Discussion

Ptdsr is required for the differentiation of multiple organ systems during development
In this study, we have generated a null mutation in the phosphatidylserine receptor (Ptdsr) gene in C57BL/6J mice. We show that ablation of Ptdsr results in profound differentiation defects in multiple organs and tissues during embryogenesis, although with variable penetrance. While this work was in progress, two other groups reported the generation of Ptdsr-deficient mice [31,32]. In all three knockout mouse lines, the first two exons ([31] and this study) or exons one to three [32] were deleted by replacement with a neomycin-selection cassette. The Ptdsr-knockout mouse lines differ in the genetic background in which the mutation was generated and maintained, however. In our case, the Ptdsr-null allele was generated in an isogenic C57BL/6J background, whereas Li et al. [31] and Kunisaki et al. [32] investigated the phenotype of their Ptdsr-knockout mice in a mixed 129 × C57BL/6 background. The ablation of Ptdsr function results in perinatal lethality in all cases, but there are interesting differences in severity or expressivities of phenotypes among the different Ptdsr-deficient mouse lines. This might be due either to differences in genetic background or because the phenotypes that have been investigated in this study have not been analyzed in such detail before.
In the Ptdsr-knockout mouse line reported here, growth retardation started from E12.5 onwards and was associated with delayed differentiation in several organs in which Ptdsr is expressed either during embryogenesis or later in adulthood. At E16.5 almost no branching morphogenesis of the lung epithelium was observed in Ptdsr -/- lungs. Similarly, epithelial structures were only partially developed in mutant kidneys, without terminal differentiation of Bowman's capsule and with a severe reduction in the number of differentiated collecting tubules. Likewise, the differentiation of the intestine was also severely delayed at this developmental stage. When compared with wild-type controls, intestinal tissues of Ptdsr knockout mice appeared unstructured, with an absence of enteric ganglia and of differentiated smooth muscle tissue. Interestingly, defects in kidney and intestine differentiation were not described in the Ptdsr-knockouts generated by Li et al. [31] and Kunisaki et al. [32]. Surprisingly, when we examined Ptdsr-/- embryos shortly before birth (E18.5) or neonatally, we found only mild differentiation delays in organs that appeared severely affected at mid-gestation. This 'recovery' was most visible in Ptdsr -/- lungs: at P0 we found expanded lungs in the knockout mice that showed normal branching patterns, with differentiated alveoli and bronchioles.
We investigated the occurrence of programmed cell death during lung development in wild-type and Ptdsr-knockout mice throughout embryogenesis (E16.5 to P0). Comparative immunohistochemistry for aCasp3 revealed that apoptosis is a rare event during lung morphogenesis. Furthermore, we failed to detect any differences in the number of apoptotic cells in Ptdsr-knockout and wild-type animals in the rare cases where we could detect apoptotic cells within lung tissues. These findings are contrary to the results reported by Li et al. [31], who suggested that impaired clearance of apoptotic mesenchymal and epithelial cells causes a failure in lung morphogenesis in Ptdsr-deficient mice. In contrast, our findings are in line with the current view on lung development during embryogenesis. Accordingly, formation of the epithelial lung via branching morphogenesis can be subdivided into a series of sequential steps that involve: first, formation of the organ anlage in the form of a placode; second, primary bud formation by placode invagination; third, branch initiation and branch outgrowth; fourth, further reiteration of the branching process; and fifth, terminal differentiation of organ-specific proximal and distal structures [34,35]. In contrast to other invagination processes during embryogenesis, such as mammary gland formation, the lumen of the lungs is expanded by successive branching events, branch outgrowth and elongation, rather than by apoptosis [34,36]. Finally, because the lungs of Ptdsr -/-neonates were almost fully expanded and appeared normal in structure in comparison to wild-type littermates, it is highly unlikely that Ptdsr mutants die of respiratory lung failure. In addition, Li and colleagues [31] demonstrated that surfactant expression is normal in Ptdsr-deficient animals, supporting the idea of normal maturation of surfactant-producing type II alveolar epithelial cells and lung function. Other defects must therefore be responsible for the death of Ptdsr-mutant mice. The frequently observed subcutaneous edema of various extents in Ptdsr-deficient homozygotes gave us a hint that Ptdsr-deficiency and lethality might be associated with cardiovascular problems. Indeed, very recently we have obtained strong evidence that Ptdsr-knockout mice die as a result of defects in heart development that are associated with specific cardiopulmonary malformations; (J.E. Schneider, J.B., S.D. Bamfort, A.D.G., C. Broadbent, K. Clarke, S. Neubauer, A.L. and S. Battacharya, unpublished observations).
