Register for Mechanobiology 2016!

Crucial role of phosphatidylinositol 4-kinase IIIα in development of zebrafish pectoral fin is linked to phosphoinositide 3-kinase and FGF signaling
Hui Ma, Trevor Blake, Ajay Chitnis, Paul Liu, Tamas Balla


Phosphatidylinositol 4-kinases (PI4Ks) catalyze the first committed step in the synthesis of phosphoinositides, important lipid regulators of signaling and trafficking pathways. Here we cloned Pik4a, one of the zebrafish PI4K enzymes, and studied its role(s) in vertebrate development using morpholino oligonucleotide-based gene silencing in zebrafish. Downregulation of Pik4a led to multiple developmental abnormalities, affecting the brain, heart, trunk and most prominently causing loss of pectoral fins. Strikingly similar defects were caused by treatment of the developing embryos with the phosphoinositide 3-kinase (PI3K) inhibitor, LY294002. To investigate the cause of the pectoral fin developmental defect, we focused on fibroblast growth factor (FGF) signaling pathways because vertebrate limb development requires the concerted action of a series of FGF ligands. Using in situ hybridization, the pectoral fin defect was traced to disruption of the early FGF signaling loops that are crucial for the establishment of the sharp signaling center formed by the apical ectodermal ridge and the underlying mesenchyme. This, in turn caused a prominent loss of the induction of one of the mitogen-activated protein kinase (MAPK) phosphatases, Mkp3, an essential intermediate in vertebrate limb development. These changes were associated with impaired proliferation in the developing fin bud due to a loss of balance between the MAPK and PI3K branch of FGF-initiated signals. Our results identify Pik4a as an upstream partner of PI3Ks in the signaling cascade orchestrated by FGF receptors with a prominent role in forelimb development.


Embryonic development relies on the coordinated interplay between several signaling pathways to ensure the right balance between cell proliferation, differentiation and apoptosis. Signaling during early embryonic development is based on external cues and their gradients that are derived from adjacent cells and differs in many ways from the classical endocrine signals that act at a distance with the help of circulating fluids. Secreted fibroblast growth factor (FGF) ligands have important roles in many developmental processes (Bottcher and Niehrs, 2005; Thisse and Thisse, 2005). These ligands act on four FGF receptor subtypes and activate several signal transduction cascades including the mitogen-activated protein kinase (MAPK) and phosphoinositide 3-kinase (PI3K) pathways (Eswarakumar et al., 2005). Very limited information is available on how these signaling pathways are linked to the developmental program during FGF action. A recent study showed that the MAP kinase phosphatase 3 (MKP3) is a crucial component in the FGF8 signaling cascade during vertebrate limb development (Kawakami et al., 2003). In that study the induction of mkp3 by the apical ectodermal ridge (AER)-derived FGF8 within the limb bud was found to occur through PI3Ks (Kawakami et al., 2003). These studies were the first to establish the importance of the PI3K signaling pathway in FGF action during vertebrate limb development.

In the present study we investigated whether phosphatidylinositol 4-kinase III alpha (Pi4ka) can be linked to PI3K signaling during pectoral fin development in zebrafish. (The nomenclature of PI4Ks is somewhat confusing. The human PI4KIIIα gene was formerly designated as PIK4CA. Currently, the human gene is named PI4KA whereas the gene encoding PI4K type II alpha is termed PI4K2A. Based on this, a more logical name for the gene encoding PI4KIIIα would be PI4K3A. However, we have followed the current nomenclature and used pi4ka for the zebrafish gene encoding Pi4ka.) PI4Ks catalyze the first step in the synthesis of phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P2), the major phosphoinositide of the plasma membrane and a precursor of the PI3K lipid product, PtdIns(3,4,5)P3. Hence, it is expected that PI4Ks are placed upstream from the PI3Ks in the cascade generating PtdIns(3,4,5)P3. There are four distinct PI4K genes in vertebrates (Balla and Balla, 2006) including zebrafish. In this study we chose to study the Pi4ka enzyme because we found recently that it contributes to the generation of the plasma membrane pool of phosphoinositides in mammalian cells (Balla et al., 2007). Our results show that PI3K inhibition as well as downregulation of Pi4ka causes a striking defect in zebrafish pectoral fin development. Detailed analysis of this phenotype showed a defect in the establishment of the sharp organizing center between the Fgf24-positive cells within the apical ectodermal ridge (AER) and the Fgf10- and Mkp3-producing cells within the underlying mesenchyme leading to greatly reduced induction of mkp3 within the fin bud. In addition, downregulation of Pi4ka and inhibition of PI3K were both associated with increased generalized apoptosis. Decreased proliferation was also prominent in the fin bud as a result of an imbalance between the MAPK and PI3K-Akt signaling pathways.


