Identification of three prominin homologs and characterization of their messenger RNA expression in Xenopus laevis tissues

Animals

Adult X. laevis and X. tropicalis were purchased from Xenopus Express, Brooksville, FL. The animals were kept in an environment of artificially controlled 12 h:12 h light-dark cycle and constant temperature of 18 °C. All experiments were conducted in accordance with the Animal Care and Use Committee of the University of Connecticut Health Center and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

Cloning of Xenopus laevis prominin-1 cDNA

A protein versus translated nucleotide BLAST search was performed against the Sanger Institute X. laevis expressed sequence tag (EST) database, using the protein sequence of human prominin-1 (GenBank accession number: NP_006008) as a query. Nested reverse-transcription (RT)-PCR was performed to isolate a partial cDNA from an X. laevis (adult) retinal cDNA library (a gift from Dr. Joseph C. Besharse, Medical College of Wisconsin, WI) [38]. Full-length cDNAs were subsequently isolated by performing 5′ and 3′ rapid amplification of cDNA ends (RACE) using cDNA synthesized from X. laevis retina total RNA as a template [39]. Sequences of primers used in nested RT–PCR and RACE are provided in Table 1. The isolated cDNAs were cloned into pCR 2.1 TOPO cloning vector (Invitrogen, Carlsbad, CA) and sequenced using the dye-termination method [40].

Cloning of Xenopus laevis prominin-2 and 3 cDNAs

We performed BLAST searches with sequences of known prominin homologs, including human prominin-1 (NP_006008) and 2 (NP_653308), and mouse prominin-1 (NP_032961) and 2 (NP_620089), against the Ensembl genomic database of Xenopus tropicalis [41]. Partial sequences resulting from this search were used to search the X. laevis EST databases at the Sanger Institute and to design primers to amplify prominin homologs from X. laevis by RT–PCR and RACE, using cDNA synthesized from X. laevis retina or testis total RNA as templates. Sequences of primers used in RT–PCR and RACE are provided in Table 1.

Sequence analysis of prominin homologs

Maximum likelihood phylogeny of prominin homologs from different species was generated using PhyML 3.0 software [42]. cDNAs of prominin homologs of the human, chimpanzee, rhesus monkey, rat, mouse, horse, cattle, dog, zebrafish, nematode, and fly were retrieved from GenBank. Prominin homologs of X. tropicalis were cloned by methods similar to those used for X. laevis; cDNAs synthesized from X. tropicalis retina or testis total RNA were used as templates for PCR and RACE. Bovine and platypus prominin homologs were predicted from genomic sequences by BLAST search and by aligning them with cDNA sequences of known prominin homologs and searching for the consensus exon/intron boundaries [43,44]. Platypus genomic sequences were retrieved from Ensembl Ornithorhynchus anatinus version 60.1n (OANA5) Chromosome 4. Bovine genomic sequences were retrieved from Ensembl Bos taurus version 60.4i (Btau_4.0) Chromosome 6. Other software used in sequence analysis included ClustalW for sequence alignments [45], SignalIP 3.0 for signal peptide prediction [46], NetOGlyc 3.1 for glycosylation site prediction [47], TMHMM Server, version 2.0 for prediction of transmembrane domains [48,49], and ScanProsite software for identification of PDZ-binding motifs [50].

RNA isolation and cDNA synthesis

RNA was extracted from the adult X. laevis retina, brain, testis, and kidney using TRIzol reagent (Invitrogen) and used to synthesize first-strand cDNAs with SuperScript II reverse transcriptase (Invitrogen) and oligo-dT primers (Amersham, Piscataway, NJ) according to protocols supplied by the manufacturers.

