Popeye domain containing gene 2 (Popdc2) is a myocyte-specific differentiation marker during chick heart development


  • Stephanie S. Breher,

    1. Department of Cell and Molecular Biology, Institute of Biochemistry and Biotechnology, Technical University of Braunschweig, Braunschweig, Germany
    Search for more papers by this author
  • Eleftheria Mavridou,

    1. Department of Cell and Molecular Biology, Institute of Biochemistry and Biotechnology, Technical University of Braunschweig, Braunschweig, Germany
    Search for more papers by this author
  • Christian Brenneis,

    1. Department of Cell and Molecular Biology, Institute of Biochemistry and Biotechnology, Technical University of Braunschweig, Braunschweig, Germany
    Search for more papers by this author
  • Alexander Froese,

    1. Department of Cell and Molecular Biology, Institute of Biochemistry and Biotechnology, Technical University of Braunschweig, Braunschweig, Germany
    Search for more papers by this author
  • Hans-Henning Arnold,

    1. Department of Cell and Molecular Biology, Institute of Biochemistry and Biotechnology, Technical University of Braunschweig, Braunschweig, Germany
    Search for more papers by this author
  • Thomas Brand

    Corresponding author
    1. Department of Cell and Molecular Biology, Institute of Biochemistry and Biotechnology, Technical University of Braunschweig, Braunschweig, Germany
    • Department of Cell and Molecular Biology, Institute of Biochemistry and Biotechnology, Technical University of Braunschweig, Spielmannstr. 7, 38106 Braunschweig, Germany
    Search for more papers by this author


The Popeye domain containing (popdc) genes constitute a novel gene family encoding proteins of the plasma membrane in muscle cells, with three N-terminal transmembrane domains and a cytoplasmic carboxy terminus. In vertebrates, three members of the Popdc gene family have been described. However, in the chick system only two cDNAs, Popdc1 and Popdc3, have been cloned previously. By screening a chick expressed sequence tag database, we report here the identification of five alternatively spliced chick Popdc2 cDNAs with different carboxy termini. Northern blot analysis revealed expression of Popdc2 predominantly in the myocardium and weaker expression in skeletal muscle. By whole-mount in situ hybridization, chick Popdc2 was first detected at Hamburger and Hamilton (HH) stage 7 within the anterior part of the heart fields. In the tubular heart, atrial and ventricular precursor cells stained positively for Popdc2. Weaker expression was observed in myocardium of the outflow tract and sinus venosus. By HH stage 18, the outer curvature myocardium was strongly stained, whereas expression in myocardium of the inner curvature was negligible. Popdc2 expression was absent from the endocardium and propepicardial organ. At HH stage 36, Popdc2 expression was confined to the compact layer myocardium. In addition to the heart, Popdc2 expression was also observed in the myotome and in the muscle-forming fields of the limbs. Our results indicate that Popdc2 is highly expressed in the developing heart and may serve as a novel marker of myocardial differentiation in the chick embryo. Developmental Dynamics 229:695–702, 2004. © 2004 Wiley-Liss, Inc.


The Popeye domain containing (popdc, previously named Pop or Bves) genes have been isolated on the basis of their muscle-restricted expression pattern (Reese et al., 1999; Andrée et al., 2000). In most vertebrates, three family members (popdc1–3) were found. However, in the chick only Popdc1 and Popdc3 cDNAs have been identified, but representing the products of a single gene (Andrée et al., 2000). Popdc genes in general produce several isoforms by alternative splicing encoding proteins of approximately 300 to 360 amino acids (Andrée et al., 2000). Computer-based secondary structure modeling predicted the existence of three transmembrane domains close to the amino terminus, followed by a conserved Popeye domain present in all family members (Reese et al., 1999; Andrée et al., 2000). An antibody raised against a peptide of chick Popdc1 revealed expression in the proepicardium, the epicardium, and smooth muscle cells of the coronary vasculature (Wada et al., 2001). In contrast, we reported expression of Popdc genes in striated and smooth muscle cells in zebrafish, frog, chick, and mouse embryos, suggesting that the expression pattern has been conserved during evolution (Andrée et al., 2000, 2002, 2003; Hitz et al., 2002). In chick cardiac myocytes, Popdc1 protein is present at the plasma membrane (DiAngelo et al., 2001) and analysis of the membrane topology suggested that the amino terminus is located extracellularly, while the carboxy terminus is in the cytoplasm (Knight et al., 2003; Breher et al., manuscript in preparation). Popdc1 null mutant mice are viable and do not display any overt phenotype. However, skeletal muscle regeneration upon cardiotoxin-induced muscle injury was impaired in the mutant (Andrée et al., 2002). Recently, it was proposed that Popdc1 might function as a cell adhesion molecule (Wada et al., 2001).

