Introduction

Although vertebrates are externally bilaterally symmetric, internal organs show a marked left–right (L–R) asymmetry, such as the left-sided presence of stomach and heart and occurrence of more lung lobes on the right side. This L–R asymmetry is genetically determined, and disturbance of the asymmetry often causes serious cardiovascular disease in humans 1, 2. Recent progress suggested that primary cilia on the primitive node have an important role in the establishment of the L–R asymmetry. Nodal cilia rotate and make a leftward current 3, 4, 5. This leftward flow subsequently induces asymmetrical gene expression such as nodal and lefty. The mouse mutant called inv (inversion of embryonic turning) shows a constant reversal of the L–R asymmetry in internal organs and develops cysts of the kidney 6. In inv mutant mouse embryos, the left side expression of nodal and lefty is inverted 7, 8 and abnormal nodal flow was reported 4, 9, indicating that inv acts upstream of nodal and lefty. The mouse inv gene encodes 1062 amino acids containing 15 tandem repeats of the ankyrin motif, two IQ motifs and two nuclear localization signals 10. In the inv mouse, only the first 91 amino acids of the inv protein remain, but the regions containing the ankyrin repeats as well as the IQ motifs were deleted 10, 11. The introduction of inv cDNA completely rescued all the inv phenotypes, not only situs inversus but also cyst formation of the kidney, indicating that the loss of the inv gene caused these two unrelated abnormalities 10. In humans, a mutation in the inv gene is also responsible for human laterality defects 12 and nephronophthisis (NPHP2) 13. Thus, the inv gene is essential in establishing normal L–R asymmetry and kidney development in both mice and humans.

Previously, we reported that the mouse inv protein interacts with calmodulin at two sites, and that microinjection of mouse inv mRNA into the right blastomere of two-celled Xenopus embryos resulted in randomized L–R asymmetry in the orientation of visceral organs 14. Furthermore, the calmodulin binding motif (the IQ motif) is critical for mouse inv mRNA in randomizing cardiac orientation, and Ca2+regulates the interaction between calmodulin and the mouse inv protein.

In this study, we isolated two distinct full-length cDNAs for Xenopus laevis inv homologs (Xinv-1 and Xinv-2, accession numbers AF321228 and AF321229 respectively), examined their expression and performed functional analysis of the conserved IQ motifs.

Materials and methods

Cloning of Xenopus inv homologs

Degenerative oligonucleotides were designed from regions encoding conserved ankyrin repeats of mouse and human inv proteins. Total RNAs of adult Xenopus kidney and liver were used as a template for cDNA synthesis. These RNAs were reverse-transcribed with Superscript II (Invitrogen, Carlsbad, California), and used as PCR templates. An amplified product was subcloned into the pGEM T-vector (Promega, Madison, WI) and sequenced. The product encoded multiple ankyrin repeats that were 70% identical in amino-acid sequence to mouse inv ankyrin repeats 3–10. The fragment was labelled with 32P-dCTP using an oligopriming kit (Amersham Pharmacia Biotech, Uppsala Sweden) and used as a probe to screen a Xenopus oocyte cDNA library (Clontech ZL5000a, Clontech, Mountain View, CA, USA). The first screening gave us four inv-related cDNA fragments. Sequences showed that three of these fragments encoded the same gene and were named Xinv-1, and the fourth cDNA was named Xinv-2. We screened at least half a million clones at each library screening.

To obtain a full coding sequence of the Xinv genes, we screened cDNA libraries, including two oocyte cDNA libraries (Clontech ZL5000a and ZL5001T), the kidney and liver cDNA libraries (Stratagene, La Jolla, CA, USA) and the stage 11.5 embryo cDNA library (a gift from Dr Bruce Blumberg). Rapid Amplification of cDNA Ends (RACE) was also performed to obtain the 3′ end of Xinv-1 using oocyte cDNA as a template.

Northern hybridization and RT-PCR

For Northern analysis, polyA RNAs were purified from total RNAs obtained from oocytes and liver. Five mg of polyA RNAs were electrophoresed and blotted onto nylon membranes (Amersham Pharmacia Biotech). Membranes were hybridized with a 32P-labelled Xenopus inv probe (Xinv-1) and washed with 2×SSC, 0.1% SDS at 60 °C and exposed to X-ray film for 1-7 days.