Analysis of alternative splicing

Alternatively spliced exons of xlProminin-1, 2, and 3 were identified by sequencing and alignment of multiple cDNA clones. Exon/intron boundaries were determined by comparing cDNA sequences with X. tropicalis genomic DNA sequences. The reference genomic sequences used were: X. tropicalis unplaced genomic scaffold, v4.2 XENTRscaffold_393, whole genome shotgun sequence, containing the prominin-1 gene, GenBank accession NW_003163719; X. tropicalis unplaced genomic scaffold, v4.2 XENTRscaffold_30, whole genome shotgun sequence, containing the prominin-2 gene, GenBank accession NW_003163356; X. tropicalis unplaced genomic scaffold, v4.2 XENTRscaffold_640 and 3098, whole genome shotgun sequence, containing the prominin-3 gene, GenBank accession NW_003163966 and NW_003166412, respectively. Alternative splicing at the C-terminus of xlProminin-1 in the retina, brain, testis, and kidney was examined by amplifying a cDNA fragment containing this region by RT–PCR. Primers used in this analysis were: 5′-ATA AGA ATG CGG CCG CCT GCA GAG AAC ATC CTT TGA TAT CGA G-3′ and 5′-CGC GGA TCC CAG ATT ATT TTC TGA AAG TCA TTT CTC C-3′. (Note that the Not I and BamH I sites were incorporated into the primer sequences to facilitate cloning.) The profiles and relative abundances of differentially spliced transcripts were analyzed by comparing the sizes and intensities of RT–PCR products resolved by electrophoresis on a 1% agarose gel. Optical densities were measured with ImageJ software [51] for quantification of different splicing isoforms of xlProminin-1. Individual bands were extracted from the gel using the GeneClean® III Kit (Qbiogene, Irvine, CA), cloned into the pCR 2.1 TOPO cloning vector (Invitrogen) and sequenced.

Semiquantitative reverse-transcription PCR

Two primer pairs were designed for specific amplification of each prominin homolog (Table 2 for sequences of primers). Reverse-transcriptase PCR amplification of a fragment of β-actin was used as a loading control to establish that an equal amount of starting materials was used (Table 2). PCR reactions were assembled using an Advantage® 2 PCR Kit (Clontech, Mountain View, CA) in a final volume of 100 μl containing 4 μl of cDNA. Aliquots (20 μl) taken at the end of the 27th, 31st, 35th, and 39th cycles were visualized on a 1% agarose gel by ethidium bromide staining, and digital images were analyzed using Image J software. The total amount of DNA in selected areas was represented with integrated optical densities. Results obtained with two different primer pairs were averaged and plotted.

Identification of three prominin homologs in Xenopus laevis

By performing a BLAST search of the X. laevis EST database, we retrieved a sequence (GenBank accession: BJ061435) with 62% similarity to human prominin-1. BLAST search of the NCBI nucleotide database and the Ensembl X. tropicalis genomic database using the available sequences of prominin homologs from different species as query sequences revealed three different prominin homologs in X. tropicalis. These sequences have the assigned GenBank accession numbers of BC127277, XM_002932561, and XM_002940570. Sequence BC127277 showed a high degree of protein sequence identity (95%) to the above mentioned X. laevis sequence BJ061435. However, X. laevis homologs of sequences XM_002932561 and XM_002940570 could not be retrieved from any public cDNA or genomic database by BLAST search, perhaps due to incomplete coverage of the X. laevis sequences by those databases. They were instead cloned by RT–PCR using specific primer pairs based on X. tropicalis prominin homolog sequences, employing cDNA synthesized from total RNA isolated from X. laevis tissues as templates. Multiple clones were obtained and sequenced. Full-length cDNA sequences of the three X. laevis prominin homologs were obtained by performing 5′ and 3′ RACE. Predicted protein sequences are aligned in Figure 1.

Phylogenetic analysis of prominin homologs

Previous taxonomic studies of prominins have divided prominin homologs into three groups. These are prominin-1 [2,31] and prominin-2 in mammals [3], and prominin-3 in zebrafish [52]. Two variants of prominin-1, namely prominin-1a and 1b, were identified in zebrafish [52]. We performed protein phylogenetic analysis using the PhyML 3.0 software to compare X. laevis prominin homologs with available prominin sequences; the results are presented in Figure 2. The sequences used to construct the phylogenetic tree were from online public databases, newly cloned cDNAs, or predicted mRNAs from genomic sequences (Methods). Since the N- and C-termini of prominin homologs were incomplete in many of the retrieved sequences, only the sequences from exon 4 to exon 24 of prominin-1 and homologous sequences of other prominins were used for the phylogenetic analysis. The analysis revealed that the first cloned prominin homolog is the X. laevis ortholog of prominin-1, and is therefore designated as xlProminin-1. The other two prominin-like cDNAs found in X. laevis were named xlProminin-2 and xlProminin-3 according to their positions on the phylogenetic tree.

All newly cloned and predicted prominin homologs were also named according to their positions on the phylogenetic tree, and full sequences (including certain splice variants) were deposited in GenBank with assigned accession numbers (see Table 3 and Table 4), including sequences from: X. tropicalis, fugu, platypus, and cattle. We made new predictions for bovine prominin-1 from genomic sequences, because previously predicted bovine prominin-1 sequences (XP_875477 and XP_002688513) contain obvious errors and cannot be used for an accurate phylogenetic analysis. Our newly predicted bovine prominin-1 was assigned with the GenBank accession HQ159409. Sequences used in the phylogenetic analysis and the alignment of these sequences are presented in Appendix 1.

