Essential functions of the Williams-Beuren syndrome-associated TFII-I in embryonic development

Badam Enkhmandakha, Aleksandr V. Makeyeva, Lkhamsuren Erdenechimega, Frank H. Ruddleb,1, Nyam-Osor Chimgea, Maria Isabel Tussie-Lunac, Ananda L. Royc, and Dashzeveg Bayarsaihana,1

aDepartment of Reconstructive Sciences, University of Connecticut Health Center, Farmington, CT 06032; 2Department of Molecular, Cellular, and Developmental Biology, Yale University, New Haven, CT 06520; and 3Department of Pathology, Tufts University School of Medicine, Boston, MA 02111

Contributed by Francis H. Ruddle, November 14, 2008 (sent for review May 15, 2008) GTF2I and GTF2IRD1 encoding the multifunctional transcription a -trap clone carrying the LacZ-neo insertion in the 22nd intron factors TFII-I and BEN are clustered at the 7q11.23 region hemizy- of Gtf2ird1. Three Gtf2i mutant lines were generated carrying gously deleted in Williams-Beuren syndrome (WBS), a complex independent insertions within intron 2 (Gtf2iRR105), intron 3 multisystemic neurodevelopmental disorder. Although the bio- (Gtf2iXE029), and intron 9 (Gtf2i XC455)oftheGtf2i locus (Fig. 1). In chemical properties of TFII-I family transcription factors have been all 4 gene-trap lines, the LacZ-neo insertion was upstream of the studied in depth, little is known about the specialized contribu- functional nuclear localization signal of BEN and TFII-I. The tions of these factors in pathways required for proper embryonic rationale to use these lines was that the ␤-galactosidase–fused development. Here, we show that homozygous loss of either polypeptides are not able to translocate to the nucleus, therefore Gtf2ird1 or Gtf2i function results in multiple phenotypic manifes- creating a KO condition in these mice. The gene-trap clones were tations, including embryonic lethality; brain hemorrhage; and injected into blastocysts to produce mice with the inactivated vasculogenic, craniofacial, and neural tube defects in mice. Further Gtf2ird1 and Gtf2i alleles. All 3 Gtf2i lines produced the same analyses suggest that embryonic lethality may be attributable to embryonic defects, indicating that disruption of the Gtf2i allele defects in yolk sac vasculogenesis and angiogenesis. Microarray causes these phenotypic manifestations. Therefore, in all our fol- data indicate that the Gtf2ird1 homozygous phenotype is mainly lowing studies to characterize Gtf2i, we used embryos mainly ␤ caused by an impairment of the genes involved in the TGF RII/ derived from the XE029 line. The Gtf2ird1XE465 and Gtf2iXE029 lines Alk1/Smad5 signal transduction pathway. The effect of Gtf2i were maintained on a mixed C57BL/6 and 129Sv background. inactivation on this pathway is less prominent, but downregula- Most mice heterozygous for the Gtf2ird1 and Gtf2i alleles ap- tion of the endothelial growth factor receptor-2 gene, resulting in peared to be normal, based on their external appearance and the deterioration of vascular signaling, most likely exacerbates the fertility. F1 heterozygous mice were intercrossed to produce severity of the Gtf2i mutant phenotype. A subset of Gtf2ird1 and Gtf2ird1Ϫ/Ϫ and Gtf2iϪ/Ϫ embryos. No homozygotes were recov- Gtf2i heterozygotes displayed microcephaly, retarded growth, and ered at weaning among 367 and 272 offspring of different het- skeletal and craniofacial defects, therefore showing that haploin- erozygous intercrosses of Gtf2ird1 and Gtf2i, respectively. These sufficiency of TFII-I causes various developmental anom- results indicated that loss of either the Gtf2ird1 or Gtf2i allele causes alies that are often associated with WBS. Ϫ Ϫ Ϫ Ϫ embryonic lethality. Gtf2ird1 / and Gtf2i / embryos appeared to be delayed in development by 12 to 24 h based on their body size embryonic ͉ GTF2I ͉ GTF2IRD1 and die between embryonic day (E) 8.5 and E12.5. A large number of homozygous embryos at E8.5 did not initiate axial turning, he TFII-I transcription factors represent a versatile reflecting retarded development. In some mutant embryos, we Tfamily with broad functional activities (1–4). The GTF2IRD1 observed abnormal development of the allantois, which appeared and GTF2I paralogs encoding BEN and TFII-I proteins, respec- swollen and failed to fuse to the chorion (Fig. 2C). At E9.5 and tively, are clustered at the 7q11.23 region hemizygously deleted in E10.5, surviving Gtf2ird1Ϫ/Ϫ and Gtf2iϪ/Ϫ embryos show growth Williams-Beuren syndrome (WBS), a disorder characterized by retardation, dilated pericardial sacs with arrested heart looping, distinctive facial features, mental disability, and growth retardation hypoplasia of branchial arches, and small frontonasal primordia (5, 6). The protein products of the syntenic Gtf2i and Gtf2ird1 on (Fig. 2 B–D). Many embryos at this stage were already resorbed. A murine 5 share considerable few embryos were found with midfacial clefts (Fig. 2 G and H). BIOLOGY within a repeated domain, which is involved in sequence-specific Embryonic hemorrhage of the mutant embryos was apparent at DEVELOPMENTAL DNA-binding properties (7–10). Comparative sequence analysis of E9.5 (Fig. 2D). The incidence and severity of the hemorrhage the TFII-I family members suggests that these genes have evolved increased during development, and at E12.5, over 60% of embryos from a single ancestor via duplication and divergence (7, 11). Our suffered serious bleeding (Fig. 2 K, N–Q). Mutant embryos dis- Gtf2ird1 Gtf2i previous studies revealed that products of and are played varying degrees of intraembryonic bleeding in the head, expressed very early in mouse development (12–14). Although neck, heart, and back areas. Blood vessels in the yolk sac of biochemical properties of these factors have been analyzed over the Ϫ Ϫ Gtf2ird1 / embryos were poorly developed, resulting in an obvi- past few years, little is known about their physiologic role during ously pale yolk sac compared with that of WT littermates (Fig. 2 L embryogenesis. To define the individual function of TFII-I pro- and M). The mutant embryos themselves were pale and showed teins, we have generated mouse lines mutant for Gtf2ird1 and Gtf2i alleles and provide evidence that TFII-I family factors are crucial