Here, we report the isolation of five alternatively spliced Popdc2 isoforms in chick. Whole-mount in situ hybridization and Northern blot analysis revealed that, chick Popdc2 is predominantly expressed in the myocardium, similar to other vertebrate species. Significant expression was also found in the myotome, in skeletal muscle, and stomach at later stages of development. In contrast to previous reports, we found no evidence for Popdc gene expression in the proepicardial organ.


To identify Popdc2 cDNAs in the chick, we searched a chick expressed sequence tag (EST) library by Blast with the mouse Popdc2 sequence (Boardman et al., 2002). A total of eight clones with significant homology to mouse Popdc2 were found. The different clones were incomplete either at the 5′ or the 3′ end. The full-length Popdc2 sequences were assembled as described in the Experimental Procedures section. Nucleotide sequence analysis of the resulting full-length clones revealed significant sequence divergence at the carboxy-terminus. A total of four Popdc2 isoforms (Popdc2A–D) were deduced from the various cDNA sequences (Fig. 1D,E). A fifth sequence, Popdc2A/B, was derived by sequencing the products of an reverse transcriptase-polymerase chain reaction (RT-PCR) reaction using primer pair Popdc2-I (Fig. 1C). The different isoforms are likely to be the result of alternative splicing (Fig. 1D). Blast searches revealed the presence of similar Popdc2 splice isoforms in mouse and zebrafish transcriptomes (data not show). All five isoforms were predicted to contain three transmembrane domains (Hirokawa et al., 1998) and a Popeye domain (PF04831, Fig. 1E). Comparison of Popdc2B protein with other vertebrate Popdc2 proteins as well as Popdc1A and Popdc3A from chick revealed high sequence similarity between Popdc2 proteins of different organisms (51% to zebrafish Popdc2 and 61% to both mouse and human Popdc2 proteins). Chick Popdc3A (48% similar) and Popdc1A (24% similar) were distantly related (Fig. 1A,B). By using cDNAs isolated form various embryonic tissues at different ages (Figs. 1C, 2B, 4), only three isoforms, Popdc2A, -A/B, and -B, could be amplified by RT-PCR with primer pair Popdc2-I that flanked the splice junctions (Fig. 1C).

Figure 1.

Bioinformatic analysis of Popdc2 sequences. A: Alignment of chick (Gg) Popdc2B protein sequence with zebrafish (Dr), mouse (Mm), and human (Hs) Popdc2 protein sequences. For comparison, the sequences of chick Popdc1 and Popdc3 were included. Amino acids that are conserved in at least three cDNAs are shaded in black. B: Dendrogram of different vertebrate Popdc2 proteins and chick Popdc1 and Popdc3, demonstrating that chick Popdc2 protein is most similar to other vertebrate Popdc2 sequences and only distantly related to Popdc1 and Popdc3. C: Reverse transcriptase-polymerase chain reaction (RT-PCR) amplification of HH stage 36 heart cDNA by using Popdc2 primers. Three PCR fragments of 593, 530, and 445 bp are visible. After sequencing, this cDNA fragments were found to represent Popdc2A, Popdc2A/B, and Popdc2B, respectively. D: Schematic representation of the splicing pattern, which generates the various Popdc2 isoforms. The different red boxes represent the various splice donor sites of exon 3, while the blue boxes represent exon IV. The genomic structure of the chick popdc2 locus has not yet been determined and is inferred from the gene structure of mouse and human popdc2, assuming a conserved exon/intron pattern (data not shown). Arrows demarcate the approximate positions of the primers (Popdc2-Ifwd and Popdc2-Irev) that were used for RT-PCR experiments shown in Figures 1C, 2B, 4. E: Deduced structure of the different Popdc2 protein isoforms. All isoforms are predicted to have three transmembrane domains (marked with roman numbers), a conserved Popeye domain (red box) in the middle of the molecule but differ at the carboxy terminus (labeled in different colors).

Figure 2.

