Three synonymous genes encode calmodulin in a reptile, the Japanese tortoise, Clemmys japonica

Three distinct calmodulin (CaM)-encoding cDNAs were isolated from a reptile, the Japanese tortoise (Clemmys japonica), based on degenerative primer PCR. Because of synonymous codon usages, the deduced amino acid (aa) sequences were exactly the same in all three genes and identical to the aa sequence of vertebrate CaM. The three cDNAs, referred to as CaM-A, -B, and -C, seemed to belong to the same type as CaMI, CaMII, and CaMIII, respectively, based on their sequence identity with those of the mammalian cDNAs and the glutamate codon biases. Northern blot analysis detectedCaM-Aand-Bas bands corresponding to 1.8 kb, with the most abundant levels in the brain and testis, while CaM-Cwas detected most abundantly in the brain as bands of 1.4 and 2.0 kb. Our results indicate that, in the tortoise, CaM protein is encoded by at least three non-allelic genes, and that the ‘multigene-one protein’ principle of CaM synthesis is applicable to all classes of vertebrates, from fishes to mammals.


Introduction
Calmodulin (CaM), a prototype of EF-hand Ca 2+ -binding protein, is ubiquitously distributed in all eukaryotic cells, and is particularly abundant in the brain and testes of mammals (Kakiuchi et al., 1982). CaM function is essential for the viability of cells and tissues, as well as for neuronal activity in the central nervous system (CNS) (Kennedy, 1989). When CaM binds to the Ca 2+ ion, the protein becomes active, and it in turn activates many other enzymes. It thus plays a pivotal role as a cofactor regulating a wide variety of calcium-dependent proteins.
The amino acid (aa) sequence of CaM protein is well conserved in organisms whose evolution diverged millions of years ago. Three non-allelic CaM genes encoding exactly the same 148 aa protein (17,000 daltons) were identified in the rat (Nojima, 1989 ), mouse (Bender et al., 1988;Danchin et al., 1989), and humans (Fischer et al., 1988); two were identified in Xenopus (Chien and Daxid, 1984), and four in the teleost fish medaka (Oryzias latipes) (Matsuo et al., 1992). Although the reason for the presence of multiple synonymous genes in animals is unclear, differences in the 5' and 3' non-coding region of the rat CaM genes suggest that each gene may be differentially regulated. The overall distribution of the three CaM transcripts was coordinated in rat brain , however, temporal and spatial differences in the expression of CaM genes have been observed in some areas of the CNS, both under physiological and experimental conditions (Michelhaugh et al., 1998;Palfi et al., 2000, Palfi et al., 2001, Shimizu et al., 1997Sola et al., 1997;Weinman et al., 1991). Knowledge of the evolutionary relationships between vertebrate CaM genes should provide useful information (in) regarding (to) this multiple-gene system. To this study, we attempted to isolate CaM cDNAs from a reptile, the Japanese tortoise (Clemmys japonica), which occupies a critical position in the phylogenetic tree of vertebrate evolution. The results indicate that reptiles also possess three active CaM genes, corresponding to CaMI, CaMII, and CaMIII, respectively, and thus the 'multigene-one protein' principle of CaM synthesis is as applicable to reptiles as it is to other vertebrates.

Isolation of total RNA
The Japanese tortoise was purchased from a commercial source. The animal was anesthetized by ethyl-ether and euthanized by decapitation on ice. Dissected organs (brain, liver, kidneys, intestines, heart, and testes) were immediately frozen in liquid nitrogen. Total organ RNA preparations were obtained by the guanidine-thiocyanate-CsCl ultra-centrifugation method (Chirgwin et al., 1979).

Amplification, subcloning and sequencing of CaM cDNA
A set of degenerate primers was synthesized by a DNA synthesizer, as described in a previous paper (Matsuo et al., 1992): Amino-terminal: 5'-ATG GCN GAY CAR CTN ATN GA and Carboxyl-terminal: 5'-GGGTCTAGA YA YTT NGC NGT CAT CAT YTG. N represents an A, G, C, or T base; Y, a C or T base; and R, an A or G base. The underlined bases indicate an XbaI site. First-strand cDNA was synthesized from 2 µg of total RNA using 50 pmol of the Carboxyl-terminal degenerate primer as described by Kawasaki (1990). RCR was performed according to the methods described in a previous paper (Matsuo et al., 1992). PCR products were treated with Klenow (large) fragment of E. coli DNA polymerase I, digested with XbaI, and fractionated in 1.5% agarose gel. About 450-bp bands were cloned between the Sma I and Xba I sites of pUC19. Nucleotide sequence analysis was performed by a DNA sequencer (373A, Applied Biosystems, CA) with Cycle Sequencing Kit (Takara, Tokyo), according to the manufacturer's instructions. Each nucleotide sequence was confirmed from both directions. The sequences of these cDNAs appeared in the DDBJ/EMBL/GenBank under accession Nos. AB055384 (CaM-A), AB055385 (CaM-B) and AB055386 (CaM-C).

Southern blot hybridization
One hundred ng of cloned CaM genes, CaM-A, -B and -C, were digested with XbaI and EcoRI restriction enzymes, submitted to electrophoresis in 0.8% agarose gel, and then blotted onto a NYTRAN membrane (Schleicher & Schuell). The membrane was pre-hybridized at 65°C in hybridization solution (6 x SSC/50 mM sodium phosphate buffer, pH7.0/2 x Denhardt's solution/0.5% SDS/0.2 mg salmon sperm DNA/mL). A random hexamer-primed 32 P-labeled DNA probe of each coding region was added and allowed to hybridize for 20 h at 65°C. The membrane was washed twice in 2 x SSC/0.1% SDS at room temperature for 5 min, twice in 0.1 x SSC/0.1% SDS at 65°C for 20 min, and exposed to X-ray film overnight at -80°C.

