Unique structure (construction and configuration) and evolution of the array of small serum protein genes of Protobothrops flavoviridis snake

Protobothrops flavoviridis (Pf) (Crotalinae snake) [1] inhabit the southwestern islands of Japan, mainly, Amami-Oshima, Tokunoshima, and Okinawa. The venom of P. flavoviridis contains a huge variety of toxic proteins. The representatives are snake venom metalloproteases (SVMPs) [2], serine proteases [3], phospholipase A₂s (PLA₂s) [4–7], and triflin [8]. High molecular weight SVMPs, called HR1a and HR1b [9], and middle molecular weight SVMPs, called HR2a and HR2b [10], are isozymes of each other and cause severe hypodermic hemorrhage [11]. Low molecular weight SVMP, called HV1, induces apoptosis of vascular endothelial cells [12]. Serine protease, called Flavoxobin [13,14], is known as coagulant factor. Triflin is a neurotoxin-like protein and blocks muscle contraction [15]. Protobothrops genus snake venom PLA₂ isozymes, which hydrolyze phospholipids [16,17], are generally classified into four groups, according to the primary structures and physiological activities [18]: that is, neutral [Asp⁴⁹]PLA₂, named PLA2 (pI 7.9, highly lipolytic and myolytic) [19,20]; basic [Asp⁴⁹]PLA₂, named PLA-B (pI 8.6, edema-inducing) [21]; highly basic [Asp⁴⁹]PLA₂, named PLA-N (pI 10.3, neurotoxic) [22]; and two [Lys⁴⁹]PLA₂s, named BPI and BPII (pIs 10.2 and 10.3, both weakly lipolytic but strongly myolytic) [19,23,24].

On the other hand, P. flavoviridis serum is known to contain inhibitory proteins to neutralize their venomous activities. Such serum proteins are thought to act defensively on occasions of the accidental bites by fellow snakes. PLA₂ inhibitors (PLIs) which suppress snake venomous PLA₂ activity [25,26], Habu serum factor (HSF) which inhibits the hemorrhagic induced by the SVMPs [27,28], and small serum proteins (SSPs) [29] are such well-known proteins. SSPs are the low molecular weight (∼10 kDa) serum proteins and the five homologs, named PfSSP-1, -2, -3, -4, and PfSSP-5, have been found from P. flavoviridis sera [30]. PfSSPs form complexes with HSF in P. flavoviridis serum [8] and have various counterparts of P. flavoviridis venom proteins; PfSSP-2 and PfSSP-5 show high affinity to triflin [8], PfSSP-1 and PfSSP-4 to HV1 [31], and PfSSP-3 to Flavorase, non-hemorrhagic SVMP [29]. Recently, we found that the homolog of PfSSP-2 binds to BPII [32].

The complementary DNA (cDNA)s encoding five PfSSPs, were isolated and the nucleotide sequences of the cDNAs were also determined [8]. It should be noted that the cDNAs encoding PfSSP-3 or PfSSP-4 are interrupted by nonsense mutations at the same site of the fourth exon and express the truncated mature protein. The genome fragment, in which the genes encoding PfSSP-1 and PfSSP-2, designated as PfSSP-1 and PfSSP-2, respectively, were arranged in tandem, was also cloned. The nucleotide sequence of the genome fragment from PfSSP-1 to PfSSP-2 including the intergenic region (I-Reg) between PfSSP-1 and PfSSP-2, named as PfI-Reg12 (in the present paper), was determined [33]. Mathematical analysis of the nucleotide sequences of the two genes showed that they have evolved in an accelerated manner to acquire the different amino acid sequences [8,33].