Pi4ka deficiency and PI3K inhibition impairs pectoral fin development in zebrafish embryos

We cloned the full-length zebrafish pi4ka and located the encoding gene on chromosome 10. (The sequences of all four fish PI4Ks have been deposited in GenBank under the following accession numbers: AY929293, PI4KIIIβ; AY929292, PI4KIIIα; AY929291, PI4KIIβ; AY929290, PI4KIIα.) The gene comprises 54 exons with an ORF of 6180 bp. No other copies of this gene were found in the zebrafish genome. Sequence comparisons indicated that the Pik4a protein shares a high degree of sequence homology with its mammalian orthologs, exhibiting 82.3% overall amino acid identity with the human sequence (AF012872). The zebrafish protein shares the same domain structure with the other type III mammalian PI4Ks; it contains a lipid kinase unique domain followed by a putative PH domain and the lipid kinase catalytic domain (Fig. 1A). In vitro translated Pik4a displayed strong phosphatidylinositol 4-kinase activity, that was sensitive to inhibition by wortmannin (Wm), confirming its identity as a type III PI4K (not shown).

Whole-mount in situ hybridization analysis showed the pi4ka transcript detectable at 1.5 hpf, indicating its probable maternal origin. It was uniformly expressed throughout the embryos from 3 to 24 hpf, with distinct expression in the somites. At later stages (36, 48 and 60 hpf), the expression was primarily restricted to the brain, the branchial arches and the pectoral fin-bud mesenchyme (Fig. 1B). To downregulate pi4ka, two morpholinos, MO1 and MO2 were designed to target the exon-intron boundaries flanking intron 49 to block splicing of exon 50, which contains the most conserved part of the catalytic domain. RT-PCR analysis showed their effectiveness when injected into freshly fertilized eggs (Fig. 1C left). Two additional non-overlapping antisense MOs (MO3 and MO4) targeted the 5′ UTR of pi4ka and their efficiency in blocking the translation of the transcript was confirmed by an in vitro translation assay (Fig. 1C right).

Injection of MO1 or MO2 produced a complex phenotype affecting several structures (supplementary material Table S1). At the 24 hpf stage, mutant brains were dark, non-transparent and atrophic. At 48 hpf the most conspicuous defects were smaller heads and eyes, hooked tails, domed heads, shortened body axis and pericardial edema (Fig. 1D). However, the most prominent feature of the mutant embryos was the abnormal development of pectoral fins most obvious at 72 hpf stage (Fig. 1D). Injection of MO3 caused similar abnormalities but these were less severe than those of the MO1-injected embryos (supplementary material Table S1). We attributed the milder defects in the MO3-injected embryos to the possibility that an alternative downstream start site exist that could yield a limited amount of partially functional enzyme in the MO3-treated embryos but this was not studied in further detail. Injection of controls did not cause any developmental defects (not shown).

These results suggested that Pi4ka plays an important role in limb or fin initiation and/or differentiation. We attempted to rescue the phenotype by injecting one-cell stage embryos with wild-type pi4ka mRNA. The translation of this mRNA is not expected to be sensitive to the MOs that interfere with mRNA splicing at the transcriptional level. As shown in Fig. 2A,B, co-injection of 1 ng (n=60) or 3 ng (n=79) pi4ka full-length mRNA with MO1 significantly alleviated the pectoral fin phenotype (55% and 81%, respectively). In addition, more than half of pi4ka morphants injected with full length mRNA together with MO1 showed normal brain size and eye development (Fig. 2A). Injection of pi4ka mRNA alone did not cause any gross abnormalities (not shown).

Fig. 1.

Expression and downregulation of zebrafish Pi4ka. (A) Pi4ka cloned from zebrafish shows high homology to other mammalian Pi4ka enzymes sharing the same domain organization and highest conservation within the C-terminal catalytic domain. (B) Expression pattern of pi4ka mRNA during zebrafish embryogenesis. Expression is ubiquitous in early embryos (1.5-24 hpf) but is primarily restricted to the brain, branchial arches (ba), and fin buds (fb, indicated by arrowheads) at later stages (36 and 48 hpf). (C) Downregulation of zebrafish Pi4ka by morpholino injection targeting the splicing of exon 50 that encodes a crucial region within the catalytic domain (MO1 and MO2). Left panels: RT-PCR analysis showed that antisense morpholinos could eliminate the transcript containing exon 50. Both MOs caused exon skipping in a dose-dependent manner, whereas control injections (FITC) were without effect. Right panel: in vitro translation assay showing the ability of MO3 and MO4 targeting the translation initiation site to reduce in vitro translation of Pi4ka (cMO, control morpholino). (D) Lateral view of MO1 (9 ng)-injected morphants show a complex phenotype affecting several structures. Note the shorter trunk with curved tail, smaller head and eye, reduced and disorganized pigmentation, and pericardial edema. Dorsal view of the embryos at 72 hpf, demonstrate the larger yolk sac, smaller head and eyes, and the loss of pectoral fins (pf, arrowheads) in morphants.