In addition, we demonstrate that eye development requires a functional Ptdsr gene. Ptdsr-deficient embryos can be roughly divided into two categories. The first, severely affected group develops anophthalmia that correlates with formation of ectopic retinal-pigmented epithelium and induction of proliferation of underlying mesenchyme in the nasal cavity. This phenotype represents a completely novel lesion that to our knowledge has not been described before in any other mouse mutant. The second group shows normal external eye structures, although in this case retinal development is temporally delayed during mid-gestation, with persistent, abnormal morphogenesis of the inner granular retinal layer at later stages of embryogenesis. A possible explanation for these two phenotypes can be found in the expression pattern of the Ptdsr gene. Initially, Ptdsr is expressed throughout the whole developing nervous system, with exceptionally high levels in the anterior part of the forebrain. Later expression becomes more restricted to the developing retina and lens. Thus, Ptdsr might play an important role in early events of ocular morphogenesis, such as establishment and bisection of eye fields and formation of optic cups. These early eye-formation steps are closely interconnected with development of the forebrain [37,38] and the nose [39-41].
Interestingly, we occasionally observed serious malformations of forebrain and nasal structures in Ptdsr-knockout embryos that were associated with bilateral anophthalmia (see for example the mutant embryo in Figure 1g). This suggests that Ptdsr is involved in the regulation of differentiation processes within forebrain regions, and that ablation of Ptdsr function might secondarily affect early eye formation. Li et al. [31] found smaller lenses in Ptdsr-knockout mice and described the formation of retinal protrusions, although anophthalmia and specific differentiation defects of retinal cell layers were not reported in their study. Li et al. proposed [31] that the eye phenotype they observed could be explained by failed removal of apoptotic cells during eye development, but we think that the observed defects are unrelated to a failure of apoptotic cell clearance. A recent comprehensive kinetic analysis of apoptosis induction during mouse retinal development described four major peaks of apoptotic cell death [42]. This study demonstrated that there is an initial phase of cell death during the invagination of the optic cup (E10.5), followed by subsequent waves of apoptosis induction immediately before and after birth (E18.5 to postnatal day P2), and from postnatal days P9 to P10 and P14 to P16 [42]. Thus, besides the formation of the inner and outer layers of the optic cup in early eye development, other major phases of retinal cell apoptosis take place only postnatally and correspond to important periods in the establishment of neuronal connections. Furthermore, cell death during normal retinal development occurs in retinal layers distinct from the inner granular layer where we observed the most pronounced differentiation defects in the Ptdsr -/- mutants described here. Other studies that connect the postnatal elimination of apoptotic photoreceptor cells to Ptdsr-mediated macrophage engulfment [43] should be interpreted with extreme caution as these studies were based on the monoclonal anti-Ptdsr antibody mAb 217G8E9 [26,43] (see below).
Consistent with the results of Li et al. [31], we found particular brain malformations in our Ptdsr -/- mice. Exencephaly and hyperplastic brain phenotypes were observed at a low penetrance in Ptdsr-mutant mice (less then 4.5% of homozygotes), but these do not resemble to any extent the brain-overgrowth phenotypes of caspase- or Apaf1-knockout mice ([44], and references therein) in that we failed to identify any differences in the number or distribution of apoptotic cells or pyknotic cell clusters in the neuroepithelium of Ptdsr -/- and Ptdsr+/+ mice. Thus, reduced cell death or diminished clearance of apoptotic neural progenitor cells is unlikely to be the cause of the brain hyperplasia.
