Skip Navigation


Human Molecular Genetics Advance Access originally published online on February 9, 2005
Human Molecular Genetics 2005 14(7):903-912; doi:10.1093/hmg/ddi083
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Material
Right arrow All Versions of this Article:
14/7/903    most recent
ddi083v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (7)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Cai, J.
Right arrow Articles by Lovett, M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Cai, J.
Right arrow Articles by Lovett, M.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2005. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions{at}oupjournals.org

Gene expression in pharyngeal arch 1 during human embryonic development

Juanliang Cai1, David Ash1, Lori E. Kotch1, Ethylin Wang Jabs1,*, Tania Attie-Bitach2, Joelle Auge2, Geraldine Mattei2, Heather Etchevers2, Michel Vekemans2, Yulia Korshunova3, Rose Tidwell3, David N. Messina3, Julia B. Winston3 and Michael Lovett3

1Institute of Genetic Medicine, Johns Hopkins University, Baltimore, MD 21205, USA, 2Department of Genetics and INSERM U-393, Hopital Necker Enfants Malades, 75743 Paris, France and 3Department of Genetics, Washington University School of Medicine, St Louis, MO 63110, USA

* To whom correspondence should be addressed at: McKusick-Nathans Institute of Genetic Medicine, The Johns Hopkins University, 733 N Broadway, Room 419, Baltimore, MD 21205, USA. Tel: +1 4109554160; Fax: +1 4105025677; Email: ejabs1{at}jhem.jhmi.edu

Received December 3, 2004; Accepted February 1, 2005


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
Craniofacial abnormalities are one of the most common birth defects in humans, but little is known about the human genes that control these important developmental processes. To identify relevant genes, we analyzed transcription profiles of human pharyngeal arch 1 (PA1), a conserved embryonic structure that develops into the palate and jaw. Using microdissected, normal human craniofacial structures, we constructed 12 SAGE (serial analysis of gene expression) libraries and sequenced 606 532 tags. We also performed Affymetrix microarray analysis on 25 craniofacial targets. Our data revealed not only genes ‘enriched’ or differentially expressed in PA1 during fourth and fifth week of human development, but also 6927 genes newly identified to be expressed in human PA1. Many of these genes are involved in biosynthetic processes and have binding function and catalytic activity. We compared expression profiles of human genes with those of mouse homologs to look for genes more specific to human craniofacial development and found 766 genes expressed in human PA1, but not in mouse PA1. We also identified 1408 genes that were expressed in mouse as well as human PA1 and could be useful in creating mouse models for human conditions. We confirmed conservation of some human PA1 expression patterns in mouse embryonic samples with whole mount in situ hybridization and real-time RT–PCR. This comprehensive approach to expression profiling gives insights into the early development of the craniofacial region and provides markers for developmental structures and candidate genes, including SET and CCT3, for diseases such as orofacial clefting and micrognathia.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
Craniofacial abnormalities, such as orofacial clefting, micrognathia, hemifacial microsomia, mandibulofacial dysostosis and craniosynostosis, are among the most common malformations in humans with frequencies as high as approximately one per 1000 live births (1Go). More than 700 craniofacial disorders have been reported (www.ncbi.nlm.nih.gov/omim). Many affected structures, including the palate and jaw, are derived from pharyngeal arches. Therefore, these malformations may be due to abnormalities or perturbations of pharyngeal arches during the first 2 months of human embryonic development. Elucidation of the etiology of these malformations depends upon knowledge of normal patterns of gene expression during the development of pharyngeal arches.

Pharyngeal arches are a prominent feature at the cephalic end of all vertebrate embryos (2Go). These arches appear as pairs of mesenchymal structures during fourth and fifth week of human development. Cells from all three germ layers and neural crest contribute to pharyngeal arch development. Ectoderm is the origin of arch-associated epidermis and sensory neurons and induces odontogenesis (3Go), whereas mesoderm develops into striated musculature and endothelial cells of the arch arteries. Endoderm gives rise to the epithelial lining of the pharynx and provides signaling molecules necessary for the development of the thymus, parathyroid and thyroid. Neural crest cells migrate from the edges of the neural folds at the levels of the lower midbrain and rhombomeric subdivisions of the hindbrain to populate pharyngeal arches (4Go,5Go). Subsequently, the first arch subdivides into maxillary and mandibular portions, which give rise to the palate and jaw, respectively. The maxillary prominence contributes to the formation of the upper midface and palate through interactions with the frontonasal prominence.

Relatively, little is known about the details of the early molecular processes during pharyngeal arch development. HOX genes are critical for the development of the neural crest (6Go). The signaling molecule BMP4 (7Go) and the class II homeobox gene MSX2 (8Go) expressed in migrating neural crest cells are involved in apoptotic separation of individual arches. Neural crest independent mechanisms of pharyngeal arch development have also been suggested, on the basis of expression patterns of molecules such as Bmp7, Fgf8, Pax1 and Shh (9Go).

To generate the first comprehensive gene expression profiles during early human craniofacial development, we utilized micro-cDNA technologies for serial analysis of gene expression (SAGE) (10Go) and Affymetrix microarrays. Both of these methods have been successfully used to identify differential gene expression between normal and disease states, but have rarely been used for human embryonic development (11Go,12Go). More importantly, because SAGE does not require a priori knowledge of the existence of a gene, we could identify novel genes. We focused our analysis on human pharyngeal arch 1 (PA1), because more craniofacial structures are derived from this pair of arches than the others. In situ hybridization and RT–PCR data were used on mouse PA1 samples to study conservation and differences between species. The data from our study also identified biological markers for pharyngeal arch development and candidate genes for cleft palate and micrognathia.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
Characterization of human PA1 SAGE libraries and microarray data
We constructed 12 SAGE libraries from various microdissected structures of early human craniofacial development (Table 1). For each library, we used only 1 µg of total RNA in our micro-SAGE protocol (13Go). A total of 606 352 tags were sequenced, with an average of 50 531 tags sequenced for each library, and 101 705 (16.7%) unique tags were identified. These libraries are of high quality because 1Go) there was no GC content bias (14Go), 2Go) the frequency of duplicate ditags and linker contamination (both <0.5%) was equivalent to or less than that of other reported SAGE libraries (15Go), and 3Go) simulation analysis using R (www.r-project.org) indicated that subpopulations of data from the same SAGE library were similar.