To determine whether inhibition of the PI3K-Akt signaling could reproduce some of the effects we observed in embryonic development in Pi4ka-deficient embryos, we exposed developing embryos to the PI3K inhibitor LY294002 (Vlahos et al., 1994). We treated embryos with 10 μM of the inhibitor from the onset of blastulation and analyzed the phenotype. This is a lower concentration than used previously (30-100 μM) to effectively block PI3K activity in Xenopus embryos (Carballada et al., 2001; Peng et al., 2004), but 30 μM LY294002 caused 76% of the embryos to die before the 24 hpf stage (not shown). In LY294002-treated embryos the size of brain and eyes was reduced and the body axis severely shortened at 24 hpf stage. At 48 hpf, pericardial edema has developed. At 72 hpf, LY294002-treated embryos also had smaller pectoral fins. These changes were very reminiscent of those observed in pi4ka morphants (Fig. 2C).

Fig. 2.

pi4ka mRNA co-injection can reverse, and PI3K inhibitors can mimic the MO1-induced phenotype. Fertilized zebrafish eggs were injected at the one-cell stage with either a control morpholino (cMO) or MO1 together with in vitro transcribed and purified pi4ka mRNA. Embryos were allowed to develop for 1-3 days (24-72 hpf). MO1 interferes with splicing of the endogenous mRNA causing exon skipping but does not affect the translation of the injected full length mRNA. (A) Co-injection of 1 ng mRNA can largely alleviate the brain defect at 24 hpf and 48 hpf. (B) The pectoral fin develops to almost normal length, but it poorly separates from the yolk sac and does not fully develop in the rescued embryos (left panels; arrowhead). Measurements of the length of the pectoral fin at the 72 hpf stage (right panel) show partial rescue by the injected mRNA (values are means ± s.e.m. of 70-110 observations). (C) The PI3K inhibitor, LY294002 (10 μM) mimics the effects of pi4ka downregulation on zebrafish development. LY-treated embryos have a smaller brain and eyes, a hooked tail and pericardial edema at 48 hpf. At 72 hpf, LY-treated embryos have larger yolk sacs and also reduced pectoral fins. These changes are very similar to those observed in pi4ka downregulated embryos (see Fig. 1D).

MKP3 induction is grossly affected in pi4ka-deficient embryos and after pharmacological inhibition of PI3K

As discussed above, the MKP3 MAPK phosphatase is induced in the mesenchyme by FGF ligands of the AER via the PI3K-Akt pathway during vertebrate limb or fin development (Kawakami et al., 2003). Therefore, we next determined the expression of mkp3 in the pi4ka morphants and in LY294002-treated embryos. In control embryos, mkp3 is detectable in the pectoral fin bud, the mid-hind brain boundary, branchial arches (Fig. 3A) and the tip of the tail (not shown) at 24 hpf. However, mkp3 expression was almost completely lost at this stage in the pectoral fin bud and expressed at a greatly reduced level in the branchial arches of morphant embryos (Fig. 3A). Expression of mkp3 was also weaker in the mid-hindbrain boundary and in the tip of the tail (not shown). At a later stage (36 hpf), the expression of mkp3 was substantially weaker in the branchial arches, pectoral fin bud (Fig. 3A) and mid-hindbrain boundary of mutant embryos, whereas its expression in the telencephalon and tail were less affected (not shown). Similar, but even stronger, reductions in mkp3 induction were observed in the fin bud area after LY294002 treatment.

Fig. 3.

Effect of pi4ka downregulation and LY294002 treatment expression of several genes on the early FGF signaling pathway. Fertilized zebrafish eggs were injected at the one-cell stage either with a control morpholino (cMO) or with MO1 or treated with 10 μM LY294002 and analyzed by whole-mount in situ hybridization for the presence of the indicated mRNA at different stages of development. (A) At 24 hpf embryos mkp3 expression is reduced in the fin-bud mesenchyme in the morphants as well as in embryos treated with 10 μM LY294002. This difference becomes very prominent by the 36 hpf stage and also affects the branchial arches. (B) Expression of tbx5 is only slightly reduced in the fin-bud mesenchyme of morphant embryos both at 24 and 32 hpf. The expression of fgf24 is comparable to controls in the fin-bud mesenchyme of pi4ka morphant embryos at 24 hpf. However, at 32 hpf, fgf24-positive cells move to the AER in control embryos but are still dispersed in the mesenchyme and fail to migrate to the AER in pi4ka morphant embryos. As a result, expression of fgf10, which is controlled by fgf24 in the AER, is greatly reduced in the fin-bud mesenchyme of Pi4ka morphant embryos. Note that some signals in the control are saturated because identical exposures were chosen to still see the weak signal in the morphant embryos.