in various aspects of mouse development. Author contributions: D.B. designed research; B.E., A.V.M., L.E., N.-O.C., and M.I.T.-L. performed research; B.E., A.V.M., F.H.R., and D.B. analyzed data; and B.E., A.V.M., F.H.R., Results and Discussion A.L.R., and D.B. wrote the paper. ؊ ؊ ؊ ؊ Multiple Embryonic Defects in Gtf2ird1 / and Gtf2i / Embryos. To The authors declare no conflict of interest. elucidate the biological functions of Gtf2ird1 and Gtf2i during 1To whom correspondence should be addressed. E-mail: [email protected] or mammalian embryonic development, we generated 4 mutant [email protected]. mouse lines. One line, designated as Gtf2ird1XE465, was derived from © 2008 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0811531106 PNAS ͉ January 6, 2009 ͉ vol. 106 ͉ no. 1 ͉ 181–186 Downloaded by guest on September 28, 2021 EcoR V 1 kb Gtf2i XE029R1 Sca I

Geo R 735 Amp bp FRT loxP

Sca I XE029F Sca I Sca I Sca I

2 3 TF2IF2 TF2IR2 4 5 6 7 8 8a 9 Fig. 1. Inactivation of Gtf2ird1 and Gtf2i using the gene-trap strategy (http:/www.genetrap.org/). Mu- 6.2 kb tant mice were produced from gene-trap lines carrying 1 2 3 4 M bp the ␤-Geo insertion within intron 22 (clone XE465 XE029 EcoR V - 500 Gtf2ird1 ) or intron 3 (Gtf2i ). Geo, lacZ-neo XF06 Sca I Gtf2ird1 insertion; NLS, nuclear localization signal. Animals Geo R - 300 were genotyped by PCR with primers specific for Gtf2i 1.13 kb Amp FRT - 200 (TF2IF2/TF2IR2 and XE029F/XE029R1) and Gtf2ird1

loxP (F11/BEN22R/XF06). The WT Gtf2i allele gives a 437-bp band (line 1), whereas the Gtf2iXE029 allele displays a Sca I EcoR V Sca I Sca I EcoR V 200-bp product (line 2). Similarly, Gtf2ird1-specific primers amplify a 479-bp fragment in WT DNA (line 3) XE465 22 F11 BEN22R 23 24 25 26 27 28 and an additional 320-bp band from the Gtf2ird1 gene-trapped allele (line 4). Positions of primers used 5.9 kb in PCR are shown in the diagram.