View this table:
[in this window]
[in a new window]
  		Table 1. SAGE libraries and microarray data generated from human embryonic craniofacial structures
 
Comparison between libraries provided further evidence of the quality of our SAGE libraries. In replicate libraries derived from independent RNA sources of microdissected fourth week frontonasal prominences, we found that the correlation coefficient was 0.96, and only 0.39% of the tags showed changes >2-fold (P<0.05). In contrast, comparison between fourth week pharyngeal arch 1 (W4PA1) and fifth week pharyngeal arch 1 (W5PA1) libraries showed that the correlation coefficient was smaller (0.86), and significantly more tags (1.6%) showed changes >2-fold (P<0.05).

Tag sequences from W4PA1 and W5PA1 libraries were matched to the SAGEmap database (www.ncbi.nlm.nih.gov/sage) for gene identification. In PA1, a few genes (<5%) are expressed at high levels and most are expressed at moderate to low levels (Supplementary Material, Table S1). This distribution is consistent with that observed in other tissues and cell types (16Go–18Go). Of the 19 698 and 21 881 unique SAGE tags from W4PA1 and W5PA1 libraries, respectively, ~24% were novel without matches to UniGene clusters. Of the SAGE tags that matched to UniGene clusters, ~40% matched to more than one cluster. The low specificity of a SAGE tag for a specific gene was largely because of the short length (10 bp) of the SAGE tag sequence.

We also performed microarray analysis on 25 craniofacial structures, using the Affymetrix Human Genome U95Av2 chip (Table 1). By hierarchical clustering (19Go), duplicates of the same PA1 targets were found to cluster together in the same subnode with a correlation coefficient of 0.97, showing reproducibility between duplicate arrays. For W4PA1 and W5PA1, we found that 29–39% of all genes tested were expressed in PA1; of the total 12 600 probe sets on a chip, 3681 and 4869 are expressed in W4PA1 and W5PA1, respectively. We consider that a gene is expressed if there is a ‘present’ call by the Affymetrix default algorithm, which has a sensitivity of at least 70% of detecting genes expressed at the level of one per 100 000 transcripts.

Genes identified to be expressed in PA1 by SAGE or microarray were assigned to 12 different molecular function categories, on the basis of information from the Gene Ontology Consortium (www.geneontology.org). The distribution of genes in the different categories was similar for both W4PA1 and W5PA1 (Table 2; Supplementary Material, Table S2). Genes classified into binding or catalytic activity groups accounted for 61.2–62.3%, with genes of transporter, signal transducer and transcription regulator activity groups accounting for 21.4–25.0%. The remaining genes were in structural molecule, enzyme regulator, chaperone, translation regulator, motor and antioxidant activity groups.


View this table:
[in this window]
[in a new window]
  		Table 2. Functional classification of PA1 genes
 
Comparison between W4PA1 and W5PA1 libraries
To detect human genes that are differentially expressed between fourth and fifth week of PA1 development, we analyzed both SAGE libraries [using the statistical method of Audic and Claverie (20Go), fold change >2 and P-value <0.01] and Affymetrix data [using the statistical method of Li and Wong (19Go) and lower boundary fold change >2]. We identified 97 genes differentially expressed between human W4PA1 and W5PA1 by SAGE (Supplementary Material, Table S3) and 62 genes by Affymetrix microarrays (Supplementary Material, Table S4). The distribution of differentially expressed genes assigned to different molecular functional categories was similar to all genes expressed in both W4PA1 and W5PA1 libraries (Supplementary Material, Table S2). These differentially expressed genes were also assigned to different biological process categories (21Go). Several biosynthesis processes, including nucleotide, protein and lipid synthesis processes, as well as the protein transport and the cell cycle processes were overrepresented in the differentially expressed genes determined by SAGE (P<0.05; Table 3), whereas no biological processes were overrepresented in the differentially expressed genes detected by microarrays.


View this table:
[in this window]
[in a new window]
  		Table 3. Biological processes
 
Few genes were identified to be differentially expressed by both our SAGE and microarray analyses. Only a few published studies were available to corroborate these differentially expressed genes. In the case of human FIGF, we detected increased expression from fourth to fifth week human PA1 SAGE libraries (zero versus eight tags). This result was confirmed by Affymetrix analysis (6.4-fold increase). Furthermore, mouse Figf expression was not detected in mouse PA1 at GD8.5, but was detected in PA1 at GD10.5 by in situ hybridization (22Go).

Direct validation of the expression changes would ideally be done with human samples. However, because of the limited availability of human embryos and generally presumed conservation of gene expression in mammals, mouse embryos were used to further study expression changes in PA1. We used real-time RT–PCR to analyze RNAs from corresponding mouse stages. Mouse PA1 obtained by microdissection of GD9.5 and GD10.5 embryos was analyzed, using primers specific to mouse homologs of several human genes. We found that 93% of the RT–PCR analyses agreed with our SAGE data in the direction of change and 86% had a fold change of ≥2 (Supplementary Material, Table S5). Although there was good agreement between our SAGE and RT–PCR differential gene expression data, complete agreement was not expected because of differences in sensitivity and specificity of the methods, intrinsic variation in gene expression and/or species differences between human and mouse.

