An Allelic Series Uncovers Novel Roles of the BRCT Domain-Containing Protein PTIP in Mouse Embryonic Vascular Development

Abstract

Pax transactivation domain-interacting protein (PTIP, or PAXIP1) is required for mouse development and has been implicated in DNA damage responses and histone modification. However, the physiological roles of PTIP during embryogenesis remain unclear due to early embryonic lethality of null mutants. We describe two N-ethyl N-nitrosourea-induced hypomorphic missense alleles of Ptip, each of which alters one of the six encoded BRCT domains. Phenotypic characterization of these mutants revealed important functions of PTIP in vasculogenesis and chorioplacental development that appear unrelated to activities in DNA repair or global histone modification. The results of gene expression profiling and in vitro angiogenesis assays indicated that PTIP modulates a transcriptional program, centered around Vegfa, that drives the migration of endothelial cells to properly form the embryonic vasculature. These and other data suggest that PTIP has multiple functions, one of which is to promote the formation of transcriptional complexes that provide specificity of developmental gene expression.

The formation of a normally patterned, functional vascular system is essential for embryogenesis. The development and physiological maturation of most organs is critically dependent on the accurate establishment of the vascular system (49). Two key sequential morphogenic processes underlie vascular development: (i) vasculogenesis, defined by the formation of primary capillary plexus from endothelial cells, and (ii) angiogenesis, characterized by neovascularization from preexisting vessels, such as the primary capillary plexus (48). Vascular formation is a dynamic and complicated process, and our understanding of the underlying mechanisms is incomplete.

Over the past 15 years, the functions of many genes involved in angiogenesis or vasculogenesis have been investigated by the application of gene targeting in mice. Most of these genes are components of signaling pathway networks, including Vegf (3, 11), Ang/Tie (10), TGFβ/Smad (16), Notch (12, 25, 26), integrin (52), and EphB/Ephrin (1, 61), which temporally and spatially regulate the initiation, progression, and remodeling of vasculature. Gene targeting is not ideal for the unbiased discovery of new genes essential for this process. Forward genetic mutagenesis screens provide an alternative to the discovery of new genes causing cardiovascular phenotypes in both mice (67) and zebrafish (57) and can yield hypomorphic alleles that uncover developmental functions that are otherwise masked by early embryonic lethality caused by null alleles.

PTIP contains six BRCT domains and is essential for mitotic progression and embryonic development in mice (5, 30). BRCT domains are most often found in proteins involved in DNA repair. Null mutations of these genes, such as Brca1 (51), Bard1 (37), Xrcc1 (58), and Lig3 (47), often lead to embryonic lethality that is partially due to failure to repair DNA damage. Cells deficient in PTIP have increased sensitivity to DNA damage induced by ionizing radiation (24) and an attenuated response to replication stress-induced DNA damage (15). The C-terminal BRCT domains appear to be responsible for phosphorylation-dependent protein localization into 53BP1- and γH2AX-containing nuclear foci following the induction of DNA breaks (34). However, it is unknown whether the embryonic lethality of PTIP null mutants is directly attributable to defective DNA damage repair. The roles of the N-terminal BRCT domains are unknown.

While the functions of many BRCT domain-containing proteins in DNA damage response have been well studied, little is known about the physiological roles of PTIP during mammalian embryogenesis. The Xenopus laevis ortholog, SWIFT, has ∼80% amino acid sequence identity with the BRCT domains of mouse PTIP. The inhibition of SWIFT results in the suppression of TGFβ-induced gene transcription and defective mesoderm formation (54). The results of recent studies showed that PTIP is a component of the histone H3K4 methyltransferase complex (22, 44). Histone H3 methylation levels were reported to be reduced in Ptip null mouse embryos (44), indicating the involvement of PTIP in the epigenetic regulation of embryogenesis. PTIP physically interacts with the Pax proteins, a group of DNA binding transcription factors that play crucial roles during organogenesis in several organisms (30). These data indicate that PTIP is a transcriptional regulator during embryonic development. Mouse Ptip is widely expressed throughout embryogenesis, including when organogenesis begins at embryonic day 8.5 (E8.5) to E10.5 (http://symatlas.gnf.org). Ptip null embryos show reduced mitosis and apoptosis beginning at E8.5 and are severely disorganized and dead by E9.5 (5), before possible organogenesis defects can be visualized.

Here, we describe the characterization and positional cloning of two N-ethyl N-nitrosourea-induced midgestation lethal mutations (L5Jcs4 and L5Jcs36) isolated from a forward genetic screen directed at proximal mouse chromosome 5 (64). The results of positional cloning and complementation studies showed that they are hypomorphic Ptip alleles with point mutations in the N- and C-terminal BRCT domains. The hypomorphic nature of these alleles allows mutant embryos to survive longer than null mutants, thus uncovering new aspects of PTIP in vascular formation and extraembryonic development.

MATERIALS AND METHODS

Cell culture.

Pooled human umbilical vein endothelial cells (HUVECs) (Lonza, Walkersville, MD) were cultured in EGM2 (Lonza). Mouse embryonic fibroblasts (MEFs) from E9.5 embryos were cultured in Dulbecco's modified Eagle's medium plus 15% fetal bovine serum. Blastocysts were cultured on gelatinized plates in Dulbecco's modified Eagle's medium supplemented with 15% fetal bovine serum, 1.0 mM l-glutamine, 0.1 mM minimal essential medium-nonessential amino acids, 0.1 mM β-mercaptoethanol, and leukemia inhibitory factor. Outgrowths were collected after 6 days of culture for DNA isolation and genotyping.

