Proliferative Defect and Embryonic Lethality in Mice Homozygous for a Deletion in the p110α Subunit of Phosphoinositide 3-Kinase*

Phosphatidylinositol 3,4,5-trisphosphate is a phospholipid signaling molecule involved in many cellular functions including growth factor receptor signaling, cytoskeletal organization, chemotaxis, apoptosis, and protein trafficking. Phosphorylation at the 3 position of the inositol ring is catalyzed by many different 3-kinases (classified as types IA, IB, II, and III), but the physiological roles played by each of the different 3-kinase isozymes during embryonic development and in homeostasis in animals is incompletely understood. Mammalian type IA kinase isozymes are heterodimers that are active at 37 °C when the catalytic 110-kDa subunit interacts through an amino-terminal binding domain with a regulatory 85- or 55-kDa subunit. Using gene targeting in embryonic stem cells, we deleted this binding domain in the gene encoding the α isoform of the 110-kDa catalytic subunit (Pik3ca) of the α isozyme of the type IA kinases, leading to loss of expression of the p110 catalytic subunit. We show that Pik3ca del/delembryos are developmentally delayed at embryonic day (E) 9.5 and die between E9.5 and E10.5. E9.5 Pik3ca del/delembryos have a profound proliferative defect but no increase in apoptosis. A proliferative defect is supported by the observation that fibroblasts from Pik3ca del/del embryos fail to replicate in Dulbecco’s modified Eagle’s medium and fetal calf serum, even with supplemental growth factors.

lipid signaling molecule involved in many cellular functions including growth factor receptor signaling, cytoskeletal organization, chemotaxis, apoptosis, and protein trafficking. Phosphorylation at the 3 position of the inositol ring is catalyzed by many different 3-kinases (classified as types I A , I B , II, and III), but the physiological roles played by each of the different 3-kinase isozymes during embryonic development and in homeostasis in animals is incompletely understood. Mammalian type I A kinase isozymes are heterodimers that are active at 37°C when the catalytic 110-kDa subunit interacts through an amino-terminal binding domain with a regulatory 85-or 55-kDa subunit. Using gene targeting in embryonic stem cells, we deleted this binding domain in the gene encoding the ␣ isoform of the 110-kDa catalytic subunit (Pik3ca) of the ␣ isozyme of the type I A kinases, leading to loss of expression of the p110 catalytic subunit. We show that Pik3ca del/del embryos are developmentally delayed at embryonic day (E) 9.5 and die between E9.5 and E10.5. E9.5 Pik3ca del/del embryos have a profound proliferative defect but no increase in apoptosis. A proliferative defect is supported by the observation that fibroblasts from Pik3ca del/del embryos fail to replicate in Dulbecco's modified Eagle's medium and fetal calf serum, even with supplemental growth factors.
Despite a wealth of knowledge of the biochemistry and cell biological effects of phosphoinositides, very little information is available on the particular physiological roles played by these different enzymes and isozymes during mammalian development. Inhibitors of PI 3-kinases such as wortmannin (17) and LY294002 (18) are very useful for probing the functions of these enzymes but as with all inhibitor studies suffer from questions of specificity as to which enzymes and isozymes are being inhibited at the dosages used in a particular study. In contrast, specific mutations in genes encoding these enzymes would provide greater specificity. We have therefore begun a systematic program of generating mice carrying deletion alleles of the PI 3-kinases. Here we report the successful targeting of the ␣ isoform of p110 to create a deletion allele that, in the homozygous state, leads to embryonic lethality at E9.5 due to a severe defect in the proliferative capacity of the embryo.

MATERIALS AND METHODS
Cloning of Pik3ca Genomic Sequence-7.5 ϫ 10 5 plaques from a 129SV mouse genomic library in the FIX II vector (Stratagene) were screened (19) with probes derived from Pik3ca cDNA. Of three positive phage clones, one (F1) was selected for further study and subcloning, and an 8-kb SalI fragment was isolated and found to contain the first three coding exons of Pik3ca (see Fig. 1).