Genes highly expressed in human PA1
Of the highly expressed tags present more than 100 times in either W4PA1 or W5PA1 SAGE library, 24 tags were uniquely matched to UniGene clusters (Supplementary Material, Table S6). Thirteen of these were among the 200 genes (~5% of all genes called ‘present’) with the highest normalized intensity values assayed by Affymetrix microarray. Many of these genes code for components of the ribosomal complex, reflecting the increased translational activity in PA1 during early human development. The percentage of highly expressed genes was 4-fold increased in the structural molecule activity group when compared with that of all genes identified in PA1. No highly expressed genes were identified in the transporter, signal transducer and transcription regulator activity groups.

Genes ‘enriched’ in human PA1
To screen for human genes that are at least 2-fold ‘enriched’ in PA1, we calculated the ratio of the number of tags from either W4PA1 or W5PA1 SAGE library to the average number of tags from 10 libraries derived from other embryonic, craniofacial structures. For a gene to be considered ‘enriched’, both ratios had to be greater than 2. Only moderately to highly expressed SAGE tags with a combined count of more than 20 in all libraries were analyzed. We also searched for genes at least 2-fold ‘enriched’ in W4PA1 and W5PA1 by analyzing our Affymetrix microarray data. We found 74 ‘enriched’ genes by SAGE (Supplementary Material, Table S7) and 96 genes by Affymetrix microarrays (Supplementary Material, Table S8). The proportion of ‘enriched’ genes assigned to different functional categories was also similar to that of all genes expressed in PA1 (Supplementary Material, Table S2). No biological processes were overrepresented in the SAGE ‘enriched’ genes, whereas the biosynthesis and energy pathway processes were overrepresented in microarray ‘enriched’ genes (Table 3). Some of these ‘enriched’ genes are known to be involved in human disorders affecting the craniofacial region such as GPC3 in Simpson-Golabi-Behmel syndrome type I (23Go,24Go), NOTCH3 in cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (25Go) and TNFRSF10B in head and neck squamous cell carcinoma (26Go).

Genes common to human and mouse PA1
To compare the gene expression between human and mouse craniofacial structures, we searched The Jackson Laboratory Mouse Genome Informatics database (www.informatics.jax.org) for previously reported mouse gene expression in PA1 and its quantitation. Three hundred and fifteen genes have been shown to be expressed in PA1 by immunohistochemistry, in situ hybridization or RT–PCR. Of these 315 genes, 228 genes have human homologs that matched uniquely to UniGene clusters. One hundred and seventy-nine (78%) were found in our SAGE analysis of human W4PA1 and W5PA1 libraries. Thirty-two percent of these genes were classified as signal transducers and transcription regulators such as HoxA1, Msx1, Msx2, Pax8, Pax9, Bmp4, Bmp5, Bmp7, Fgfr11 and Tgfb1 (Table 2). Of the 228 human genes, 174 have probe sets on Affymetrix HG-U95Av2 chip. One hundred and six (61%) were called ‘present’, when human PA1 targets were analyzed. Ninety-six genes were detected by both unique SAGE tags and microarray analysis. Both multiple matched SAGE tags and Affymetrix microarray data identified additional genes, such as TWIST and TCOF1, which are known to be involved in craniofacial disorders affecting pharyngeal arch development in humans (Saethre-Chotzen and Treacher Collins syndromes, respectively) and relevant mouse models (27Go–30Go).

To discover additional PA1 genes expressed in human and mouse, we performed microarray analysis on mouse PA1 microdissected at GD9.5 and compared these results with our human data. We identified an additional 7307 genes expressed in mouse PA1. Of all the genes expressed in human PA1, 2174 have mouse homologs and 1408 (64.8%) were also expressed in mouse PA1 as determined either by microarray analysis or present in The Jackson Laboratory Mouse Genome Informatics database (Supplementary Material, Table S9). These results further suggest that the large portion of genes and molecular pathways that are relatively conserved between human and mouse PA1 development account for the basic craniofacial structures present in mammals.

Temporospatial expression of mouse homologs of human genes expressed in PA1
To screen for genes that may be involved in craniofacial disorders and to investigate the temporospatial expression pattern of PA1 genes, we identified genes that are more highly expressed in pharyngeal arches than that in other embryonic, craniofacial structures as listed in Table 1. The mouse homologs for human genes previously not known to be expressed in PA1 were studied by whole mount in situ hybridization analysis using mouse embryos from GD8.5 through GD10.5, the stages corresponding to fourth and fifth week human embryonic development.

One of these is the SET gene, which was previously identified because it was disrupted by a translocation breakpoint in human chromosome 9 associated with a subtype of acute myeloid leukemia (31Go). Human SET, matched by multiple SAGE tags, was present in 75 of 59 959 tags and 101 of 67 982 tags sequenced in W4PA1 and W5PA1 libraries, respectively. This gene was relatively ‘enriched’ in PA1, because it was present on average only 47 of 50 531 tags in all other libraries. Its PA1 expression was further confirmed by both human and mouse Affymetrix microarray analysis. By whole mount in situ hybridization, we found that at GD8.5, mouse Set was highly expressed in cells of the neural crest folds and less expressed in PA1 (Fig. 1). At GD9.5, a high level of Set expression was in the pharyngeal arches, especially in PA1, suggesting that the Set-expressing neural crest cells had migrated from the hindbrain folds to the arches. Upon further development at GD10.5, Set expression was still strong, but restricted to the maxillary portion of PA1. During this stage, Set was also expressed in the frontonasal prominence, specifically in the lateral nasal prominence and, interestingly, in the limb buds.