We compared the results of our phylogenetic analysis of prominin homologs with previous findings and found that X. laevis and X. tropicalis are the only species verified so far to possess all three prominin orthologs. Only two prominin homologs, prominin-1 and 2, were found in human and mouse. We conducted BLAST searches of the NCBI nucleotide database and the Ensembl database for prominin homologs in more species of mammals, including the human, chimpanzee, rhesus monkey, cattle, horse, dog, and rat. All retrieved prominin homolog sequences from these mammals were better grouped with prominin-1 or 2 than with prominin-3. A total of three prominin homologs were found in zebrafish (Danio rerio) and fugu (Takifugu rubripes). However, these fish do not have a prominin-2 homolog; they instead have two prominin-1 homologs, prominin-1a and 1b, and a prominin-3 homolog [52]. Incomplete sequences of prominin homologs were retrieved from EST or genomic databases of the chicken (Gallus gallus) and platypus (Ornithorhynchus anatinus, a primitive egg-laying mammal). GenBank accession numbers of these sequences are given in Table 3. They seem to group with prominin-1, 2, or 3 by sequence alignment. However, further taxonomic studies are needed to precisely position them in the phylogeny of prominins. Three C. elegans prominin homologs and two D. melanogaster prominin homologs were used in constructing the prominin phylogeny. These sequences did not belong to any of the three groups of prominins in our analysis, suggesting the possibility of other prominin homologs.

All three prominin homologs are predicted to be pentatransmembrane glycoproteins

Similar to mammalian prominins, hydropathy analysis using TMHMM software predicted that all X. laevis homologs have five hydrophobic domains, which likely constitute membrane-spanning domains that divide the sequence into two small intracellular loops and two large extracellular domains. However, the last hydrophobic sequence in xlProminin-2 did not reach the set threshold of TMHMM for prediction of a transmembrane domain (Appendix 2). Alignment of translated protein sequences from three cloned X. laevis prominin homolog cDNAs reveals that the most conserved regions are the transmembrane domains, as illustrated in Figure 1. A cysteine-rich region at the junction of the first transmembrane domain and intracellular loop (M1 and I1) is conserved in xlProminin-1 and 3, but is less conserved in xlProminin-2. Multiple leucine residues on the extracellular domains (E2 and E3) are conserved in all three xlProminin homologs. The conserved cysteine-rich region and leucine residues were all previously identified in mammalian prominin-1 and 2 [3,6,53,54]. The extracellular domains of all three xlProminins contain consensus sequences for N-linked glycosylation. In contrast to mouse and human prominin-1, no PDZ-binding motifs were identified in the C-terminal splice variants of xlProminin-1, 2, or 3 using ScanProsite software. PDZ-binding motifs were identified on the C-termini of certain splice variants of prominin-1 in the rodent (s1, s2, s7, s8, and s11) and human (s3, s4, s5, and s9) models, [31] and could potentially modulate interactions of prominin-1 with other proteins.

Three prominin homologs are differentially expressed in Xenopus laevis tissues

Although prominin-1 is broadly expressed, the principle phenotype observed in patients is retinal degeneration [19,20,22-24]. It was proposed that there may be functional and spatial overlap of prominin homologs in most tissues, such that they compensate for one another if a loss of function mutation occurs [3]. We therefore analyzed expression of the three xlProminin homologs in different X. laevis tissues to assess their relative expression levels in those tissues. We were particularly interested in determining whether any prominin homolog is predominantly expressed or absent in particular tissues among the four we studied.

We performed semiquantitative RT–PCR with mRNAs isolated from the X. laevis retina, brain, testis, and kidney to analyze the expression profiles of xlProminin-1, 2, and 3, because these four tissues are mRNA rich and differ significantly in their function and metabolism. The results are presented in Figure 3. Precautions were taken to ensure specificity of the PCR reaction and accuracy of the analysis, including ensuring that measurements were consistent when using two completely different primer pairs for each prominin homolog. We also used relatively large PCR products (>1.4 kb), such that small nonspecific products could easily be recognized and disregarded. For the primer design, we carefully chose primers that only complement the desired homolog. We found that mRNAs of three xlProminin homologs were differentially expressed in these tissues: xlProminin-1 was the predominant form found in the retina, brain, and testis, while xlProminin-3 was the predominant form found in the kidney. The expression level of xlProminin-2 was the lowest of the three prominin homologs in all examined tissues in this frog.

