The complete DNA sequence and analysis of R27, a large IncHI plasmid from Salmonella typhi that is temperature sensitive for transfer

Abstract

Salmonella typhi, the causative agent of typhoid fever, annually infects 16 million people and kills 600 000 world wide. Plasmid-encoded multiple drug resistance in S.typhi is always encoded by plasmids of incompatibility group H (IncH). The complete DNA sequence of the large temperature-sensitive conjugative plasmid R27, the prototype for the IncHI1 family of plasmids, has been compiled and analyzed. This 180 kb plasmid contains 210 open reading frames (ORFs), of which 14 have been previously identified and 56 exhibit similarity to other plasmid and prokaryotic ORFs. A number of insertion elements were found, including the full Tn10 transposon, which carries tetracycline resistance genes. Two transfer regions, Tra1 and Tra2, are present, which are separated by a minimum of 64 kb. Homologs of the DNA-binding proteins TlpA and H-NS that act as temperature-regulated repressors in other systems have been located in R27. Sequence analysis of transfer and replication regions supports a mosaic-like structure for R27. The genes responsible for conjugation and plasmid maintenance have been identified and mechanisms responsible for thermosensitive transfer are discussed.

INTRODUCTION

Multiple antibiotic-resistant Salmonella typhi has contributed significantly to the persistence of typhoid fever. Plasmid-mediated resistance to chloramphenicol in S.typhi was first identified in the 1970s. Strains of S.typhi containing H incompatibility group (IncHI1) plasmids resulted in typhoid fever epidemics throughout Mexico, India, Vietnam and Thailand (1). Salmonella typhi has remained endemic in the Indian subcontinent and IncHI1 plasmids have contributed to the pathogen’s persistence by conferring resistance to multiple antibiotics (2). Harnett et al. (3) have recently reported that strains of S.typhi resistant to up to nine antibiotics were isolated from travellers returning to Ontario, Canada. Resistance is encoded by IncHI1 plasmids and highlights the importance of these plasmids in the emergence of multidrug-resistant S.typhi.

R27, the prototype IncHI1 plasmid, is a large (180 kb), self-transmissible plasmid which is capable of transfer between members of the Enterobacteriaceae and several Gram-negative organisms of environmental significance (4). Maintenance of R27 in bacterial cells is provided by one of two replicons, RepHI1A or RepHI1B (5,6), which are specific to IncHI plasmids. A third replicon is responsible for one-way incompatibility with the F factor and has been designated RepFIA (5). IncH1 plasmids are characterized by an unusual thermosensitive mode of transfer, in which transfer occurs optimally between 22 and 28°C but is inhibited at 37°C. The advantage conferred by thermosensitive transfer remains unknown; however, it suggests that IncH1 plasmids are potential vectors in the dissemination of antibiotic resistance among pathogenic and indigenous bacteria in water and soil environments (4).

Another distinguishing feature of these plasmids is the slow growth rate for the flexible H pili, which are required to facilitate initial contact with the recipient cell during conjugation. For example, 15 min is required for H pilus maturation, in contrast to 30 s needed for the F factor pilus (7). This slow rate of pilus biosynthesis in combination with binding of H pili-specific bacteriophage allowed the demonstration that the pilin subunits are added at the base of the pilus and not by extrusion through the pilus lumen to the tip (8). This study also showed that although H pili maintain their structure at 37°C, they are not produced and are non-functional at this temperature, suggesting a basis for the thermosensitive mode of transfer.

Restriction analysis and transposon mutagenesis has shown that two separate and distinct regions of R27 are responsible for conjugative transfer, Tra1 and Tra2 (9). Recently Tra2 has been sequenced and characterized (10). This region contains genes for pilus production and mating pair formation (10,11). The gene organization within Tra2 resembles the transfer region of the F plasmid that codes for those same functions (10). In the F factor, as well as all conjugative systems, these gene products are postulated to form a membrane structure that functions in translocating pilin subunits across the inner membrane and assembling them into a mature pilus on the cell surface. In addition, these transfer proteins are believed to function in pilus extension and retraction, formation of stable mating pairs and formation of a lumen in which plasmid DNA transfers between cells (12). The exact roles of these proteins, however, remain poorly understood in all conjugative systems.