In the present study, 30534 bp of P. flavoviridis genome segment containing five PfSSPs; PfSSP-4, PfSSP-5, PfSSP-1, PfSSP-2, and PfSSP-3 in this order, was deciphered. The particular nucleotide sequences of the long interspersed nuclear elements (LINEs), the DNA transposons, and the repetitive sequences were identified in the introns of PfSSPs and the I-Regs of the array of PfSSPs. The comparison analysis of the configuration of the fragments of chicken repeat-1 (CR1) LINE between P. flavoviridis and Ophiophagus hannah (Oh) (Elapidae snake) [34] showed that the chromosome inversion of the genome segment encompassing from PfSSP-1 to PfSSP-2 occurred and the inactive form of another PfSSP-1, PfSSP-1δ(Ψ), was formed at that site. Moreover, according to the types of the nucleotide sequences inserted into the intron of the gene, five PfSSPs were divided into two subgroups. The Long SSP subgroup, which consisted of PfSSP-1, PfSSP-2, and PfSSP-5, contained the large fragment of L2 LINE in the third intron of the gene. The Short SSP subgroup, which consisted of PfSSP-3 and PfSSP-4, contained the particular repetitive sequences in the second intron of the gene and no L2 LINE in the third intron. The mathematical analysis of the nucleotide sequences of the genes also showed that PfSSPs of the Short SSP subgroup evolved in an accelerated manner, whereas those of the Long SSP subgroup evolved alternately in an accelerated and neutral manner. The ortholog analysis of various snake SSPs suggested that the evolutionary emerging order of PfSSPs was reflected in the order of their configuration on the chromosome. In addition to the unique interpretation about accelerated evolution, a novel idea was proposed that the transposable elements such as LINEs and DNA transposons are involved in maintaining the host genome besides their transposition natures.

P. flavoviridis specimen of Amami-Oshima island was provided from the Institute of Medical Sciences of the University of Tokyo. High molecular weight genomic DNAs were prepared from the liver of the snake according to the method of Blin and Stafford [35]. Restriction endonucleases and KOD plus DNA polymerase were purchased from Nippon Gene (Tokyo, Japan) and TOYOBO (Osaka, Japan), respectively. The other reagents and antibiotics were from Nacalai Tesque (Kyoto, Japan) and Takara Bio (Shiga, Japan). Specific oligonucleotide primers were synthesized by GENNET (Fukuoka, Japan) (Table 1).

From the BLAST analysis against HabAm1 (Habu Amami version 1) [36], we found that PfSSP-5, PfSSP-1, PfSSP-2, and PfSSP-3 are harbored in this order on the scaffold 2858. Furthermore, PfSSP-4, which had not been identified via gene prediction so far, was also found in the 5′ upstream region of this arrangement of four PfSSPs on the scaffold 2858. In order to acquire the complete nucleotide sequence of this genome segment, genomic polymerase chain reactions (PCRs) against P. flavoviridis genome were performed with the specific sense and antisense primers referring to the nucleotide sequences of transcripts and genes of PfSSP or HabAm1 (Table 1).

The sense primer, JUS1, 5′-ATT CCT CCC TAC CAA gAg TCT-3′, which can anneal specifically to the first exon of PfSSP-5, and the antisense primer, JUS2, 5′-TCT ATg TgA Agg gAT gAg AAT C-3′, which can anneal specifically to the fourth exon of PfSSP-5, referring to the nucleotide sequence of the cDNA encoding PfSSP-5 (AB360910), amplified the 4065-bp genome fragment, named Pfjb-I. The Pfjb-I fragment was cloned into pCR®-Blunt II-TOPO® vector (Life Technologies, Carlsbad, CA, U.S.A.) and transformed with DH5α competent cells (Takara Bio) and sequenced. The nucleotide sequences were determined with an ABI 3130xl capillary sequencer. The Pfjb-I was found to encompass from the first exon to the fourth exon of PfSSP-5. In the present study, PfSSP-5 contained extra Val at position 89 encoded by ²⁶⁵GTG and Asp at position 91 encoded by ²⁷¹GAT was also substituted to Asn encoded by ²⁷¹AAT.

In order to acquire the nucleotide sequence of the I-Reg between PfSSP-5 and PfSSP-1, named as PfI-Reg51, genomic PCR was carried out against P. flavoviridis genome with the sense primer, JUS5, 5′-CAT gCC AAC ATg AAT CCT ATA gg 3′, which can anneal to the fourth exon of PfSSP-5, and the antisense primer, JUS8, 5′-ACC CAC Tgg AAT AAA TTT CTC AT-3′, which can anneal to the fourth exon of PfSSP-1, referring to the nucleotide sequence of SSP-1 gene (AB769881), amplified the 2513 bp genome fragment, named Pfjb-II. The Pfjb-II fragment was also cloned and sequenced. The Pfjb-II was found to encompass from the fourth exon of PfSSP-5 to the fourth exon of PfSSP-1 including PfI-Reg51. The 4065 bp Pfjb-I overlapped 77 bp with the 2513 bp Pfjb-II. The physical structure of 6501 bp segment encompassing from the fourth exon of PfSSP-5 to the first exon of PfSSP-1 was deciphered.