Fig. 4.

Fluorescent in situ hybridization (FISH) and confocal analysis of fgf24 and mkp3 expression and localization. (A) Time sequence of the position of the fgf24-positive cells showing their migration to the AER that is very clear by 36 hpf. At the same time, the number of mkp3-positive cells and their signal progressively increases but remains beneath the AER. The DAPI staining is not shown for better clarity but the contours of the ectoderm based on the DAPI staining are indicated by a dashed line (see panel B). (B) Representative images showing the position of the developing fin bud in a lateral view of a control fish embryo at the 36 hpf stage analyzed by FISH for mkp3 (orange) and counterstained with DAPI (green). The boxed region of each image is enlarged in the panel to the right. (C) Left: schematic illustration of the interplay between the AER and the underlying mesenchyme showing the position of fgf24- and mkp3-positive cells at the 30 hpf stage. Right: at the 36 hpf stage the fgf24-expressing cells do not approach the apical layer of the budding fin and the AER is not developing. As a consequence the mkp3 expression remains weak and the cells are scattered in the pi4ka MO1-injected and LY294002-treated embryos. Note that these pictures do not reflect the true relative signals in the control and morphant embryos, because the images were adjusted to make the position of the positive cells clear. LPM, lateral plate mesenchyme.

The severe defect in mkp3 induction prompted us to investigate the upstream steps in the origination of the developing fin bud. One of the earliest genes activated in the lateral plate mesenchyme (LPM) in fin-bud development is the Tbx5 transcription factor, a master regulator of many genes in the limb-developmental program (Capdevila and Izpisua Belmonte, 2001). The first member of the Fgf family that appears at this location is fgf24 followed by fgf10 (Fischer et al., 2003). As shown in Fig. 3B, there was only a slight reduction in the expression of tbx5 in the lateral plate mesenchyme at the 24 hpf stage in pi4ka morphants. At this point the expression of fgf24 was slightly reduced but the fgf24-positive cells appeared to be more scattered in the morphant embryos. Expression of Fgf10, however, was already more severely affected even at this early stage. Analysis of embryos at a slightly later stage (32 hpf) showed more prominent differences. Here, the migration of fgf24-positive cells toward the AER could be clearly observed in control embryos (shown by the arrows) together with high expression of fgf10. By contrast, fgf24-positive cells failed to concentrate in the AER and fgf10 induction had been almost abolished in the pi4ka morphant embryos (Fig. 3B). The same changes were observed after LY294002 treatment (not shown).

This process was followed at higher spatial resolution using fluorescent in situ hybridization (Clay and Ramakrishnan, 2005) and confocal microscopy to better resolve the distribution of the fgf24 and mkp3-positive cells. Fig. 4A shows that in control embryos, the fgf24-positive cells that had already appeared at 24 hours gradually moved toward the ectoderm and populated the AER. By 36 hpf, these fgf24-positive cells were clearly separated from the mkp3-positive cells that were located in the mesenchyme below the AER (Fig. 4A,B). Both in pi4ka morpholino-treated embryos and after LY294002 treatment, the fgf24-positive cells appeared at the early stages but remained scattered at the later stages when the AER is supposed to form. These cells did not move toward the ectoderm and the AER did not develop properly (Fig. 4C, upper row). Since the AER is the signaling center that controls fin/limb outgrowth, the primordial limb area remains rudimentary. This is also reflected in the mkp3 distribution and expression, as the early mkp3-positive cells do not show proliferation and remain weak and scattered by 36 hours (Fig. 4C, lower row). By contrast, in control embryos the mkp3-positive cells are present in great number lining up against the AER within the mesenchyme (Fig. 4B). These results suggested that the early mkp3-positive cells do not undergo proliferation and are probably subject to apoptosis because of the lack of an appropriate AER.

Loss of Pi4ka leads to widespread cell death and inhibits cell proliferation in the fin bud

These data prompted us to test whether apoptosis and proliferation is affected by these manipulations. For this, injected embryos at different stages were stained with Acridine orange (AO) or were subjected to TUNEL analysis to detect apoptotic cells. Fig. 5A shows that only a few AO-staining or TUNEL-positive cells were observed in control embryos injected with negative MOs at all development stages tested. However, embryos injected with pi4ka MOs had high numbers of apoptotic cells throughout all tissues at 24 hpf stage, especially in the brain, eyes and the trunk. Increased cell death was not particularly apparent in the lateral plate mesoderm that later gives rise to the pectoral fin. LY294002 treatment caused similar increases in apoptotic cells at all stages examined (Fig. 5A). It is important to note that parallel downregulation of p53 by morpholino injections failed to prevent the increased general apoptosis (or the pectoral fin defect) evoked by Pi4ka downregulation (data not shown).