arrested growth at E9.5, suggesting the possibility of defective 10% of Gtf2iϩ/Ϫ mice were significantly smaller than their WT hematopoiesis in both the embryo body and the yolk sac attribut- littermates despite being viable and fertile (Fig. 4 C and D). The able to defects in blood vessel formation. body weight was 65–75% of the normal weight of adult mice. To An open anterior neural tube (exencephaly) was displayed by evaluate the growth potential of newborn pups, we analyzed WT Ϸ15% of Gtf2iϩ/Ϫ and 2% of Gtf2ird1ϩ/Ϫ mice (Fig. 4B). In and Gtf2ird1ϩ/Ϫ mice over several months. Heterozygotes grew homozygous embryos, this number is larger: 60% of Gtf2iϪ/Ϫ significantly slower than control pups, indicating poor growth rate embryos have exencephaly compared with 9% of Gtf2ird1Ϫ/Ϫ (not shown). We also observed pigmentation defects in some embryos (Fig. 2 C, F, I–K). Normally, the neural tube starts to close heterozygous embryos, albeit at a low frequency. The lack of along the ventral midline around E8.5–9.0 and is completed by E9.5. pigment was observed as white patches of variable size on the belly This process of neurulation is a complex task, which involves cell (Fig. 4E). migration as well as extensive neuroectoderm proliferation and/or Among other abnormalities in Gtf2ird1ϩ/Ϫ animals, we observed cell death. At E9.5, either Gtf2i or Gtf2ird1 homozygotes displayed hydrocephalus and kyphosis, a pronounced arched spine (Fig. 4C). open and everted cranial neural folds. In most cases, closure defects The skeletal staining confirmed that the dorsal ‘‘hump’’ of het- occurred at the forebrain/midbrain boundary and extended ros- erozygous animals resulted from an increased backward curvature trally to the anterior neuropore, caudally to the cervical/hindbrain of the spine (Fig. 4F). There were no apparent differences between boundary, or both (Fig. 2 F, I–K). the skulls of mutant and normal adult mice except for the small size and defects in the bones associated with the snout in heterozygotes. Expression Pattern of Gtf2ird1 and Gtf2i. To understand the cause of The frontonasal suture located between the nasal and frontal bones embryonic defects, we compared the expression of Gtf2ird1 and was abnormal, lacking the degree of interdigitation observed in the Gtf2i using whole-mount in situ hybridization and LacZ staining of corresponding WT sutures (Fig. 4G). In contrast, other sutures of gene-trap embryos (Fig. 3A). We confirmed the findings of previ- the cranial vault looked normal. ous immunohistological studies that protein products of Gtf2ird1 and Gtf2i exhibit dynamic expression during embryogenesis (13, Microarray Studies Revealed Altered Gene Expression Pattern in In E8.5 embryos, the majority of Gtf2ird1 transcripts are in the Gtf2ird1؊/؊ and Gtf2i؊/؊ Embryos. In an effort to identify possible .(14 rostral parts of developing embryo (Fig. 3Ag). Transcripts are also molecular changes caused by the absence of TFII-I proteins, we detected in somites and primitive streak. At E9.5, Gtf2ird1 displays compared the gene expression profiles of WT and mutant E9.5 a more distinctive pattern of expression, with a high level in the embryos by microarray analysis. A total of 217 upregulated and neural crest derivatives, including the frontonasal area, branchial 2,356 downregulated genes with more than 1.5-fold difference were arches, and dorsal root ganglia (Fig. 3Ah, i). Additional expression identified in the Gtf2i embryonic array (Fig. 5A). Numbers of genes was detected in the neural tube, limb buds, and tail. From E10.5 that changed their expression in the Gtf2ird1 embryonic array were until E12.5, Gtf2ird1 expression is maintained in the regions of considerably lower: 38 upregulated and 498 downregulated genes. frontonasal primordia, downregulated in somites (Fig. 3Aj–l), and It is remarkable that activities of a substantial portion of these genes increased in the tail tips and appendages. Gtf2i exhibits a more showed similar changes in both arrays (Fig. 5B) despite the differ- diffused expression pattern in comparison to Gtf2ird1 (Fig. 3Aa–f). ence in total numbers of affected genes between Gtf2ird1Ϫ/Ϫ and Gtf2iϪ/Ϫ embryos, with the exception that only 7 genes were A Subset of Gtf2ird1؉/؊ and Gtf2i؉/؊ Embryos Demonstrate Bitem- upregulated in the Gtf2i array but downregulated in the Gtf2ird1 poral Narrowing and Growth, Skeletal, Craniofacial, and Pigmentation array. Gene sorting according biological functions Defects. In a small subset of Gtf2ird1ϩ/Ϫ and Gtf2iϩ/Ϫ embryos at (15) revealed another difference between Gtf2i and Gtf2ird1 em- E11.5 and E12.5, the head is smaller (Fig. 3Bb, d, f, h), with signs bryonic arrays (Fig. 5B). Namely, downregulated genes in the Gtf2i of abnormal brain development compared with that of control array are statistically significantly overrepresented by many genes embryos. Given that hemizygosity in these genes are causal to a involved in core biological processes such as oxidative metabolism, WBS phenotype, we investigated these mice further. The parasag- cell division, transcription, translation, ubiquitin cycle, reorganiza- ittal histological sections revealed bitemporal narrowing in mutant tion of cytoskeleton, and cell motility. In contrast, Gtf2ird1 loss of embryos (Fig. 3Bf, h). About 15% of Gtf2ird1ϩ/Ϫ mice displayed function did not result in any appreciable clustering of affected craniofacial defects (Fig. 4 A and D). Ϸ11% of Gtf2ird1ϩ/Ϫ mice and genes in these functional categories.