Interrupting the Tra2 gene cluster is a region coding for plasmid partitioning and a regulator of conjugation, htdA (13). htdA mutants are derepressed for transfer, with a 1000-fold increase in mating frequency in 2 h conjugation experiments, and produce between one and four H pili per cell, as opposed to <10^–3 per cell for the wild-type. Our studies on R27 genetic organization and transfer highlight both the uniqueness and complexity of the transfer system of IncHI1 plasmids.

The aim of this project was to compile and analyze the complete nucleotide sequence of R27 with the goals of identifying transfer determinants and any genes involved in pathogenesis.

MATERIALS AND METHODS

Sequencing of R27

Plasmid DNA was isolated from Escherichia coli (J53-1) by alkaline lysis and purified by cesium chloride density gradient centrifugation. DNA was prepared for library construction by nebulization, end repair and size fractionation (14). Recovered DNA fragments were ligated into M13Janus (15). Library subclones were picked as plaques, from which template DNAs were prepared and then sequenced by Prism dye-terminator cycle sequencing chemistry and analyzed on ABI377 automated sequencers (16). The annotated sequence is deposited in GenBank, accession no. AF250878.

Annotation

Sequence assembly and identification of open reading frames (ORFs) was performed as described previously (17). Further identification of the ORF start sites was based on ribosomal binding site (RBS) matches upstream of potential starts. In the absence of strong RBSs prior to putative start sites, codon usage data was utilized to identify likely starts.

Searches of the protein databases for amino acid similarities were performed using the DeCypher II system (TimeLogic Inc.) to assign known functions or suggest functions for new ORFs (16). Further analysis of the ORFs was performed using BLAST sequence analysis tools (18) with subsequent comparison of ORFs showing significant homology (>10^–3 significance) performed using a Lipman–Pearson algorithm (19). Significant similarity was defined as at least 30% identity observed over 60% of the ORF, although those ORFs showing <30% identity over >60% of the protein were also included. ORFs are labeled ORF001–ORF210.

G+C content was calculated using the Window program from the GCG Software Suite v.9.1 (Genetics Computer Group Inc.) with the following parameters: window of 1000 nt with a 100 nt shift.

Prediction of transmembrane helices was completed using the TMPred Internet-based program (www.ch.embnet.org/software/TMPRED_form.html ).

Analysis of ORF158 and comparison to TlpA was done using the COILS2 program (http://www.ch.embnet.org/software/COILS_form.html ) (20) (28 residue window, MDITK sequence profile, weighting on) and PARCOIL (http://nightingale.lcs.mit.edu/cgi-bin/score ) (21).

RESULTS AND DISCUSSION

General properties of the plasmid sequence

The IncHI1 plasmid R27 contains 180 461 bp which correspond to 210 ORFs initiated by 186 putative ATG, four TTG and 20 GTG start codons. Table 1 lists the exact positions and sizes of the previously identified ORFs and those ORFs from this study that show similarity to protein sequences available in the public databases. For the latter, the percentage of amino acid identity between the R27 ORF and the prospective homolog is also included. Excluded from Table 1 are the 140 ORFs (66.7%) that do not demonstrate similarity to any known ORFs. A linear map of R27 indicating functional organization is shown in Figure 1 with all ORFs labeled. The putative ORFs account for 84.4% of the total plasmid, leaving 28 098 bp of intergenic regions. No ORFs directly related to pathogenesis were identified.

Table 1. Previously characterized ORFs (termed Tra, Trh and Htd) and R27 ORFs that have significant homology to other prokaryotic ORFs.

^aPredicted number of amino acids of R27 ORF.

^bPreviously identified R27 ORFs are in bold, while those R27 ORFs being analyzed for the first time have homologs listed as well as the source organism or genetic element of the homolog.

^cThe number of amino acids in a contiguous stretch from which the identity was calculated.

^dIf the ORF is previously uncharacterized then the accession number for the homologous ORF is listed.

Figure 1.

Linear representation of R27, showing all ORFs, their orientation and with color coding by functional groups. The upper rectangles are ORFs transcribed from right to left and the lower rectangles are transcribed left to right. Color coding of functional groups is as follows: yellow, Tn and IS elements; black, replication and stability; red, drug resistance; blue, citrate utilization; magenta, conjugal transfer. The GenBank entry of ORFs were assigned unique identifiers in the form Rxxxx. This figure was created using GeneScene (DNASTAR).