View larger version (58K):
[in this window]
[in a new window]
  		Figure 1. Mouse in situ hybridization of the Set gene. Whole embryo staining for GD8.5 (A and B), GD9.5 (C and D) and GD10.5 (E and F) shows Set expression throughout early craniofacial development. (B, D and F) are the higher magnifications of the boxed regions in (A, C and E), respectively. Set expression on GD8.5 is localized to the frontonasal prominence, crest of the neural folds and PA1. At GD9.5, expression is present throughout PA1 and also at the posterior half of PA2. By GD10.5, Set is expressed mostly in the maxillary portion of PA1. Set is expressed in the frontonasal prominence at all stages observed (B, D and F). At GD10.5, Set is expressed in the lateral nasal prominence (F). Set is expressed in the limb bud (C). {ddi08303}, PA1; {bigtriangleup}, frontonasal prominence; {blacktriangledown}, neural folds; {ddi08304}, PA2; {ddi08305}, limb bud.

 
Another gene that we studied was CCT3, which encodes a chaperone protein (32Go). Human CCT3, matched by multiple SAGE tags, was present in 338 of 59 959 tags and 230 of 67 982 tags sequenced in W4PA1 and W5PA1 libraries, respectively. This gene was highly expressed in PA1, being present on average only 91 of 50 531 tags in all other libraries. CCT3 was also determined to be ‘enriched’ in both PA1 and PA2 using a Wilcoxon rank sum test (33Go), in which the normalized expression level of this gene in W4PA1, W5PA1, W4PA2 and W5PA2 was at least twice that in the other eight libraries tested with a one-sided P-value <0.05. Furthermore, Affymetrix microarray data on both human and mouse confirmed that CCT3 was highly expressed (top 5% of genes) in PA1. In whole mount in situ hybridization at GD8.5, mouse Cct3 was predominantly expressed in PA1, as well as in the frontonasal prominence (Fig. 2). Its level of expression in PA1 and the frontonasal prominence persisted, and it was also present in PA2. By GD10.5, the predominant expression of Cct3 had moved to the mandibular portion of PA1 and the limb buds.



View larger version (62K):
[in this window]
[in a new window]
  		Figure 2. Mouse in situ hybridization of the Cct3 gene. Whole embryo staining for GD8.5 (A and B), GD9.5 (C and D) and GD10.5 (E and F) shows Cct3 expression throughout early craniofacial development. (B, D and F) are the higher magnifications of the boxed regions in (A, C and E), respectively. Cct3 expression on GD8.5 is localized to PA1 and frontonasal prominence. At GD9.5, Cct3 expression is present in the frontonasal prominence and is predominant in PA1. By GD10.5, Cct3 expression is observed throughout PA1, but most prominently in the mandibular portion of PA1. It is also expressed in PA2 (D and F) and frontonasal prominence (B, D and F) at these stages. Cct3 is expressed in the limb bud (C and E)., PA1; {bigtriangleup}, frontonasal prominence;, PA2;, limb bud.

 
Genes newly identified to be expressed in human PA1
We identified 6927 genes previously not known to be expressed in human PA1. Of these genes, 4842 and 6096 were detected by SAGE and microarray, respectively. In addition, these genes had not been identified to be expressed in mouse PA1 by The Jackson Laboratory Mouse Genome Informatics database (www.informatics.jax.org). Among these new genes are all of the highly expressed (Supplementary Material, Table S6) and most of the differentially expressed or the ‘enriched’ human PA1 genes (Supplementary Material, Tables S3, S4, S7 and S8). A comparison was made between our human versus mouse microarray data on PA1, using the default Affymetrix algorithm of present/absent call. We found that 35% (766 of 2174 mouse homologs) were expressed in only human PA1 (Table 4), whereas 40% (940 of 2348 human homologs) were expressed in only mouse PA1 (Supplementary Material, Table S10). The distribution of these genes that are more specific to human or mouse into molecular functional categories is similar to all genes identified in PA1 (Table 2). Therefore, species differences in gene expression levels as well as timing and localization, rather than function, may account for some of the obvious morphological differences of the craniofacial region at later stages. However, further studies will need to be performed to substantiate this hypothesis. Our data also suggest the limitation of using mouse models to study some human conditions and provide a means by which genes can be prioritized for mouse knockout studies.


View this table:
[in this window]
[in a new window]
  		Table 4. Genes newly identified to be expressed in human PA1 by SAGE and microarray
 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
This study is unique in that gene expression profiling was performed on RNAs from normal human embryos. Our analysis allowed us to identify a large number of transcripts previously not known to be expressed in the human craniofacial region. We focused on PA1, because it is essential to the development of many craniofacial structures. Although technically challenging, we were able to microdissect PA1 because of its distinct morphology and generate sufficient cDNA by adapting micromethods using very small quantities of total RNAs.

Previous studies of gene expression during normal craniofacial development have been carried out in mouse or zebrafish, using microarrays (34Go) and phenotype-based techniques such as ENU mutagenesis (35Go) or subtractive libraries (36Go). Microarray results were restricted to those genes present on a chip or filter and may not be as sensitive as SAGE or RT–PCR. Both mutagenesis screens and subtractive libraries were also less comprehensive.

Fowles et al. (36Go) used subtractive hybridization of a mouse PA1 cDNA library against an adult mouse liver cDNA library to identify genes ‘enriched’ in PA1. Their subtraction method aimed to remove ubiquitously expressed housekeeping genes and to enrich for those genes with a specific role in PA1 development. However, genes common to both liver and craniofacial development may still be important to the latter and would be missed. After sequencing 453 clones, 273 non-redundant cDNA clones were identified. One hundred and twenty-three of the latter clones had sequences homologous to unique UniGene clusters. When we compared their results with our data, we found that all were identified in our SAGE and microarray analyses. In addition, we identified thousands of more genes expressed in mouse PA1, as well as in human PA1, by SAGE and microarrays.