Splice variants of Xenopus laevis prominin-1

A total of seven full-length splice variants (s1–s7) of xlProminin-1 were cloned and sequenced, and the presence or absence of certain alternatively spliced exons is summarized in Table 4. These splice variants were not named following the nomenclature for splice variants of human and mouse prominin-1 [31], because some alternative exons in xlProminin-1 do not have corresponding exons in human and mouse, and vice versa. Comparison of exon organization of X. laevis, mouse, and human prominin-1 is described in detail below. Alignment of these cloned cDNAs reveals that there are at least seven alternatively spliced exons in the xlProminin-1 gene (Figure 4). We numbered exons following the established nomenclature [31]. Exons 3a (“a” for “alternative”), 8a, 11a, and 28a are newly discovered alternative exons found in X. laevis. They were numbered based on their positions relative to the constitutive exons. Exons 19, 26b, and 27 are alternative exons in the human and mouse that were alternatively spliced in X. laevis. Alternatively spliced exons 26b to 28a of xlProminin-1 reside in a region encoding the C-terminus of the protein. Exons 3 and 28 of xlProminin-1 were always retained, despite the fact that they appear to be alternative in the mouse and human [2,31]. We noticed that the inclusion or deletion of these alternatively spliced exons does not change the downstream reading frame, except for exon 28a, within which an in-frame stop codon resides. Inclusion of exon 28a results in a longer but different protein sequence for the C-terminus of xlProminin-1 relative to other isoforms.

Xenopus laevis prominin-2 and 3 are also alternatively spliced

Multiple variants arising from alternative splicing of all three prominin homologs were identified among mRNAs isolated from different tissues (Figure 4). A region near the 5′ end of xlProminin-2 transcript was alternatively spliced. Exon 1 of xlProminin-2 was predicted to contain the translation initiation site (ATG). Surprisingly, exon 1 was skipped in one of the sequenced clones, implying an alternative start codon in exon 2. This proposed alternative start codon was associated with a suboptimal Kozak consensus sequence (−3 CAAATGG +4). The sequence of one full-length xlProminin-2 was obtained; this was assigned GenBank accession FJ358519. The 5′ end of xlProminin-3 transcript and its exon 24a were found to be alternatively spliced. Two full-length splice variants (s1 and s2) of xlProminin-3 were obtained, and assigned GenBank accession numbers FJ358520 and JF274982, respectively. xlProminin-3.s1 contains an alternatively spliced exon 24a, while xlProminin-3.s2 does not. Details of exon organization, alternative 5′ untranslated region (UTR) or alternative polyA sites at 3′ UTR of xlProminin-1, 2, and 3 are provided in Appendix 3. The functional significance of the alternatively spliced variants is unknown.

Characterization of alternative splicing of Xenopus laevis prominin-1 in exons 20 to 28

Fargeas et al. previously reported that splicing of the prominin-1 gene in mice results in great variations in the sequences of alternative C-termini [32]. We therefore examined the splicing profile of xlProminin-1 within a region from exon 20 to 28 by RT–PCR (Figure 5). We found that the two most abundant isoforms of xlProminin-1 in the X. laevis retina did not contain exon 27; their exon compositions were: (20–24)+26b-27-28a-28+ (ε in Figure 5) or (20–24)+26b+27-28a-28+ (δ in Figure 5). Isoforms containing exon 28a were also detected in the retina; their exon compositions were: (20–24)+26b-27-28a+28+ (γ in Figure 5), (20–24)+26b+27-28a+28+ (β in Figure 5) and (20–24)+26b+27+28a+28+ (α in Figure 5). However, isoforms containing exon 28a only accounted for a very small proportion of all xlProminin-1 in the retina (Figure 5). This result was unlikely to be biased by preferential amplification of smaller PCR products, since profiling of xlProminin-1 alternative splicing in this region gave very different results in the kidney, testis, and brain: Those tissues preferentially express the isoforms with exon 28a (Figure 5).

Comparison of the exon organization of Xenopus laevis, mouse, and human prominin-1

The organization and splicing of prominin-1 exons in the mouse and human have been well characterized [30-32]. To determine whether the gene structure is conserved in xlProminin-1, we aligned all identified exons of prominin-1 from X. laevis, mice, and humans (Figure 6). Some exons found in one species were absent in another, but these were always alternatively spliced exons (Figure 6). We also correlated the positions of all alternatively spliced exons of xlProminin-1 to its membrane topology and found that all alternatively spliced exons represent sequences coding for the extracellular domains (E1, E2, and E3) or the C-terminus of the protein, as was previously observed in prominin-1 from mouse and human [3,30-32].