The calculated G+C content of R27 is 45.8%, compared with 50–53% for S.typhi (22), the usual host for this plasmid, and 50.8% for E.coli (16). Within R27, regions of notably different G+C content can be identified (Fig. 2). A peak located at 26 300 bp indicates the high G+C content corresponding to repeats involved in plasmid incompatibly (see below). A pair of troughs in the range 76 300–82 900 bp correspond to the transposon Tn10. The intergenic region between ORF077 and ORF078 and the 5′-end of ORF083 (tetC) are responsible for these two troughs in Figure 2 (29.9 and 31.4% G+C, respectively), possibly representing the composition of the original source of the transposon. The peak at 132 400–133 900 bp corresponds to the genes encoding citrate utilization proteins (citA and citB) and a gene whose product shows significant similarity to a hypothetical Rhodobacter capsulatum protein.

Figure 2.

G+C profile of the complete DNA sequence of R27. This figure was plotted using data generated from the GCG program Window.

Transposable elements

R27 contains five insertion sequence elements and one transposon, Tn10. In addition, ORF148 shows 47.1% identity to a putative transposase from the large Rhizobium sp. plasmid pNGR234 (23). Three copies of IS1 (designated a–c) are present. IS1a is located ~13 kb from the Tra1 region on the putative lagging strand. One of the coding regions within IS1a (insB) has accumulated several nonsense mutations. As this would result in premature termination of the protein, this copy of IS1 is probably inactive. IS1b and IS1c are found on either side of a DNA segment containing both ORF179 and the repeat sequences that flank ORF179, which display 94.8% similarity to the F plasmid replication protein E (24).

Two other intact elements, IS30 and IS2, share 100 and 99% identity, respectively, with previously reported counterparts. Both appear to encode full-length transposases and are predicted to be active. All the IS elements are oriented in the same direction in R27 with respect to their transposases, although the significance of this is unknown.

The tetracycline resistance determinant of R27 is encoded within the full Tn10 transposon by the tetACD genes and is controlled by the Tet repressor (ORF082–ORF084 and ORF081, respectively) (25). Both the tetracycline determinants in Tn10 and the transposases (ORF076 and ORF085) are well conserved with respect to their counterparts in E.coli, although transposase activity has not been demonstrated within R27.

Replication regions

R27 contains two IncHI1-specific replicons and a partial IncFIA replicon (6,26), with copy number control dependent on an iteron-binding mechanism, in which the replication initiation protein binds to repeated oligonucleotide sequences in the origin. The two IncHI1-specific replication determinants are both able to support active replication of the plasmid and both operate independently of one another (27). Each replicon independently specifies incompatibility (27), where incompatibility is defined as the inability of two related plasmids to replicate in the same host cell. The one-way incompatibility with IncF plasmids is due to ORF179, the R27 replication initiator protein RepE. R27 contains multiple replicons which could give it a competitive advantage over plasmids with only a single replicon and thus increase its host range within the Enterobacteriaceae.

The specific IncHI replicons, RepHI1A and RepHI1B, are similar in their overall organization, but the amino acid sequences of their replication initiation proteins are only 34.9% similar (27). The two RepHI proteins are comparable in size (292 and 294 amino acids, respectively) and organization of flanking replication protein binding sites: four direct repeat sequences upstream and three downstream for RepHIA and four upstream and four downstream for RepHIB. The repeats surrounding the gene for RepHIA have been shown to act as iterons for the RepHIA protein (26). The same mechanism is presumed to occur between the RepHIB protein and eight surrounding repeats. The independence of the two replicons in incompatibility is believed to be due to differences in the repeats. The repeat sequences of RepHIA and RepHIB are 5′-AAAAGCTTTGnATGAATG-3′ and 5′-ATCCACTATACCGGGTA-3′, respectively. Both replicons contain recognition sequences for DNA adenine methylases (GATC) (28) upstream of the Rep genes and in both cases the recognition sequence is embedded in a 7 nt repeat (AGGATCAA).

