Phosphatidylserine receptor is required for the engulfment of dead apoptotic cells and for normal embryonic development in zebrafish

Apoptotic cell death occurs by a mechanism that is conserved from nematodes to humans (Meier et al.,2000). In vivo, the typical fate for apoptotic cells is rapid engulfment and degradation by phagocytes(Savill, 1998). Amongst higher organisms, the removal of apoptotic cells by phagocytes suppresses inflammation, modulates the macrophage-directed deletion of host cells, and critically regulates the immune response of an individual(Savill and Fadok, 2000). For vertebrates, the phagocyte engages the dying cells through specific receptors that include the phospatidylserine receptor (PSR) (Fadok, 2000; Hong et al., 1998; Li, 2003), Fc receptors,complement receptors 3 and 4, the ABC1 transporter, and members of the scavenger-receptor family (Platt et al.,1998). For lower vertebrate systems such as the zebrafish, the cell corpses generated developmentally are quickly removed, although which specific type(s) of engulfment genes are involved still remains largely unknown.

Recent advances in the study of the nematode Caenorhabditis elegans (Chung et al.,2000; Wang et al.,2003) and Drosophila melanogaster (France et al., 1999a)illustrate the power of using genetically tractable systems to identify necessary phagocytic genes. Major efforts to understand crucial pathways that mediate programmed cell death have also led to the genetic and molecular characterization of a number of genes involved in the recognition and engulfment mechanisms of cells amongst invertebrates(Chung et al., 2000; Lauber et al., 2003; Arur et al., 2003;Ravichandran, 2003). For C. elegans it is important to recognize that phagocytosis is performed by cells that are non-specific phagocytes rather than by specialized phagocytes such as macrophages, as tends to be the case in D. melanogaster (Savill and Fadok, 2000).

Cell death that is morphologically and genetically distinct from apoptosis is strongly implicated in some human disease(Meier et al., 2000). Little is known regarding the molecular mechanisms by which the resulting (cell)corpses are eliminated, and the clearance of defective events for zebrafish. We sought to define the genetic requirements for a potentially distinct death paradigm associated with a loss-of-function of the cell-corpse receptor PSR in the zebrafish model system, to elucidate gene function related to organogenesis.

Morpholinos are chemically modified oligonucleotides with base-stacking abilities similar to those of natural genetic material(Summerton and Weller, 1997). Morpholinos have been shown to bind to and block translation of mRNA during cell genesis in zebrafish (Nasevicius and Ekker, 2000). Initially, we cloned the psr gene from the 24 hours post-fertilization (hpf) cDNA library, and designed PSR morpholinos for the knockdown of PSR expression during the embryonic development of zebrafish. We found that a lot of apoptotic cells accumulated in the furrow or individual boundary of the whole somite, and interfered with the normal interior and posterior somitic formation at the early segmentation stage. At the post-segmentation and organogenesis developmental stages, the embryos were phenotypically defective in the brain, heart, somite and notochord. In addition, injection of psr mRNA with PSR morpholinos could compensate for the defective phenotype. These observations are consistent with an evolutionarily conserved pathway involving PSR that is responsible for the removal of cell corpses during cell migration, and for cell-cell interaction processes that are tightly linked to normal development during morphogenesis and organogenesis in zebrafish.

Techniques for the care and breeding of zebrafish have been previously described in detail (Westerfield,1993). Embryos were collected from natural matings and maintained in embryo medium (15 mM NaCl, 0.5 mM KCl, 1 mM CaCl₂, 1 mM MgSO₄, 0.15 mM, 0.05 mM Na₂HPO₄, 0.7 mM NaHCO₃) at 28.5°C. Embryos were staged according to standard morphological criteria (Kimmel et al.,1995).

Embryos at different developmental stages (30% epiboly, 50% epiboly,shield, 90% epiboly to tailbud) were collected with plastic droppers, placed in microtubes, washed twice with phosphate buffered saline (PBS), and then fixed with 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) for 2 hours. They were then washed with sodium cacodylate buffer, post-fixed in 1%aqueous osmium tetroxide for 2 hours, and washed with the same buffer. The embryos were dehydrated in a series of ethanol solutions and embedded in a Spurr's resin mixture using standard protocols. Semi-thin sections were cut,stained with Toluidine Blue, and examined by light microscopy (Nikon Eclipse E600, Nikon Corporation, Japan) to identify morphological patterns. Ultrathin sections, cut using a microtome, were stained with standard preparations of lead citrate and uranyl acetate, and observed using an electron microscope(Hitachi H-7000, Japan) (Hong et al.,1998).