As the nucleotide sequences of the first exons of PfSSP-4 and PfSSP-3 are completely identical, it is hard to design the specific primer at the first exon which can amplify each gene differentially. So, the sense primer, TOY6, 5′-ggC TgC ACA TCT ggC TgT TTC AA-3′, which can anneal specifically to 220 bp 5′ upstream of the first exon of PfSSP-4, and the antisense primer, TOY7, 5′-TTC CTC CTg gCA gTg TTA gAC C-3′, which can anneal specifically to the second intron of PfSSP-4, avoiding the nucleotide sequence of the open reading frame (ORF) of PfSSP-4 (AB360909), amplified the 1421 bp genome fragment, named Pfjb-IV. The Pfjb-IV fragment was cloned and sequenced. The Pfjb-IV was found to encompass from 220 bp 5′ upstream of the first exon to the second intron of PfSSP-4.

Then, the sense primer, TOY1, 5′-ggA gTA TTC CTT TAC CTg AAA Tgg-3′, which can anneal specifically to the second exon of PfSSP-4, and the antisense primer, TOY2, 5′-ggT Agg gAg gAA TTA CCg ggA g-3′, which can anneal specifically to the first exon of PfSSP-5, referring to the nucleotide sequence of the cDNA encoding PfSSP-5 (AB360910), amplified the 5888 bp genome fragment, named Pfib-III. The Pfjb-III fragment was cloned and sequenced. The Pfjb-III was found to encompass from the second exon of PfSSP-4 to the first exon of PfSSP-5 including the I-Reg between PfSSP-4 and PfSSP-5, named as PfI-Reg45.

The 5888 bp Pfjb-III overlapped 104 bp with the 1421 bp Pfjb-IV. The physical structure of 7205 bp segment encompassing from 220 bp 5′ upstream of the first exon of PfSSP-4 to the first exon of PfSSP-5 was deciphered. The nucleotide sequence of the genome segment containing from PfSSP-1 to PfSSP-2 has been already reported by Tanaka et al. ([33], AB769881). Moreover, the 5888 bp Pfjb-III overlapped 13 bp with the 4065 bp Pfjb-I. The physical structure of 23202 bp segment encompassing from 220 bp 5′ upstream of the first exon of PfSSP-4 to 5′ terminal of the first exon of PfSSP-2, including the nucleotide sequence of the genome segment encompassing from PfSSP-1 to PfSSP-2, was deciphered.

The sense primer, prs-2, 5′-TTg TCA TTC TCT gAg AAg Tgg-3′, which can anneal specifically to 571 bp 3′ downstream of the fourth exon of PfSSP-3, and the antisense primer, prs-1, 5′-gAg TgT TCC TCT ACC TAT Ag-3′, which can anneal specifically to the second exon of PfSSP-3, referring to the nucleotide sequence of the cDNA encoding PfSSP-3 (AB360908), amplified the 2967 bp genome fragment, named Pfjb-V. The Pfjb-V fragment was cloned and sequenced. The Pfjb-V was found to encompass from 571 bp 3′ downstream of the fourth exon to the second exon of PfSSP-3.

Then, the sense primer, prs-6, 5′-AAg AgC AgC ACC TCT CTg TgA Ag-3′, which can anneal specifically to the first exon of SSP-2, and the antisense primer, prs-5, 5′-CgC TTg CAC TgA AgA TgC AAT gg-3′, which can anneal specifically to 378 bp 3′ downstream of the fourth exon of PfSSP-3, avoiding the nucleotide sequence of the ORF of PfSSP-3 (AB360908), amplified the 3470 bp genome fragment, named Pfjb-VI. The Pfjb-VI fragment was cloned and sequenced. The Pfjb-VI was found to encompass from 401 bp 5′ upstream of the fourth exon of PfSSP-3 to the first exon of PfSSP-2. The 2967 bp Pfjb-V overlapped 193 bp with the 3470 bp Pfjb-VI.