RNA interference.

Small interfering RNA (siRNA) was purchased from Dharmacon (Lafayette, CO), and the sequences were described (24). HUVECs were seeded on six-well plates at 2 × 10⁵ cells/well and grown to 70 to 80% confluence. The cells were then transfected with siRNA against Ptip (UGC ACU AGC CUC ACA CAU AdTdT) or scrambled siRNA (CCU ACU AAG CGA CAC CAU UdTdT) at a final concentration of 100 nM using Lipofectamine 2000 (Invitrogen). After 6 h, the transfection medium was replaced with EGM2. The cells were harvested 48 h posttransfection.

Mice, genetic mapping, and complementation testing.

The isolation and mapping of L5Jcs4 and L5Jcs36 have been described previously (64). For positional cloning, C3HeB/FeJ-L5Jcs4/Rw females were outcrossed to males that carried the CAST/Ei chromosome 5 semicongenic on the same strain background. Rw (rump white) is a visibly marked, recessive lethal inversion. Non-Rw F₁-hybrid siblings were intercrossed, and the resulting F₂ offspring were screened for recombination along the Rw region by PCR of microsatellite markers polymorphic between C57BL/6J and CAST/EiJ. Primer pairs for “D5Jcs” markers are listed in the Mouse Genome Informatics database (informatics.JAX.org).

Complementation testing was performed by crossing L5Jcs4/Rw to L5Jcs36/Rw animals. All 18 offspring were Rw, indicating that there were no L5Jcs4/L5Jcs36 animals born (six expected; chi square = 9, P = 0.0027).

Histological analyses.

Deciduae were fixed overnight at 4°C in 4% paraformaldehyde-phosphate-buffered saline (PBS) (137 mM NaCl, 2.7 mM KCl, 4.3 mM Na₂HPO₄, 1.4 mM KH₂PO₄) and then sequentially dehydrated in 50%, 70%, 95%, and 100% ethanol; cleared in xylene; and embedded in paraffin. The sections were 5 μm thick. Feulgen staining was performed as described previously (62). Terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) staining was performed with an ApopTag peroxidase in situ apoptosis detection kit (Intergen Co.). VIP (Vector Laboratories) was used as the peroxidase substrate, and counterstaining was performed with 0.5% (wt/vol) methyl green in 0.1 M sodium acetate (pH 4.0).

RNA in situ hybridization and PECAM-1 staining.

Ptip probes were synthesized by using a DIG RNA labeling kit (primer sequences ATGACCCAGCAGTGGAGATT and GCGACCTCATTTGGTTTCAG; Roche) and hybridized as described previously (63). Embryos for PECAM-1 (platelet-endothelial cell adhesion molecule-1) staining were fixed in 4% paraformaldehyde in PBS overnight at 4°C and stained with a monoclonal anti-PECAM-1 antibody (clone MEC 13.3; Pharmingen). 3,3-Diaminobenzidine (Sigma) was used as the substrate for horseradish peroxidase, using avidin-biotinylated enzyme complex as a signal amplifier (Vector Laboratories). For fluorescent immunodetection, Alexa Fluor 594 goat anti-rat immunoglobulin G (Invitrogen) was used as the secondary antibody.

Quantitative real-time PCR.

Amounts of 5 μg of total RNA, from the same preparations used for microarray analysis, were reverse transcribed using a Superscript III system (Invitrogen). ABI TaqMan gene expression assays were used to quantify the expression of Vegfa, Meox2, and Stc1. PCRs were performed on an ABI 7500 real-time PCR system and analyzed with sequence detection software version 1.3 (Applied Biosystems). Gapdh was used as the endogenous control for normalization of cDNA input. The relative quantification of gene expression was determined by the 2^−(ΔΔCT) method (32).

Microarray analysis.

E9.5 embryos were dissected from Ptip^L5Jcs4 intercrosses, and the yolk sacs were harvested for genotyping. Total RNA was isolated using an RNeasy mini kit (Qiagen). The RNA purity and integrity were examined on an Agilent Bioanalyzer 2100. RNA was labeled by using a MessageAmp II-biotin enhanced single-round RNA amplification kit from Ambion (Austin, TX), and cRNA was hybridized to Affymetrix mouse genome 430 2.0 GeneChips and read on a GeneChip scanner 3000 (Affymetrix, Santa Clara, CA). The raw data was processed by using Affymetrix GCOS software to obtain detection calls and signal values. The signals of each array were scaled to a target value of 500 using GCOS software.

Western blot analysis.

Protein was extracted from E9.5 embryos or MEFs using radioimmunoprecipitation assay buffer, and the concentration was quantified with a BCA kit (Pierce). Amounts of 15 μg of total protein were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, electrotransferred onto a pure nitrocellulose membrane (Bio-Rad), and probed with the relevant antibodies. Binding was detected by using a Pierce ECL kit. The band intensities were quantified by using NIH Image J software. For the detection of histone components, E9.5 embryos were lysed in TEB (1× PBS containing 0.5% Triton X-100, 2 mM phenylmethylsulfonyl fluoride, and 0.02% NaN₃) on ice for 10 min, pelleted at 2,000 rpm for 10 min at 4°C, washed once with TEB, and resuspended in 0.2 N HCl. After overnight incubation at 4°C, the material was repelleted. Amounts of 3 μg of extracts were separated by using 12% precise protein gels in Tris-HEPES-sodium dodecyl sulfate buffer (Pierce). The antibodies employed were as follows: rabbit anti-PTIP (Bethyl Laboratories); rabbit anti-HIF-1α (Santa Cruz Biotechnology); mouse anti-α-tubulin (Sigma); rabbit anti-cleaved caspase 3 (Cell Signaling); rabbit anti-H4, H3K9M2, -H3K9M3, -H3, -acetyl-H4K12, and -γH2AX (Upstate); and anti-H2Ax, -H3K4M2, -H3K4M3, and -H3K14Ace (Abcam). Chicken anti-PTIP was kindly provided by Gregory Dressler.