182 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0811531106 Enkhmandakh et al. Downloaded by guest on September 28, 2021 A B C D E F

G H I J K

L MN O P Q

Fig. 2. Phenotypes of Gtf2ird1Ϫ/Ϫ (B, F–H, M, and N) and Gtf2iϪ/Ϫ (C, D, I–K, and O–Q) embryos at E9.5–12.5 (I–K). (B–D) Mutants are much smaller compared with a WT littermate (A), and some are resorbed. The morphology of the cranial region of mutants is distinct from that of WT littermates; they lack prominent mesencephalic and telencephalic vesicles. Homozygotes have a small first branchial arch (BA) and a reduced second BA. The dilated pericardial sacs are highlighted by arrowheads. (C) Malformation of the allantois in an E8.5 embryo is shown by a white arrow. The allantois appeared short and swollen and failed to fuse to the chorion in mutant embryos. Gtf2ird1Ϫ/Ϫ (F–H) and Gtf2iϪ/Ϫ (I–K) embryos showed open neural tube defects (exencephaly) and midfacial clefts (black arrow). The embryo in K has hemorrhaging in the head. The WT E10 embryo is shown in E.(L and M) Yolk sac angiogenesis defects in E10.5 Gtf2ird1Ϫ/Ϫ embryos. Gross morphology of yolk sac of WT and mutant embryos at E10.5. The mutant yolk sac exhibits a pallid appearance because of the vast reduction in circulating erythrocytes. Large blood vessels and extensive branching of the vessels can be seen in the control yolk sac (black arrowhead in L), but there is absence of an organized yolk sac vasculature in the mutant embryo, although blood islands are readily detectable. (N–Q) Vascular defects in E10.5 and E12.5 Gtf2ird1Ϫ/Ϫ and Gtf2iϪ/Ϫ embryos. Mutant embryos with hemorrhage in the head, the neck, and the pericardial cavity are shown by black arrowheads.

TFII-I Transcription Factors Regulate TGF␤RII/Alk1/Smad5 and Vegfr2 promoter of the VEGFR2 gene is bound and activated by ectopically Signal Transduction Cascades. Defects in establishing a proper expressed TFII-I and counterregulated by BEN/TF2IRD1 in pul- cardiovascular system is a frequent cause of early embryonic monary artery endothelial cells (20, 21). Consistent with these lethality at midgestation (16), which might explain the observed cell-based observations, our current results indicate significant phenotype of our mutant embryos. Therefore, we analyzed the downregulation of the Vegfr2 gene in Gtf2i mutant embryos but not BIOLOGY expression of specific genes that are indispensable in early vascu- in Gtf2ird1 embryos (Fig. 6). DEVELOPMENTAL logenesis and angiogenesis. It is important to note that phenotypes of Gtf2iϪ/Ϫ and Vascular defective embryos described in the literature can be Gtf2ird1Ϫ/Ϫ were similar to each other but different from the Vegfr2 generally divided into 2 groups based on their phenotypes in murine KO phenotype. Although the vessels in the yolk sac failed to form models. One group has an early defect in the development of both a robust network, Gtf2i or Gtf2ird1 mutants contained red blood hematopoietic and endothelial cells as a result of disruption of the cells and had enlarged embryonic blood vessels (Fig. 2). This Flk1/vascular endothelial growth factor receptor (VEGFR)/ligand phenotype strongly resembles Smad5Ϫ/Ϫ mutants (22) as well as system (17), whereas the other shows abnormal angiogenesis as a phenotypes resulting from inactivation of genes for receptors consequence of a broken TGF␤RII/Alk1/Smad5 signal transduc- involved in activation of Smad5: Tgfbr1/Alk5 (23), Tgfbr2 (24, 25), tion cascade but maintains intact hematopoietic potential (18). and Alk1 (26). In line with this phenotypic observation, our The VEGFR2 (Vegfr2/Kdr/Flk1) functions as the primary me- microarray results show a dramatic decrease in the expression of diator of vascular endothelial growth factor activation in endothe- Smad5 and significant downregulation of both TGF␤ receptor lial cells. The critical role of VEGFR2 in development was revealed genes: Tgfbr1 and Tgfbr2 (Fig. 6). by targeted disruption of this gene in mice (19): yolk-sac blood Because the TGF␤ pathway is one of the signaling cascades that islands were absent at 7.5 days, organized blood vessels could not is initiated during embryonic stem (ES) cell differentiation, we used be observed in the embryo or yolk sac at any stage, and hemato- in vitro differentiating mouse ES cells to test our hypothesis that the poietic progenitors were severely reduced. It has been shown that TFII-I family of transcription factors is involved in regulation of

Enkhmandakh et al. PNAS ͉ January 6, 2009 ͉ vol. 106 ͉ no. 1 ͉ 183 Downloaded by guest on September 28, 2021 Fig. 3. Morphological analysis of the head in mutant embryos and expression of Gtf2i and Gtf2ird1.(A) Expression of Gtf2i (a–f) and Gtf2ird1 (g–l) was detected by the ␤-galactosidase staining and whole-mount in situ hybridization of developing embryos. (B) Brain defects in Gtf2ird1 embryos at E11.5 (b) and E12.5 (d, f, and h) are compared with WT embryos (a, c, e, and g), respectively. (e–h) Parasagittal sections of WT and Gtf2ird1ϩ/Ϫ embryos. (b, d, f, and h) Mutant brain shows bitemporal narrowing. lv, lateral ventricle; 3v, third ventricle; 4v, fourth ventricle.