For RT-PCR, RNA was extracted from embryos at various stages by the AGPC method. Reverse transcription was performed using M-MLV reverse transcriptase (Invitrogen) following the manufacturer's protocol. Primers specific to each Xinv clone were as follows:

Xinv-1 forward: 5′-GCT GGT GGA AAA GGA GAA AGC TAG AA-3′

Xinv-1 reverse: 5′-CCT ACA TTA CAA CAC TGG AGA GCG CA-3′

Xinv-2 forward: 5′-CCT ACA TTA CAA CAC TGG AGA GCG CA-3′

Xinv-2 reverse: 5′-CTA CAC CAC ACC AAA CGT CAA TTG AAA-3′.

PCR reaction was performed using Ex Taq (Takara, Kyoto).

We amplified β-actin to verify the RT reaction. Primers for β-actin amplification were actinF: 5′-GGA CTT GGC TGG ACG TGA CCT GAC-3′, actinR: 5′-CAG ACT CAT CAT ACT CCT GCT TGC TG-3′ 18 cycles of PCR were performed.

Whole-mount in situ hybridization

The plasmid pYY94 contained 1453 bp of 3′ terminal region of Xinv-1 ORF (open reading frame) in pBluescript SK(-) (Stratagene, La Jolla, CA, USA). The plasmid was linearized at either the Not I site (for the antisense probe) or the Xho I site (for the sense probe) and subjected to in vitro transcription using T7 or T3 polymerase to produce an antisense or sense probe, respectively, in the presence of digoxigenin-modified UTP. Albino X. laevis embryos at various stages were fixed in MEMPFA (0.1 M MOPS pH 7.4, 2 mM EGTA, 1mM MgSO4, 4% paraformaldehyde) and processed for whole mount in situ hybridization as described previously 15. Coloring reaction was performed overnight at 4 °C, then overnight at room temperature using BM purple (Roche Diagnostics, Basel, Switzerland).

DNA construction, site-directed mutagenesis, in vitro transcription for mRNA injection and in vitro translation

The ORFs of Xinv-1 and -2 were PCR-amplified and ligated into the transcriptional vector pCS2+ 16 between the EcoR I and Stu I sites using a ligation kit (Takara). The sequence was confirmed not to have a mutation. For Xinv-1 transcription, we substituted the T3 promoter for the SP6 promoter of pCS2+. The plasmids, pYY158 (Xinv-1) and pYY159 (Xinv-2), were linearized at the Not I site, and transcribed with T3 RNA polymerase using MEGAscript kit (Ambion).

Mouse inv cDNA (pGEM-mINV) 10 was digested with EcoR I (Takara) and the 3 kb fragment obtained was ligated into the EcoR I site of pCS2+. The resultant plasmid pYY34 was linearized at the Not I site and transcribed with SP6 RNA polymerase. Site-directed mutagenesis of Xenopus inv IQ motifs was performed using GeneEditor kit (Promega). The mutated ORFs were subcloned to pCS2+ (modified its promoter from SP6 to T3) and transcribed with T3 RNA polymerase. To obtain GFP mRNA, pbGFP/RN3P 17 was linearized with SfiI and transcribed with T3 RNA polymerase. All in vitro transcription was performed in the presence of a cap analog (New England Biolabs).

Two cDNA clones, Xinv-1 and -2, were subjected to in vitro translation using rabbit reticulocyte lysate (Promega) in the presence of 35S-methionine. Subsequently, lysates were analyzed with SDS-PAGE. The gel was dried and examined by autoradiography.

Yeast two-hybrid assay

The following plasmids were used to examine the interaction with calmodulin in the MATCHMAKER yeast two-hybrid system (Clontech):

GBT-158 (Xinv-1 full-length cDNA)

GBT-182 (IQ1 mutated in Xinv-1)

GBT-183 (IQ2 mutated in Xinv-1)

GBT-184 (IQ1 and IQ2 mutated in Xinv-1)

The vector was derived from the pGBT vector (Clontech). Plasmids were introduced into the Y187 yeast strain with pCaM8 that contained the calmodulin gene fused to an activating domain for the yeast two-hybrid assay. Colonies, grown on minimal media lacking tryptophan and leucine, were tested for β-galactosidase activity with a colony-lift filter assay.