Expression analysis of Popdc2 in the chick embryo. A: Northern blot analysis of Popdc1–3 gene expression in organs of Hamburger and Hamilton stage 39 (day 13) chick embryos. Exposure times for the different cDNAs were 3 days (Popdc1 and Popdc2) and 5 days (Popdc3). B: Reverse transcriptase-polymerase chain reaction analysis of Popdc2 isoform expression and as a loading control glyceraldehyde-3-phosphate dehydrogenase (GAPDH). M, muscle; H, heart; Lu, lung; T, tongue; Li, liver; B, brain; K, kidney; G, gut; S, stomach.

Figure 4.

Popdc genes are not expressed in the proepicardial organ. Reverse transcriptase-polymerase chain reaction analysis using primer pairs specific to Popdc1–3, Tbx18, AMHC1, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and cDNAs synthesized from RNA isolated from outflow tract (OT), ventricle (VT), atrium (AT), or proepicardial organ (PEO) of Hamburger and Hamilton (HH) stage 17 hearts or from total hearts of HH stage 30 embryos.

Popdc2 Is Predominantly Expressed in the Heart of Midgestation Chick Embryos

To study the mRNA expression of Popdc2 during chick organogenesis, Northern blot and RT-PCR analyses were performed by using total RNA isolated from various organs of day 13 chick embryos (Fig. 2A,B). Northern blot probed with Popdc2A revealed the presence of at least three mRNA species of 2.4, 3.5, and 4.4 kb that were predominantly expressed in heart. Weak expression was detectable in skeletal muscle. After longer exposure of the Northern blot, Popdc2 expression was also detectable in tongue, brain, and stomach (data not shown). The same blot was also probed for Popdc1 and Popdc3, which were also predominantly expressed in heart and weakly in skeletal muscle (Fig. 2A). RT-PCR analysis with primer pair Popdc2-I that would amplify all currently known Popdc2 isoforms (Fig. 1C) revealed the presence of three Popdc2 isoforms (Popdc2A, -2A/B, and -2B; Fig. 2B) in heart, skeletal muscle, stomach, and tongue. Some tissue-specific differences were also observed, i.e., thePopdc2A isoform was the predominant isoform in tongue, whereas the Popdc2B isoform was the predominant isoform in stomach. (Fig. 2B). In heart and skeletal muscle, all three isoforms appeared to be presented at similar concentrations. Popdc2C and Popdc2D isoforms were not detected in our analysis.

Popdc2 Is Expressed in Differentiated Myocardial Tissue During Chick Heart Formation

Expression of Popdc2 was first detected at Hamburger and Hamilton (HH) stage 7 by whole-mount in situ hybridization (Fig. 3A). Expression was confined to the anterior half of the bilateral heart fields (Redkar et al., 2002; Hochgreb et al., 2003) and appeared frequently asymmetrically distributed. In many embryos, the right heart field displayed weaker expression than the left one. This asymmetry was still visible at HH stage 8 (Fig. 3B). Transverse sections revealed that Popdc2 expression was confined to the precardiac mesoderm (Fig. 3I). At HH stage 10, the entire tubular heart showed Popdc2 expression that was weaker at the anterior end (Fig. 3C). Expression was limited to the myocardium and absent from the endocardium (Fig. 3J). At HH stage 11, the tubular heart was intensely labelled; however, both inflow and outflow tracts did not or only marginally express Popdc2 (Fig. 3D). At HH stage 15, the entire myocardium displayed expression of Popdc2 with the exception of the sinus venosus and conus (Fig. 3E). At HH stage 18, all myocardial segments were positive for Popdc2; in addition, some weak expression appeared in rostral somites. Transverse sections through the heart revealed that the outer curvature displayed stronger expression than the inner curvature (Fig. 3K), and the expression gradually increased toward the inflow tract of the heart. Of interest, the proepicardial organ did not show any hybridization signals. At HH stage 25, both the heart and the myotome expressed Popdc2 (Fig. G,L). In embryos with more intense staining, muscle cells within the limbs also expressed Popdc2 (data not shown). At HH stage 29, both atrial and ventricular myocardium were intensely labeled but weaker expression was found in the outflow tract myocardium (Fig. 3H). Sections through the ventricular myocardium revealed that the expression was confined to the subepicardial compact layer, while the epicardium, the endocardium, and the trabecular layer failed to express Popdc2 (Fig. 3 M,N).