Endothelial cell migration and tube formation assay.

Cell migration was assessed by a wound healing assay. Two days after siRNA transfection, confluent cultures of HUVECs were scraped with a micropipette tip. The endothelial cell migration from the edge of the injured monolayer was quantified by measuring the distance between the wound edges before and 16 h after incubation. To measure capillary tube formation, HUVECs were seeded on six-well plates at 2 × 10⁵ cells/well. On the following day, the cells were transfected with siRNA against Ptip or scrambled siRNA. Forty-eight hours later, the cells were trypsinized and counted. No difference was observed between knockdown and control groups. Amounts of 1 × 10⁴ cells were then seeded on a 96-well plate covered with polymerized ECMatrix (Chemicon) and incubated at 37°C for 12 h in a CO₂ incubator. Images were exported to the Image J program for the identification of endothelial cell tubule-like networks.

RESULTS

Positional cloning of two Ptip alleles.

As previously reported (64), the L5Jcs4 and L5Jcs36 embryonic lethal mutations reside on chromosome 5 within the deletion Dpp6^df4J (Fig. 1A). To further narrow the interval, recombination mapping was conducted by intercrossing L5Jcs4/+^CAST F₁ heterozygotes and genotyping viable F₂ progeny with microsatellite markers. Regions that were homozygous for the parental (C57BL/6J) markers could not contain the L5Jcs4 mutation. The recombination mapping data localized L5Jcs4 to a 966-kb region (between D5Jcs89 and D5Jcs75) containing four RefSeq genes (Fig. 1A). Of these, Htr5a and Dpp6 were eliminated as L5Jcs4 candidates because homozygous disruptions of them are viable (17, 64). Speer4b is a member of a gene family that is expressed in testis specifically (56). Since homozygous knockout of Ptip (officially Paxip1) causes embryonic lethality (5), this gene was sequenced, revealing a missense mutation (Ser51→Pro) in the N-terminal BRCT domain of PTIP in L5Jcs4 (Fig. 1B and C). L5Jcs4 and L5Jcs36 failed to complement (see Materials and Methods), demonstrating that they are alleles of the same gene. Sequencing uncovered a missense mutation (Leu857→Pro) in the C-terminal BRCT domain of Ptip in L5Jcs36 mice (Fig. 1B and C). Finally, and as shown below, neither L5Jcs4 nor L5Jcs36 complemented the lethality of a Ptip null allele (Ptip⁻) (5), proving that L5Jcs4 and L5Jcs36 are alleles of Ptip. Henceforth, these alleles are designated Ptip^L5Jcs4 and Ptip^L5Jcs36.

FIG. 1.

Genetic mapping and positional cloning of L5Jcs4 and L5Jcs36. (A) Map of the L5Jcs4 region. The L5Jcs4 mutation lies within the Dpp6^df4J deletion (blue bar; the thin lines extending from the bar indicate deletion breakpoint regions), and recombination mapping further refined the critical region as indicated. RefSeq genes within the L5Jcs4 critical region are shown as purple bars. (B) Identification of missense mutations in L5Jcs4 and L5Jcs36 heterozygotes. DNA sequencing traces are shown with mutated bases shaded blue. (C) Schematic structure of mouse PTIP protein, adapted from Pfam. The six BRCT domains and the mutations in PTIP are indicated by the green areas and the red dots, respectively. (D) Western blot of PTIP protein in E9.5 embryos (emb) and MEFs. Duplicate blots were probed with anti-α-tubulin as a loading control. The predicted PTIP protein is 130 kDa.

To determine if the mutations affect PTIP levels or stability, Western blot analysis of protein extracted from E9.5 embryos and MEFs was performed. Whereas PTIP levels in Ptip^{L5Jcs4/L5Jcs4} samples were indistinguishable from the levels in controls, there was markedly decreased PTIP in Ptip^{L5Jcs36/L5Jcs36} embryos and MEFs (Fig. 1D). The reduction in PTIP protein was not accompanied by decreased mRNA levels (data not shown). Therefore, the mutated C-terminal BRCT domain of PTIP^L5Jcs36 apparently renders it unstable or more susceptible to degradation.

Ptip^L5Jcs4 and Ptip^L5Jcs36 are hypomorphic Ptip alleles.