TGF␤RII/Alk1/Smad5 signal transduction. We performed post- during embryonic development. These observations also suggest transcriptional silencing (knockdown) of Gtf2i or Gtf2ird1 in ES that both alleles of Gtf2ird1 and Gtf2i are necessary for normal cells by gene-specific siRNAs and observed downregulation of embryonic growth and development. Taken together, we conclude Smad5, Tgfbr1, and Tgfbr2, which mirrors the KO effects (Fig. 6). that TFII-I transcription factors are indispensable during mouse Conversely, doxycycline induction of Gtf2ird1 overexpression re- embryonic development. sults in a considerable increase in mRNA levels for Smad5, Tgfbr1, The results reported here are different from those for the and Tgfbr2 in our Tet-on ES cell line (Fig. 6). recently described transgenic, knock-in and KO Gtf2ird1 mice that Our current studies revealed the importance of TFII-I family are viable and have either a mild craniofacial or behavioral phe- transcription factors during mouse embryonic development. The notype (27–29). However, comprehensive identification of tran- majority of Gtf2ird1Ϫ/Ϫ and Gtf2iϪ/Ϫ embryos die by E10.5. Gross scription start sites by Cap Analysis Gene Expression, together with anatomical analysis indicated that these homozygotes die from our preliminary data on alternative splicing in the beginning of multiple developmental defects. Most mutant embryos develop to Gtf2ird1 (data not shown), suggests that targeted disruption of the the gastrulation stage but die before and during midgestation, first coding exons can leave the possibility for expression of an exhibiting severe defects in the embryonic and extraembryonic alternative but functional protein isoform. In such case, the residual vasculature. The finding that disruption of Gtf2ird1 or Gtf2i causes expression of Gtf2ird1 in the previously described mouse lines could embryonic lethality is consistent with the expression patterns of explain observed discrepancies in phenotypes. In addition, the these genes. Gtf2ird1 and Gtf2i display dynamic expression during mouse background used in our studies versus the earlier studies may both fetal and postnatal development (12–14). The fact that play a significant role in phenotypic manifestations. Gtf2ird1Ϫ/Ϫ and Gtf2iϪ/Ϫ embryos can survive until midgestation The characteristic facial appearance and dental problems in might be explained by the presence of maternal proteins that individuals with WBS (5, 6) suggest that some critical genes within supported early stages of embryonic development before implan- the deleted region may be involved in neural crest development. tation. Alternatively, some functional redundancy may exist be- Our earlier work as well as studies by other investigators has shown tween Gtf2ird1 and Gtf2i. The incomplete penetrance of that murine Gtf2ird1 and Gtf2i are expressed in embryonic precur- Gtf2ird1Ϫ/Ϫ and Gtf2iϪ/Ϫ phenotypes such as neural tube and sors to many structures affected in WBS, including neural crest– craniofacial defects could be explained by these variables. These derived tissues (12–14, 27). Here, we show that mice heterozygous results also indicate that Gtf2i cannot compensate for the loss of for these genes are often growth retarded and exhibit hypoplasia of Gtf2ird1 activity or vice versa at later stages of development. A the mandible and other craniofacial defects. There are several subset of mice haploinsufficient for Gtf2ird1 or Gtf2i displayed reports in which spinal defects, scoliosis, and kyphosis were de- various brain, growth, craniofacial, neural, frontonasal suture, and tected in WBS (30, 31). Collectively, these results suggest that skeletal defects, indicating that a dose of TFII-I proteins is critical haploinsufficiency for GTF2IRD1 and GTF2I could be responsible