Figure 3.

Expression pattern of Popdc2 during chick embryogenesis. A–H: Ventral view (A–D,H) and left lateral view (E–G) of embryos subjected to whole-mount in situ hybridization using Popdc2A as a probe. I–N: Transverse sections through the developing heart (I–K,M,N) or trunk (L) region of chick embryos. In each panel, the developmental stage is indicated at the lower left. AT, atrium; CL, compact layer; DM, dorsal mesocardium; EC, endocardium; EP, epicardium; IC, inner curvature; MT, myotome; NC, notochord; NF, neural folds; NT, neural tube; OC, outer curvature; PCM, precardiac mesoderm; PEO, proepicardial organ; SV, sinus venosus; TL, trabecular layer.

Popdc Genes Are not Expressed in the Proepicardial Organ

Conflicting data have been published regarding the expression of Popdc genes in the proepicardial organ. Immunohistochemical data in chick embryos and RT-PCR expression analysis in a rat epicardial cell line suggested that Popdc1 (Bves) is expressed in the proepicardial organ (Reese et al., 1999; Wada et al., 2001, 2003). However, previous analysis of our group by using whole-mount in situ hybridization on different vertebrate species could not support this observation (Andrée et al., 2000, 2002, 2003). To clarify whether any of the three Popdc genes was expressed in the proepicardial organ, we performed RT-PCR analysis on microdissected cardiac segments of hearts from HH stage 17 embryos (Fig. 4). All three Popdc genes were detected in atrial, ventricular, and outflow tract myocardium but not in the proepicardial organ. The splicing pattern of Popdc2 within the different heart segments was identical at this stage of development. In agreement with previous in situ hybridization data, the atrial segment expressed higher levels of Popdc1 and Popdc3 than the ventricle and outflow tract. Another myocardial differentiation marker, AMHC1, used as a control, was not expressed in the proepicardial specimen. In contrast, Tbx18, a marker for proepicardial development (Kraus et al., 2001), was strongly expressed in the proepicardial organ and to a lesser extent in myocardial tissue. Thus, we conclude that Popdc genes are not expressed in the proepicardial organ, in contrast to previous reports.


The Popdc gene family constitutes a novel gene family, which encodes transmembrane proteins with no homology to any other known proteins. Two conserved structural features are present in all chick Popdc2 proteins like in any other member of the Popdc family. First, the sequence of Popdc2 predicts the presence of three transmembrane domains, two of which constitute “primary” transmembrane helices, whereas the third one may be considered as being “secondary,” due to some charged amino acids within this domain (Hirokawa et al., 1998; Mitaku et al., 2002). Second, Popdc2 proteins contain the Popeye domain, that is highly characterized by several invariant amino acid residues (Andrée et al., 2003). The function of the Popeye domain is presently unknown, but it probably is involved in protein–protein interactions. In this regard, it is noteworthy that oligomeric complexes of Popdc proteins have been reported recently (Knight et al., 2003).

We and others have recently determined the membrane topology of Popdc1 with the N-terminus located extracellularly and the carboxy terminus inside the cell (Knight et al., 2003; Breher et al., manuscript in preparation). As the number of transmembrane helices present in the protein probably affects the membrane topology, it will be important to verify experimentally the predicted transmembrane domains in the Popdc2 proteins.

As shown previously for chick Popdc1 and Popdc3, multiple isoforms are also generated from the Popdc2 gene by alternative splicing (Andrée et al., 2000). We present evidence for five Popdc2 isoforms, generated by the utilization of several alternative splice sites. The resulting proteins differ in length and sequence of the carboxy terminus, which may alter the spectrum of proteins interacting with Popdc2 and the cellular function of individual isoforms. Splice variants of Popdc2 are not only present in the chick but also in zebrafish, mouse, and human transcriptomes (data not shown).

Popdc2 Displays a Specific Expression Pattern During Important Steps of Heart Development

Between HH stage 7 and 9, Popdc2 expression was found to be transiently asymmetric being stronger on the left than on the right side. The asymmetric expression may reflect differences in temporal progression of cardiac differentiation between the left and the right side (Satin et al., 1988). Asymmetric gene expression within the heart fields have been reported previously for several extracellular matrix molecules, including hLAMP and flectin on the left and JB3 on the right side (Smith et al., 1997; Linask et al., 2003). The Nodal-Pitx2 pathway, which governs many but not all decisions of left–right asymmetry in the vertebrate body is not responsible for the direction of cardiac looping (Brand, 2003). While it is currently not fully understood what controls early asymmetry of the heart fields, retinoic acid has been mechanistically linked to the asymmetric expression of JB3 and hLAMP and may also be involved in Popdc2 expression (Smith et al., 1997).