Two pairs of degenerate primers derived from human and Drosophilasequences (GenBank) (Fadok, 2000) were used to synthesize a 0.5 kb probe (by RT-PCR) to screen a 24-hour-old wild-type zebrafish (Danio rerio)embryo λ cDNA library (Stratagene). The degenerate primers used were as follows:

1. PSR-5′P1,5′-gAA(gA)TATCg(Cg)AACCAgAAgTTCAAgTg(CT)gg(CT)gg(CT)gAgg-3′(35mer);

2. PSR 5′P2, 5-gt(Cg)AagATgAAgATgAA(gA)TACTAC(gA)T(gC)gAgTACATg-3′(36mer);

3. PSR-3′P1, 5′CtggAAgAgTCgCTggAgCTgTC-3′; and (4)PSR-3′P2,5′-TtgAg(Cg)AC(AC)AC(Ag)TgCCAggAgCC(Tg)CC(Tg)gg-3′.

The cDNA library was screened using low stringency conditions. The positive clones obtained were sequenced using a single-base reaction with the ABI Prism 377 DNA sequencer (Applied Biosystems, Foster City, CA, USA), according to the manufacturer's protocol. The following PSR homologs were acquired from GenBank: the human homolog KIAA 0585 (Nagase, 1998) (Accession number BAA25511); the mouse homolog AAH 06067 (Accession number AAG27719); the D. melanogaster homolog CG5383 (Accession number AF401485); the zebrafish homolog (Accession number AF401485. The C. elegans homolog was acquired from the cosmid F29B9.4 (Accession number AAF99922).

Morpholinos were obtained from Gene Tools, LLC (Corvallis, OR, USA). All morpholinos were arbitrarily designed to bind to sequences flanking and including the initiating methionine. We selected sequences based on design parameters according to the manufacturer's recommendations (21-25mer antisense), and tested each design sequence for representation elsewhere in the genome (Nasevicius and Ekker,2000). Sequences were as follows (the sequence complimentary to the predicted start codon is shown in bold in all cases): PSR-MO,5′-TCCgTTTCTTgCTTTTATggTTCAT-3′; and control-MO,5′-TCCCTTACTTgCATTTATCgTACAT-3′. The five sites in the control sequence that are subject to point mutation in the PSR-MO sequence are underlined.

Morpholino oligonucleotides were solubilized in water at a concentration of 1mM, and diluted with water to 0.5, 0.25 and 0.125 mM prior to injection(1.5-3 nl) into the yolk (Ekker et al.,1995).

Digoxigenin-labeled antisense RNA probes were synthesized from linearized DNA templates, including psr, pax2.1 and nkx2.5, using T7 RNA polymerase (Boehringer Mannheim, Germany). Whole-mount in situ hybridization was performed as previously described(Xu et al., 1994). The in situ hybridization assay used embryos injected with PSR-MO or control-MO at 12 hpf,36 hpf or 3 days post-fertilization (dpf).

Embryos at the one- or two-cell stage were injected with PSR-MO or control-MO. They were harvested at 12 and 24 hpf, and fixed with 4%paraformaldehyde in PBS (pH 7.4) at room temperature for 30 minutes. The embryos were stained with Acridine Orange (1 μg ml^-1) for 3-5 minutes, washed twice in PBS, and evaluated under fluorescence microscopy(using incident light at 488 nm excitation, with a 515 nm longpass filter for detection) (Hong et al.,1998). For the TdT-dUTP labeling step, the embryos were fixed in paraformaldehyde at the end of the incubation period (12 and 24 hpf),dechorionated, and incubated in blocking solution (0.1%H₂O₂ in methanol) for 30 minutes at room temperature. Embryos were rinsed with PBS, incubated on ice in a solution of 0.1% Triton X-100 in 0.1% sodium citrate for 30 minutes, to increase permeability, and rinsed twice with PBS. Afterwards, 50 μl of TUNEL reaction mixture (in-situ cell death-detection Kit, Boehringer Mannheim, Germany) was added and the embryos were incubated in a humidified chamber for 1 hour at 37°C. Embryos were analyzed for positive apoptotic cells under a fluorescence microscope equipped with a spot II cool CCD (Diagnostic Instruments, Sterling Heights,MI, USA).