Transfer regions 1 and 2

The Tra1 region of R27 encodes proteins required for DNA translocation across the membrane during conjugation as well as initial replication events after the plasmid has entered the recipient cell. This is an energy-dependent process in which R27 utilizes host DNA replication machinery and Tra1-encoded proteins. The limits of the Tra1 region were identified by transposon mutagenesis (26). A major element of the Tra1 region is the origin of transfer (oriT) from which transfer is initiated by nicking, unwinding and replication of the leading strand (29). A potential oriT is present between ORF122 and ORF123 (a 758 bp untranslated region). Within this region are a series of four 33 bp inverted repeats, with two potential transcription initiator sequences oriented in opposite directions near the repeats, a common pattern for oriT sequences (29).

Although there are no significant database homologies in the region surrounding oriT, the R27 Tra1 region has clear organizational similarities to both the IncP and IncW plasmid transfer regions. On the lagging strand (the last genes to enter in conjugation) of the IncP and IncW plasmids are genes encoding for a relaxase/nickase (traI and trwC, respectively) and a coupling protein (traG and trwB) (30). The relaxase/nickase protein introduces a site-specific single-stranded nick in oriT, with the coupling protein stabilizing the complex and possibly aiding in the recruitment of the cellular DNA replication machinery. Plasmid DNA transfer is thought to utilize the host DNA replication machinery to drive the single-stranded replication product across the membranes into the recipient cell, where it is circularized and the complementary strand synthesized by host-encoded proteins (30).

A pattern of motifs has been reported in the class of related anchoring/coupling proteins of which TrwB of R388, TraD of R100, VirD4 of Ti and TraG of RP4 are members (30,31). All of these proteins contain N-terminal transmembrane segments followed by one or more ATP-binding domains (30). BLAST analysis of ORF119 identified low similarity (BLAST score 0.78) to the TrwB protein of the IncW plasmid R388. The deduced amino acid sequence of ORF119 contains two predicted transmembrane segments spanning residues 22–40 and 46–64, with an ATP/GTP-binding motif ‘A’ spanning residues 205–212. The location of this ORF near oriT, its requirement for transfer as indicated by transposon mutagenesis (26), and the presence of the transmembrane segments followed by the NTP-binding domain lead us to propose that ORF119 is an anchoring/coupling protein and to name it TraG following the IncP nomenclature.

Immediately following TraG in the IncP system is TraI, the protein responsible for nicking and cleavage of the double-stranded DNA at oriT (31). A number of studies have identified a catalytic tyrosine residue in the nicking enzymes found in different families of plasmids (30,31). Llosa et al. have identified three regions present in all the proteins mediating nicking and initiation of DNA transfer: (I) an active tyrosine residue, (II) the residue sequence H(X_n)S, followed by (III) H(X₁₀)HXHXXXXN (30). In ORF120 we have found regions in the N-terminus corresponding to each of the three essential motifs for nicking proteins (Fig. 3). We have named the gene for ORF120 traI based on the three motifs identified in nicking enzymes and the necessity of this ORF for transfer indicated by transposon mutagenesis (26). TraI is more closely related to the nicking enzymes of the IncP system since it does not contain an identifiable C-terminal helicase-specific motif (31) seen in IncW TrwC and IncF TraI.

Figure 3.

Alignment of TraI (ORF120) with nickase/relaxase proteins from other conjugative plasmids (IncW R388, IncP RP4, IncI R64, Ti and IncHI1 R27). Amino acid residues that follow the guidelines of the nickase/relaxase distinguishing motifs, as set out in the discussion, are highlighted.

ORF126 shows similarity (24.1% over 212 amino acids) to a Sphingomonas aromaticivorans pilus synthesis and assembly protein and is necessary for transfer (26). ORF127 has 22.9% identity over the entire length compared to TraH of IncF. traG and traH of the F plasmid are thought to encode proteins involved in pilus assembly and aggregate stabilization in the F system (31). The similarity between ORF127 and IncF TraH and the necessity of ORF127 for transfer, based on transposon mutagenesis (26), led us to designate it TrhH. This ORF was not called Tra because this designation is reserved for the DNA processing proteins of Tra1, whereas all proteins responsible for mating pair formation and pilus assembly will be named Trh. Adjacent to trhH is ORF128, which was also identified as necessary for transfer by transposon mutagenesis (26). This ORF encodes a protein of 1329 amino acids predicted to have eight strong transmembrane helices. These characteristics are similar to that of TraG of the IncF plasmids, encoded immediately downstream of traH (12). We did not expect to encounter pilus assembly proteins (ORF126, TrhH and ORF128) within the DNA processing region of R27.