Genetic Mapping of the Pik3ca Gene-A tetranucleotide repeat (CCTT) within an intron of Pik3ca was used for chromosome mapping using the Jackson Lab C57BL/6J and Mus spretus backcross DNA panel map service. Two PCR primers flanking the tetranucleotide repeat (forward, GTA GGG ATG AAG GTG GAG AGG; backward, AGC TGA CTC AAT GGA AGT AGG G) were used for the amplification. PCR reaction was performed on 50 ng of DNA in a 20-l volume containing 400 nM each primer, 200 nM of dNTP, 10 mM Tris-HCl (pH 8.3), 1.5 mM MgCl 2 , and 0.5 unit of Taq polymerase. Reactions were denatured at 94°C for 5 min and then subjected to 30 step cycles consisting of 94°C for 20 s, 57°C for 60 s, and 72°C for 30 s. PCR products were loaded on 3.5% MetaPhore agarose gel and separated at 70 V for 3 h.
Vector Construction and Pik3ca Gene Targeting in 129SV ES Cells-The targeting construct was made in the pPNT vector (20) using an 8-kb SalI fragment from phage F1 and resulting in replacement of the first coding exon with a neomycin resistance gene transcribed in reverse orientation from a mouse PGK promoter (see Fig. 1). Electroporation into TC1 ES cells (21) and selection for homologous recombination were by published procedures. Recombinant ES cell lines were screened by PstI genomic Southern blot using probe B, located outside the targeted region of homologous recombination, and positive cell lines were confirmed by SacI genomic Southern blot using probe E (see Fig. 1). All studies described in this report were performed on mice carrying the mutant alleles on a 129/SV inbred background. Animal experiments were performed under protocols approved by the National Human Genome Research Institute's Animal Care and Use Committee under National Institutes of Health guidelines, "Using Animals in Intramural Research." Prenatal and Postnatal Genotyping of Wild Type and Deletion Alleles * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
of Pik3ca-DNA was extracted from yolk sac (22) and tails (19) by published methods. Two PCR reactions were performed for genotyping. One pair of primers derived from the exon deleted in the deletion allele (P110N, forward, CTC CCC AAT GGA ATG ATA GTG, and P110N, reverse, TCT TTT CTT CAC GGT TGC CT) amplifies a 260-bp fragment only from the wild type. The second pair of primers derived from the neomycin resistance gene (Neo1, 5Ј-AGA GGC TAT TCG GCT ATG ACT G, and Neo2, 3Ј-TTC GTC CAG ATC ATC CTG ATC) amplifies a 430-bp fragment only from the disrupted Pik3ca gene. PCR amplification was performed per the manufacturer's protocols (Life Technologies, Inc.).
Western Blot Analysis-Cells obtained from Pik3ca ϩ/ϩ , Pik3ca ϩ/del , and Pik3ca del/del E9.5 embryos were collected by trypsinization on the third day of culturing. Cultured cells were washed three times with 1ϫ phosphate-buffered saline, and the cell pellets were frozen on dry ice immediately. Cell pellets were lysed in 25 l of deionized H 2 O on ice, the lysates were separated on 6% SDS/polyacrylamide electrophoresis gels, and the proteins were transferred to Immobilon-P transfer membrane according to the manufacturer's protocols (Millipore). Filters were used for Western blotting (23) with either rabbit anti-human p85␣ serum (Upstate Biotechnology Inc.) or rabbit anti-human p110␣ (Santa Cruz Biotechnology sc-7174) in 0.1% Tween-20/1ϫ phosphate-buffered saline for 1 h at RT, and the signal was detected using peroxidaseconjugated donkey anti-rabbit antibody and chemiluminescence according to the manufacturer's protocols (ECL, Amersham Pharmacia Biotech). Filters were exposed to x-ray films (Kodak) for 25 min at room temperature.