The sense primer, prs-14, 5′-TTC TCC Tgg CgT TAT TAg ACA-3′, which can anneal specifically to the second intron of PfSSP-3, and the antisense primer, prs-13, 5′-TTC CTT CTg gCA gTg gAT TC-3′, which can anneal specifically to 59 bp (the present study) 5′ upstream of the first exon referring to the nucleotide sequence of PfSSP-3 (AB360908), amplified the 1269 bp genome fragment, named Pfjb-VII. The Pfjb-VII fragment was cloned and sequenced. The Pfjb-VII was found to encompass from the second intron to 59 bp 5′ upstream of the first exon of PfSSP-3. The 1269 bp Pfjb-VII overlapped 93 bp with the 2967 bp Pfjb-V. The physical structure of 7420 bp segment encompassing from the first exon of PfSSP-2 to 59 bp 5′ upstream of the first exon of PfSSP-3 was deciphered. The 3470 bp Pfjb-VI overlapped 37 bp with the first exon of PfSSP-2 (AB769881). Finally, the physical structure of 30534 bp segment encompassing from 220 bp 5′ upstream of the first exon of PfSSP-4 to 59 bp 5′ upstream of the first exon of PfSSP-3 was completely deciphered.

The nucleotide sequence and the detailed annotations of the genome domain composed of Pfjb-IV, Pfjb-III, Pfjb-I, Pfjb-II, Pfjb-VI, Pfjb-V and Pfjb-VII, are available in the Genbank/EMBL/DDBJ databases under accession number MK574076.

The draft nucleotide sequences of O. hannah [34,37], Python bivittatus (Pythonidae snake) [38,39], P. mucrosquamatus (Viperidae snake) [40,41], and Thamnophis sirtalis (Colubridae, Naticinae snake) [42,43], were downloaded to make personal genome databases. Referring to the nucleotide sequences and the deduced amino acid sequences of PfSSPs via tblastn or blastn, the nucleotide sequences encoding the orthologs of PfSSPs and the flanking regions of them in each snake genome data were deciphered.

The personal database was constructed with the repetitive sequences of the genomes of various organisms collected from the Repbase of the Genetic Information Research Institute [44]. RepeatMasker was carried out the nucleotide sequences of the genome segments containing SSPs of O. hannah, P. bivittatus, P. flavoviridis, P. mucrosquamatus, and T. sirtalis, against the database via BLAST+, RMBlast (NCBI), and Tandem Repeats Finder (Boston University) [45].

Alignment of the amino acid sequences of PfSSPs was performed using ClustalX software. The nucleotide sequences of ORFs encoding the mature proteins of PfSSPs were rearranged, removing the gaps, by PAL2NAL according to the aligned amino acid sequences. The rates of synonymous (K_S) and nonsynonymous (K_A) substitutions per site between the ORFs of the genes were calculated using Nei-Gojobori method as implemented with PAML [46]. The rates of substitutions of the introns (K_N) were calculated from the aligned sequence data.

The nucleotide sequence of 30534 bp of P. flavoviridis genome segment was deciphered as described above and found to contain the array of five PfSSPs, that is, PfSSP-4, PfSSP-5, PfSSP-1, PfSSP-2, and PfSSP-3 in this order (Figure 1A). The precise construction of each of the five PfSSPs including the promoter and four exons was revealed (MK574076) 3733 bp of PfSSP-4, 4198 bp of PfSSP-5, 2796 bp of PfSSP-1, 3619 bp of PfSSP-2, and 3513 bp of PfSSP-3.