Microinjection of synthetic mRNA

Unfertilized eggs of X. laevis were manually ovulated from gravid females that had been injected with a human chorionic gonadotropic hormone, gonatropin (Teikoku Zoki, Kawasaki Kanagawa). Eggs were artificially fertilized, dejellied in 2% cysteine-HCl (pH 8.0) and kept in 0.1×Steinberg's solution 18. The mRNAs to be tested were microinjected at a dosage of 2000 pg/embryo unless otherwise noted, together with GFP mRNA (200 pg/embryo in all the experiments) in 10 nl of distilled water. As injection of 5 ng mRNA/embryo resulted in a frequent occurrence of malformed embryos (abnormal cleavage and gastrulation failure), we decided to inject 2 ng of inv mRNAs per embryo. Microinjection was performed at the late two-cell stage or the early four-cell stage. The mRNA solutions were injected into the right or left half of each embryo in the animal hemisphere at the presumptive dorsal and ventral regions. All of the embryos were placed in 1×modified Barth's solution (MBS), containing 3% Ficoll 400 and 50 μg/ml gentamycin during microinjection. Injected embryos were kept in 1×MBS until stage 7 19, and subsequently transferred into 0.1×Steinberg's solution for further culture. Embryos were kept at 23 °C in 0.1×Steinberg's solution, containing 50 μg/ml gentamycin, before further observations were made. Embryos were cultured until stage 22 19, and grouped into left-side injected and right-side injected embryos according to the distribution of the GFP signal under a fluorescent microscope (Olympus Model BHS-RFK, with a filter set for FITC observation). At stage 45, embryos were paralyzed with Tricaine (0.1% in 0.1×Steinberg's solution) and the orientations of heart and gut were examined under a dissection microscope. Statistical analyses were performed using a standard χ2 test, and p<0.05 was considered a significant difference.

Results

Two Xenopus inv homologs contain conserved ankyrin repeats and two IQ motifs

Two Xenopus cDNA clones, named Xinv-1 and -2, contained an open reading frame (ORF) homologous to mouse inv, and encode 1007 and 1002 amino acids, respectively (Figure 1A). Both ORFs started with an ATG initiation codon that agrees with the Kozak rule 20, and have an in-frame stop codon in the 3′ end. The clones are 89% identical to each other in amino-acid sequence. Nucleotide sequences of ORFs of Xinv-1 and Xinv-2 are 58.2% and 57.8% identical to that of the mouse inv ORF. At amino-acid level, Xinv-1 and Xinv-2 have 50% and 52% overall amino-acid identity to mouse inv. Both Xinv-1 and Xinv-2 have 15 repeats of the ankyrin motif in the N-terminus (a yellow box in Figure 1A), and two calmodulin-binding IQ motifs, named IQ1 and IQ2 (16 amino-acid residues), that are highly homologous to those of mouse and human inv genes. Amino-acid sequences of the ankyrin repeats of Xinv-1 are 73.4%, 71.8% and 74.5% identical to that of mouse, chick and zebrafish inv, respectively. Amino-acid sequences of the ankyrin repeats of Xinv-2 are 74.7%, 75.5% and 72.7% identical to that of mouse, chick and zebrafish inv, respectively. A search in the X. tropicalis database identified a sequence that was homologous to Xinv-1 and -2. The sequence encodes 15 repeats of the ankyrin motif and two IQ motifs. The sequence has 83.5% and 83.1% overall amino-acid identity to Xinv-1 and -2, respectively. We could not find other related sequences in mouse, human and X. tropicalis databases. IQ1 is just behind the ankyrin repeats, and IQ2 is located near the C-terminus (red boxes in Figure 1A). We also found two basic amino-acid residues, one next to IQ1 and the other at the N-terminal side of IQ2, conserved between mouse and Xenopus (blue boxes), which are named Basic Region (BR)-1 and -2, respectively. There is a conserved hydrophobic region (designated HR) in the amino-terminal side of BR2 (green box in Figure 1A). Despite the general similarity of the deduced protein structure, Xinv-1 has only one monopartite (SV40T antigen type) nuclear localization signal (NLS), and Xinv-2 has no typical NLS, while there are two typical bipartite (nucleoplasmin type) nuclear localization signals (NLS) in mouse inv protein (Figure 1B). These structures are well conserved among species (Figure 1B). Figure 2 shows a phylogenetic relationship among mouse, human, chick, zebrafish and Xenopus (tropicalis and two laevis) inv genes.