Proliferation Assay-Pregnant mice from a Pik3ca ϩ/del ϫ Pik3ca ϩ/del cross were injected intraperitoneally at day E9.5 with BrdUrd (Sigma, 100 g/g body weight). E9.5 embryos were separated from their yolk sacs, which were used for determining Pik3ca ϩ/ϩ and Pik3ca del/del genotypes. Embryos were fixed in 4% paraformaldehyde at 4°C overnight and then embedded in paraffin. Sagittal sections were mounted onto silanized slides without staining, dewaxed, hydrated in decreasing ethanol solutions, treated with 3% H 2 O 2 , and washed in phosphate-buffered saline. Sections were exposed to anti-BrdUrd antibody (Becton Dickinson) and developed in goat anti-mouse IgG (Sigma) using peroxidase-conjugated Elite ABC (Vector Lab) and Sigma Fast DAB as described (24).
Apoptosis Assay of E9.5 Embryos Sections-Pik3ca ϩ/ϩ and Pik3ca del/del E9.5 embryos were separated from their yolk sacs, which were used for genotyping. Embryos were fixed in 4% paraformaldehyde at 4°C overnight then transferred into 70% EtOH. Sagittal and parasagittal sections from the midline of embryos were stained by standard hematoxylin and eosin staining and the number of apoptotic cells assayed by TUNEL assay (25).

RESULTS
Pik3ca Gene Maps to Mouse Chromosome 3-A novel tetranucleotide repeat (CCTT) n within the fourth intron of Pik3ca was identified when this fragment was sequenced to determine the exon/intron boundaries. The repeat was found to be polymorphic between strains of C57BL/6J and M. spretus. A backcross between these two strains provided by the Jackson Laboratories Backcross DNA Panel Map Service was used to map the locus. Pik3ca maps cleanly to chromosome 3 at 12.9 centimorgans between Il7 and D3Mit21, with no crossovers with markers D3Bir4 and #Sno. This location is syntenic with the region between D3S3682 and D3S1564 on the long arm of human chromosome 3 where human PIK3CA maps (SHGC-12912).

Construction of Targeting Vector and Gene
Targeting-A clone (F1) isolated from a 129SV mouse genomic library (Stratagene) contained the first four coding exons of Pik3ca (Fig. 1). The first coding exon contains 76 bp of 5Ј-untranslated sequence and the first 352 bp of coding sequence including the translation initiation codon (numbering according to Gen-Bank TM accession number U03279).
The targeting construct ( Fig. 1) contains 4.8 kb of sequence 3Ј and 3 kb 5Ј to the first coding exon. Homologous recombination of the targeting construct shown in Fig. 1 results in deletion of this exon. The transcript of the deleted allele lacks the translation initiation codon as well as the domain that mediates the binding to the inter-Src homology 2 domain of a regulatory subunit that is required for catalytic activity of p110 in mammalian cells (11). The next downstream, in-frame ATG occurs at base 15 of the second coding exon (ATTGGCATGC-CAGT); this ATG would allow the rest of the transcript to be translated in frame, although it sits within an incomplete Kozak consensus context (26) that lacks a G in position ϩ4.
Eight out of seventy-two (ϳ11%) ES colonies showed the expected recombinant pattern by genomic Southern blot analysis (Fig. 1). The recombinant ES cells were injected into blastocysts (C57BL6) to generate chimeric males. The chimeras transmitted the disrupted Pik3ca gene to the next (F1) generation.
To identify the stage of embryogenesis at which lethality is occurring in Pik3ca del/del embryos, embryos at different stages of embryonic development were examined and genotyped (Table I). At 10.5 days of embryonic development, no viable Pik3ca del/del homozygote embryos were observed among 19 genotyped. Genotyping of 60 E9.5 embryos showed 17 Pik3ca ϩ/ϩ , 31 Pik3ca ϩ/del , and 12 Pik3ca del/del as well as 9 apparent resorptions that could not be genotyped. Although there appeared to be a deficiency of Pik3ca del/del embryos, a deficiency that could be accounted for by the nine resorptions, the 17:31:12 ratio was not significantly different from the expected ratios of 15:30:15 (chi-square with Yates' correction). At E8.5, 17 embryos showed normal Mendelian ratios (Table I).
From these data, we conclude that homozygote embryos survive to E8.5, begin to die at E9.5, and do not survive to E10.5.