In summary, our studies demonstrate that Ptdsr is required for normal tissue differentiation, especially during the mid-gestation period when we observed the most severe differentiation delays in several organs of Ptdsr-knockout mice. The multiple defects in tissue differentiation cannot be explained by failure of apoptotic cell clearance, as this process is normal in our Ptdsr-knockout line. This result therefore indicates that Ptdsr has a novel, hitherto unexpected, role in promoting tissue maturation and terminal differentiation. Additional studies with conditionally targeted Ptdsr-deficient mice are required to investigate the role of spatial and temporal Ptdsr expression and function during tissue differentiation.

Ptdsr is not essential for the clearance of apoptotic cells
Our studies demonstrate that Ptdsr is not a primary receptor for the uptake of apoptotic cells. Investigation of apoptotic cell clearance in vivo in Ptdsr -/- embryos conclusively showed that removal of apoptotic cells is not compromised by ablation of Ptdsr function. Comparative analysis of ten different tissues and organs in Ptdsr+/+ and Ptdsr -/- animals at several stages of embryonic development and in neonates failed to identify impaired uptake of apoptotic cells at any time during development. Furthermore, phagocytosis assays in vitro demonstrated a completely normal uptake of apoptotic cells by Ptdsr -/- macrophages, with some knockout macrophages showing loads even higher than wild-type of engulfed dead cells. These results are contrary to the expected role of Ptdsr in apoptotic cell clearance and to the reported findings of Li et al. [31] and Kunisaki et al. [32], as well as to a study done with a phosphatidylserine receptor null allele in C. elegans [45]. In previous studies in the mouse, the distribution and amount of apoptotic cells in Ptdsr-knockout and control animals were investigated in only a few tissues and at one [31] or two [32] developmental stages. Li et al. [31] examined lung, midbrain and retina at day E17.5 of gestation and identified apoptotic cells by TUNEL staining. Their findings must be interpreted with caution because remodeling of cellular structures by apoptosis in specific retina layers is known to occur mainly postnatally [42], and apoptosis plays an important physiological role in the maintenance and homeostasis of lung epithelium after birth or in pathological conditions involving pulmonary inflammation and not during lung development [46]. This postnatal role for apoptosis is in accordance with our data, as we rarely observed apoptotic cells in retina or lung tissue throughout embryogenesis in Ptdsr+/+ and Ptdsr -/- mice. Kunisaki et al. [32] analyzed TUNEL-stained sections of liver and thymus at days E13.5 and E16.5 of development in Ptdsr+/- and Ptdsr -/- embryos and found reduced rather than increased numbers of TUNEL-positive cells in Ptdsr-deficient embryos. Using co-localization of TUNEL-positive cells with F4/80-positive macrophages they suggested that Ptdsr -/- embryos exhibited a three-fold increase in the frequency of unphagocytosed TUNEL-positive cells together with a severely reduced number of F4/80-positive cells. These results must be interpreted very carefully, however, as it is technically difficult to unambiguously identify engulfed target cells in individual macrophages in solid tissues by fluorescence microscopy.
In addition, our data suggest that during embryogenesis, macrophage-mediated clearance of apoptotic cells is not the only - or even the primary - mechanism for the removal of apoptotic cells. In many tissues where programmed cell death occurs as a prominent event during embryogenesis, such as remodeling of the genital ridge during gonad morphogenesis and differentiation of the neural tube, we found almost no co-localization of apoptotic cells and macrophages. This indicates that in these cases clearance of apoptotic cells is directly mediated by neighboring 'bystander' cells rather than by macrophages that have been recruited into areas where apoptosis occurs. Obviously these in vivo clearance mechanisms are not compromised by Ptdsr-deficiency in our knockout mutant. This finding is in line with studies in macrophageless Sfpi1-knockout embryos that are deficient for the hematopoietic-lineage-specific transcription factor PU.1. Here, the phagocytosis of apoptotic cells during embryogenesis is taken over by 'stand-in' mesenchymal neighbors [47]. As recognition of phosphatidylserine is thought to be a universal engulfment mechanism for all cells that are able to phagocytose apoptotic cells, it is very striking that apoptotic cell clearance mediated by non-professional bystander cells is also not compromised by Ptdsr-deficiency.
