the transplanted cells exhibited elevated levels ofapoptosis. In the wild-type host environment a focal region ofapoptosis was observed coinciding with the DiI-labeled trans-planted Tcof1�/� cells (Fig. 5G). In the Tcof1�/� mutant hostenvironment, despite high levels of endogenous apoptosis through-

out the neural plate, extensive DiI and TUNEL costaining wasobserved at the axial level of transplantation (Fig. 5H). Therefore,Treacle plays key cell-autonomous roles in neural crest cell gener-ation and proliferation through its critical role in regulating theproduction of mature ribosomes that are integral to these processes.

DiscussionHaploinsufficiency of Tcof1 in mice causes major craniofacialmalformations, including hypoplasia of the frontonasal and maxil-lary regions, cleft palate, and mandibular hypoplasia. On a mixedDBA�C57 background Tcof1�/� embryos reproducibly recapitu-late the midfacial anomalies characteristic of severe cases of TCS inhumans, and using this mouse model we have uncovered itsmechanistic etiology and pathogenesis. Our results clearly revealthat the cranioskeletal hypoplasia that is characteristic of TCScraniofacial abnormalities arises as a direct result of a deficiency inthe number of cranial neural crest cells, which occurs because ofdefects in neural crest cell formation and proliferation. Tcof1�Treacle elicits its proliferation function cell-autonomously throughdynamically regulating the spatiotemporal production of matureribosomes in neuroepithelial and neural crest cells.

In support of a link between Tcof1 and ribosome biogenesis,biochemical evidence has recently demonstrated that Treacle bindsupstream binding factor, which, together with promoter selectivityfactor SL1, forms a complex that is important for the activity ofRNA polymerase I (RNA pol I in Fig. 8, which is published assupporting information on the PNAS web site) (21). Interestingly,we observed a reduction in Ubf activity in Tcof1�/� embryoscompared with wild-type littermates (data not shown), suggestingthat Treacle may regulate proliferation directly through Ubf. Be-cause this is a key component and rate-limiting step of ribosomebiogenesis (22), it strongly supports a cell proliferation function forTcof1�Treacle. Furthermore, down-regulation of Treacle by usingspecific short interfering RNA in HeLa cell culture assays resultedin the inhibition of ribosomal DNA transcription and cell prolifer-ation (21). These in vitro cell culture data correlate well with our invivo observations in Tcof1�/� embryos and validate a role forTreacle in driving the formation and proliferation of neural creststem cells through the regulation of ribosome biogenesis. Interest-ingly, it has been estimated that in proliferating cells up to 95% ofall transcription is dedicated to ribosome biogenesis (23).

The revelation that Treacle is a critical regulator of neural crestcell formation and proliferation and that genetically it functionsbetween BMP�WNT and Snail1 is significant because BMP, WNT,and Snail have all been shown to regulate the cell cycle duringneural crest cell induction. The formation and delamination ofneural crest cells depend on the successful transition from G1 to Sphase, and blocking BMP, WNT, or Snail1 signaling disrupts thistransition, resulting in the inhibition of neural crest cell formationand delamination (17, 24) This finding argues in favor of a func-tional role for Treacle and the generation of mature ribosomes inthe control of cell-cycle progression between BMP�WNT andSnail1. In support of this idea, a recent study in frog oocytesdemonstrated that Treacle methylates pre-RNA during G1 (25).Thus, in the absence of Treacle, insufficient ribosome biogenesisrestricts cell-cycle progression, causing reduced proliferation, cell-cycle arrest, and apoptosis.

In summary, TCS is a rare congenital birth defect caused bymutations in TCOF1 and whose characteristic craniofacial abnor-malities arise uniquely as a consequence of a specific spatiotem-poral disruption of ribosome biogenesis in the neural plate andneural crest cells. The lack of mature ribosomes compromises theability of neuroepithelial cells to proliferate and leads to neuroep-ithelial apoptosis. Furthermore, this anomaly results in a deficiencyin the number of migrating cranial neural crest cells, the effect ofwhich is compounded by their reduced proliferative capacity. Theseeffects are caused cell-autonomously through haploinsufficiency ofTcof1, which highlights the essential novel roles played by Treacle

Dixon et al. PNAS � September 5, 2006 � vol. 103 � no. 36 � 13407

DEV

ELO

PMEN

TAL

BIO

LOG

Y

in the formation and proliferation of neural crest cells through itsinvolvement in generating mature ribosomes and possibly promot-ing cell-cycle progression. The early mechanistic onset of TCSabnormalities during the formation and proliferation phases ofneural crest cell development highlights the difficulty in detectingand treating the problem in humans. However, our observationspotentially open up avenues for rescuing TCS abnormalities byenhancing cell-cycle progression and inhibiting neuroepithelialapoptosis.