Northern blot analysis
Five µg of total tortoise RNA were denatured with glyoxal and dimethyl sulfoxide, submitted to electrophoresis in 1.0% agarose gel, and then blotted onto a NYTRAN membrane. The membrane was pre-hybridized for 4 h at 68°C in hybridization solution (50% formamide/5 x SSC/50 mM sodium phosphate buffer, pH7.0/2 x Denhardt's solution/0.5% SDS/0.2 mg salmon sperm DNA/mL). A random hexamer-primed 32 P-labeled DNA probe of each coding region was added and allowed to hybridize for 20 h at 68°C. The membrane was washed three times in 0.2 x SSC/0.1% SDS at 68°C for 20 min, and exposed to X-ray film overnight at -80°C.

Isolation of three CaM-coding regions
Our previous study showed that the degenerate primers encoding the Amino-and Carboxyl-terminal six aa of CaM are useful for reverse transcription and PCR of total teleost fish RNA (Matsuo et al., 1992). In this study, we also employed the same pair of degenerate primers to isolate CaM cDNA from transcripts of tortoise brain ( Figure  1). Eleven of the 60 cDNA clones obtained fell into three groups of distinctive CaM sequences, designated A, B, and C. The nucleotide sequences of these three groups of cDNA are shown in Figure 1. The deduced aa sequences were 44 Shimoda et al. exactly the same, and they were identical to those of the vertebrate CaM genes. Since the nucleotide substitutions were scattered throughout the coding regions, the isolated coding regions were unlikely to have been generated by alternative splicing from a single gene. We therefore concluded that at least three distinct CaM genes, CaM-A, -B, and -C, were transcribed in the tortoise. The existence of three different CaM genes was further confirmed by Southern blot hybridization (Figure 2). Each probe of CaM-A, -B, and -C could detect its own clone, but not others, in stringent conditions (Tm-7°C).

Comparative analysis of the tortoise CaM sequences and codon bias of glutamate
The sequence homologies between any two of the three tortoise cDNAs were 82% to 85%, which is very low considering that they encode CaM, i.e., a gene product with identical aa sequences. Comparisons with the CaM sequences of other animals showed that CaM-A, -B, and -C exhibited the highest homology to CaMI, CaMII, and CaMIII, respectively, within each species (Table I). We therefore concluded that CaM-A, -B, and -C should be classified as CaMI, CaMII, and CaMIII genes, respectively. This conclusion was also supported by the fact that the codon usage of glutamate, which accounts for 21 residues of the CaM protein, was biased to the GAA codon in CaMI and CaMII, as opposed to the GAG codon in CaMIII (Table  I). However, further confirmation of sequence homology in the non-coding and regulatory regions will be necessary.

Northern blot analysis
Brain, liver, kidney, intestine, heart, and testis were analyzed for expression of tortoise CaM genes by Northern blot hybridization (Figure 3). The CaM-A and -B probes detected bands of about 1.8 kb, respectively. The strongest expression of both genes was observed in the brain and testis. The possibility remained that the 1.8kb band was at least partially the result of cross-hybridization between CaM-A and -B, however, the CaM-A and -B genes were specifically detected by Southern blot hybridization, and the expression pattern in various tissues was slightly different. The CaM-C probe detected two bands of about 1.4 and 2.0 kb, most prominently in the brain. The results of other studies indicated that utilization of multiple polyadenylation sites yielded CaM RNAs of different sizes in other organisms (Lagace et al., 1983;Nojima, 1989;Nojima et al., 1987).
Calmodulin genes in tortoise 45  a) The pairwise sequence alignment was performed in the 409 nucleotide indicated in Figure 1. b) Sequence identities (%) with corresponding mammalian genes are shown. Asterisks (*) indicate the highest homology with CaM genes in the species. c) CaM protein contains 21 glutamate residues.
Our results clearly demonstrate that the CaM protein of the tortoise is encoded by at least three independent CaM transcripts.

Discussion
The 'multigene-one protein' principle of CaM synthesis is conserved in the reptile, and thus seems to be common to all classes of vertebrates (Toutenhoofd and Strehler, 2000). The nt sequence homologies between any two of the three types of CaM cDNA are very low, considering that the aa sequence of CaM protein is identical in all vertebrates. Comparisons with the nt sequences of other animals may provide some information concerning the type of CaM cDNA (Table I). Our results also indicate that the codon usage of glutamate is an additional landmark for the classification of CaM genes.
Although we do not know why organisms that lost the multiple CaM gene system failed to survive natural selection, this multigene system may confer proper expression of CaM protein in the cells of vertebrate tissues. While all three CaM genes are coordinately expressed in adult rat brain , slight but definite differences in individual gene expression have been observed in CNS neurons under normal conditions . Differential responses of CaM genes in the rat brain after administration of psychotropic drugs (Michelhaugh et al., 1998;Shimizu et al., 1997) and a convulsant neurotoxic agent (Sola et al., 1997) have also been reported. Our previous studies indicated that the disparate promoters of rat CaMII and CaMIII direct similar, but not identical, expression patterns in the CNS of adult and fetal transgenic mice Shimoda et al., 1995). Further analysis of the CaM 5' and 3'-flanking regions in reptiles and other vertebrates is needed to confirm both the regulatory mechanisms and the evolutionary relationships of CaM genes.