184 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0811531106 Enkhmandakh et al. Downloaded by guest on September 28, 2021 Fig. 4. Reduced growth, exencephaly, and craniofacial and pigmentation defects A B C in heterozygous animals. (A and B) Lateral views of WT and mutant embryos at E18. (A, C, and D) Subset of Gtf2ird1ϩ/Ϫ mice dis- played craniofacial defects (A, Top), indi- cated by the white arrow. (A, Bottom) The WT control is also shown. (B) Defects in the D cranial vault in Gtf2i heterozygotes. Skele- tal preparations of a WT mouse (Left) and a Gtf2iϩ/Ϫ littermate (Right). The mutant mouse is missing the frontal (f), parietal (p), interparietal (i), and supraoccipital (s) bones. (E) Pigmentation defects in Gtf2ird1ϩ/Ϫ animals. A ventral white patch E F G (melanocyte defects) is noticeable on the belly of some of the heterozygous mice. (C) Lateral view of Gtf2ird1ϩ/Ϫ mouse at 4 weeks showing kyphosis, an increased back- ward curvature of the spine in the mutant (Top) compared with the WT littermate (Bottom). (D) Frontal view of the same mice showing the hydrocephalus, the character- istic domed head (a wider but longitudi- nally shorter skull) and lateral displacement of the ears, and a short narrow snout (black bracket) in the Gtf2ird1ϩ/Ϫ mouse (Right) compared with the WT littermate (Left). (F and G) Skeletal staining of Gtf2ird1ϩ/Ϫ mouse to show kyphosis, the curvature of the spine (F, Bottom), compared with the WT littermate (F, Top) and the defective frontonasal suture (G, Right). (G) Dorsal view of the snout. The white box marks the frontonasal suture, and the frontopremaxillary suture area is shown in detail in the panel below. The black arrow indicates the control frontonasal suture (Left) and the black arrowhead marks the defective frontonasal suture in the heterozygote (Right). The mutant is a different mouse from what is shown in C and D. f, frontal bone; fns, frontonasal suture; n, nasal bone; nc, nasal cartilage; pm, premaxillary bone.

for some of the skeletal and craniofacial pathogenesis observed in ray data revealed changes in the expression of a wide variety of patients with WBS. genes in both Gtf2i and Gtf2ird1 mutants. It is of note that in the In line with the complexity of the observed phenotypes, microar- Gtf2i array, the number of affected genes was considerably higher than in the Gtf2ird1 array (Fig. 5), and in contrast to the Gtf2ird1 response genes, the vast majority of the Gtf2i response genes were A Gtf2ird1 associated with basal cellular processes (Fig. 5). However, a more Gtf2i array Gtf2i array dramatic shift of general gene expression in homozygotes does 217 30 38 not necessarily mean a less significant effect of the Gtft2ird1 loss of function in developing embryos, because changes in the expression 80 498 of some selected genes are equally strong in both Gtf2i and Gtf2ird1 2356 arrays. Conversely, expression changes of the largest group of genes 16619 in the Gtf2i array (genes involved in oxidative phosphorylation) are

increase B Gtf2i KO Gtf2ird1 KO Anatomical structure morphogenesis Tgfbr1 Tgfbr2 Smad5 angiogenesis BIOLOGY nervous system development 1.5 skeletal development Vegfr2 DEVELOPMENTAL eye development Alk1 Cellular metabolic process * 1.0 carbohydrate metabolic process * protein metabolic process * 1.5 lipid metabolic process oxidative phosphorylation * Folds of changes Cell division * 2.0 Cell differentiation Apoptosis decrease KO KD * Gtf2i Gtf2i Cytoskeleton and cell mobility 2.66 Cell adhesion Gtf2ird1 KO Gtf2ird1 KD Transport * Transcription * Gtf2ird1 over-expression Translation * Ubiquitin cycle Cell communication / signal transduction Fig. 6. TFII-I transcription factors regulate TGF␤RII/Alk1/Smad5 and Vegfr2

-25 0 25 50 75 100 0 25 50 75 100% signal transduction cascades. Changes in expression of genes involved in early vasculogenesis and angiogenesis in Gtf2i and Gtf2ird1 KO embryos are based Fig. 5. Microarray functional profiling with Gene Ontology (GO). (A) Number on microarray data, Gtf2i and Gtf2ird1 knockdown data were obtained by of genes up-regulated (green) and downregulated (red) and their overlapping quantitative (q) RT-PCR after posttranscriptional silencing of Gtf2i or Gtf2ird1 sets among total genes represented on the OMM16K chip. (B) GO functional in ES cells by gene-specific siRNAs, and Gtf2ird1 overexpression results are enrichment analysis. Changes are shown as percentages of downregulated genes based on qRT-PCR data obtained after doxycycline induction of Gtf2ird1 in a calculated from total number of genes with the same GO represented on the chip. Tet-on ES cell line. Error bars represent the SD calculated from 3 or 2 biological Statistically significant enrichments (P Ͻ 0.05) are indicated by an asterisk. repeats.