Embryos were injected with PSR-MO or control-MO at the one- or two-cell stage. They were harvested at 24 hpf and lyzed in 150-200 μl sodium dodecyl sulfate (SDS) sample buffer [0.63 ml 1 M Tris-HCl (pH 6.8), 1.0 ml glycerol,0.5 ml β-mercaptoethanol, 1.75 ml 20% SDS, 6.12 ml H₂O in a total of 10 ml]. Protein from 40 μg of 24 hpf embryos was loaded on to each lane. Standard western-blot analysis was conducted using a human anti-PSR monoclonal antibody (Cascade Bioscience, Winchester, MA, USA) and a mouse anti-actin monoclonal antibody (Chemicon, Temecula, CA, USA). PSR was visualized using horseradish peroxidase-conjugated anti-mouse immunoglobulin(IgG) and the ECL detection kit (Amersham Pharmacia Biotech, Denmark)(Hong et al., 1999).

Zebrafish psr was cloned into the pCDNA3 vector, which contains a T7 RNA polymerase promoter site. Linearized plasmid DNA was used a template for in vitro transcription with the Message Machine Kit (Ambion, Austin, TX,USA), according to the manufacturer's instructions. For rescue of defective morphants, 0.1 nl of 200 ng/μl mRNA encoding soluble psr mRNA and PSR-MO (0.5 μM) were injected into the one-cell stage of each embryo using a gas-driven microinjector (Medical System Corporation)(Ekker, 1995).

Apoptotic death occurred during shield stage(Fig. 1A). Approximately four percent of embryonic zebrafish cells showed apoptotic death(Fig. 1B). Apoptotic cells possessed a multitude of membrane-enclosed micronuclei. Fig. 1C shows an apoptotic body displaying these micronuclei, as well as a normal-looking nucleus, being engulfed by a neighboring cell. The dying cells migrated to the margin of the yolk sack, either at the shield stage or at the two-to-three segmentation stage.

psr was cloned using a zebrafish-specific 500 bp probe to screen the 24 hpf cDNA library. The three positive clones that were obtained were used to search GenBank databases. Some significant matches were made after translating the nucleotide sequences to amino acid residues(Fig. 2A).

When we compared the nucleotide sequence identity of zebrafish psr(zfpsr) (AF401485) with other species, the match was 76% for D. melanogaster, 74% for human (KIAA0585) and 72% for mice (AF304118). The corresponding figure for C. elegans, from the hypothetical protein F29B9.4 (U70848), was 60%.

The predicted molecular weight (Fig. 2A) for F29B9.4 was 44.3 kDa, based on sequence analysis(Fig. 2A). This is similar to the corresponding molecular weight for the mouse species(Fadok et al., 2000), slightly smaller than the corresponding figure for the gene products in humans (45.5 kDa) and D. melanogaster (45 kDa), but larger than the analogous figure for C. elegans (38.4 kDa). In addition, the consensus sequence for the PSR-binding motif (FxFxLKxxxKxR) found in protein kinase C isoforms indicates that a 12-amino acid peptide motif is responsible for the specific interaction with PSR (Igarashi et al.,1995). A potential tyrosine phosphorylation site is indicated by box A (Fig. 2A), corresponding to residues 100-108 (KCGEDNDGY), which is well within the predicted intracellular domain (Schultz et al.,2000). We found that the sequence indicated by box B (residues 143-206) resembled the jmjC domain that is part of the cupin metalloenzyme superfamily that can regulate the chromatin-reorganization process(Clissold and Ponting, 2001). Presently, the function(s) of the PSR region is unknown.