Figure 1
figure 1figure 1

(A) Alignment of putative Xenopus laevis inv amino-acid sequences along with mouse, chick (Gallus gallus) and zebrafish (Danio rerio), and putative X. tropicalis inv protein sequences. Xinv-1, Xinv-2, X. tropicalis inv, zebrafish inv, chick inv and mouse inv proteins are 1007, 1002, 1020, 1021, and 1269 and 1138 amino acids, respectively. The conserved motifs are indicated as follows. Yellow boxes: ankyrin repeats, red boxes: IQ motifs, blue boxes: conserved basic regions (BRs), green box: conserved hydrophobic region (HR). IQ motifs and BRs were numbered from the N-terminal (IQ1, 2 and BR1, 2, respectively). (B) Schematic description of the structure of mouse and Xenopus inv proteins.

Full size image
Figure 2
figure 2

A phylogenic tree of inv related genes. The tree is constructed by the neighbor-joining method 21 using the amino-acids sequences of ankyrin repeat domains. Values indicate expectation of nucleotide substitution on each residue.

Full size image

Xinv-1 is expressed maternally and in early embryos of Xenopus

The Northern hybridization analysis detected two signals of Xinv mRNA (about 4.4 and 5.8 kb) in the ovary. Only a larger band (about 5.8 kb) was detected in the liver (Figure 3A). As we obtained two Xinv cDNAs, we made specific primers in their 3′ ends to distinguish Xinv-1 from Xinv-2 mRNAs, and examined their expression. The signal for Xinv-1 was constantly obtained in RNA from the ovary, cleavage stage embryos through stages 3–27 (Figure 3B). Although the signal for β-actin was detected at 18 PCR cycles, the signal for Xinv-2 was not detected from the unfertilized egg (ovary) to the hatching stage (St.33), even at 50 PCR cycles (Figure 3B). The detection of Xinv signals in the ovary suggested a possibility that the inv message is maternally provided in the egg.

Figure 3
figure 3

(A) Northern hybridization for Xenopus inv cDNA expression. In all, 2 μg of polyA+RNA is loaded in each lane, and exposed for 7 days at −80 °C. A 4.4 kb transcript (arrowhead) was detected in ovary while not in liver. Size markers (in kb) are shown in the left side of the autoradiograph. (B) Temporal expression patterns of Xenopus inv examined with RT-PCR. Xinv-2 is not detected by PCR with up to 50 cycles, while Xinv-1 is detected at 46 cycles. RT-PCR with β-actin specific primers was carried out in parallel to control the amount of input RNA. RT (-), RT-PCR without reverse transcriptase. Stages are shown over each lane. (C) Expression pattern of Xinv from stage 6 to stage 27 shown by whole-mount in situ hybridization using digoxigenin-labeled Xinv-1 cRNA as a probe. No significant unilateral expression of Xinv is detected.

Full size image

Whole-mount in situ hybridization (Figure 3C) showed that the Xinv mRNA signal was present in the morula (St. 7) to the tailbud stage. After the neurula stages, Xinv signals were detected in the dorsal side of embryos. In the tailbud stage (St. 22), Xinv signals were also in the dorsal side with a little anterior bias. No obvious asymmetrical pattern was observed in the examined embryos. Throughout the stages, the signal was not detected using a sense probe.

A single amino-acid substitution of isoleucine to glutamic acid eliminates calmodulin-inv protein interaction

We reported that the region encoding the IQ motifs plays a crucial role in the activity of mouse inv mRNA to randomize the L–R asymmetry of Xenopus embryos by right-sided mRNA microinjection 14. Sequence analysis shows that Xenopus inv proteins have IQ motifs that are homologous to mouse inv protein (Figure 1A and 4A), and yeast two-hybrid assay showed that Xinv-1 protein also interacts with calmodulin (Figure 4B).