Most Pik3ca del/del homozygote embryos at E9.5 appeared to be viable, with obvious cardiac function, and histological analysis of E9.5 embryo sections stained with hematoxylin and eosin revealed no defect in organogenesis. They were, however, smaller than normal embryos, with fewer somites (average 22) compared with the normal embryos (average 25) from the same litter. The head region of E9.5 Pik3ca del/del homozygote embryos was smaller as compared with normal, and hemorrhage was observed at forehead, snout, and other parts of the body of the Pik3ca del/del embryos (Fig. 2). Heterozygous embryos were normal in appearance.
Pik3Ca Transcripts-Northern blot analysis of RNA from wild type embryos and adult mouse tissues revealed three transcripts (7.5, 5.4, and 4.4 kb) detected by a cDNA probe containing the coding region of Pik3ca cDNA (data not shown). The different transcripts result from alternative splicing in untranslated portions of the gene and are not the result of alternative splicing of coding exons (data not shown). All three transcripts appeared to be expressed at similar levels from E7 to E17 of embryonic development and in different tissues of adult mice, although there was variability in the expression levels of the different sized transcripts among various tissues. Whole mount in situ hybridization at E9.5 showed that the Pik3ca transcript was expressed uniformly throughout the whole embryo (data not shown).
RT-PCR analysis of RNA obtained from Pik3ca ϩ/ϩ and Pik3ca del/del embryos revealed that the deletion allele was transcribed. As shown in Fig. 3, Pik3ca transcripts were present in both Pik3ca ϩ/ϩ and Pik3ca del/del embryos, when assayed by RT-PCR of two segments downstream of the deletion (primer pair RT-1 and RT-2 and primer pair RT-3 and RT-4), whereas the portion of the transcript deleted by the targeted event was missing from Pik3ca del/del RNA (primer pair p110-F and p110-  a Three del/del embryos were found at 10.5 days but were clearly nonviable and undergoing tissue autolysis. b NS, not significant. R). As control, the neomycin resistance gene transcript was amplified from Pik3ca del/del embryos. Proliferation and TUNEL Assay in Embryos-The PI 3-kinases have been implicated both in signaling through growth factors (8,27,28) and in blocking apoptosis through signaling through the Akt Ser/Thr kinase (29,30). The delayed embryonic development and lethality of Pik3ca del/del embryos and the inability of cells derived from these embryos to expand in culture could be due either to failure to proliferate or to excessive apoptosis or both. We tested E9.5 Pik3ca del/del embryos for their ability to proliferate using BrdUrd incorporation in vivo (24) as well as examining them for excess apoptosis by TUNEL assay (25). In wild type mice, the cells in three rapidly dividing regions of an E9.5 embryo revealed heavy BrdUrd labeling, whereas the same regions of Pik3ca del/del littermate embryos incorporated very little BrdUrd (Fig. 4). In contrast to the severe proliferative defect, no abnormality in apoptosis was seen (Fig. 5). The number of cells undergoing apoptosis was counted and averaged in six sagittal and parasagittal sections; wild type embryos contained an average of 58.8 apoptotic cells per section, whereas Pik3ca del/del contained an average of 48.0 apoptotic cells per section (p Ͼ 0.05, Student's t test). The apoptotic cells were mostly located in forebrain and hindbrain with a similar distribution in both genotypes (Fig. 5).
Mouse Embryo Fibroblast Culture-Cells from wild type, heterozygote, and homozygote embryos were obtained by explanting E9.5 embryos. Mouse embryonic fibroblasts grew normally from explants of wild type and Pik3ca ϩ/del heterozygous embryos. In contrast, when explants were obtained from E9.5 Pik3ca del/del embryos, cells would adhere to tissue culture plates but could not be induced to expand in culture in Dulbecco's modified Eagle's medium supplemented with 20% fetal calf serum and a variety of growth factors (bombesin, phorbol 12myristate 13-acetate, and lysophosphatidic acid), used either FIG. 4. Proliferation assay of wild type (panels A-C) and Pik3ca del/del (panels D-F) embryos. Sagittal sections of E9.5 embryos harvested 1 h after intraperitoneal injection of the pregnant mouse with BrdUrd. A and D are from midbrain region, B and E are from region of rhombomere, and C and F are from neural tube. Heavy brown staining of most cells in A-C from wild type embryos indicates that proliferation (as evidenced by BrdUrd incorporation) was taking place. The nearly complete absence of similar staining in the Pik3ca del/del embryos shows that despite generally normal development up to E9.5, embryonic cells had ceased to proliferate. KO, knockout; N, normal.