The draft genome data of O. hannah (Elapidae snake) was also investigated to decipher the nucleotide sequences of the genome segment harboring the orthologs of PfSSPs (see the details in the ‘Determination of the nucleotide sequences and the chromosomal configurations of 221 the genes encoding the orthologs of PfSSPs of various snakes’ subsection of ‘Experimental’ section). Then, four nucleotide sequences of the genes which encode SSP-1, SSP-2, SSP-4, and SSP-5, were identified. As the draft genome data contained many unidentified nucleotides which were described as ‘N’, only the nucleotide sequences of four exons of the gene encoding SSP-4 were identified. The nucleotide sequences of the third exons of the genes encoding SSP-1 and SSP-2 were partly identified, and that of the fourth exon of the gene encoding SSP-5 could not be identified (Figure 1B). The nucleotide sequences encoding SSP-1 and SSP-2 were named as OhSSP-1 and OhSSP-2, respectively. As the nucleotide sequences of the third exons of both genes encoding SSP-4 and SSP-5 contained the insertion of 98 and 8 nucleotides to cause nonsense mutation, they were named as OhSSP-5(Ψ) and OhSSP-4(Ψ), respectively. It should be noted that the directions of the transcription of OhSSPs were all the same, in contrast with those of PfSSP-4 and PfSSP-5 were opposite to those of the other three PfSSPs, PfSSP-1, PfSSP-2, and PfSSP-3. It is likely that the genome fragment harboring PfSSP-1, PfSSP-2, and PfSSP-3, has been inverted.

Detailed analysis showed that the I-Reg between PfSSP-5 and PfSSP-1, PfI-Reg51, was an interesting structure. First, the nucleotide sequence which encodes another fragmented PfSSP-1 was found in the middle portion of PfI-Reg51. The nucleotide sequence, named PfSSP-1δ(Ψ), consisted of 48 bp of the 3′ portion of the third intron and the fourth exon with five nucleotides of the 3′ untranslated region (UTR) of PfSSP-1 (Figure 1C). Second, the Repeatmasker revealed that the fragments of three types of LINE, L2, R4, and Gypsy, were inserted so as to sandwich the PfSSP-1δ(Ψ) (Figure 1D). The two fragments of L2 and R4 LINEs were located in the 3′ downstream of PfSSP-1δ(Ψ) and that of Gypsy LINE was located in the 5′ upstream of PfSSP-1δ(Ψ). Each of the three fragments encoded most of the reverse transcriptase (RT) domain of each LINE.

Two DNA transposon fragments hAT [47], Mariner [48], and another Gypsy LINE fragment were inserted in the region between PfSSP-5 and the L2 fragment-R4 fragment-PfSSP-1δ(Ψ) arrangement. Both DNA transposons are known to carry out gene conversion via double-strand break [49,50]. In addition, the 30 bp nucleotide sequence predicted to form the stem-loop structure (Figure 1E) was found immediately next to the L2 fragment-R4 fragment-PfSSP-1δ(Ψ) arrangement. The stem-loop structure is also known to be the scaffolding of the gene conversion [51,52]. PfSSP-1δ(Ψ) should be the remnant of the amplified PfSSP-1 which was destroyed by the plural times of insertions of LINEs and DNA transposons after being amplified into PfI-Reg51.

Further investigation of the nucleotide sequences of the arrays of SSPs of P. flavoviridis and O. hannah showed that there were two pairs of the particular nucleotide sequences. One pair was 140 nucleotides in the 3′ downstream of PfSSP-1 and 140 nucleotides in the 3′ downstream of OhSSP-1, the other pair was 937 nucleotides in the 5′ upstream of PfSSP-2 and 961 nucleotides in the 5′ upstream of OhSSP-2 (Figure 2A,B). The nucleotide sequence of the former pair was designated as ‘α’ and that of the latter pair as ‘β’. The identity between the nucleotide sequences of α or β pairs was 69 or 65%, respectively, but the direction of the nucleotide sequences of the α or β pairs was opposite. These findings showed that the P. flavoviridis genome segment encompassing from the α sequence to the β sequence had been inverted, moreover the tandem arrangement of PfSSP-1 and PfSSP-2 had already been formed before the inversion occurred.