Tra2, previously characterized by Rooker et al. (10), contains the genes responsible for mating pair formation and pilus biosynthesis. Nomenclature of the Tra2 region was established previously (10), with gene names following the F plasmid transfer system to which the R27 Tra2 region shows the greatest similarity with regard to gene organization. Several ORFs within the Tra2 region show extremely high similarity to the H transfer determinants (Htd) identified in the Serratia marcescens IncHI2 plasmid R478 (32). The organization of four R27 genes (trhB, htdT, trhV and trhC) is similar to that of the R478 htd homolog. Though Page et al. were unable to identify two separate transfer regions in R478 via transposon mutagenesis, we find that the Tra2-equivalent regions of the two IncHI plasmids are related (32).

The Tra2 region is interrupted by partitioning and stability genes, leaving the trhWUN transfer genes separately located from the rest of the Tra2 genes that encode transfer proteins. In other prototypical transfer systems the trhWUN homologs are contiguous with the other transfer-related genes. The 3127 bp segment interrupting Tra2 is found between the trhC and parA genes. This region has also been shown to be responsible for IncH incompatibility (5). A DNA segment containing five 31 bp direct repeats (27541–27786) is sufficient for incompatibility (5). Another twenty-one 50 nt repeats (25597–26746), each containing an identical 31 bp direct repeat, is located downstream. When subcloned, the five 31 bp repeats are sufficient for some level of incompatibility (5). These repeats, along with the three adjacent partitioning system homologs (ORF014, ORF019 and ORF020), constitute a partitioning module that ensures each daughter cell receives R27, which has a very low copy number, during division of the host cell. ORF014 is very similar to the StbA protein of plasmid R478 (Table 1) and also StbA encoded on the enteropathogenic E.coli EAF plasmid (38% identity over 324 amino acids) (33). ORF019 and ORF020 are similar to the ParA and ParB partitioning proteins of the E.coli prophage/plasmid P1. During P1 partitioning ParB interacts with direct repeats (parS) forming a nucleoprotein structure (34,35), which is functionally equivalent to a plasmid centromere (35). The role of ParA during partitioning of P1 is less clear, but it is known to regulate its own expression (36). Interruption of the Tra2 region by the partitioning region indicates that R27 could have undergone an intramolecular rearrangement during its evolution.

The IncP plasmid RP4 has been found to encode a sequence-specific peptidase, TraF, that cleaves the pre-pilin subunit TrbC at the C-terminus (37). This activity is required for pilin processing. The peptidase activity is attributed to the catalytic domains I and II, whereas domain III is believed to play a role in the conformation of the mature protein. Comparison of ORF008 and the traF gene product from RP4 shows conservation of catalytic domains II and III, along with an N-terminal transmembrane segment, potentially anchoring the protein to the membrane (Fig. 4). In particular, the catalytic lysine (Lys89 in TraF and Lys73 in ORF008) of domain II is conserved. These similarities have prompted us to term ORF008 TrhF. Nevertheless, in domain I the catalytic serine residue (Ser37) found in peptidases of the IncP and Ti systems is replaced by a tyrosine (Tyr17) in TrhF. The predicted transmembrane helical domain of TrhF spans residues 17–35 (data not shown), effectively burying the domain I-equivalent region of TrhF in the cytoplasmic face of the inner membrane. This would itself prevent catalytic function, if any indeed remained after the serine substitution to tyrosine. If domains II and III of TrhF alone are sufficient for activity is not yet clear. Haase and Lanka found that when the catalytic Ser37 of RP4 domain I is replaced by an alanine, this cloned traF construct is able to partially complement a traF mutant (37). Like the homologs in the Agrobacterium tumefaciens Ti conjugative transfer region, the TraF protein in the IncP system is required for C-terminal processing of the pilin subunits leading to synthesis of the mature pili, thus the R27 homolog may also have an important role in pilin maturation.

Figure 4.

PILEUP comparison of the TraF homologs identifying a conserved motif between the proposed TrhF and other proteins related to the TraF protein of the IncPα plasmid RP4 (30,31). The illustrated sequences are from R27 (TraF_R27), the IncPα plasmid RP4 (TraF_PAlph), the IncPβ plasmid R751 (TraF_PBeta), the A.tumefaciens octopine-type Ti plasmid (TraF_OctTi), the A.tumefaciens nopaline-type Ti plasmid (TraF_NopTi) and ORF2 from the Vir2 region of the Ti plasmid (Orf2_TiVir). Catalytic residues described in the text are denoted above the alignment with an asterisk. Catalytic domains I, II and III as described by Haase and Lanka (37) are denoted above the alignment with a line.