We identified genes that are ‘enriched’ in PA1 from fourth to fifth week of human development. These PA1-enriched genes are good candidates for craniofacial diseases. Some of these PA1-enriched genes when mutated could lead to significant craniofacial abnormalities or even result in embryonic lethality. As a validation, some of these genes have already been implicated in several human craniofacial disorders. The presence of some PA1-enriched genes in our libraries that cause human diseases without craniofacial manifestations may be due to their functional redundancy and/or their functional importance in other organ systems (Supplementary Material, Tables S7 and S8).

These profiling data, in combination with in situ hybridization, have also identified genes that might have a role in specific craniofacial abnormalities. The high expression of the Set gene in PA1 and its localization in the maxillary portion by GD10.5 suggests this gene as a candidate for orofacial clefting, because the palate normally forms by the fusion of the maxillary prominences. Similarly, the localization of Cct3 expression to the mandibular portion of the PA1 suggests this gene as a candidate for micrognathia, as well as for other jaw abnormalities. Furthermore, these genes may serve as biological markers for PA1. For example, at GD10.5, cranial derivatives for neural crest cell populations are limited to pharyngeal arches and sensory ganglia of the face (upper only). At this stage, both Set and Cct3 appear to be expressed in neural crest derivatives such as both arches.

Our results show that a large portion (65%) of human genes expressed in PA1 is also expressed in mouse PA1, providing further evidence for the conservation of developmental pathways between species. Real-time RT–PCR on mouse samples confirmed several expression changes detected by human SAGE libraries. These differentially expressed genes may shed light on the pathways involved in temporal development of PA1. In addition, many of these genes are involved in biosynthetic processes of early development. To date, a significant proportion of genes studied in both human and mouse PA1 are signal transducers and transcription regulators, but our functional analysis of all genes identified by our SAGE and Affymetrix data suggest future studies should focus on the genes which have binding or catalytic activities, because they represent the majority of PA1 genes.

We also found that Set and Cct3 are expressed in the limb bud, an observation that may have clinical significance. Many patients with craniofacial syndromes also manifest limb abnormalities. Classic examples include craniosynostosis syndromes such as Pfeiffer, Apert, Crouzon, Greig cephalopolysyndactyly and metabolic defects such as Smith-Lemli-Optiz syndromes (www.ncbi.nlm.nih.gov/omim). Identifying molecules and signaling pathways shared by different organ systems will reveal common factors among diverse birth defects.

We observed limited agreement between our SAGE and microarray data. Only one of the ‘enriched’ genes identified by SAGE was identified by Affymetrix analysis, whereas only two genes that were differentially expressed by SAGE were confirmed by Affymetrix analysis. However, when highly expressed genes were analyzed, there was better agreement. More than half of the transcripts with SAGE tag counts greater than 100 were also among the highly expressed genes by Affymetrix analysis (Supplementary Material, Table S6). These discrepancies are similar to observations reported in other comparative studies (37Go–39Go) and are highlighted by issues of sensitivity (e.g. the number of sequenced SAGE tags, Affymetrix hybridization and detection procedures), specificity (multiple and non-matched SAGE tags and Affymetrix probe sets for alternative transcripts) and analytical methodologies. These results emphasize the need to use more than one global gene expression method and gene-specific expression assays for validation.

This large-scale expression profile of normal human craniofacial development is an important initial step toward elucidating this complex process. Our data can now be used to design microarrays with genes expressed in PA1 to provide further insights into early embryonic development. Correlating gene expression patterns and abnormal craniofacial phenotypes with linkage, association and mutation analyses is a powerful integrated approach to identify the causes of malformations.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
SUPPLEMENTARY MATERIAL
 REFERENCES
 
Human RNA acquisition
Morphologically normal human embryos were obtained through legalized abortions induced by Mifepristove (RU-486), according to the recommendations of the French National Ethics Committee. The developmental stage of each embryo was estimated, according to the Carnegie classification. Common chromosomal abnormalities were excluded by fluorescent quantitative PCR on chromosomes 13, 18, 21, X and Y, which are among the most commonly occurring aneuploidies. Embryos were microdissected from the whole trophoblasts stored in Tyrode's solution. Microdissected structures were suspended in Trizol, and total RNA was isolated according to manufacturer's protocol (Invitrogen, Carlsbad, CA, USA). Structure-specific RNAs collected from three to five different embryos at the same developmental stage were pooled and subjected to gene expression analysis.

Construction and analysis of human SAGE libraries
SAGE libraries were generated using a modified micro-SAGE protocol as previously reported (13Go). We constructed SAGE libraries from 1 to 2 µg of total RNA isolated from pooled microdissected tissues, without any preamplification steps that would potentially compromise the quantitative nature of this method. To increase the yield of SAGE ditags, we use a single-tube procedure for all steps prior to tag release, Dynal magnetic beads (Dynal, Brown Deer, WI, USA) and PhaseLock Gel (Eppendorf AG, Hamburg, Germany). All the SAGE data are available at hg.wustl.edu/cogene. Comparisons between libraries were carried out using the SAGE2000 (www.sagenet.org) and eSAGE softwares (14Go).

Human Affymetrix microarray hybridization and analysis
Human cDNA was generated using the same total RNA sources for SAGE (discussed earlier) with a modified SMART oligo method (hg.wustl.edu/cogene). In vitro transcription was performed with 100 ng of cDNA, using a MEGAscript High Yield Transcription Kit (Ambion, Austin, TX, USA). The resulting RNA was treated with DNase I, passed through a Sephadex G50 column and ethanol precipitated. This RNA was spiked with control RNAs and used for synthesis of cDNA primed with oligo(T)18-T7, according to the Affymetrix GeneChip instructions (Affymetrix, Santa Clara, CA, USA). The resulting DNA was used in an in vitro transcription reaction with labeled ribonucleotides. These targets were used on HG-U95Av2 chips. Hybridization, washing and image scanning were performed, according to the Affymetrix protocol. Duplicate experiments were performed for each target. Estimation, normalization of gene expression values and hierarchical clustering were performed, using dChip software (19Go). All our Affymetrix microarray data are available at hg.wustl.edu/cogene.