Interestingly, RepeatMasker analysis also showed the five fragments of CR1 LINE, which is the most major LINE contained in the reptilian genome, were found in all the PfI-Regs except for PfI-Reg51. They are CR1_45, the CR1 fragment inserted in the PfI-Reg45, and CR1_12_i and CR1_12_ii, or CR1_23_i and CR1_23_ii, the CR1 fragments inserted in this order at the middle portion of PfI-Reg12 or PfI-Reg23, respectively (Figure 2B). CR1 is composed of two ORFs, ORF1 and ORF2 [53]. ORF1 encodes RNA binding protein and ORF2 encodes two-domain protein which is composed of endonuclease (EN) and RT domains. The RT domain of CR1 consists of ten subdomains from 0 to IX and a carboxy-terminal conserved region (CTCR), which is known to be the scaffold of reverse-transcription of CR1 LINE [54–56]. The fragments CR1_45 and CR1_12_ii contained four subdomains from III to VI and from IV to VII of the RT domain, respectively. On the other hand, each of the fragments CR1_12_i, CR1_23_i, and CR1_23_ii contained only the CTCR of the RT domain. Interestingly, the direction of the transcription of the fragments CR1_12_i, CR1_12_ii, and CR1_23_i, was opposite to that of the transcription of the fragments CR_45 and CR_23_ii (Figure 2B). Namely, these findings further showed that the inversion of the genome segment encompassing from the α sequence with CR1_23_i to the β sequence at the 3′ terminal of PfSSP-1 had occurred. If the inversion had not occurred, the direction of the transcription of all five CR1 fragments would be the same and the five CR1 fragments should have been located 3′ downstream of all PfSSPs except for PfSSP-3 (Figure 2C). Ikeda et al. (2010) [57] also found that the genes encoding P. flavoviridis venom PLA₂ isozymes are linked to the fragments of CR1 LINE, named PLA2 gene-coupled RT fragment (PcRTF), in their 3′ downstream. Thus, CR1 LINE seems to be involved into the amplified genes in P. flavoviridis genome.

On the other hand, the types of the nucleotide sequences inserted in the second or third intron of the gene classified five PfSSPs into two subgroups. One subgroup consisted of three PfSSPs, PfSSP-1, PfSSP-2 and PfSSP-5, was characterized by the nucleotide sequence of the fragment inserted into the third intron of the gene, which encoded the RT domain of L2 LINE [56,58]. As the three SSPs belonging to this subgroup encoded the full-length proteins [8], this subgroup was designated as the Long SSP subgroup. Interestingly, the inserted fragments were truncated according to the order of the name of each gene. The lengths and the constructions of the three inserted fragments were as follows. The fragment inserted into PfSSP-5 was 1011 bp, which encoded nine subdomains from 0 to VIII of RT domain. The fragment inserted into PfSSP-2 was 431 bp, which encoded four subdomains from 0 to III of RT domain. The fragment inserted into PfSSP-1 was 320 bp, which encoded three subdomains from 0 to II of RT domain. Though LINEs are known to be generally truncated from the 5′ terminal region and become inactive, the fragments of L2 LINE inserted into PfSSP-1, PfSSP-2, and PfSSP-5 truncated from the 3′ terminal. The first PfSSP in this subgroup should have been PfSSP-5 with the inserted L2 LINE fragment. As the amplifications occurred from PfSSP-5 to PfSSP-2 and then from PfSSP-2 to PfSSP-1, the inserted L2 fragment became truncated every time at each. The nucleotide sequence between L2 fragment and Mariner-ii in the third intron of PfSSP-2 and PfSSP-5 was considered to be an irrelevant nucleotide sequence brought in from the genome site where L2 LINE had been retrotransposed just before. In PfSSP-1, this ‘orphan’ nucleotide sequence is thought to have disappeared accompanying the transposition of Mariner-ii.

The other subgroup consisted of PfSSP-3 and PfSSP-4 was characterized by three nucleotide sequences of the fragments of DNA transposons inserted into the second and third introns of the gene. One was the fragment of Mariner, named as Mariner-i, inserted into the same site of the second intron of the gene. The other two were the fragments of hAT and another Mariner, named as Mariner-iii. Two juxtaposed fragments were inserted into the same site between Mariner-ii and Gypsy-i in the third intron of the gene. In addition, the particular repetitive nucleotide sequences were also found at the same site of the second intron of the gene (see the details in the next section). As PfSSP-3 and PfSSP-4 encoded the truncated proteins [8], the subgroup was designated as the Short SSP subgroup. The positions and the nucleotide sequences of the eight inserted fragments, L1 and CR1 fragments in the first intron, Mariner-i in the second intron, Mariner-ii, hAT, Mariner-iii, Gypsy-i and Gypsy-ii in the third intron were almost the same (Table 2). Namely, the insertion of them should have occurred before the branching of PfSSP-3 and PfSSP-4 and not much time must have passed since two genes were branched.