Enkhmandakh et al. PNAS ͉ January 6, 2009 ͉ vol. 106 ͉ no. 1 ͉ 185 Downloaded by guest on September 28, 2021 relatively low, indicating that these changes are possibly indirect. Materials and Methods Although the microarray results suggest a broad and more basic Information about the targeting vector, gene-trap clones, PCR probes, protocols, function for Gtf2i in comparison to Gtf2ird1, Gtf2ird1 is likely to be generation of mutant mouse lines, siRNA knockdown, and Tet-on system is important and indispensable in more specific developmental pro- available from the authors on request. Whole-mount in situ staining, skeletal cesses. A notable overlap between Gtf2i and Gtf2ird1 affected genes preparations, and histology were carried out by standard methods (32). Mouse embryos for expression microarray studies were obtained at the age of E9.5. Total is consistent with the phenotypic similarity of Gtf2i and Gtf2ird1 RNA was isolated with RNeasy columns (Qiagen). The RNA integrity and quality mutants. for each embryonic sample were ensured by electrophoresis, using the Agilent In summary, our studies show the importance of the Gtf2i and Bioanalyzer 2100 (Agilent Technologies). Two mutant and 2 WT embryonic RNAs Gtf2ird1 genes that encode the TFII-I family transcription factors. from each female mouse (total of 4 mutant and 4 WT embryos from 2 different The inactivation of each individual locus causes mouse embryonic female mice) were pooled for the expression microarray analysis. The mouse 16K lethality by midgestation, which may be attributable to defects in microarray slide (OMM16K) was the same platform that we used in our recent studies of the expression profiling of BEN and TFII-I in mouse embryo fibroblast yolk sac vasculogenesis and angiogenesis, although the precise cells (33, 34). Two micrograms of total RNA was used for each hybridization molecular effects caused by the inactivation of these genes may be experiment. The 3DNA Array 900 kit (Genisphere) designed to provide increased different. We speculate that in Gtf2ird1 homozygous embryos, the sensitivity for small quantities of RNA was used for labeling reactions. The phenotype is mainly caused by an impairment of the genes involved labeling, hybridization, and scanning were done by the Keck facility at Yale in the TGF␤RII/Alk1/Smad5 signal transduction pathway. In con- University. Scanning was performed with a GenePix 4000A scanner (Axon Instru- trast, the effect of Gtf2i inactivation on this pathway is less prom- ments), and the acquired images were analyzed with GENEPIX PRO 5.1 (Molec- ular Devices). Microarray functional profiling with Gene Ontology was per- inent, but deterioration of VEGFR signaling is most likely to formed using FatiGOϩ and FatiScan software (http://babelomics.bioinfo.cipf.es/) exacerbate the severity of the Gtf2i mutant phenotype. In addition (15). Quantitative PCR was carried out on an ABI PRISM 7300 detection system. We to the homozygous phenotype, a small subset of Gtf2i or Gtf2ird1 used predeveloped TaqMan gene expression assays for quantitative detection of heterozygotes exhibits microcephaly, retarded growth, and skeletal Gtf2i (Mm00494841࿝ml), Gtf2ird1 (Mm00465654࿝ml), Smad5 (Mm01341687࿝g1), and pigmentation defects, suggesting that even haploinsufficiency Tgfbr1 (Mm00436971࿝ml), and Tgfbr2 (Mm00436978࿝ml) and TaqMan Gene of these alleles may cause significant anomalies in developing Expression Master Mix (Applied Biosystems). Data from triplicates are expressed mouse embryos. Further analysis of the mutant embryos, along with as normalized expressions by using the delta-delta-Ct calculation method (35) and the Gapdh reference gene (Mm99999915࿝g1). validation of target genes of TFII-I family factors, will undoubtedly lead to uncovering of other signaling pathways and corresponding ACKNOWLEDGMENTS. This work was supported by National Institutes of Health biological processes that these transcription factors control. grants K02DE18412 and R01DE017205 (to D.B.) and R01HD046034 (to A.L.R).