Differential expression of Popdc2 was also observed at HH stage 18 between the inner and outer curvature. Of interest, myocardium of the inner curvature is believed to retain the gene expression profile of the primitive tubular heart and undergoes remodelling, while the outer curvature myocardium is involved in atrial and ventricular chamber formation and is associated with a different pattern of gene expression (Christoffels et al., 2000). Whether Popdc2 expression is part of this process must await further analysis.

At HH stage 29, Popdc2 expression was confined to the subepicardial compact layer similar to Popdc1 and Popdc3 in chicken and mouse (Andrée et al., 2000, 2002). The compact layer containing less-differentiated cardiac myocytes, which maintain a high level of proliferative activity (Kastner et al., 1997). In contrast, cardiac myocytes of the trabecular layer are characterized by low levels of proliferative activity (Kochilas et al., 1999). Myocardial differentiation is controlled by the epicardium and depends on an autocrine loop in epicardial cells that involves retinoic acid and erythropoietin (Stuckmann et al., 2003). The epicardium secretes an unknown signaling factor that promotes myocardial proliferation and survival (Chen et al., 2002). Whether myocardial Popdc2 expression is dependent on trophic factors or affected by selective inhibition within the trabecular myocardium is currently unknown. Survival of trabecular myocytes depends on the endocardium and may be mediated by the neuregulin erbB2/erbB4 and BMP10 signaling pathways (Gassmann et al., 1995; Lee et al., 1995; Meyer and Birchmeier, 1995; Neuhaus et al., 1999).

Comparison of the three Popdc family members reveals distinct and overlapping expression patterns. Among the three genes, Popdc2 is the first to be expressed at HH stage 7, while Popdc3 becomes detectable at HH stage 10, and Popdc1 at HH stage 11 (Andrée et al., 2000). After heart looping has started, Popdc1 and Popdc2 are coexpressed in the entire tubular heart with low levels in the newly formed segments of outflow tract and sinus myocardium. In contrast, Popdc3 is predominantly expressed in the atrial compartment and elevated levels in the ventricles are only seen after epicardialization has started (Andrée et al., 2000). All three genes display stronger expression in the outer curvature, than in the inner curvature myocardium. At late stage of heart development, all three family members are specifically expressed in the compact layer myocardium and absent from the trabecular layer. Apart from the heart, each family member is also expressed in the myotome and subsequently in forming skeletal muscle. In conclusion, the different family members share similar expression patterns suggesting common regulatory mechanisms. Significantly, very similar expression profiles to the one observed for the chick were found in the case of zebrafish (Mavridou et al., in preparation), Xenopus (Hitz et al., 2002), and mouse (Andrée et al., 2000), suggesting that the expression pattern of Popdc genes has been conserved during evolution.

Absence of Popdc Gene Expression in the Proepicardial Organ

Conflicting data exist in the literature with regard to the presence of Popdc in the proepicardial organ and its derivatives. Immunohistochemical staining of chick embryonic hearts with a monoclonal Popdc1 antibody suggested expression in the proepicardial organ, delaminated mesenchyme, and coronary vascular smooth muscle cells (Reese et al., 1999). In contrast, a different monoclonal Popdc1 antibody specifically marked the myocardium and did not stain the epicardium (DiAngelo et al., 2001). We recently have documented the expression pattern of Popdc1 and Popdc3 during chick and mouse embryogenesis and found no evidence for Popdc expression in the propicardial organ or in its derivatives (Andrée et al., 2000, 2002). Here, we extended this analysis and show that Popdc1–3 are present in the myocardium but not in the proepicardial organ. There remains the possibility that very low levels of Popdc RNA are not detected by in situ hybridization or RT-PCR. In addition, it remains possible that Popdc isoforms are generated by alternative splicing, which are not amplified by the primer sets that have been used here. Recent immunohistochemical and RT-PCR data suggested the presence of Popdc1 in a rat epicardial cell line (Wada et al., 2003). However, proepicardial explants from the chick are able to differentiate into cardiac myocytes and smooth muscle cells, possibly explaining the observed Popdc1 gene expression in these cells (Kruithof, 2003).