To characterize the timing and nature of the embryonic lethality, we analyzed the morphology of mutant embryos at different stages of gestation. At E9.5, Ptip^L5Jcs4 homozygotes were smaller than their wild-type (WT) siblings, but the gross morphology, including the head, heart, and trunk, did not show obvious abnormalities (Fig. 2B). However, Ptip^L5Jcs36 mutants were severely growth retarded. They had shortened trunks and poorly developed headfolds, and they failed to complete embryonic turning and heart looping (Fig. 2C). Consistent with the first two phenotypes, the results of RNA in situ hybridization revealed that Ptip transcripts are most abundant in the brain and posterior regions at E8.5 (Fig. 2K). Ptip^−/− embryos were disorganized and degenerating at E9.5 (Fig. 2D). At E10.5, the most striking feature of Ptip^{L5Jcs4/L5Jcs4} embryos, aside from the smaller size, was a swollen pericardial sac (Fig. 2F). At this stage, Ptip^{L5Jcs36/L6Jcs36} embryos had a pale color, were disorganized, and were degenerating. All Ptip^L5Jcs36 mutants had a failure of chorioallantoic fusion (Fig. 2G). Ptip^−/− embryos were undergoing resorption by E10.5 (Fig. 2H). With death characterized by tissue degeneration and lack of a beating heart, Ptip^{L5Jcs4/L5Jcs4}, Ptip^{L5Jcs36/L5Jcs36}, and Ptip^−/− embryos died at E12.5, E10.5, and E9.5, respectively. The results indicate that Ptip^L5Jcs4 is a milder allele than Ptip^L5Jcs36, which in turn is less severe than the null. The hypomorphic nature of Ptip^L5Jcs4 and Ptip^L5Jcs36 was also supported by the phenotypic resemblance of Ptip^L5Jcs4/Ptip⁻ and Ptip^L5Jcs36/Ptip⁻ compound mutant embryos to the corresponding missense homozygotes (Fig. 2I and J).

FIG. 2.

Ptip^L5Jcs4 and Ptip^L5Jcs36 are hypomorphic Ptip alleles. Shown are whole-mount embryos at indicated gestational ages and genotypes (L4, Ptip^L5Jcs4; L36, Ptip^L5Jcs36; −, Ptip knockout allele). Arrowheads in panels F and I highlight the enlarged pericardia. The defective allantois (“al”) is indicated in panels D and G. Panel H shows a yolk sac. (K, L) In situ hybridization of antisense Ptip RNA to WT E8.5 embryo and chorionic placenta, respectively. Size bars = 500 μm.

Ptip^L5Jcs4 and Ptip^L5Jcs36 cause defects in vascular formation.

At E9.5, all Ptip^L5Jcs4 homozygotes were growth retarded, but only ∼40% of embryos had an inflated pericardium. To explore if the cardiac defect is secondary to a failure in vasculogenesis, whole-mount embryos were stained with the pan-vessel-specific marker PECAM-1. Unlike control siblings, Ptip^{L5Jcs36/L5Jcs36} embryos had a deficiency of blood vessels throughout their bodies (Fig. 3E and F). Ptip^{L5Jcs4/L5Jcs4} embryos did not show vascular formation defects at E9.5 (data not shown). However, angiogenesis impairment was obvious in E10.5 Ptip^L5Jcs4 homozygotes. Blood vessel networks failed to elaborate properly in the head (Fig. 3B). In somitic regions, there was a lack of small sprouting blood vessels compared to the blood vessel development in controls (Fig. 3C and D). In WT littermates, aortic arches were visible and connected with the dorsal aorta (Fig. 3A), whereas a well-developed aortic arch was not formed in the Ptip^L5Jcs4 mutants (Fig. 3B). To rule out the possibility that the vasculogenesis and developmental abnormalities in Ptip^L5Jcs4 and Ptip^L5Jcs36 mutants (Fig. 2 and 3) were due to a generalized failure of mesoderm formation, we examined the expression of Brachyury (T) by in situ hybridization in E8.5 and E9.5 embryos. T expression appeared normal in the mutants (Fig. 4). However, because Ptip^{L5Jcs36/L5Jcs36} embryos show caudal truncation, it is possible that there is an impairment in the continual supply of mesoderm.

FIG. 3.

Vascular defects in Ptip mutants. Shown are whole-mount embryos stained with PECAM-1. (A to D) E10.5 embryos. (A) In WT littermates of Ptip^L5Jcs4 mutants, the head contains large, fused blood vessels (inset). (B) In Ptip^L5Jcs4 mutants, the blood vessels are not remodeled properly (inset). The arrowhead in panel A points to aortic arches, which are missing in Ptip^L5Jcs4 mutants. (C, D) Vascular formation in somite regions. (C) WT littermate of Ptip^L5Jcs4 mutants. (D) In Ptip^L5Jcs4 mutants, there is a lack of small sprouting vessels. (E, F) E9.5 embryos. (E) WT littermate of Ptip^L5Jcs36 mutants. (F) In Ptip^L5Jcs36 mutants, blood vessels are sparse. Size bars = 500 μm.

FIG. 4.

Brachyury expression in embryos. An antisense Brachyury (T) riboprobe was hybridized to whole embryos of the indicated gestational ages. Note that the E9.5 embryo in panel D has the developmental appearance of an E8.5 embryo (failed to undergo turning) due to the mutation. L4, Ptip^L5Jcs4; L36, Ptip^L5Jcs36. Size bars = 500 μm.

PTIP dysfunction impairs extraembryonic tissue development.

Yolk sac defects and placental insufficiency are two primary causes of embryonic lethality in midgestation (E8.5 to E12.5) embryos (43). E10.5 WT yolk sacs exhibited hierarchically organized vessel architecture with numerous large blood-filled vitillene vessels (Fig. 5A). In contrast, the yolk sacs of Ptip^{L5Jcs4/L5Jcs4} embryos were pale and lacked prominent vasculature (Fig. 5B). Whole-mount PECAM-1 staining of yolk sacks further revealed the lack of a properly developed capillary network or branching of large vessels (Fig. 5C to F). Histological analysis revealed dense, large blood vessels filled with primitive nucleated blood cells in E9.5 controls (Fig. 5G). In contrast, the blood vessel diameters in Ptip^L5Jcs4 and Ptip^L5Jcs36 mutants were smaller (Fig. 5H and I) and there were fewer blood vessels in the latter.

FIG. 5.