In contrast to Li et al. [31], we did not observe any impairment in the uptake of apoptotic cells by Ptdsr -/- macrophages in vitro. We performed phagocytosis assays in vitro with fetal-liver-derived macrophages, while in their assays, Li and colleagues used thioglycollate-elicited peritoneal macrophages after adoptive transfer of Ptdsr -/- hematopoietic stem cells. The different results obtained in the two studies are puzzling; they might be due to the use of different macrophage or cell populations. We and Kunisaki et al. [32] found that Ptdsr-deficiency is to some extent associated with defects in hematopoiesis. Thus, it seems possible that recruitment and activation/differentiation of macrophages after adoptive transfer and thioglycollate elicitation are affected by Ptdsr-deficiency. We do not think that the different results observed in Ptdsr-knockout mice in a mixed C57BL/6 × 129 background and in a pure C57BL/6J background can be attributed to genetic background effects: comparison of apoptotic cell engulfment efficacies of thioglycollate-elicited macrophages from 129P2/OlaHsd and C57BL/6J mice did not show any differences in apoptotic cell uptake (J.B. and A.L., unpublished observations). Moreover, in contrast to our studies, neither Li et al. [31] nor Kunisaki et al. [32] determined phagocytotic engulfment indexes for Ptdsr-deficient macrophages.
Interestingly, we observed differences between Ptdsr+/+ and Ptdsr -/- macrophages in the secretion of pro- and anti-inflammatory cytokines after stimulation with LPS and apoptotic cells. This provides evidence that cellular activation and effector mechanisms are impaired in Ptdsr-deleted macrophages. It remains to be determined which classical pathways of macrophage activation and function involve Ptdsr. This is especially important in light of recent findings that demonstrated nuclear localization of the Ptdsr protein [29].
Most strikingly, the recently published data regarding the genetic ablation or perturbation of phosphatidylserine receptor function in C. elegans are also contradictory. Wang et al. [45] reported that psr-1, the C. elegans homolog of Ptdsr, is important for cell-corpse engulfment, whereas psr-1 RNAi studies performed by Arur et al. [25] yielded, in this respect, no phenotype. Moreover, Wang and colleagues hypothesized on the basis of their data that psr-1 might act to transduce an engulfment signal upstream of Ced-2 (Crk II), Ced-5 (Dock 180), Ced-10 (Rac 1) and Ced-12 (Elmo) in one of the two cell-corpse engulfment pathways in the worm [45]. But the loss-of-function phenotype of psr-1 mutants and the complementation phenotypes in overexpressing transgenic worms shown by Wang et al. [45] are rather weak as compared to the classical C. elegans engulfment mutants [8].
Many previous functional studies that reported a requirement for Ptdsr for the phagocytosis of apoptotic cells used the monoclonal anti-Ptdsr antibody mAb 217G8E9 [26]. This antibody was used in Ptdsr binding and blocking experiments, as well as in subcellular localization studies, which led to the conclusion that Ptdsr is a transmembrane receptor critical for signal transduction at the engulfment interface. More recently it was used in binding assays to show that the human and worm Ptdsr molecules can recognize phosphatidylserine [45]. In the course of the study presented here, we stained immunohistochemically for Ptdsr with mAb 217G8E9 on wild-type and Ptdsr-deficient macrophages and fibroblasts (see Additional data file 1, Figure S2 and data not shown). To our surprise, we observed similar staining patterns with cells of both genotypes. Furthermore, using a Ptdsr-peptide array we found that mAb 217G8E9 can bind weakly to a Ptdsr peptide, explaining the original isolation of Ptdsr cDNA clones by phage display [26]; but the antibody mainly recognizes additional, as-yet unknown, membrane-associated protein(s) (see Additional data file 1, Figure S2). Experiments that have used this antibody should therefore be interpreted with great caution as they might come to be viewed in a different light.