Figure 4
figure 4

(A) A single amino-acid substitution of isoleucine (I) by glutamic acid (E) eliminates calmodulin–inv protein interaction. (B) Description of the mutated Xinv IQ motifs. Substitution of isoleucine (I) with glutamic acid (E) is indicated in red. IQ motif mutated Xenopus inv's lose calmodulin-binding activity. The interaction between a wild-type Xinv protein or mutated Xinv proteins and calmodulin or SV40 large T antigen (CaM and SV40T, shown in the upper row as 'prey') was assayed using a yeast two-hybrid system. A schematic description of the constructs indicated in the left column as 'bait'. The white letters 'E' in IQ motifs (depicted as red blocks) represent the amino-acid substitution. In this experiment, wild-type Xinv proteins or mutated Xinv proteins were used as baits. Blue staining indicates the interaction between bait and prey. Only wild-type Xinv-1 shows calmodulin binding activity. SV40 large T antigen was a prey as a negative control of the yeast two-hybrid assay, and no interaction was observed.

Full size image

To verify that Xenopus inv protein interacts with calmodulin at the IQ motifs, we performed site-directed mutagenesis at the regions encoding IQ motifs (Figure 4A). We made three mutant constructs of Xinv-1, containing a single amino-acid substitution from isoleucine (I) to glutamic acid (E) in one of their IQ motifs or both (Figure 4A). Xinv-1/IQ1E, Xinv-1/IQ2E and Xinv-1/2IQE contain the mutation in IQ1, IQ2 and in both IQ1 and IQ2 regions, respectively (Figure 4B, left column). A full-length ORF of these mutant constructs were cloned into yeast expression vector, and a two-hybrid assay was performed using mouse calmodulin (CaM) as prey. The yeast two-hybrid assay indicated that all these mutants had lost their calmodulin-binding activity (Figure 4B). Interestingly, mutation in either one of the IQ motifs sufficiently eliminated the whole calmodulin binding activity of the Xinv protein, indicating that the Xinv protein–calmodulin interaction needs two intact IQ motifs.

IQ-mutated but not wild-type Xinv mRNA injection reverses cardiac orientation

Microinjection of mouse inv mRNA into the right blastomere of two-cell stage Xenopus embryos causes randomization of L–R asymmetry and Xnr-1 expression pattern 14. To examine whether Xinv mRNA has the same effect on L–R asymmetry, we injected synthetic mRNA of Xinv-1 (total 2 ng per embryo) into one of the blastomeres of two-cell-stage Xenopus embryos with GFP mRNA (Figure 5). Unexpectedly, the microinjection of Xinv-1 mRNAs did not cause an L–R inversion (Figure 6). We performed in vitro translation experiment of Xinv-1 and Xinv-2 to verify that Xinv mRNAs were translated, using rabbit reticulocyte lysate in the presence of 35S-methionine. Resulted lysates were analyzed with SDS-PAGE and autoradiograph, giving signals at the expected molecular size (not shown).

Figure 5
figure 5

Injection of synthetic mRNAs into the blastomere of the two-cell-stage Xenopus embryo and randomization of the L–R asymmetry. (A) Microinjection was performed at two-cell or early four-cell stage, and mRNA solutions were injected into the left or right half of the embryos. (B) Dorsal view of injected embryo observed under a fluorescent microscopy. GFP fluorescence occurs in the right half of the embryo. The injected embryos were divided into right- and left-injected groups according to the occurrence of GFP fluorescence at St. 23–24. (C) A noninjected embryo showing normal visceral L–R asymmetry. (D) An example of reversal of L–R asymmetry. (E) and (F) The outer appearance of normal heart and gall. Abbreviations: ot, outflow tract; v, ventricle; g, gall bladder. Inverted L–R asymmetry of heart and gall in the embryo. Note that the positions of outflow tract and gall bladder are completely inverted.