Extraembryonic tissue abnormalities in Ptip^L5Jcs4 and Ptip^L5Jcs36 mutant embryos. (A, B) Whole-mount yolk sacs at E10.5. The Ptip^L5Jcs4 (L4) mutant (B) has sparse blood vessels compared to the WT (+/+) control (A). Size bars = 500 μM. (C to F) PECAM-1-stained whole-mount yolk sacs. The genotypes and gestational ages are indicated. Size bars = 200 μM. Dense capillary networks and maturing, large, branching vessels are evident in controls (C, E), but the mutants (D, F) are not well developed. (F) The Ptip^L5Jcs4 mutant yolk sac displays a number of incomplete sprouts, indicated by arrowheads. (G to I) Feulgen-stained sections of E9.5 yolk sacs from WT, Ptip^L5Jcs4, and Ptip^L5Jcs36 embryos, respectively. (J, K) Hematoxylin and eosin-stained sections of E10.5 placentas of WT (J) and mutant (K) embryos. D, decidua; Tr, trophoblast giant cells; Sg, spongiotrophoblast; La, labyrinth layer. In the mutant, the labyrinth and spongiotrophoblast layers are difficult to discern and are smaller, in aggregate, than in the WT. Size bars = 100 μm.

The vascular portion of the placenta is derived from extraembryonic mesoderm (allantois) that extends from the posterior end of the embryo at E8.0. At E8.5, the allantois and the chorion fuse in a process called chorioallantoic attachment. Soon thereafter, the chorion folds to form villi, creating a space into which the fetal blood vessels grow from the allantois (7). The abnormal allantois in Ptip^{L5Jcs36/L5Jcs36} embryos led to failure of chorioallantoic attachment (Fig. 2G). Although chorioallantoic attachment occurred in Ptip^L5Jcs4 mutants, the chorion-derived labyrinth layer was much smaller than in WT littermates and was less defined (Fig. 5J and K). Ptip expression is abundant in the chorion (Fig. 2L).

Ptip mutation causes deregulation of angiogenesis genes.

PTIP physically interacts with transcription factors and is a component of a histone methyltransferase complex (22, 44), suggesting the involvement of PTIP in transcriptional regulation. Therefore, we investigated the impact of the Ptip^L5Jcs4 mutation on gene expression during embryogenesis by microarray analysis of mutant and WT E9.5 embryos. This revealed 298 differentially expressed genes, of which 58 had a >50% change (P < 0.01). Among these 58 genes, 14 were reduced in Ptip^{L5Jcs4/L5Jcs4}, and 44 were increased (Fig. 6A).

FIG. 6.

Gene expression profiling of Ptip mutants. (A) Cluster analysis of the top 58 differentially expressed genes. Each has a >50% change from the WT. Genes were clustered in columns, with each array experiment in a row. The 15 genes included in the VEGF gene network are indicated by blue bars and are listed above the heat map. Red equals upregulation, and green equals downregulation. (B) Gene network analysis of 58 differentially expressed genes. The genes shaded pink (induced expression) and green (repressed expression) are PTIP dependent. (C) Reverse transcription-PCR analysis of Ptip isoforms in E9.5 embryos. GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was used as a control. (D) Quantification of Vegfa, Meox2, and Stc expression by real time PCR. Error bars show standard errors of the means. (E) Western blot analysis of HIF-1α in E9.5 embryos. The same blot was probed with anti-α-tubulin.

To identify possible relationships among the 58 differentially expressed genes, network analysis was performed using the Ingenuity database and software (www.ingenuity.com). This revealed one network of 15 genes containing the critical regulator of vasculogenesis, VEGF, at its core (Fig. 6B). The results of quantitative PCR verified the upregulation of VEGFA in Ptip^{L5Jcs4/L5Jcs4} embryos (Fig. 6D), and the results of semiquantitative reverse transcription-PCR indicated that all three isoforms of mouse VEGFA were elevated in the mutant (Fig. 6C). To exclude the possibility that VEGFA upregulation is secondary to impaired blood circulation and consequent hypoxia, the levels of HIF-1α, which is induced and promotes Vegfa transcription under hypoxic conditions (50), were assessed by immunoblotting. No difference in HIF-1α was found between WT, Ptip^{L5Jcs4/L5Jcs4}, and Ptip^{L5Jcs36/L5Jcs36} embryos (Fig. 6E). In the VEGFA network, Meox2 and Stc1 are also implicated in angiogenesis. Meox2 heterozygous mutants have an ∼50% reduction in total brain capillary length (65). STC1 may play a selective modulatory role in angiogenesis, possibly serving as a “stop” signal or stabilizing factor contributing to the maturation of newly formed blood vessels (68). The alterations in Meox2 and Stc1 expression levels were also confirmed by quantitative real-time PCR (Fig. 6D). These combined data suggest that PTIP is associated with angiogenesis signaling pathways.

Ptip knockdown impairs endothelial cell migration and capillary tube-like formation.