The evolutionary process of the Long SSP subgroup was not plain. The mathematical analysis showed that the branching between PfSSP-1 and PfSSP-2 of the Long SSP subgroup had occurred in an accelerated manner (Table 3) [8,33]. In addition, the fact that the rate of K_N between the introns of PfSSP-1 and PfSSP-2 was 0.0649 also suggested that the time passed after branching of PfSSP-1 and PfSSP-2 was very short (Table 4). On the other hand, the rate of K_A/K_S between the ORFs of PfSSP-1 and PfSSP-5 or PfSSP-2 and PfSSP-5, which is the relative ratio of synonymous substitution rate to nonsynonymous substitution rate, was 0.625 or 0.646 (Table 3) and the rate of K_N between the introns of PfSSP-1 and PfSSP-5 or PfSSP-2 and PfSSP-5 was 0.328 or 0.312, respectively (Table 4). These results suggested that PfSSP-1 or PfSSP-2 and PfSSP-5 had been diverged in a ‘neutral manner’ long time ago and that PfSSP-1 and PfSSP-2 branched in a very short time long time later.

As concerning about the Short SSP subgroup, the nucleotide sequence between PfSSP-3 and PfSSP-4 including the fragments of LINEs and DNA transposons was almost identical except for the number of repetition of the nucleotide sequences in 2nd intron (Figure 3). The repetitive sequences were of two types. One was the repetition of five nucleotides of TAAAA, which was repeated 32 times for PfSSP-3 and 36 times for PfSSP-4. The other was that of five nucleotides of AATAA immediately next to the repeat of TAAAA, which was repeated 42 times only for PfSSP-4. Without consideration of the repeats, the rates of K_A/K_S and K_N between PfSSP-3 and PfSSP-4 were 1.54 (Table 3) and 0.0488 (Table 4), respectively. These results showed that PfSSP-3 and PfSSP-4 has been branched very recently in an accelerated manner.

Detailed tblastx analysis against the draft genome databases of four snakes P. bivittatus, T. sirtalis, O. hannah, P. mucrosquamatus, in addition to that of P. flavoviridis has revealed the orthologs for PfSSPs and their chromosomal configurations as follows (Figure 4). The genome of non-venomous Pythonidae snake, P. bivittatus, contained three orthologs of PfSSP-5, named PbSSP-5α, PbSSP-5β, and PbSSP-5g(Ψ), that of Colubridae snake, T. sirtalis, contained the ortholog of PfSSP-4, named TsSSP-4, and two orthologs of PfSSP-5, named TsSSP-5a, and TsSSP-5β, that of neurotoxic Elapidae snake, O. hannah, contained the orthologs of PfSSP-4, PfSSP-5, PfSSP-1, and PfSSP-2 on one chromosome in this order, named OhSSP-4(Ψ), OhSSP-5(Ψ), OhSSP-2, and OhSSP-1, respectively (see the details in the first section of this chapter), that of Viperidae Taiwan Habu snake, P. mucrosquamatus, contained the orthologs of PfSSP-5, PfSSP-1, PfSSP-2 and PfSSP-3 on one chromosome in this order, named PmSSP-5, PmSSP-1, PmSSP-2, and PmSSP-3, respectively (in the present study). In addition, the ortholog of PfSSP-4, named PmSSP-4, was also found in P. mucrosquamatus in another scaffold.

In the present study, the names of four orthologs were renamed. As a result of our detailed investigation based on the deduced amino acid sequences, the nucleotide sequences which have been annotated as PbSSP-2 and TsSSP-2 in the original databases should be renamed as PbSSP-5β and TsSSP-5β. In addition, the nucleotide sequence newly found from the genome of P. bivittatus in the present study, which encoded the ortholog of PfSSP-5 but contained the deletions of 34 nucleotides and 7 nucleotides causing frameshifts at the second and third exons, respectively. It was named as PbSSP-5g(Ψ). Therefore, those already annotated as PbSSP-5 and TsSSP-5 in the original databases should be renamed as PbSSP-5α and TsSSP-5α, respectively. The relationship between PbSSP-5α, PbSSP-5β and PbSSP-5g(Ψ), or TsSSP-5α and TsSSP-5β was paralog. Furthermore, the nucleotide sequence, which was newly found from the genome of T. sirtalis in the present study, encoded the ortholog of PfSSP-4 then was named as TsSSP-4.