Full size image
Figure 6
figure 6

IQ-mutated but not WT Xinv mRNA injection reverses cardiac orientation. Cardiac reversal in Xenopus embryos injected with the mRNAs coding WT ORF or mutated construct of Xinv-1. Schematic description of the constructs indicated in the left lane. The red letters 'E' over the IQ motifs (depicted as red blocks) represent the amino-acid substitution. In the right graph, the percentages of embryos with cardiac reversal are shown. All three mutant Xinv mRNAs affected L–R asymmetry through the right-sided microinjection of mRNA, while the wild-type Xinv mRNAs did not. *Statistically different from that of the group injected with the Xinv-1WT RNA into a right blastomere.

Full size image

However, when the IQ mutated Xinv mRNAs were microinjected into the right blastomere of two-cell stage Xenopus embryos, the reversal of cardiac orientation was seen in a significant number of the embryos (Figure 6), though the percentage of cardiac reversals was lower than that caused by mouse inv, 18% (Xinv-1/2IQE) to 22% (Xinv-1/IQ2E). Throughout the experiment, left-sided microinjection had no effect on the cardiac reversal rate when compared with noninjected siblings (not shown).

Discussion

In the present study, we identified two X. laevis genes that were homologous to the mouse inv gene. The overall similarity between mouse and Xenopus inv's seems to be relatively low, but there are some distinct conserved domains between them: 15 ankyrin repeats, two CaM-binding motifs and NLSs (though NLS is absent in Xinv-2). These features are highly conserved among previously identified inv homologs of various kinds of vertebrates (mouse, chick and zebrafish). In addition, a search in the X. tropicalis database identified a sequence that was more than 80% identical to the two X. laevis genes. Although pylogenetic analysis showed that the distance between mouse and Xenopus inv genes is larger than that between mouse and zebrafish inv genes, this is not significant. Therefore, it is unlikely that the cloned genes are psedoalleles of the inv gene. These results suggest that the genes are X. laevis inv genes, and a possibility that their biological functions are shared.

The temporal expression pattern of Xinv-1 indicated that Xinv mRNAs exist maternally and are expressed constantly during embryogenesis, similar to that of mouse inv 22. Whole-mount in situ hybridization showed that Xenopus inv mRNA expresses evenly in both left and right sides of embryos. These results are also consistent with the previous observation in mouse embryos 10. Although the present study identified two Xenopus inv genes (Xinv-1 and -2), Xinv-2 mRNA was not detected by RT-PCR, suggesting that Xinv-1 is transcribed mainly during Xenopus development, and that Xinv expression patterns shown by whole-mount in situ hybridization (Figure 3) are likely to reflect the distribution of Xinv-1 mRNA.

A previous study showed that the injection of mouse inv mRNA into Xenopus two-cell-stage embryos reverses cardiac orientation 14, and suggested that calmodulin signalling was involved in the inv pathway for the establishment of L–R asymmetry. Unexpectedly, mRNAs encoding wild-type Xenopus inv proteins did not affect cardiac orientation when injected into Xenopus two-cell stage embryos. As the N-terminal ankyrin repeats are highly homologous, the difference between mouse and Xenopus invs on cardiac inversion is likely to lie in the C-terminal half of inv proteins, probably by the interaction between calmodulin and IQ motifs. IQ-mutated Xinv-1 mRNA randomized cardiac orientation and lost interaction with calmodulin (Figure 6), supporting this idea. The results suggest that the interaction with calmodulin also regulates the Xenopus inv function. The mutation in only one of the two IQ motifs eliminated the overall CaM binding activity of Xinv protein (Figure 6), suggesting an intramolecular mechanism controlling the interaction between calmodulin and inv protein.

Randomization of the L–R asymmetry was induced only when these inv mRNAs were injected into the right blastomere, whereas injection into the left blastomere was ineffective. Our results suggest a factor(s) that activates inv protein differently between two blastomeres. The present study suggests that calmodulin is an important factor in the control of inv function. Calmodulin is a conserved Ca2+-binding protein involved in a variety of cellular calcium-dependent signalling pathways 23, 24, 25. Ca2+regulates the interaction between calmodulin and inv protein 14. As recent studies suggest that there is a difference in Ca2+ concentration between the right and left sides in mice 26 and chickens 27, Ca2+ might be a candidate as a factor in making the difference in cardiac reversal induced by inv mRNA injection between right and left Xenopus blastomeres.