Angiogenesis is regulated by a tight balance between pro- and antiangiogenic agents. It involves a cascade of events, of which capillary endothelial cell migration is an essential component (27). To investigate whether the angiogenesis defects in the hypomorphic Ptip alleles are caused by dysfunction of endothelial cells, we performed siRNA knockdown of PTIP in HUVECs and observed the effects on cell migration and capillary tube-like formation. The results of Western blot analysis showed that PTIP protein was reduced >80% in siRNA-transfected HUVECs compared to its level in scrambled siRNA controls (Fig. 7A and B). The results of a wound healing assay showed that the migration rate of the knockdown cells was only 40% of the migration rate of controls (Fig. 7C and D). In vitro culture of HUVECs on Matrigel can mimic sprouting and tube formation during angiogenesis in vivo (46). Figure 7E illustrates the markedly reduced tubule formation in PTIP knockdown cells compared to the tubule formation in controls. Quantification of the results of multiple experiments indicated that both the length of the tubule structures and the number of branch points were reduced in the siRNA-treated cells (Fig. 7F and G). Although no decrease in cell number was observed 48 h after Ptip siRNA treatment (not shown; see Materials and Methods), there was an increase in cleaved caspase 3 (Fig. 7H). In the results of similar assays, decreased tube formation and cell migration were associated with increased apoptosis in cells conditionally inactivated (by Cre-recombinase) for focal adhesion kinase (52). Thus, increased apoptosis may have a role in the vascular formation defects in Ptip mice.

FIG. 7.

PTIP regulates endothelial cell migration and tube formation. (A, B) HUVECs were transfected with siRNA complementary to Ptip or with double-stranded RNA containing scrambled versions of the complementary siRNA and then cultured for 48 h. (A) Proteins were extracted for Western analysis with the indicated antibodies. Lanes: 1, mock transfected; 2, scrambled siRNA; 3, siRNA-Ptip. (B) Quantification of band intensities (% of control values) is shown as means ± standard errors of the means (SEM) (n = 3; *, P < 0.05). (C, D) Endothelial cell migration after a scratch wound 48 h after siRNA transfection. (D) Quantification of the data as means ± SEM (n = 10; *, P < 0.05), as described in Materials and Methods. (E to G) Tube formation assay in siRNA-treated HUVECs. Images of representative fields are shown. The lengths of tubes (F) and the numbers of branch points (G) were quantified (% of control values) from the results of five independent experiments and are shown as the means ± SEM. *, P value is <0.05 versus the result for scrambled siRNA. (H) Assessment of apoptosis in Ptip knockdown HUVEC cells by Western blot analysis of cleaved caspase 3. The cells (three replicates) were harvested for protein extraction 48 h after siRNA treatment. scr, scrambled siRNA.

Knockdown of MLL, an H3K4 methyltransferase, impaired the migration and sprout formation of umbilical endothelial cells, suggesting a role of histone methylation in regulating angiogenic function (9). Since PTIP is a component of a histone H3K4 methyltransferase complex (44), alteration in histone H3K4 methylation due to PTIP dysfunction might result in the observed deregulation of angiogenesis genes described earlier. However, we did not observe a decrease in H3K4 di- or trimethylation (or several other histone modifications) in the results of Western blot analysis of protein from E9.5 embryos (Fig. 8A), in contrast to the results of in situ studies of knockout embryos that reported global decreases (44). Furthermore, siRNA-mediated knockdown of Ptip in HUVECs did not alter the levels of H3K4 di- or trimethylation (Fig. 8B).

FIG. 8.

Western blot analysis of histone H3 modifications. (A) Protein from E9.5 embryos probed with the indicated antibodies. Anti-pan-H3 served as a loading control. (B) HUVECs were transfected with siRNA complementary to Ptip or with double-stranded RNA containing scrambled versions of the complementary siRNA and then cultured for 48 h. Extracted proteins were probed with the indicated antibodies. Lanes: 1, mock transfected; 2, scrambled siRNA; 3, siRNA-Ptip. Anti-pan-H4 served as a loading control.

The essential roles of PTIP in embryogenesis are not attributable to apoptosis or generalized defects in cell proliferation.

To determine if the lethality of mutant embryos is a consequence of generalized cellular dysfunction, we examined apoptosis levels by both TUNEL staining and Western blot analysis of cleaved caspase 3. Additionally, we assessed cell proliferation in MEF cultures and by blastocyst outgrowth assays. Compared to those of WT littermates (Fig. 9A), histological sections of E9.5 Ptip^{L5Jcs4/L5Jcs4} embryos did not indicate elevated apoptosis (Fig. 9B and F) or an activated caspase 3 pathway (Fig. 9E), while Ptip^{L5Jcs36/L5Jcs36} embryos showed increases in both (Fig. 9C to F). Since Ptip^L5Jcs36 mutants experience severe vasculogenesis defects at E9.5, we surmise that this enhanced apoptosis is a secondary consequence of decreased blood supply. Ptip^{L5Jcs4/L5Jcs4} (not shown) and Ptip^{L5Jcs36/L5Jcs36} blastocysts were able to attach to culture plates, and their inner cell masses proliferated as did those of WT controls (Fig. 9G and H). Similarly, MEFs of these genotypes propagated for >5 passages in vitro (data not shown). These results indicate that PTIP has specific cellular functions during embryogenesis but that these hypomorphic alleles are not required for general cell proliferation.

FIG. 9.

Apoptosis and cell proliferation in Ptip mutants. (A to D) TUNEL-stained paraffin sections of E9.5 embryos. WT littermate of Ptip^L5Jcs4 mutants (A); Ptip^L5Jcs4 mutant (B); WT littermate of Ptip^L5Jcs36 mutants (C); Ptip^L5Jcs36 mutant (D). TUNEL-positive apoptotic cells are darkly stained compared to the methyl green-counterstained cells. In each panel, selected clusters of apoptotic cells are enclosed in a rectangle. (E) Western blot analysis of cleaved caspase 3, a marker for apoptosis. Lanes: 1, WT littermate of Ptip^L5Jcs4 mutants; 2, Ptip^L5Jcs4 mutant; 3, WT littermate of Ptip^L5Jcs36 mutants; 4, Ptip^L5Jcs36 mutant. (F) Quantification of apoptotic cells in TUNEL-stained sections. For each genotype, apoptotic cells from eight whole-embryo sections were counted, and the number is represented as a relative value (controls were set at 1). Error bars show standard errors of the means. *, P = 0.0001; L4, Ptip^L5Jcs4; L36, Ptip^L5Jcs36. (G, H) Inner cell mass outgrowths of WT (G) and Ptip^L5Jcs36 mutant (H) blastocysts after 6 days in culture. The size bars in panels A to D and panels G and H indicate 100 μm and 300 μm, respectively.