Fig. 5.

Loss of Pi4ka function leads to widespread cell death and reduced cell proliferation. (A) Cell death detected by TUNEL analysis showed few positive cells in control embryos (cMO) at 24 hpf (i, iii) and 36 hpf stage (v). Embryos injected with pi4ka MO1 had large numbers of apoptotic cells throughout the embryo at both 24 and 36 hpf stages especially in the brain, eyes and the trunk. Notably, there was significant cell death in the lateral plate mesoderm region (iv, vi) the early precursor of the pectoral fin bud. Acridine orange (AO) staining (vii-x) showed similar pattern of apoptosis at 24 hpf in pi4ka MO1-injected (vii versus viii) or LY294002-treated (ix versus x) embryos. (B) Proliferating cells were detected by immunohistochemistry using anti-H3P antibody. Proliferation was significantly lower in the brain of pi4ka MO1-injected embryos (MO1) than in that of controls (cMO). Proliferation was also greatly reduced at the somites and lateral plate mesenchyme at 24 hpf and almost completely disappeared in the pectoral fin buds region of pi4ka morphants at 48 hpf. (C) H3P-positive cells in the mesenchyme of fin buds at 48 hpf were quantified.Values are means ± s.d. of at least 20 measurements on independent embryos.

We also identified the proliferating cells using an anti-H3P antibody, widely used as a mitosis marker (Wei et al., 1998). H3P-positive cells were broadly detected in the brain region and trunk of control MO-injected embryos at 24 hpf (Fig. 5B). By contrast, they were seldom observed in the same parts of pi4ka morphants. This suggested that Pi4ka-depleted mutants underwent lower level of mitosis. H3P-positive cells were also detected in the mesenchyme of fin buds of control embryos at 48 hpf, but their number in the mesenchyme of fin buds of the MO1-injected embryos was greatly reduced by 72% (Fig. 5C). These results showed that cell proliferation in the mesenchyme of the Pi4ka-depleted fin buds was considerably decreased. Together these data indicated that reduction of Pi4ka function during early stages of zebrafish development leads to a major imbalance between apoptosis and proliferation with a decreased proliferation dominating the picture in the case of the developing fin bud.


The experiments described above were designed to clarify the link between PI4KIIIα and PI3K signaling using the zebrafish as a model. This model offered several advantages. First, the developmental processes are thoroughly explored and mapped in this species with several genes and signaling pathways characterized and their testing reagents available. Second, the morpholino-induced downregulation of genes is not an all or none process and, therefore, even essential genes can be studied when remaining small expression permits survival of the cells, and pinpoint the pathways that most depend on the expression of the gene. The recently identified role of PI3K signaling in vertebrate limb development (Kawakami et al., 2003) also offered a biological readout that could be studied in a whole organism.

Downregulation of Pi4ka caused a number of developmental abnormalities consistent with a presumed pleiotropic role of the enzyme in trafficking and signaling (Balla and Balla, 2006). However, one of the most prominent defects was the lack of pectoral fin development. Notably, the brain and pectoral fin buds were also the areas showing highest expression of the transcript in normal fish embryos. The impaired fin development pointed to a defect in FGF signaling. There is ample literature suggesting that sequential activation of members of the FGF family are the key to the development of pectoral fin in zebrafish (Grandel and Schulte-Merker, 1998), a process analogous to forelimb development in birds and mammals. Vertebrate limb/fin development has two crucial phases: first is the establishment of a limb field at the lateral plate mesenchyme and second, the limb outgrowth that is controlled by the interplay between the AER and the underlying mesenchyme. In the fish, Fgf24, a member of the FGF8/17/18 family, is the earliest Fgf expressed in the pectoral fin bud, and it promotes migration of tbx5-positive cells towards the fin field and helps maintain tbx5 expression (Fischer et al., 2003) (Fig. 4C). In pi4ka morphant embryos the establishment of the limb field appears normal, as indicated by the early markers, tbx5 and fgf24. By contrast, limb outgrowth does not occur and the AER does not form in the morphant embryos. This is clearly manifested in the lack of fgf10 induction in the pi4ka morphant embryos. Disruption of the mouse Fgf10 gene, which is expressed in the lateral plate mesoderm even before any visible outgrowth occurs (Ohuchi et al., 1997), also results in complete loss of limbs (Min et al., 1998; Sekine et al., 1999). The AER-derived Fgf8 is the main signal to the underlying mesenchyme to direct outgrowth of the limb bud and maintain fgf10 expression as well as the further induction of Mkp3, which is also an essential intermediate in limb development (Kawakami et al., 2003) (Fig. 6). The substantial reduction both in fgf10 and mkp3 expression in the pectoral fin bud in the morpholino-injected embryos suggested that the reduced Pi4ka activity impacts AER development and functions.