Mouse Affymetrix microarray hybridization and real-time RT–PCR analysis
Mouse GD9.5 and GD10.5 embryos were collected. PA1s were microdissected from the embryos, and 10–12 PA1s (each ~200 µm in size and from the same stages) were pooled and stored in RNAlater solution (Invitrogen). Total RNA was extracted using an RNeasy Mini Kit (Qiagen, Valencia, CA, USA) and cDNA was synthesized using an Omniscript RT Kit (Qiagen), according to the manufacturer's protocol for both real-time PCR and Affymetrix microarray hybridization to MOE430A and B chip set. Analyses were performed as for the human chips (discussed earlier).

For real-time PCR, PCR primers were designed with Primer3 software (frodo.wi.mit.edu/) to amplify intron-spanning amplicons. Real-time PCR reactions were carried out with QuantiTect SYBR Master Mix (Qiagen), as described previously (40Go). Specific amplification from cDNA and no amplification from genomic DNA were confirmed by melting curve analysis and subsequent gel electophoresis. External standards were generated using mouse brain Poly(A) RNA (Ambion) and the OmniScript kit for reverse transcription (Qiagen). Ten-fold serial dilutions of the products were made from 1 : 1 to 1 : 10–4 to create a standard curve. Triplicate reactions were performed for both the samples and the standards. Transcript concentrations of both GD9.5 and GD10.5 PA1s were inferred from the standard curve, and fold changes were calculated after normalizing against a reference gene. The Tmsb4x gene (UniGene Hs.75968) was chosen as a reference gene, because it was among the top 1% of genes with respect to constant expression level in all SAGE libraries, and a robust real-time RT–PCR assay could be developed.

Mouse whole mount in situ hybridization
PCR primers, designed with Primer3 software (frodo.wi.mit.edu/), were used to amplify cDNA-specific products, which were sequenced in both directions to ensure no mutations were incorporated. DIG-labeled RNA probes were generated from the PCR products using a DIG RNA transcription kit (Roche, Indianapolis, IN, USA), according to the manufacturer's protocol. Normal C57BL/6J mice were mated for 2 h. GD8.5, GD9.5 or GD10.5 mouse embryos were dissected in Ringer's solution and were fixed in 4% paraformaldehyde in PBS overnight at 4°C. Mouse whole mount in situ hybridization was then performed as described previously (41Go). Images were taken with DXM1200 digital camera (Nikon, Melville, NY, USA).

GO analysis
Functional categories for genes were assigned to top-level GO terms under the ‘molecular function’ hierarchy, on the basis of information from the Gene Ontology Consortium (www.geneontology.org). Gominer software (discover.nci.nih.gov/gominer) was used with data source set to ‘UniProt (H. sapiens et al.)’ and organisms set to ‘H. sapiens’ (42Go). To find statistically overrepresented biological process GO terms, GOstat program was used (gostat.wehi.edu.au) (21Go).


   SUPPLEMENTARY MATERIAL
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
REFERENCES
 
Supplementary Material is available at HMG Online.


   ACKNOWLEDGEMENTS
 
We thank Drs Aravinda Chakravarti, Kenneth Kinzler and Alan F. Scott for helpful discussions on this study and Dr Barbara R. Migeon for suggestions on the manuscript. The work represents the effort of the Craniofacial and Oral Gene Expression Network (COGENE) consortium. This study was supported by National Institutes of Health grants T32 GM07814 (J.C.) and NO1DE92630.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 

  1. Gorlin, R.J., Cohen, M.M., Jr and Hennekam, R.C.M. (2001) Syndromes of the Head and Neck. Oxford University Press, New York.

  2. Graham, A. and Smith, A. (2001) Patterning the pharyngeal arches. Bioessays, 23, 54–61.[CrossRef][ISI][Medline]

  3. Neubuser, A., Peters, H., Balling, R. and Martin, G.R. (1997) Antagonistic interactions between FGF and BMP signaling pathways: a mechanism for positioning the sites of tooth formation. Cell, 90, 247–255.[CrossRef][ISI][Medline]

  4. Lievre, C.L. (1974) Role of mesectodermal cells arising from the cephalic neural crest in the formation of the branchial arches and visceral skeleton. J. Embryol. Exp. Morphol., 31, 453–477.[ISI][Medline]

  5. Noden, D.M. (1983) The role of the neural crest in patterning of avian cranial skeletal, connective, and muscle tissues. Dev. Biol., 96, 144–165.[CrossRef][ISI][Medline]

  6. Trainor, P. and Krumlauf, R. (2000) Plasticity in mouse neural crest cells reveals a new patterning role for cranial mesoderm. Nat. Cell Biol., 2, 96–102.[CrossRef][ISI][Medline]

  7. Graham, A., Francis-West, P., Brickell, P. and Lumsden, A. (1994) The signalling molecule BMP4 mediates apoptosis in the rhombencephalic neural crest. Nature, 372, 684–686.[CrossRef][Medline]

  8. Winograd, J., Reilly, M.P., Roe, R., Lutz, J., Laughner, E., Xu, X., Hu, L., Asakura, T., van der Kolk, C., Strandberg, J.D. et al. (1997) Perinatal lethality and multiple craniofacial malformations in MSX2 transgenic mice. Hum. Mol. Genet., 6, 369–379.[Abstract/Free Full Text]