PTIP-deficient cells do not exhibit hypersensitivity to spontaneous DNA damage, but mutation of the C-terminal BRCT domain causes moderate sensitivity to induced DSBs.

BRCT domains are most often found in proteins involved in DNA repair and cell cycle checkpoint control. These functions are particularly important during early embryogenesis, when cell proliferation and DNA replication are highly active. PTIP was reported to participate in a DNA damage response induced by double-strand break (DSB)-inducing agents (5, 24, 34). Therefore, to determine if PTIP deficiency causes excessive spontaneous genome damage, we examined the frequency of chromosomal aberrations in metaphase spreads prepared from Ptip mutant MEFs (Fig. 10A). No chromosome abnormalities were detected in 60 spreads from each mutant genotype tested.

FIG. 10.

DNA damage and growth characteristics in Ptip mutant cells. (A) Metaphase spreads were prepared from primary cultures of E9.5 embryos and stained with Giemsa. Size bar = 10 μm. (B) Western blot analysis of γH2Ax and H14K12 acetylation. H2Ax and H4 served as loading controls. (C) WT and mutant MEFs were treated with the indicated doses of ionizing radiation (IR) and camptothecin (CPT). Cell viability was assessed 72 h after treatment, and the results are plotted as the percentage of viable cells relative to the results for untreated control cultures. Each point represents the mean ± standard error of the mean of the result. The plots represent three independent MEF cultures for each genotype.

Phosphorylation of H2AX (33) and acetylation of H4 (41) occur on damaged DNA and are biomarkers for assessing DNA damage levels in cells. In mutant E9.5 embryos, no increase in these two modified proteins was detected by Western blot analysis (Fig. 10B). The results of other studies have shown that PTIP is involved in the response to stress-induced DNA damage (5, 24). Consistent with these reports, we found that Ptip^{L5Jcs36/L5Jcs36} MEFs had increased sensitivity to the DSB-inducing agents camptothecin and ionizing radiation (Fig. 10C). The response of PTIP to DNA damage, including its recruitment to DNA damage foci and interaction with 53BP1, occurs via its C-terminal BRCT domain, which is disrupted in Ptip^L5Jcs36. Ptip^{L5Jcs4/L5Jcs4} MEFs did not have significantly elevated sensitivity to DNA damage, indicating that the N-terminal BRCT domain (which is mutated in this allele) has a role in embryonic development, but not in DNA damage responses.

DISCUSSION

PTIP plays essential but poorly understood roles in mouse embryonic development. The results of a previous study suggested that this protein is required for cell proliferation (5). Using the two hypomorphic PTIP alleles obtained by a forward genetic screen, we uncovered novel functions of this protein in embryonic vascular formation and extraembryonic tissue development. Defects in the cardiovascular system and extraembryonic tissues appear to be the primary causes of embryonic lethality in the two hypomorphic Ptip mutants we describe.

TGFβ signaling controls cell proliferation, cell recognition, differentiation, apoptosis, and specification of developmental fate in species ranging from flies to mammals (53). Smad family proteins lie at the core of this signaling pathway and regulate the expression of genes involved in embryonic development. A connection between PTIP and TGFβ-Smad signaling has been demonstrated in Xenopus. Inhibition of SWIFT, the Xenopus ortholog of mammalian PTIP, results in the suppression of TGFβ-induced gene transcription and defective trunk development (54), the latter of which occurs in Ptip^L5Jcs36 mutants. SWIFT physically interacts with Smad1/Smad2 and activates target gene transcription in a TGFβ-dependent manner. Since the BRCT domains of PTIP and SWIFT are 80% identical and there is >95% identity between mouse and Xenopus Smad1/Smad2, it is likely that the functions of vertebrate PTIP during embryogenesis are conserved. Knockout of TGFβ signaling pathway components, including the ligand (Tgfβ1) (8), receptors (Alr1 and Alr5) (28, 42, 60), and effectors (Smad1, Smad2, and Smad5) (4, 20, 29, 59, 66), causes lack of normal vasculature in the yolk sac, marked impairment of allantois formation, failure of embryonic turning and anterior morphogenesis, or defects in embryonic vascular formation. We found that the regions of highest Ptip expression, such as brain, posterior streak, and chorion, showed obvious developmental defects in the hypomorphic alleles.

The anatomic characterization, gene expression profiling, and in vitro angiogenesis data presented here demonstrate that PTIP is involved in vasculogenesis and angiogenesis during embryogenesis. Our data lead us to hypothesize that deregulation of Vegfa underlies these defects in Ptip mutants. VEGFA is a crucial regulator of vascular development during embryogenesis, participating in the specification of blood islands and the proliferation, survival, and migration of endothelial cells. Proper VEGFA levels are tightly controlled during angiogenesis. The loss of a single VEGFA allele in mice causes embryonic lethality at E9 due to cardiovascular defects (3, 11), while the overexpression of VEGFA induces embryonic cardiac failure (39) and impaired retinal vascular patterning (40). In zebrafish, the Vegfa promoter contains a number of bone morphogenetic protein-activated Smad binding elements (SBEs), implicating Smad1 and Smad5 in the regulation of bone morphogenetic protein-induced Vegfa expression (18). In mouse macrophages, two SBEs are critical for TGFβ1-induced Vegf promoter activity (23). Since Vegfa transcription increases in the Ptip mutants, perhaps PTIP interaction with Smads inhibits their ability to bind the SBEs. Alternatively, PTIP may recruit a negative regulator to the Vegf promoter, via sequence-specific binding driven by Smads (see below). The mechanism by which Vegf is upregulated by the absence of PTIP requires further investigation.