Most of the effects of Pi4ka morpholino injections in fin development were phenocopied by the PI3K inhibitor, LY294002, confirming previous claims that the PI3K branch of FGF signaling is important for limb development (Kawakami et al., 2003). Although LY294002 inhibits mammalian casein kinase 2 (Davies et al., 2000) and at higher concentration the type III PI4Ks (Balla et al., 2008), the Kawakami study together with the results of another study in zebrafish identifying PI3Ks necessary for process formation and cell polarization and migration of mesendodermal cells during gastrulation (Montero et al., 2003) makes it likely that the observed defects are the results of PI3K inhibition. Importantly, higher concentrations of LY294002 used in the latter study (30 μM) almost completely stopped development, but at a reduced concentration (10 μM) we could reproducibly observe the pectoral fin developmental defect. It is possible that a migration defect preventing the fgf24-positive cells from migrating and forming the AER contributes to the observed limb phenotype. However, it is more likely that PI3K is required for the proper sequence of FGF ligand induction, which sets up the activator-inhibitor loops between the AER and underlying mesenchyme sharpening an initially broader signaling domain to form a tighter organizing center that controls limb outgrowth. Another possible explanation for the lack of fin bud outgrowth is the increased apoptosis that is observed in the embryos treated with LY294002 or injected with the pi4ka morpholinos. Increased apoptosis is not unexpected when the PI3K signaling pathway is inhibited. However, we did not observe a particularly high apoptotic activity in the fin-bud area mesenchyme and it was the decreased rate of proliferation that dominated in the morphant and LY294002-treated embryos. Nonetheless, it is probable that increased apoptosis also contributed to the defect in limb outgrowth. In this regards it is worth noting that a generalized increased apoptosis reported in perp (a p53 downstream target gene) overexpression did not result in a pectoral fin developmental defect (Nowak et al., 2005). The strong functional connection between Pi4ka and PI3K in FGF signaling at the pectoral fin suggests that this PI4K is important for the generation of the PtdIns(4,5)P2 pool of the plasma membrane that is used as substrate by the PI3Ks (Fig. 6).

Fig. 6.

Schematics of the proposed participation of Pi4ka and PI3Ks in the FGF signaling cascade in the developing fin bud mesenchyme. Pi4ka contributes to the generation of the PtdIns(4,5)P2 pool that is converted by the PI3Ks to PtdIns(3,4,5)P3 in the plasma membrane. Production of this latter lipid activates the Akt signaling cascade that regulates a host of downstream genes important for proliferation or protection of the cells from apoptosis. One of the most important genes regulated by this pathway is the MAPK phosphatase, MKP3 that is a key intermediate in driving pectoral fin development (Kawakami et al., 2003). This phosphatase keeps a right balance between MAPK and the Akt pathways. The right panel shows the FGF signaling cascade in which the Fgf24-positive cell migration appears to be the first step showing impairment after Pi4ka depletion.

An overall decrease in PI4K activity or changes in endogenous PtdIns4P, PtdIns(4,5)P2 or PtdIns(3,4,5)P3 levels is not expected to be demonstrable after these manipulations. Such changes have been difficult or impossible to show after downregulation of type III PI4Ks even in cultured cells with well established lipid labeling or imaging methods. It is most likely that the lipid changes affect specific cellular compartments and only in certain cells and tissues during development. Generation of transgenic fish lines expressing phosphoinositide-specific fluorescent reporters might help us identify such localized and subtle changes. These efforts are in progress in several laboratories. The importance of specific PI4Ks at the level of the whole organism was shown by a recent study describing a late onset spino-cerebellar degeneration in a mouse line with gene-trapped PI4K type-IIα (Simons et al., 2009). Those studies showed that a complete elimination of the PI4KIIα caused no major developmental abnormalities and the mice started to develop the above pathology only at a later stage of life causing a reduced life span of the animals. This mouse study exemplifies the difficulty of linking a complex phenotype to a specific biochemical process at the level of the whole organism and the unpredictability of a phenotype based on our knowledge of the cell biology of the particular enzyme. Nevertheless, such studies provide important pieces of information to advance our understanding of the role of the lipid kinases in vertebrate physiology and development.