The protein sequence of topology programs varied slightly in their specific assignments (see box C; FVPGGWWHVVLNLDTTIAITQNF, residues 257-287 of PSR-F),based on an assessment of topology and possible hydrophobicity (SMART-TMHMM2)(Schultz et al., 2000). The predicted extracellular domain (indicated by box D) of the membrane-associated domain (residues 340-359) contains a serine-rich sequence (342-355) that may be glycosylated sites.

psr was expressed in embryos from the one-cell developmental stage(30 minutes; data not shown) to the 3 dpf larval stage(Fig. 2B, panels a-f). After the somite segmentation period, psr was apparent throughout the embryo (Fig. 2B, panels c-e)and the hatching grand (Fig. 2B, panel e). At the larval (3 dpf) stage, psr expression was detected in the heart and kidney (Fig. 2B, panel f).

PSR morpholino oligonucleotides (MO; 40 ng) were injected into embryos to accomplish knockdown of psr expression(Fig. 3). psrknockdown at the epiboly stage strongly affected embryonic morphological formation at 12 hpf, and produced a severe delay in epiboly formation(Fig. 3F), when compared with control embryos (Fig. 3E). In addition, cell corpses accumulated between the somite boundaries, close to the notochord, at the six-to-seven segmentation stage(Fig. 3D). Embryos treated with control-MO did not display cell corpses in the somite or near the notochord(Fig. 3B). PSR-MO embryos appeared to be thin from an anterior lateral view, with the loss of furrow from within the somite (Fig. 3C), when compared with control embryos(Fig. 3A).

Observation of embryonic development at different stages revealed that the initial accumulation of the corpse cells in the brain, interior somite and tail bar (Fig. 3G,H, 12 hpf) is followed by the gradual expansion of the zone of cell death(Fig. 3I,J, 36 hpf).

We next examined whether the cell corpses underwent apoptotic death(Fig. 4). Apoptotic corpses were covered in the furrow (indicated by arrows in Fig. 4), and lay close to the surface of individual somites at 14 hpf(Fig. 4C). After this time, a gradual increase in numbers and accumulation throughout embryos treated with psr oligonucleotides was evident(Fig. 4A,B, 17 hpf). Cell corpses leaked out from the 17 hpf embryos(Fig. 4B) that displayed chromatin condensation (Fig. 4D,E) in the corpse cells. Positive apoptoic cells were also evident in whole embryos at 17 hpf upon examination by the TUNEL assay(Fig. 4F-I).

We next tested the ability of the PSR-MO and the control-MO to block translation. As shown in Fig. 5A (panel a), injections of 10 and 40 ng of PSR-MO blocked between 70% (lane 4) and 90% (lane 5) of the normal protein expression. While the predicted molecular weight of the zebrafish PSR is 44.3 kDa, the protein evident in lane 4 migrated as two bands that were approximately 62 and 59.7 kDa species. This apparent discrepancy was likely due to the glycoprotein nature of PSR. Glycosylation could be subject to interference following PSR-MO-mediated psr knockdown. At the 36 hpf stage, the PSR-MO (40 ng) samples showed a defective phenotype that reflected a loss of normal morphogenesis and a distorted psr expression pattern(Fig. 5B, panels b,c; as indicated by arrows), when compared with control-MO(Fig. 5B, panels a). Embryos at the 3-dpf stage displayed similar alterations(Fig. 5B, panels e,f; as indicated by arrows), as compared with control(Fig. 5B, panel d).

Acridine Orange staining of abnormal 3 dpf embryos revealed both weakly and severely defective phenotypes (Fig. 5B, panels h,i), compared with the control-MO group(Fig. 5B, panel g). In the relatively few embryos that displayed the weakly defective phenotype, bending of the notochord or a slight delay in heart development prior to hatching was evident (Fig. 5B, panels e,h). Conversely, the severely phenotypically defective embryos exhibited marked developmental impediments, such as shrinkage of the brain, loss of posterior somite development, and failure to hatch out(Fig. 5B, panels f,i).

We estimated the effect of different individual doses of PSR-MO (5 ng, n=209; 10 ng, n=238; 20 ng, n=160; 40 ng, n=224; control-MO, 40 ng, n=194) on embryonic development. As summarized in Table 1,developmental effects correlated to PSR-MO dosage, and were significant when compared with control-MO-injected embryos.