Although it was reported that PTIP is a component of a histone H3K4 methyltransferase complex and knockout embryos exhibited markedly decreased H3K4 di- and trimethylation by in situ analysis (44), we did not observe a decrease in H3K4 di- or trimethylation (or any of several other histone modifications) in Western blot analysis of protein from E9.5 Ptip^L5Jcs4 or Ptip^L5Jcs36 mutant embryos (Fig. 8A). One explanation for this disparity may be that the BRCT domain mutations do not affect binding of the hypomorphic Ptip alleles to PAX2, which promotes H3K4 methyltransferase complex formation at specific sites (44). However, this would necessitate that the reduced PTIP^L5Jcs36 levels are sufficient for complex formation. Also, PTIP knockdown inhibited HUVEC migration and capillary tube formation but did not reduce H3K4 methylation (Fig. 8B) or affect the expression of the genes regulated by this histone modification (data not shown). Perhaps our observations reflect a cell type difference in PTIP's role in chromatin modification. Patel et al. noted decreased H3K4 methylation in presumptive neural tissues of the headfolds and mesodermal derivatives of E9 embryos, but not in extraembryonic tissues (44).

PTIP is associated with active chromatin but does not itself possess transactivation activities (30). The results of transient transfection studies revealed that PTIP is a strong inhibitor of the transactivation activities of PAX2A and PAX2B on the glucagon promoter. The binding of p8, a nuclear DNA binding protein, to PTIP prevents this inhibition (21). PTIP might also interact with CCAAT/enhancer-binding proteins (C/EBP) via phosphoserine-containing motifs localized in a highly conserved region of C/EBP transactivation domains (38). The transactivation domain regions of PAX and C/EBP contain a PTIP binding motif, suggesting that PTIP may serve as a transcriptional coregulator.

As discussed above, PTIP could regulate embryonic development via TGFβ/Smad signaling pathways. The transcriptional regulatory activity (both positive and negative) and specificity of Smad proteins are regulated by Smad complex cofactors that affect SBE binding affinity (36). PTIP contains six BRCT domains which provide multiple binding sites for its interacting partners (14). We propose that PTIP participates in the organization of transcriptional complexes via the recruitment of transcription factor cofactors at promoters, thereby controlling the specificity and activity of gene expression during embryogenesis.

The four C-terminal tandem BRCT domains of PTIP are required for mediating the response to DSBs. The function of the N-terminal BRCT domains is unknown. The mutation of an N-terminal BRCT domain in Ptip^L5Jcs4 did not impair PTIP stability or the response to radiation-induced DNA damage, suggesting that this mutation might only disturb the conformation of the N-terminal BRCT domains. Therefore, Ptip^L5Jcs4 may be a separation-of-function allele that is useful for dissecting the essential physiological roles of PTIP mediated through the N-terminal BRCT domains. Since the Ptip^L5Jcs36 C-terminal BRCT domain mutation severely reduces the amount of PTIP, we do not know if these BRCT domains are required for embryogenesis. Given the binding of Xenopus PTIP/Swift to Smad2 via C-terminal BRCT domains (54), the N- and C-terminal BRCT domains might cooperate to interact with molecules involved in embryonic processes.

Although mouse PTIP was first identified by its physical interaction with Pax family proteins, it is unclear whether the essential roles of PTIP in embryogenesis are linked with Pax during organogenesis. Among nine Pax family members in mice, only the deletion of Pax3 caused embryonic lethality, due to defective cardiac neural crest migration and consequent malformation of cardiac outflow tracts (6). The other Pax mutants were not demonstrated to undergo defects in vasculogenesis and extraembryonic tissue, as shown in Ptip mutants, suggesting that the roles of PTIP during early embryonic development are mediated via interactions with other developmental regulators. However, Pax family proteins might share some redundancy (35) or function synergistically (45) in organogenesis. Provided that PTIP interacts with more than one Pax member in specific tissues, the loss of PTIP could cause developmental defects ascribed to the combined dysfunction of multiple Pax proteins.

Although BRCT domain-containing proteins are generally associated with DNA damage responses, our studies demonstrate that PTIP has essential functions in development distinct from and in addition to DNA damage repair. The multiple BRCT domain-containing protein TOPBP1 has roles in transcription regulation, DNA replication, and DNA damage checkpoint responses (13). Its three C-terminal BRCT domains directly modulate the activities of several transcription factors (2, 19, 31), including Stromelysin-1 platelet-derived growth factor responsive element binding protein (55), which has no apparent involvement in DNA damage responses. Therefore, it is becoming evident that BRCT domain-containing proteins do not function exclusively in the canonical roles of DNA repair and replication.

The hypomorphic PTIP alleles enabled the uncovering of a wide spectrum of developmental phenotypes, suggesting multiple functions of this protein in regulating embryonic organogenesis and patterning. We have provided evidence that the N- and C-terminal domains are involved in distinct processes. It is intriguing to consider if there is any evolutionary advantage for these functions to reside in the same protein or if there are synergistic functions that are not yet evident.