FIG. 5. TUNEL assay of wild type and Pik3ca del/del embryos at E9.5. A-C are from wild type, with B and C showing higher magnification; D-F are from the Pik3ca del/del embryo, with E and F showing higher magnification of similar locations as shown in B and C. No significant difference in BrdUrd incorporation into nuclei by terminal transferase was seen in Pik3ca del/del embryos as compared with wild type embryos, indicating that there is no significant difference in apoptosis between wild type and Pik3ca del/del embryos. singly or in combination. These growth factors were chosen because they were reported to bypass a block in mitogenic signaling in cultured cells caused by antibodies to Pik3ca (28).
Western Analysis of Regulatory p85/p55 in Homozygous Deletion Embryos-The deletion mutation in Pik3ca eliminates the p85/p55 binding site, but the transcript containing the deletion could theoretically still be translated into a protein lacking the first 122 residues of the full-length p110␣. Antibody directed against residues 189 -390 of p110␣ was used for Western blot analysis of fibroblasts from wild type and Pik3ca del/del embryos. Although this region of p110␣ was not deleted in the knock-out construct, no p110␣ protein was seen (Fig. 6A). Furthermore, by comparing two different amounts of protein (18 and 36 g) from wild type and heterozygous embryos, the amount of p110␣ was clearly reduced in the heterozygotes, consistent with loss of expression of p110␣ resulting from the targeting event. Another difference between Pik3ca ϩ/ϩ and Pik3ca del/del embryos was also apparent when rabbit anti-human p85 polyclonal antibody (Upstate Biotechnology Inc.) was used for Western blotting of cells derived from Pik3ca ϩ/ϩ and Pik3ca del/del embryonic tissue. An 85-kDa fragment recognized by rabbit anti-human p85 (Upstate Biotechnology Inc.) polyclonal antibody in the wild type appeared to be substantially increased in Western blot of Pik3ca del/del embryo tissues (Fig.  6B). Other proteins detected by the antibody were either unchanged or decreased in intensity in lysate from Pik3ca del/del embryos.

DISCUSSION
PtdIns 3,4,5-trisphosphate has been implicated in the response to growth factor signaling through tyrosine kinase receptors for platelet-derived growth factor, insulin, insulin-like growth factor 1, colony-stimulating factor 1, nerve growth factor, hepatocyte growth factor, and stem cell growth factor (8). Phosphoinositides phosphorylated at the 3 position of the inositol ring, including PtdIns 3,4,5-trisphosphate, are generated by the type I A PI 3-kinases, which also interact with Ras proteins in a GTP-dependent manner (31). In addition, Akt Ser/Thr protein kinases, p70 ribosomal S6 kinase and protein kinase C are considered to be downstream targets of PI 3-kinases by receptor-stimulated signaling (6). Other PI 3-kinases have also been found to participate in protein trafficking (7), cytoskeletal organization (32), and regulation of apoptosis (29,30). Given that there are so many PI 3-kinases involved in so many cellular processes, it is important to ask to what extent each of the isozymes of PI 3-kinase is either necessary or sufficient for carrying out these many functions of PI 3-kinases. In particular, is the ␣ isozyme of PI 3-kinase essential for normal embryonic development or is it redundant because a deficiency could be compensated for by other activities, such as the ␤ isozyme?