Brain and heart development were monitored by the use of the marker genes pax2.1 (Klaus and Brand,1998) and nkx2.5(Schwartz and Olson, 1999), at 12, 24, 36 and 72 hpf. Significant differences were apparent at 12 and 24 hpf(data not shown) between PSR-MO- and control-MO-injected embryos. The developing mid- and hindbrains of PSR-MO-injected embryos appeared shrunken and displayed an abnormal pax2.1 expression pattern at 36 hpf(Fig. 6B), when compared with the control (Fig. 6A). At 3 dpf, there was about a 2-fold shortening of the mid- and hindbrain regions(indicated by the open squares in Fig. 6D), when compared with the control-MO group(Fig. 6c). Knockout of psr affects brain development in mice(Li et al., 2003).

Probing with nkx2.5, or staining with Acridine Orange, allowed the monitoring of heart development and its overall morphogenesis. Based on an in situ assay conducted at the 36 hpf stage, the heart suffered a severe developmental delay that was manifest as an absence of the atria and ventricles (Fig. 6E,F). At the 3 dpf stage, the formation of a tube-like heart and an abnormal blood-circulation rate was noted (Fig. 6G,H). Defective embryos had a heart cavity that was enlarged up to 60% more than normal size, and displayed abnormal heart formation(Fig. 6I-L). The different phenotypes of the notochord are shown in Fig. 6M-U. The top view of PSR-MO larvae revealed a substantial bending of the notochord(Fig. 6N), when compared with control-MO larvae (Fig. 6M). From a lateral view, the defective embryos showed bending of the notochord between the interior-posterior and posterior positions(Fig. 6N,P,R,T), when compared with the control-MO group (Fig. 6O,Q,S). Acridine Orange staining revealed a substantial number of apoptotic cells located in the posterior of the somite(Fig. 6R), when compared with controls (Fig. 6Q).

In a previous study (Bauer et al.,2001), 10-20 pg of psr mRNA compensated for the developmental blockage imposed by PSR-MO when co-injected with 20 ng of PSR-MO at the embryonic one- or two-cell stage. We observed that a similar application of 20 ng of PSR-MO and 20 pg of psr mRNA reproduced the earlier findings. In the PSR-MO group, embryos had accumulated many corpse cells in the whole embryo by 12 hpf (Fig. 7C). Accumulation was particularly evident in the furrow between somites. Conversely, in rescued embryos only a few corpse cells were evident in the interior somite (Fig. 7D). Wild-type or control-MO embryos did not show accumulated corpse cells (Fig. 7A,B).

At 48 hpf, PSR-MO embryos still displayed both weakly and severely defective developmental phenotypes (Fig. 7G). However, in rescued embryos examined at 48 hpf, only the weakly defective phenotype was detectable(Fig. 7H). Compare with these the normal phenotype of the wild-type and control-MO embryos(Fig. 7E,F). The two defective developmental phenotypes were evident, even at 3 dpf(Fig. 5B, panels e,f,h,i; Fig. 7K), in PSR-MO embryos. By comparison, rescued `normal-like' morphants(Fig. 7L) showed a slightly different phenotype at 3 dpf, namely a bending of the trunk and heart cavity,when compared with wild type (Fig. 7I) and the control-MO group(Fig. 7J).

We estimated the survival ratio and morphants phenotype ratio in the PSR-MO plus psr mRNA group (n=64), and in the PSR-MO group(n=93) at 2 dpf. Addition of psr mRNA reduced mortality from 12.5 to 1.5% (Fig. 7M). Addition of psr mRNA reduced the morphants phenotype ratio from 38 to 28% for the strongly defective embryos, and from 9 to 1.5% for the weakly defective embryos (Fig. 7M). The rescued group showed a normal morphogenesis and kidney-development pattern(Fig. 7O,R), when compared to the control-MO group (Fig. 7N,Q) at the 3-dpf stage. At the same stage, the weakly defective embryos in the PSR-MO group displayed patterns of morphogenesis and psr expression that were indicative of delayed kidney development(Fig. 7P,S).

Phagocytosis, the phenomenon caused by inflammation and autoimmune responses (Savill and Fadok,2000; Wyllie et al.,1980; Henson et al.,2001), is an important process for modeling tissue during development (France et al., 1999). The link between phagocytosis and development thus provides a useful system with which to identify genes that are important for phagocytosis. These genes have been difficult to identify in more complex mammalian systems (France et al., 1999). Although some questions remain, PSR appears to play a central role in the clearance of apoptotic cells(Fadok et al., 2001).