Animals and Tissues

Fertilized White Leghorn eggs (Charles River) were incubated, and embryos were staged according to Hamburger Hamilton (Hamburger and Hamilton, 1951). For RNA preparation, organs were dissected and excised tissues were frozen in liquid nitrogen and stored at −80°C. For in situ hybridization, embryos were dissected in phosphate buffered saline and fixed overnight in 4% paraformaldehyde and stored dehydrated in methanol at −20°C.

Whole-Mount In Situ Hybridization

Whole-mount in situ hybridization and photographic documentation of the results were as previously described (Andrée et al., 1998). For the detection of Popdc2, a 990-bp fragment of Popdc2A (ChEST clone 986a14 [BU439958]) was linearized with EcoRI, and an antisense probe was synthesized with T7 RNA polymerase.

Northern Blot Analysis

Total RNA was isolated as previously described (Schultheiss et al., 1995). A total of 20 μg of RNA from different organs of day 13 chick embryos were loaded onto a 1% formaldehyde denaturing agarose gel, size separated, and transferred onto a nylon membrane (GeneScreen, NEN). The membrane was sequentially hybridized with a 1.2-kb EcoRI-NotI fragment of Popdc1, a 921-bp EcoRI-NotI fragment of Popdc2A (ChEST clone 986a14, accession no. BU439958), and a 1.4-kb NotI fragment of Popdc3. Finally, a chick glyceraldehyde-3-phosphate dehydrogenase probe served as a loading control.


cDNA was synthesized from DNase treated total RNA by using AMV reverse transcriptase. PCR was performed by using the primer pairs shown in Table 1. The PCR products were size separated on 2% agarose gels. PCR products were sequenced to prove identity.

Table 1. Primer Pairs Used
GenePrimer sequence

Cloning of Popdc2 Isoforms

A Blast search of the chick EST database (http://www.chick.umist.ac.uk/; Boardman et al., 2002) was performed by using the mouse Popdc2 sequence. A total of eight clones (accession nos. BU360286, BU439958, BU308467, BU354920, BU250832, BU313571, BU219238, BU314261) were identified, which showed significant homology to the mouse Popdc2 cDNA and were different from Popdc1 and -3 cDNAs (Andrée et al., 2000). The clones were subjected to sequence analysis and were found to be incomplete, either at the 5′ end or at the 3′ end. Moreover, the cDNA clones that mapped to the 3′ end of the Popdc2 sequence revealed significant sequence divergence. Four variant cDNA sequences, named Popdc2A–D were identified. The sequence for Popdc2A was represented by clones BU308467, BU250832, BU313571, and BU439958, Popdc2B by clone BU360286, Popdc2C by clone BU354920, and Popdc2D was represented by EST clone BU219238. The full-length Popdc2 sequences were assembled from the sequence of clone BU314261, which contained the 5′ end of the Popdc2 cDNA and terminated prematurely at an A-rich sequence, the different EST clones that mapped to the 3′ end of the Popdc2 cDNA, and an overlapping PCR fragment, which was amplified using primer pair Popdc2-II the sequences of which were derived from the 3′ end sequence of clone BU314261 and the consensus 5′ sequence of the other Popdc2 clones. The PCR fragment was amplified by using cDNA of HH stage 39 chick hearts. The sequence of Popdc2A/B was obtained by an RT-PCR reaction with primer pair Popdc2-I by using HH stage 39 heart cDNA. The full-length Popdc2A/B sequence was assembled as outlined above for the other splice isoforms. The cDNA sequences were deposited at NCBI with the following accession numbers: Popdc2A, AY388621; Popdc2B, AY388622; Popdc2C, AY388624; Popdc2D, AY388623; Popdc2A/B, AY427076.


The cDNA alignments and dendrogram were done with the Clustal algorithm using the Lasergene program. The prediction of the transmembrane domains were done with the Sosui program (http://sosui.proteome.bio.tuat.ac.jp/sosuimenu0.html) (Hirokawa et al., 1998). The presence of a Popeye domain in Popdc2 was identified by the CDART algorithm (http://www.ncbi.nlm.nih.gov/Structure/lexington/lexington.cgi).


We thank Fabienne Lütge for expert technical assistance.