We created mice homozygous for a deletion within Pik3ca by gene targeting and demonstrated loss of expression of p110␣ by Western analysis using an antibody directed against a region of the protein not involved in the deletion event. Pik3ca del/del homozygotes have delayed development and undergo intrauterine death between E9.5 and E10.5 of embryonic development. The developmental delay and growth failure were due to an inability of Pik3ca del/del embryos to proliferate in response to mitogenic stimulation. Consistent with this severe proliferation defect was our finding that all attempts to grow cells obtained from explants of E9.5 Pik3ca del/del embryos from 129/SV inbred matings failed. Cells from Pik3ca del/del embryos failed to expand in culture and appeared resistant to the mitogenic effects of fetal calf serum as well as defined growth factors such as bombesin, lysophosphatidic acid, and phorbol 12-myristate 13-acetate. These results were consistent with the findings of Nave et al. (33), who reported PI 3-kinase is involved in signaling through phorbol 12-myristate 13-acetate, but contrasts with the findings of Roche et al. (28), who reported that neither bombesin nor lysophosphatidic acid uses signaling pathways involving PI 3-kinase to stimulate growth.
Because of the observation that PI 3-kinases activate the AKT protein kinase and inhibit apoptosis in the nervous system (29,30), we were particularly interested in determining whether there was increased apoptosis in the Pik3ca Ϫ/Ϫ embryos, particularly in the developing brain. In fact, embryos did not demonstrate increased numbers of apoptotic cells, either in the region of the developing brain or elsewhere. We conclude from these experiments that the PI 3-kinase encoded by Pik3ca is absolutely required for normal embryonic growth and development; the phenotype seen in Pik3ca del/del mice results from an inability to proliferate in response to growth factor signaling rather than from unregulated apoptosis. However, we cannot eliminate the possibility that abnormalities in other phosphoinositides phosphorylated at the 3 position of inositol by the class I A PI 3-kinase we disrupted in these mice may play a role in the embryonic lethality through other unknown mechanisms.
Given the wide variety of cellular processes in which PI 3-kinases are involved, a puzzling aspect of the phenotype of Pik3ca del/del homozygotes is that the embryos manage to survive to E9.5 of embryonic development. There may be an alternative supply of PI 3-kinase that is either lost or becomes inadequate by E9.5 in face of a sharp increase in demand for PtdIns 3,4,5-trisphosphate that occurs normally at this stage of development. For example, a maternal contribution of p110␣ in the oocyte may only be sufficient to support limited early development. Alternatively, embryonic forms of PI 3-kinase with limited temporal expression might compensate early in development but not after E9.5. Finally, other PI 3-kinase catalytic subunits, such as the p110␤ subunit, might be sufficient to allow development to occur prior to E9.5.
The period between E9.5 and E10.5 is a time of markedly increased proliferation and differentiation as organogenesis and placentation accelerate rapidly in the embryo (34). A fourchambered heart begins to beat, and angiogenesis allows the vascular system to expand in order to deliver nutrients and remove waste products. We might hypothesize that as the E9 -E9.5 embryo faces markedly increased demands for proliferation, growth, and differentiation, intrauterine death of Pik3ca del/del embryos occurs at this stage because the increased demand exceeds generate residual capacity of PI 3-kinase pathways to phosphatidylinositols required for proliferation. It is interesting to note that the Pik3ca del/del embryos demonstrated areas of extravasated blood, suggestive of defective angiogenesis. A role in angiogenesis for phosphatidylinositols phosphorylated by PI 3-kinase at the 3 position of inositol is supported by studies in which wortmannin inhibited angiogenesis in a chick embryo chorioallantoic membrane system (35).
One interesting finding in the Pik3ca del/del homozygous embryos was an apparent increase in p85/p55 as determined by Western blot analysis of embryo tissue obtained at E9.5. These data raise the intriguing possibility that one component of the deleterious effect of the Pik3ca del allele on embryonic development may be mediated through excess p85/p55 rather than simple loss of function of p110␣. Overexpression of p85 can interfere with the binding of PI 3-kinases to other proteins by disrupting the binding between the Src homology 2 domains of the p85 subunits of the PI 3-kinases with their targets (36). There is very little information on the regulation of transcription and turnover of p85/p55 and an investigation into the mechanism of this elevation of p85/p55 is greatly hampered by our inability to propagate Pik3ca del/del cell lines. The increase in p85 protein seen in the Pik3ca del/del embryos points to a previously unrecognized aspect of p85 regulation that deserves further investigation.