  9. Veitch, E., Begbie, J., Schilling, T.F., Smith, M.M. and Graham, A. (1999) Pharyngeal arch patterning in the absence of neural crest. Curr. Biol., 9, 1481–1484.[CrossRef][ISI][Medline]

  10. Velculescu, V.E., Zhang, L., Vogelstein, B. and Kinzler, K.W. (1995) Serial analysis of gene expression. Science, 270, 484–487.[Abstract/Free Full Text]

  11. Richards, M., Tan, S.P., Tan, J.H., Chan, W.K. and Bongso, A. (2004) The transcriptome profile of human embryonic stem cells as defined by SAGE. Stem Cells, 22, 51–64.[Abstract/Free Full Text]

  12. Sperger, J.M., Chen, X., Draper, J.S., Antosiewicz, J.E., Chon, C.H., Jones, S.B., Brooks, J.D., Andrews, P.W., Brown, P.O. and Thomson, J.A. (2003) Gene expression patterns in human embryonic stem cells and human pluripotent germ cell tumors. Proc. Natl Acad. Sci. USA, 100, 13350–13355.[Abstract/Free Full Text]

  13. Cai, J., Ash, D. and Jabs, E.W. (2002) SAGE analysis from 1 µg of total RNA. In Yamada, K.M. (ed.), Current Protocols in Cell Biology. John Wiley & Sons, Inc., Vol. 2.

  14. Margulies, E.H. and Innis, J.W. (2000) eSAGE: managing and analysing data generated with serial analysis of gene expression (SAGE). Bioinformatics, 16, 650–651.[Abstract/Free Full Text]

  15. Velculescu, V.E., Zhang, L., Zhou, W., Vogelstein, J., Basrai, M.A., Bassett, D.E., Jr, Hieter, P., Vogelstein, B. and Kinzler, K.W. (1997) Characterization of the yeast transcriptome. Cell, 88, 243–251.[CrossRef][ISI][Medline]

  16. Velculescu, V.E., Madden, S.L., Zhang, L., Lash, A.E., Yu, J., Rago, C., Lal, A., Wang, C.J., Beaudry, G.A., Ciriello, K.M. et al. (1999) Analysis of human transcriptomes. Nat. Genet., 23, 387–388.[ISI][Medline]

  17. Margulies, E.H., Kardia, S.L. and Innis, J.W. (2001) A comparative molecular analysis of developing mouse forelimbs and hindlimbs using serial analysis of gene expression (SAGE). Genome Res., 11, 1686–1698.[Abstract/Free Full Text]

  18. Zhou, G., Chen, J., Lee, S., Clark, T., Rowley, J.D. and Wang, S.M. (2001) The pattern of gene expression in human CD34(+) stem/progenitor cells. Proc. Natl Acad. Sci. USA, 98, 13966–13971.[Abstract/Free Full Text]

  19. Li, C. and Wong, W.H. (2001) Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection. Proc. Natl Acad. Sci. USA, 98, 31–36.[Abstract/Free Full Text]

  20. Audic, S. and Claverie, J.M. (1997) The significance of digital gene expression profiles. Genome Res., 7, 986–995.[Abstract/Free Full Text]

  21. Beissbarth, T. and Speed, T.P. (2004) GOstat: find statistically overrepresented gene ontologies within a group of genes. Bioinformatics, 20, 1464–1465.[Abstract/Free Full Text]

  22. Avantaggiato, V., Orlandini, M., Acampora, D., Oliviero, S. and Simeone, A. (1998) Embryonic expression pattern of the murine figf gene, a growth factor belonging to platelet-derived growth factor/vascular endothelial growth factor family. Mech. Dev., 73, 221–224.[CrossRef][ISI][Medline]

  23. Pilia, G., Hughes-Benzie, R.M., MacKenzie, A., Baybayan, P., Chen, E.Y., Huber, R., Neri, G., Cao, A., Forabosco, A. and Schlessinger, D. (1996) Mutations in GPC3, a glypican gene, cause the Simpson-Golabi-Behmel overgrowth syndrome. Nat. Genet., 12, 241–247.[CrossRef][ISI][Medline]

  24. Veugelers, M., Cat, B.D., Muyldermans, S.Y., Reekmans, G., Delande, N., Frints, S., Legius, E., Fryns, J.P., Schrander-Stumpel, C., Weidle, B. et al. (2000) Mutational analysis of the GPC3/GPC4 glypican gene cluster on Xq26 in patients with Simpson-Golabi-Behmel syndrome: identification of loss-of-function mutations in the GPC3 gene. Hum. Mol. Genet., 9, 1321–1328.[Abstract/Free Full Text]

  25. Joutel, A., Vahedi, K., Corpechot, C., Troesch, A., Chabriat, H., Vayssiere, C., Cruaud, C., Maciazek, J., Weissenbach, J., Bousser, M.G. et al. (1997) Strong clustering and stereotyped nature of Notch3 mutations in CADASIL patients. Lancet, 350, 1511–1515.[CrossRef][ISI][Medline]

  26. Pai, S.I., Wu, G.S., Ozoren, N., Wu, L., Jen, J., Sidransky, D. and El-Deiry, W.S. (1998) Rare loss-of-function mutation of a death receptor gene in head and neck cancer. Cancer Res., 58, 3513–3518.[Abstract/Free Full Text]

  27. Howard, T.D., Paznekas, W.A., Green, E.D., Chiang, L.C., Ma, N., Ortiz de Luna, R.I., Garcia Delgado, C., Gonzalez-Ramos, M., Kline, A.D. and Jabs, E.W. (1997) Mutations in TWIST, a basic helix–loop–helix transcription factor, in Saethre-Chotzen syndrome. Nat. Genet., 15, 36–41.[CrossRef][ISI][Medline]