More studies are required to substantiate the validity of the tentative model shown in Fig. 6. However, the complex involvement of the phosphoinositide cascade at many levels of zebrafish development might make this a rather difficult task, and we need to emphasize that even the partial inactivation of the fish Pi4ka resulted in a complex set of developmental defects. The importance of other elements of the phosphoinositide signaling cascade such as phospholipase C-mediated PtdIns(4,5)P2 breakdown and InsP3 production was shown in the very early four-cell stage development (Ashworth et al., 2007) as well as in dorsoventral patterning and Wnt signaling (Westfall et al., 2003). PLCγ1 was also found to be important in zebrafish primitive hematopoiesis (Ma et al., 2007) and vasculogenesis (Lawson et al., 2003). However, it will be important to determine whether kinase-dead versions of Pi4ka can rescue the pectoral-fin defect and whether alternative induction or activation of the putative downstream signaling molecules such as Akt and MKP3 can rescue this phenotype.

In summary, the present studies identify a principal component of the PI3K branch of FGF signaling cascade, namely Pi4ka. The probable importance of this enzyme is to supply the plasma membrane with PtdIns4P that is further converted to PtdIns(4,5)P2 by a phosphatidylinositol 4-phosphate 5-kinase and, hence, supply the PI3K with its lipid substrate. This role of the fish Pi4kα enzyme is consistent with its evolutionary conservation as its homologs are responsible for the maintenance of plasma membrane phosphoinositide pools in yeast (Audhya and Emr, 2002) and probably in mammalian cells (Balla et al., 2007). Pi4ka as well as PI3Ks are vital for the induction of fgf10 and maintenance of fgf24 in the AER and expression of the MKP3 MAPK phosphatase gene, a previously identified FGF-regulated gene that is essential for limb development. Further studies are needed to identify additional elements of this signaling pathway as well as their exact contribution to the complex roles of FGF in vertebrate development.

Materials and Methods

Zebrafish maintenance

Zebrafish (Danio rerio) were raised and handled at the Fish Core Facility of the National Human Genome Research Institute under an approved National Institutes of Health animal use protocol as previously described (Blake et al., 2000). After breeding, embryos were maintained in egg water [0.006% Instant Ocean sea salt (Aquarium Systems) in distilled water] with 2 ppm methylene blue to prevent fungal growth. Fish of the wild-type strain EKW was used for the production of wild-type embryos for in situ hybridization and RNA isolation. Embryos were staged according to morphology and hours post-fertilization (hpf).

Isolation and sequence analysis of zebrafish PI4K cDNAs

Pieces of the zebrafish PI4K IIIα gene were identified by BLAST searches of the zebrafish EST and genomic sequence databases using the bovine PI4KIIIα (U88532) amino acid sequence. The full-length pi4ka cDNA was cloned from two fragments obtained by two RT-PCR reactions using the following primer pairs: (5′-ATGTGTCCTGTGGACATCCGTGG-3′ and 5′-GTAGGTGTGATTTGGTGACCGATG-3′; 5′-TGGACAGCATAGTGAAGGACTTTGC-3′ and 5′-GTCTCAGTATGGGATTTGGTTCTGG-3′). The full-length pi4ka cDNA was then assembled by combining these two overlapping sequences in the pGEM T-Easy vector (Promega).

Assay of phosphatidylinositol 4-kinase

Full-length Pi4ka protein was obtained by in vitro translation using the TNT quick coupled transcription/translation systems (Promega). The activity of the enzyme was then measured with the standard kinase reaction method described previously (Downing et al., 1996) in the absence or presence of 1 μM wortmannin, an inhibitor of PI3Ks and the type III PI4K isoforms.

Whole-mount in situ hybridization and other histochemical methods

In situ hybridization was performed as previously described (Lyons et al., 2001). Antisense RNA probes were generated with UTP-digoxigenin according to the manufacturer's instructions (Roche Diagnostics). The antisense probe (from 371 bp to 823 bp of the coding region) was used to analyze zebrafish Pi4ka expression. The plasmid was digested with SacII, and SP6 RNA polymerase was used to synthesize antisense RNA. In addition, the following RNA probes were used: fgf8 (Reifers et al., 1998), mpk3 (Tsang et al., 2004) and sef (Tsang et al., 2002). Probes for tbx5, fgf10, fgf16 and fgf24 were generated from the full-length coding sequences amplified from 24 hpf stage embryos and cloned in the pGEM-T-Easy T/A cloning plasmids. (The following primers were used: fgf10 Fw, 5′-CAATGTGCAAATGGAAAGTGAC-3′; fgf10 Rev, 5′-CCAAGTCTTTCCTCAGTGCAG-3′; fgf16 Fw, 5′-ATGGCAGAGGTGGCTGGATTTC-3′; fgf16 Rev, 5′-TCACCTATGCTGGAACAATTCTC-3′; fgf24 Fw, 5′-GATGTCTGTTCTGCCGTCAAG-3′; fgf24 Rev, 5′-GTCCTTTGTGAACTTGACTCAG-3′; tbx5 Fw, 5′-CATGGCGGACAGTGAAGACAC-3′; tbx5 Rev, 5′-TCTGCATGTTAGCTGGCTTCG-3′. Fluorescent in situ hybridization was performed as previously described (Clay and Ramakrishnan, 2005). In brief, after digoxigenin (DIG)-probe binding, the embryos were hybridized with anti-DIG-POD antibody (Roche) and co-stained for 1 hour with TSA-AlexaFluor 555 (1:200, from Invitrogen) and DAPI (1.5 nM). Embryos were then mounted in 0.8% low melting point agarose for fluorescent analysis. Images were collected with a Zeiss LSM 510 Meta Inverted microscope and analyzed by Zeiss Image Browser and Adobe Photoshop softwares.