The configuration of the paralogs of SSPs in each snake genome (Figure 4) seemed to show their emerging order. Before the branching of Colubridea and Booidea snakes, SSP-5 had already existed in advance. After the branching, the genome of Colubridea snakes acquired SSP-4 derived from the paralog of SSP-5. On the other hand, the formation of the paralog of SSP-5 occurred twice in P. bivittatus genome or once in T. sirtalis genome. They became PbSSP-5β and PbSSP-5g(Ψ) or TsSSP-5β, respectively. In the genome of Elapidae and Viperidae snakes, the derivation of the paralogs of SSP-5 occurred twice at least and then they have become SSP-1 and SSP-2. The comparative analysis of the construction of the L2 LINE fragment in the third intron of the gene has already showed that SSP-5, SSP-2, and SSP-1 appeared in this order (see details in the fourth section of the present study). And then, in the genome of Viperidae, the inversion of the genome segment encompassing from SSP-2 to SSP-1 occurred (see details in the third section of the present study). Interestingly, the mathematical analysis between SSP-1 and SSP-2 showed that those of Elapidae snake have been evolved in a neutral manner in contrast with those of Viperidae snakes have been evolved in an accelerated manner. It is only speculation, the nucleotide substitutions dominant at the nonsynonymous sites only occurred immediately after duplication and then random mutations accumulated over time and the selective pressure to preserve the ‘neutral’ mutations at the synonymous sites have reduced the traces of the accelerated evolution. But the inversion of Viperidae genome segment encompassing from PfSSP-1 to PfSSP-2 might have avoided the accumulation of random mutations. The emerging process of the most newcomer, SSP-3, which is structurally highly related to SSP-4, is an issue to be addressed in the next study. It is also interesting that the positions and the nucleotide sequences of the fragments of LINEs and DNA transposons seem to be conserved rather than the nucleotide sequences of the introns and the I-Regs in which those transposable elements are inserted. It is likely that such transposable elements are involved in maintaining the construction of the host genome besides their transposition natures.

Takahito Chijiwa, Ami Takeuchi and Marie Maeda designed experiments. Takahito Chijiwa, Kento Inamaru, Ami Takeuchi, Marie Maeda, Kazuaki Yamaguchi, Hiroki Shibata, Shousaku Hattori, Naoko Oda-Ueda and Motonori Ohno prepared materials from live specimens. Ami Takeuchi and Marie Maeda performed sequencing and manual assembly of SSP gene array. Hiroki Shibata performed whole genome sequencing and provided HabAm1. Kento Inamaru, Ami Takeuchi, Marie Maeda and Kazuaki Yamaguchi performed bioinformatics analyses. Takahito Chijiwa supervised the project. Takahito Chijiwa and Kento Inamaru wrote the manuscript with contributions from all the other authors.

This work was supported in part by the Ministry of Education, Science, Sports and Culture, Grant-Aid for Science Research (C), 2014–2016 of Japan [grant number 26340095 (to Takahito Chijiwa)].

The authors declare that there are no competing interests associated with the manuscript.

cDNA

complementary DNA

CR1

chicken repeat-1

CTCR

carboxy-terminal conserved region

HabAm1

Habu Amami version 1

hAT

hobo-Ac-Tam3

HR1

hemorrhagic factor 1

HR2

hemorrhagic factor 2

I-Reg

intergenic region

LINE

long interspersed nuclear element

Oh

Ophiophagus hannah

ORF

open reading frame

PAML

Phylogenetic Analysis by Maximum Likelihood

PCR

polymerase chain reaction

Pf

Protobothrops flavoviridis

PLA₂

phospholipase A2

RT

reverse transcriptase

SSP

small serum protein

SVMP

snake venom metalloprotease

UTR

untranslated region