  28. Bourgeois, P., Bolcato-Bellemin, A.L., Danse, J.M., Bloch-Zupan, A., Yoshiba, K., Stoetzel, C. and Perrin-Schmitt, F. (1998) The variable expressivity and incomplete penetrance of the twist-null heterozygous mouse phenotype resemble those of human Saethre-Chotzen syndrome. Hum. Mol. Genet., 7, 945–957.[Abstract/Free Full Text]

  29. The Treacher Collins Syndrome Collaborative Group (1996) Positional cloning of a gene involved in the pathogenesis of Treacher Collins syndrome. Nat. Genet., 12, 130–136.[CrossRef][ISI][Medline]

  30. Dixon, J., Brakebusch, C., Fassler, R. and Dixon, M.J. (2000) Increased levels of apoptosis in the prefusion neural folds underlie the craniofacial disorder, Treacher Collins syndrome. Hum. Mol. Genet., 9, 1473–1480.[Abstract/Free Full Text]

  31. von Lindern, M., van Baal, S., Wiegant, J., Raap, A., Hagemeijer, A. and Grosveld, G. (1992) Can, a putative oncogene associated with myeloid leukemogenesis, may be activated by fusion of its 3' half to different genes: characterization of the set gene. Mol. Cell. Biol., 12, 3346–3355.[Abstract/Free Full Text]

  32. Sevigny, G., Joly, E.C., Bibor-Hardy, V. and Lemieux, N. (1994) Assignment of the human homologue of the mTRiC-P5 gene (TRIC5) to band 1q23 by fluorescence in situ hybridization. Genomics, 22, 634–636.[CrossRef][ISI][Medline]

  33. Wilcoxon, F. (1945) Individual comparisons by ranking methods. Biometrics Bull., 1, 80–83.[CrossRef]

  34. Li, H., Chen, Y., Jiang, Y., Bittner, M., Meltzer, P., Trent, J., Yamada, K. and Yamada, Y. (1999) Profiling expression patterns of 2214 unigenes in mouse craniofacial development by cDNA microarray analysis. Nat. Genet., 23, 57–58.

  35. Neuhauss, S.C., Solnica-Krezel, L., Schier, A.F., Zwartkruis, F., Stemple, D.L., Malicki, J., Abdelilah, S., Stainier, D.Y. and Driever, W. (1996) Mutations affecting craniofacial development in zebrafish. Development, 123, 357–367.[Abstract]

  36. Fowles, L.F., Bennetts, J.S., Berkman, J.L., Williams, E., Koopman, P., Teasdale, R.D. and Wicking, C. (2003) Genomic screen for genes involved in mammalian craniofacial development. Genesis, 35, 73–87.[CrossRef][ISI][Medline]

  37. Ishii, M., Hashimoto, S., Tsutsumi, S., Wada, Y., Matsushima, K., Kodama, T. and Aburatani, H. (2000) Direct comparison of GeneChip and SAGE on the quantitative accuracy in transcript profiling analysis. Genomics, 68, 136–143.[CrossRef][ISI][Medline]

  38. Haverty, P.M., Hsiao, L.L., Gullans, S.R., Hansen, U. and Weng, Z. (2004) Limited agreement among three global gene expression methods highlights the requirement for non-global validation. Bioinformatics, 20, 3431–3441.[Abstract/Free Full Text]

  39. Lu, J., Lal, A., Merriman, B., Nelson, S. and Riggins, G. (2004) A comparison of gene expression profiles produced by SAGE, long SAGE, and oligonucleotide chips. Genomics, 84, 631–636.[CrossRef][ISI][Medline]

  40. Cai, J., Goodman, B.K., Patel, A.S., Mulliken, J.B., van Maldergem, L., Hoganson, G.E., Paznekas, W.A., Ben-Neriah, Z., Sheffer, R., Cunningham, M.L. et al. (2003) Increased risk for developmental delay in Saethre-Chotzen syndrome is associated with TWIST deletions: an improved strategy for TWIST mutation screening. Hum. Genet., 114, 68–76.[CrossRef][ISI][Medline]

  41. Henrique, D., Adam, J., Myat, A., Chitnis, A., Lewis, J. and Ish-Horowicz, D. (1995) Expression of a delta homologue in prospective neurons in the chick. Nature, 375, 787–790.[CrossRef][Medline]

  42. Zeeberg, B.R., Feng, W., Wang, G., Wang, M.D., Fojo, A.T., Sunshine, M., Narasimhan, S., Kane, D.W., Reinhold, W.C., Lababidi, S. et al. (2003) GoMiner: a resource for biological interpretation of genomic and proteomic data. Genome Biol., 4, R28.[CrossRef][Medline]


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 J. Med. Genet.Home page
J W Park, J Cai, I McIntosh, E W Jabs, M D Fallin, R Ingersoll, J B Hetmanski, M Vekemans, T Attie-Bitach, M Lovett, et al.
High throughput SNP and expression analyses of candidate genes for non-syndromic oral clefts
J. Med. Genet., July 1, 2006; 43(7): 598 - 608.
[Abstract] [Full Text] [PDF]

Home page
 J. Med. Genet.Home page
T Frebourg, C Oliveira, P Hochain, R Karam, S Manouvrier, C Graziadio, M Vekemans, A Hartmann, S Baert-Desurmont, C Alexandre, et al.
Cleft lip/palate and CDH1/E-cadherin mutations in families with hereditary diffuse gastric cancer
J. Med. Genet., February 1, 2006; 43(2): 138 - 142.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Material
Right arrow All Versions of this Article:
14/7/903    most recent
ddi083v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (7)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Cai, J.
Right arrow Articles by Lovett, M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Cai, J.
Right arrow Articles by Lovett, M.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's thi