The zebrafish (Danio rerio) has several advantages as a model for studying development (ZFIN website, http://zfish.uoregon.edu/)(Dooley and Zon, 2000). In our temporal study, we traced the path of abnormal development during the knockdown of psr. We detected three phenotypically different embryos(weakly defective, strongly defective and death type) during examinations at 12 and 36 hpf and 3 dpf. Although the weakly defective embryo was not apparent at 12 hpf, the defect was apparent at 3 dpf as a mild abnormality that included an enlarged heart cavity and defective notochord formation. The strongly defective embryo type at 12 hpf was characterized by the accumulation of a large number of apoptotic corpses in the posterior section of the embryo that interfered with the posterior development. This was also the case at 36 hpf, and the embryo failed to hatch out at 3 dpf. The death type embryo was characterized by an accumulation of a large number of cell corpses in the whole embryo at 12 hpf. The embryos were developmentally delayed at 36 hpf and died before the 3-dpf stage. Interestingly, the rescue studies that demonstrate that the death type, and the weakly and strongly defective phenotypes could be corrected or compensated for by the injection of psr mRNA (Fig. 7)suggest that psr mRNA could potentially be used to correct diseases arising from psr gene defects.

The present study offers support for our idea that the accumulation of cell corpses interferes with normal embryonic development by altering cell movement(Fig. 3E,F) and cell-cell interaction (Fig. 8).

The zebrafish embryo undergoes the first cell cycle at zero hours post-fertilization. Cell cycles 2-7 occur rapidly and synchronously. The embryo enters the blastula stage at 2.25 hpf, and the metasynchronous cell cycles become longer at the 8- to 10-cell cycle stage. Asynchronous cell cycles begin at the midblastula transition stage, when epiboly commences(Kimmel, 1995) (ZFIN website, http://zfish.uoregon.edu/). At the midblastula stage, the gastrula starts to emerge at 5.25 hpf and all cells acquire the ability to move. Movement is required by cells to achieve their developmental goals. These goals include the morphogenetic movement of involution, convergence and extension from the epiblast, hypoblast and embryonic axis through to the end of epiboly(Woo et al., 1995; Sampath et al., 1998; Heisenberg et al., 2000). If the apoptotic cell corpses are not quickly removed, they can impede cell movement. At the onset of organ development, cells are typically associated with one of three germ layers (ectoderm, mesoderm or endoderm) that participate in the development of somites, primordial pharyngea and neuromeres, and in primary organogenesis and tailbud formation(Kimmel et al., 1995; Woo et al., 1995). At this stage cell-cell interaction is also important(Mellitzer et al., 1999; Xu et al., 1999; Jiang et al., 2000). Finally,cell corpses gradually affect morphogenesis of normal organs at the pharyngula(24 hpf), hatching (48 hpf) and early laval (72 hpf) stages.

PSR mediates the engulfment of apoptotic cells in mice(Fadok et al., 2000). A defect of PSR in the early stages of organogenesis may be involved in respiratory distress syndromes and congenital brain malformation(Li et al., 2003). In addition, studies in C. elegans(Wang et al., 2003) and D. melanogaster illustrate the power of using genetically tractable systems to identify essential phagocytic genes(Chung et al., 2000; Fares and Greenwald, 2001; Conradt, 2002). It is important to recognize that phagocytosis is performed by cells that are classified as`non-professional' phagocytes, rather than by specialized phagocytes, such as macrophages in D. melanogaster. Here, we shed more light upon the role of this emerging phagocytosis receptor (PSR) in the vertebrate system(Savill and Fadok, 2000; Henson et al., 2001; Schlegel and Williamson, 2001; Li et al., 2003). Our results indicate that phagocytosis during the development of zebrafish is performed by PSR-mediated phagocytes, which are distributed throughout the entire embryo,especially between the 2- and 3-somite stage and 24 hpf, and that these are`professional' phagocytes rather than non-specialized phagocytes.

We thank Mr H.-Y. Gong for the pax2.1 plasmid and Dr J.-N. Chen for the nkx2.5 plasmid. This work was partially supported by grants from the National Science Council, Taiwan, Republic of China awarded to J.-L.W. (NSC-90-2311-B-001-018) and J.-R.H. (NSC-91-2311-B-006-007;NSC-92-2313-B-006-005).