Materials and MethodsGenotyping of Mouse Embryos. Congenic DBA Tcof1 heterozygotesand wild-type C57BL�6 mice were bred to generate F1 progeny ona mixed DBA � C57BL�6 background. Mutant embryos wereobtained by timed matings, the morning of the vaginal plug beingconsidered E0.5. Genotyping was performed by using previouslydescribed methods (13).

Skeletal, Histological, and Expression Analyses. Skeletal staining,histological analysis, and whole-mount and section in situ hybrid-ization were performed as described previously (13, 26, 27).

Isolation, Culture, Labeling, and Transplantation of Embryos. E8.25–E8.75 embryos were dissected, cultured, and labeled, and tissuetransplants were all performed as previously described (28, 29). Aminimum of 10 wild-type and Tcof1 heterozygous embryos wereused in each of the labeling and transplantation experimentsdescribed.

Flow Cytometry and Neural Crest Cell Quantification. DBA Tcof1heterozygotes were bred with C57BL�6 Pax3GFP homozygousmice. Pax3GFP is a phenotypically normal targeted knockin thatlabels the neuroepithelium and migrating cranial neural crest cellswith GFP. E9 embryos were harvested, and, after a brief incubationof the cranial region in 1 mg�ml dispase at room temperature,tungsten needles were used to completely remove the cranialneuroepithelium. The remaining cranial tissue was incubatedbriefly in 0.25% trypsin with EDTA at 37°C and gently trituratedto generate a single-cell suspension. Single-cell suspensions wereresuspended in PBS with 2% FBS and incubated with 2 �g�ml7-aminoactinmycin D for 5 min on ice to discriminate live from deadcells. The 488-nm laser-excited fluorescence was measured by usinga CyAn flow cytometer (DakoCytomation, Fort Collins, CO). TheGFP log signal was detected by using FL1 (530�40 nm) PMT, andthe 7-aminoactinmycin D log signal was evaluated by using FL4(680�30 nm) PMT. The data were analyzed by using FlowJosoftware (Tree Star, Ashland, OR) with a gating strategy thatremoved nonnucleated and 7-aminoactinmycin D positively stained

events before enumerating the percentage of cells that were ex-pressing GFP. This percentage represents the number of migratingneural crest cells as a proportion of the total number of cells in thecraniofacial mesenchyme in individual wild-type (n � 7) and Tcof1heterozygous mutant (n � 4) embryos.

Analyses of Cell Death. E7.5–E10.5 embryos were assessed forneuroepithelial cell death by TUNEL staining by using the FITCCell Death Detection Kit (Roche, Palo Alto, CA) according to themanufacturer’s instructions. E8–E9.5 embryos were also assessedfor neuroepithelial cell death by wax section immunostaining withan anti-caspase3 antibody (R & D Systems, Minneapolis, MN).Briefly, embryos were fixed in 4% paraformaldehyde and paraffin-processed routinely. Embryos were sectioned sagittally at 10–12�m, deparaffinized, and incubated with a 1:50 dilution of caspase3.Sections were counterstained with hematoxylin.

Analysis of Proliferation. E8.5 embryos were cultured in mediumcontaining 10 �g��l BrdU (Roche) for �2 h. For immunohisto-chemistry studies embryos were fixed in 4% paraformaldehyde,paraffin-processed, and sectioned at 10–12 �m in transverse planes.BrdU incorporation was revealed by using a standard antigen-retrieval method with citrate buffer (pH 6.0), and the slides weresubsequently incubated with a 1:50 dilution of mouse anti-BrdU(Amersham Pharmacia Biosciences, Piscataway, NJ). Detection ofBrdU signal was accomplished by using Alexa Fluor-conjugatedanti-mouse antibody (Molecular Probes, Carlsbad, CA) and thencounterstained with DAPI by using established protocols.

Analysis of rRNA Integrity. Deparaffinized transverse sections(10–12 �m) prepared from E8.5 embryos were processed forimmunohistochemical staining with Y10B, a mouse monoclonalantibody to rRNA (19). To determine the extent of rRNA degra-dation, the slides were incubated with a 1:500 dilution of superna-tant from the Y10B hybridoma and subsequently a 1:200 dilutionof biotinylated horse anti-mouse antibody (Vector Laboratories,Burlingame, CA) as described previously (19).

We thank Dr. Robb Krumlauf and the M.J.D. and P.A.T. groups forcomments on the manuscript. We greatly appreciate the technicalassistance of Sharon Beckham, Teri Johnson, and Mike Morgan. Weacknowledge Dr. Rubel (University of Washington, Seattle, WA) forgenerously providing the Y10B antibody. Research in the P.A.T. labo-ratory is supported by March of Dimes Research Grant 6FY05-82,National Institute of Dental and Craniofacial Research Grant R01 DE016082-01, the Hudson Foundation, and the Stowers Institute forMedical Research. Research in the M.J.D. laboratory is supported byNational Institutes of Health Grant P50 DE 016215 and MedicalResearch Council (U.K.) Grant G81�535.