1. Roy AL (2007) Signal-induced functions of the transcription factor TFII-I. Biochim 17. Carmeliet P, Collen D (2000) Transgenic models in angiogenesis and cardiovascular Biophys Acta 1769:613–621. disease. J Pathol 190:387–405. 2. Ashworth T, Roy AL (2007) Cutting edge: TFII-I controls B cell proliferation via regu- 18. Goumans MJ, Mummery C (2000) Functional analysis of the TGFbeta receptor/Smad lating NF-kappaB. J Immunol 178:2631–2635. pathway through gene ablation in mice. Int J Dev Biol 44:253–265. 3. Hakre S, et al. (2006) Opposing functions of TFII-I spliced isoforms in growth factor- 19. Shalaby F, et al. (1995) Failure of blood-island formation and vasculogenesis in Flk-1- induced gene expression. Mol Cell 24:301–308. deficient mice. Nature 376:62–66. 4. Caraveo G, van Rossum DB, Patterson RL, Snyder SH, Desiderio S (2006) Action of TFII-I 20. Wu Y, Patterson C (1999) The human KDR/flk-1 gene contains a functional initiator outside the nucleus as an inhibitor of agonist-induced calcium entry. Science 314:122– element that is bound and transactivated by TFII-I. J Biol Chem 274:3207–3214. 125. 21. Jackson TA, Taylor HE, Sharma D, Desiderio S, Danoff SK (2005) Vascular endothelial 5. Morris CA, Mervis CB (2000) Williams syndrome and related disorders. Annu Rev growth factor receptor-2: Counter regulation by transcription factors, TFII-I and TFII- Genomics Hum Genet 1:461–484. IRD1. J Biol Chem 280:29856–29863. 6. Bellugi U, Lichtenberger L, Mills D, Galaburda A, Korenberg JR (1999) Bridging cogni- tion, the brain and molecular genetics: Evidence from Williams syndrome. Trends 22. Yang X, et al. (1999) Angiogenesis defects and mesenchymal apoptosis in mice lacking Neurosci 22:197–207. SMAD5. Development 126:1571–1580. 7. Makeyev AV, et al. (2004) GTF2IRD2 is located in the Williams-Beuren syndrome critical 23. Larsson J, et al. (2001) Abnormal angiogenesis but intact hematopoietic potential in region 7q11.23 and encodes a protein with two TFII-I-like helix-loop-helix repeats. Proc TGF-beta type I receptor-deficient mice. EMBO J 20:1663–1673. Natl Acad Sci USA 101:11052–11057. 24. Oshima M, Oshima H, Taketo MM (1996) TGF-beta receptor type II deficiency results in 8. Bayarsaihan D, et al. (2002) Genomic organization of the genes Gtf2ird1, Gtf2i, and defects of yolk sac hematopoiesis and vasculogenesis. Dev Biol 179:297–302. Ncf1 at the mouse chromosome 5 region syntenic to the human chromosome 7q11.23 25. Goumans MJ, et al. (1999) Transforming growth factor-beta signalling in extraembry- Williams syndrome critical region. Genomics 79:137–143. onic mesoderm is required for yolk sac vasculogenesis in mice. Development 126:3473– 9. Vullhorst D, Buonanno A (2003) Characterization of general transcription factor 3, a 3483. transcription factor involved in slow muscle-specific gene expression. J Biol Chem 26. Oh SP, et al. (2000) Activin receptor-like kinase 1 modulates TFG-beta signaling in the 278:8370–8377. regulation of angiogenesis. Proc Natl Acad Sci USA 97:2626–2631. 10. Vullhorst D, Buonanno A (2005) Multiple GTF2I-like repeats of general transcription 27. Palmer SJ, et al. (2007) Expression of Gtf2ird1, the Williams syndrome-associated gene, factor 3 exhibit DNA binding properties. Evidence for a common origin as a sequence- during mouse development. Gene Expr Patterns 7:396–404. specific DNA interaction module. J Biol Chem 280:31722–31731. 28. Tassabehji M, et al. (2005) GTF2IRD1 in craniofacial development of humans and mice. 11. Hinsley TA, Cunliffe P, Tipney HJ, Brass A, Tassabehji M (2004) Comparison of TFII-I gene Science 310:1184–1187. family members deleted in Williams-Beuren syndrome. Protein Sci 13:2588–2599. 29. Young EJ, et al. (2007) Reduced fear and aggression and altered serotonin metabolism 12. Bayarsaihan D, Ruddle FH (2000) Isolation and characterization of BEN, a member of in Gtf2ird1-targeted mice. Genes Brain Behav 7:224–234. the TFII-I family of DNA-binding proteins containing distinct helix-loop-helix domains. 30. Osebold WR, King HA (1994) Kyphoscoliosis in Williams syndrome. Spine 19:367–371. Proc Natl Acad Sci USA 97:7342–7347. 31. Cohen DB, Quigley MR (2006) Thoracolumbar syrinx in association with Williams 13. Bayarsaihan D, et al. (2003) Expression of BEN, a member of TFII-I family of transcription syndrome. Pediatrics 118:522–525. factors, during mouse pre- and postimplantation development. Gene Expr Patterns 3:579–589. 32. Hogan B, Beddington R, Constantini F, Lacy E (1994) Manipulating the Mouse Embryo. 14. Enkhmandakh B, Bitchevaia N, Ruddle FH, Bayarsaihan D (2004) The early embryonic A Laboratory Manual (Cold Spring Harbor Lab Press, Cold Spring Harbor, New York). expression of TFII-I during mouse preimplantation development. Gene Expr Patterns 33. Chimge NO, Mungunsukh O, Ruddle FH, Bayarsaihan D (2007) Expression profiling of 4:25–28. BEN regulated genes in mouse embryonic fibroblasts. J Exp Zool 308:209–224. 15. Al-Shahrour F, et al. (2006) BABELOMICS: A systems biology perspective in the func- 34. Chimge NO, Mungunsukh O, Ruddle FH, Bayarsaihan D (2007) Gene expression analysis tional annotation of genome-scale experiments. Nucleic Acids Res 34:472–476. of TFII-I modulated genes in mouse embryonic fibroblasts. J Exp Zool 308:225–235. 16. Ihle JN (2000) The challenges of translating knockout phenotypes into gene function. 35. Livak K, Schmittgen T (2001) Analysis of relative gene expression data using real-time Cell 102:131–134. quantitative PCR and the 2(-Delta Delta C(T)) method. Methods 25:402–408.

186 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0811531106 Enkhmandakh et al. Downloaded by guest on September 28, 2021