For detection of apoptotic cells, the TUNEL assay was used or fresh embryos were stained with Acridine orange as described by previously (Nowak et al., 2005). Both bright-field images and fluorescent images of Acridine orange staining were taken on a Leica MZ16 F stereomicroscope (Leica). Bright-field and fluorescent images were superimposed with Openlab software (Improvision, Lexington, MA).

Morpholino antisense oligonucleotides and RT-PCR

Morpholino antisense oligonucleotides (morpholinos or MOs) were purchased from Gene Tools (Philomath, OR). Two MOs interfering with pi4ka mRNA splicing were designed. MO1 (5′-aatgtgtgtaacCTTCTGGAAAGCC-3′; lower case indicates intron nucleotides and uppercase indicates exon nucleotides) was to target the exon49-intron49 junction whereas MO2 (5′-AGTTATACCGAGCctagaaatgagc-3′) was to target the intron49-exon50 boundary. MOs were resuspended in Danieau solution (5 mM HEPES, 58 mM NaCl, 0.7 mM KCl, 0.4 mM MgSO4, and 0.6 mM Ca(NO3)2, pH 7.6) and stored at –20°C. Subsequent dilutions were made in distilled water. MOs were injected at doses ranging from 2 to 10 mg/ml. To confirm the splicing defects following MO injection, RT-PCR was carried out using the Superscriptase II kit (Invitrogen). For this, RNA was prepared from the injected embryos at 24 hpf using RNA-stat 60 (Tel-Test, Friendswood, TX). RNA was then reverse transcribed by the reverse transcription system of Promega. PCR was performed on the cDNA with the following primers: 5′-GATGGCTCAAAGGGTCTGCTGGCAG-3′ and 5′-GTCTCAGTATGGGATTTGGTTCTGG-3′.

Two additional non-overlapping antisense MOs targeting the 5′ UTR of pi4ka were also designed to block translation of the mRNA (MO3: 5′-CACCACGGATGTCCACAGGACACAT-3′ and MO4: 5′-GACCTATATTTAACAAATGTGCATT-3′). Their effectiveness was confirmed by an in vitro translation assay (Blasiole et al., 2005).


Microinjections were performed with a glass needle and a Narishige micromanipulator linked to a nitrogen gas pressure system. Working dilutions of MOs and mRNA for zebrafish Pi4ka were prepared with Danieau solution and injected in the yolk of one-stage embryos.

Rescue of Pi4ka phenotype

The full-length PI4KIIIα coding region was cloned into the expression vector pCS2+. Capped sense mRNA was synthesized from this construct using the ribomax large-scale RNA production system (Promega). After purification, PI4KIIIα mRNA (1-3 ng) was co-injected with MO1 (9-10 ng) into one-cell stage zebrafish embryos. Embryos were collected at 3 dpf. Pectoral fin lengths were measured and compared with those of controls or PI4KIIIα morphants.

Inhibitor treatment

To block PI3K activity, we used LY294002 (Calbiochem), a specific inhibitor of PI3K and treated embryos from 50% epiboly stage in 10 μM LY294002 and fixed in 4% paraformaldehyde at different stages.

Whole-mount antibody staining

Phosphohistone H3 was detected in fixed embryos by immunohistochemistry using an anti-H3P rabbit polyclonal antibody (1:100) (Upstate Biotechnology, Lake Placid, NY) and HRP-conjugated anti-rabbit IgG (1:500) (KPL).


  • Supplementary material available online at

  • We would like to thank Gregory Palardy for his help with the injections and Milton English for his technical suggestions. This research was supported by the Intramural Research Program of the National Institute of Child Health and Human Development and the National Human Genome Research Institute of the National Institutes of Health. Deposited in PMC for release after 12 months.

  • Accepted September 15, 2009.


View Abstract