1. Trainor, P., Bronner-Fraser, M. & Krumlauf, R. (2004) in Handbook of Stem Cells:Embryonic Stem Cells, eds. Lanza, G., Weissman, I., Thomson, J., Pedersen, R.,Hogan, B., Gearhart, J., Blau, H., Melton, D., Moore, M., Verfaillie, C., et al.(Academic, Boston), pp. 219–232.

2. Selleck, M. A. & Bronner-Fraser, M. (1995) Development (Cambridge, U.K.) 121,525–538.

3. Bronner-Fraser, M. & Fraser, S. (1989) Neuron 3, 755–766.4. Moury, J. D. & Jacobson, A. G. (1990) Dev. Biol. 141, 243–253.5. Rollhauser-ter Horst, J. (1977) Anat. Embryol. 151, 309–316.6. Cano, A., Perez-Moreno, M. A., Rodrigo, I., Locascio, A., Blanco, M. J., del Barrio,

M. G., Portillo, F. & Nieto, M. A. (2000) Nat. Cell Biol. 2, 76–83.7. Gans, C. & Northcutt, R. (1983) Science 220, 268–274.8. Jones, N. C. & Trainor, P. A. (2004) Expert Opin. Biol. Ther. 4, 645–657.9. Treacher Collins Syndrome Collaborative Group (1996) Nat. Genet. 12, 130–136.

10. Poswillo, D. (1975) Br. J. Oral Surg. 13, 1–26.11. Wiley, M. J., Cauwenbergs, P. & Taylor, I. M. (1983) Acta Anat. (Basel) 116, 180–192.12. Herring, S. W., Rowlatt, U. F. & Pruzansky, S. (1979) Am. J. Med. Genet. 3, 225–229.13. Dixon, J., Brakebusch, C., Fassler, R. & Dixon, M. J. (2000) Hum. Mol. Genet. 9,

1473–1480.14. Dixon, J. & Dixon, M. J. (2004) Dev. Dyn. 229, 907–914.15. Kanzler, B., Foreman, R. K., Labosky, P. A. & Mallo, M. (2000) Development

(Cambridge, U.K.) 127, 1095–1104.16. Ittner, L. M., Wurdak, H., Schwerdtfeger, K., Kunz, T., Ille, F., Leveen, P., Hjalt,

T. A., Suter, U., Karlsson, S., Hafezi, F., et al. (2005) J. Biol. 4, 11.

17. Marsh, K. L., Dixon, J. & Dixon, M. J. (1998) Hum. Mol. Genet. 7, 1795–1800.18. Lerner, E. A., Lerner, M. R., Janeway, C. A., Jr., & Steitz, J. A. (1981) Proc. Natl.

Acad. Sci. USA 78, 2737–2741.19. Garden, G. A., Canady, K. S., Lurie, D. I., Bothwell, M. & Rubel, E. W. (1994)

J. Neurosci. 14, 1994–2008.20. Garden, G. A., Hartlage-Rubsamen, M., Rubel, E. W. & Bothwell, M. A. (1995) Mol.

Cell. Neurosci. 6, 293–310.21. Valdez, B. C., Henning, D., So, R. B., Dixon, J. & Dixon, M. J. (2004) Proc. Natl.

Acad. Sci. USA 101, 10709–10714.22. Poortinga, G., Hannan, K. M., Snelling, H., Walkley, C. R., Jenkins, A., Sharkey,

K., Wall, M., Brandenburger, Y., Palatsides, M. & Pearson, R. B., et al. (2004)EMBO J. 23, 3325–3335.

23. Martin, D. E., Soulard, A. & Hall, M. N. (2004) Cell 119, 969–979.24. Burstyn-Cohen, T., Stanleigh, J., Sela-Donenfeld, D. & Kalcheim, C. (2004) Devel-

opment (Cambridge, U.K.) 131, 5327–5339.25. Gonzales, B., Henning, D., So, R. B., Dixon, J., Dixon, M. J. & Valdez, B. C. (2005)

Hum. Mol. Genet. 14, 2035–2043.26. Kaufman, M. H. (1992) The Atlas of Mouse Development (Academic, London).27. Nagy, A., Gertsenstein, M., Vintersten, K. & Behringer, R. R. (2003) Manipulating

the Mouse Embryo (Cold Spring Harbor Lab. Press, Cold Spring Harbor, NY).28. Trainor, P. A. & Tam, P. P. L. (1995) Development (Cambridge, U.K.) 121,

2569–2582.29. Trainor, P. & Krumlauf, R. (2000) Nat. Cell Biol. 2, 96–102.

13408 � www.pnas.org�cgi�doi�10.1073�pnas.0603730103 Dixon et al.