The pilin subunit of R27 is most likely to be ORF034 (TrhA). This is encoded by the first gene in the Tra2 operon and has a deduced post-cleavage size similar to TraA, the F pilin subunit (7.6 kDa for TrhA compared to 7.2 kDa for TraA) (12). TrhA shows 19.4% identity over 93 of its 121 amino acids to TrbC, the RP4 pilin (38). Peptidase cleavage sites are predicted to be present in TrhA similar to those found in TrbC.

Temperature-sensitive conjugation

A process that resembles bacterial conjugation and is permissive at low temperature is the T-DNA transfer system of A.tumefaciens (38,39), where crown gall tumor formation in the recipient plant is induced at 19 but not above 29°C (40,41). Studies on T-DNA transfer show that thermosensitivity is a property of the T-DNA transfer proteins, in which elevated temperatures result in repression of T-DNA transfer gene expression and instability of at least one essential transfer protein (37,41,42). In addition, conjugal transfer of the Ti plasmid is also thermosensitive (43).

Transfer of IncHI1 plasmids, including R27, is also temperature sensitive and is inhibited at and above 37°C (44). Transfer of R27 is dependent on the temperature in which the donor cells have been grown prior to mating (44). There are several possible underlying mechanisms of thermosensitivity, including transfer gene mRNA instability, transfer protein or pilus instability, the loss of requisite protein–protein or protein–DNA interactions or differential binding of a repressor protein(s) through a range of temperatures leading to transfer gene repression.

Several observations support these hypotheses. First, Maher et al. found that donor cultures grown exclusively at 37°C are non-piliated, but when 27°C cultures are switched to 37°C for 1 h H pili are retained (8). This implies that there is either repression of transfer gene expression or transfer gene mRNA instability at 37°C, rather than instability of H pilus structure. This is further supported by the observation that when cultures grown at 37°C are switched to 27°C, pili are not observed until 3 h have elapsed (8). A 3 h regeneration period might be necessary to transcribe transfer genes, translate transfer mRNA into protein and assemble a functional transfer protein complex and pilus. Support for a second hypothesis is given by the fact that when mating cultures are shifted to 37°C for 30 min, aggregation of donor and recipient cells no longer occurs (8). Also, when a mating culture is briefly exposed to high temperature (i.e. a 1 min pulse at 37°C) there is a reduction in transfer frequency (26). This rapid response to temperature suggests inactivation of preformed transfer elements, possibly pili or other transfer proteins involved in cell–cell contact.

Sequence analysis of R27 has identified ORFs whose homologs are involved in thermorepressed expression in other systems. Homologs of the temperature-regulated DNA-binding proteins TlpA and H-NS have been discovered in the R27 sequence. The TlpA protein encoded on the Salmonella typhimurium virulence plasmid is a thermosensing repressor (45). The C-terminus of TlpA forms coiled-coils, while deletion of N-terminal residues 31–43 prevents DNA binding (46). ORF158 shows 21.9% amino acid identity over 370 amino acids and distinguishable similarity to the TlpA DNA-binding domain (Fig. 5A). Both ORF158 and TlpA show very similar profiles from coiled-coil prediction algorithms (Fig. 5B; 21,47). The TlpA and H-NS proteins both bind DNA at low temperature and permit expression at higher temperatures, the opposite of the thermosensitivity seen in R27. Additional studies are required to determine the roles played by TlpA and H-NS homologs in R27, including their potential influence on transfer. As in other recently sequenced plasmids (34,48), we observed a mosaic-like structure in the R27 conjugative plasmid.

Figure 5.

(A) Amino acid alignment of the TlpA DNA-binding domain (45), underlined, with surrounding sequences, and the proposed equivalent in ORF158. (B) The COILS2 output for ORF158 and TlpA indicating the probability (P) of distinct regions of each protein having a coiled-coil conformation. ORF158 and TlpA also have similar outputs using the PARCOIL program (21) (data not shown).

DDBJ/EMBL/GenBank accession no. AF250878