INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

Ribonucleases H (RNH) endonucleolytically cleave the RNA strand of an RNA-DNA hybrid. Because of their unique enzymatic activity, RNH domains are believed to have played an important role in the transition from the RNA world to the DNA world. A remnant of that activity is proposed to be the role of RNH domains in removing RNA primers at the 5' ends of lagging strand synthesis in DNA replication (e.g., Qiu et al. 1999). RNH enzymes have also been implicated in DNA repair and RNA transcription (Crouch and Toulme 1998).

There appear to be three broadly distributed lineages of RNH enzymes: RNase HI (rnhA gene), HII (rnhB), and HIII (rnhC; Ohtani et al. 1999). Common evolutionary ancestry has been firmly established for rnhB and rnhC, whereas rnhA may represent a case of convergent evolution (see Lai et al. 2000). These three lineages also differ in their phylogenetic distribution among the three kingdoms. Archaea only possess rnhB genes, whereas all Eukarya appear to have both rnhA and rnhB genes. Eubacteria can possess all three genes, but most encode either rnhA-B or rnhB-C. In cases of eubacterial genomes that have all three genes, (e.g., Bacillus subtilis), one of the encoded proteins might lack enzymatic activity (Ohtani et al. 1999). Among cellular rnhA genes, the gene from Escherichia coli has been most extensively studied, both with respect to cellular function, as well as the structural aspects of its encoded enzymatic activity (Katayanagi et al. 1993; Goedken and Marqusee 2000). The key residues involved in the catalytic mechanism have been identified and found to be the same in all rnhA proteins (Johnson et al. 1986; Davies et al. 1991).

rnhA domains (hereafter referred to as simply RNH) have also been observed as adjunct domains to the RT as part of the pol gene in retroviruses and in other LTR-bearing retrotransposons (for review, see Boeke and Stoye 1997). In the retroviral and LTR retrotransposon life cycles, RNH performs three related functions: Degradation of the original RNA template, generation of a polypurine tract (the primer for plus-strand DNA synthesis), and final removal of RNA primers from newly synthesized minus and plus strands. RNH domains can be readily aligned between Eubacteria, Eukarya, retroviruses and other LTR-retrotransposons (Johnson et al. 1986; Doolittle et al. 1989). The three-dimensional structures of the HIV-1 and E. coli enzymes are strikingly similar, with the positions of the core catalytic residues virtually invariant (Davies et al. 1991).

RNH domains have also been found in several lineages of non-LTR retrotransposons (Fawcett et al. 1986; Doolittle et al. 1989; Blesa and Martinez-Sebastian 1997). One early study proposed the presence of RNH domains in a wider range of non-LTR element lineages (McClure 1991). However, a recent more comprehensive analysis of all available non-LTR retrotransposons has suggested that only a limited number of lineages possess this domain, and the lineages that do possess it contain many examples in which it has been lost (Malik et al. 1999). The position of the RNH domain of non-LTR retrotransposons carboxy terminal to the RT domain is similar to that of LTR retrotransposons. Because non-LTR retrotransposons reverse transcribe their RNA template directly on to the chromosomal target site (target-primed reverse transcription; (Luan et al. 1993), the cellular RNH activity present in the nucleus may suffice (Malik et al. 1999). In contrast, LTR retrotransposons require RNH activity in RNA-protein particles in the cytoplasm, which may account for the rigid requirement to encode their own RNH domain.

Two of the outstanding questions in the evolution of reverse-transcriptase-bearing elements (retroelements) are: When and from where did retrotransposons arise? Previous phylogenetic analyses based on the RT domain have attempted to address this issue (Xiong and Eickbush 1988,1990; Doolittle et al. 1989; Eickbush 1994; Nakamura et al. 1997). Despite some success with outlining the evolution of retroelements, the LTR retrotransposons have proven especially difficult to place phylogenetically. The RT domain of the LTR retrotransposons are severely truncated compared with all other known examples of RTs; non-LTR retrotransposons and telomerases from Eukarya, and retrons, plasmids, and group-II introns from Eubacteria, making it difficult to unambiguously align the sequences (Malik and Eickbush, in prep.).

In this report we have employed a phylogenetic analysis of RNH domains to address the origin of the LTR retrotransposons. In contrast to the ambiguous RT phylogeny, the RNH phylogeny clearly suggests that the LTR retrotransposons evolved from a late-branching lineage of non-LTR retrotransposons. It also suggests that the lineage of LTR retrotransposons leading to vertebrate retroviruses has "replaced"; its original RNH domain. The present-day tether domain (connection) found in vertebrate retroviruses represents a molecular fossil of the original RNH domain.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

LTR-Retrotransposons Lack an Important Catalytic Motif of RNH

We performed a multiple alignment of RNH domains with representative sequences from Eubacteria, Eukarya, non-LTR retrotransposons and seven different lineages of LTR-retrotransposons. The different RNH domains were chosen from published GenBank entries. In addition, we included several previously unreported sequences available in public databases because these sequences expanded the distribution of the BEL and DIRS groups of elements. This multiple alignment is presented in Figure 1. Overlaid on to this alignment are the secondary structures from the RNH domains from E. coli and HIV1 (PDB structures 1RDD and 1RDH respectively; SCOP database, http://scop.berkeley.edu).

View larger version (82K): [in this window] [in a new window]   Figure 1   Alignment of the Ribonuclease HI (RNH) domains. Representative RNH domains from Eubacteria, Eukarya, non-LTR retrotransposons and each of the seven lineages of LTR retrotransposons were aligned using CLUSTALX and PSI-BLAST. Highlighted in bold are the residues believed important for the catalytic mechanism of RNH, including the four carboxylate (dark arrows) and the single histidine residue (white arrow) that are numbered according to their position in the Escherichia coli RNH domain. Also overlaid are the secondary structures of E. coli and HIV-1 RNH domains (above and below the alignment, respectively). Note the missing histidine residue in all lineages of the LTR retrotransposons except the vertebrate retroviruses.

The catalytic residues for RNH enzymatic activity, indicated by dark arrows (three aspartatic acid and one glutamatic acid residue), are unvaried across all RNH domains. Also indicated is a histidine residue (white arrow) that is believed to be essential for the enzymatic mechanism of RNH (Oda et al. 1993; Kashiwagi et al. 1996). Surprisingly, although the RNH of vertebrate retroviruses have this histidine, all other LTR retrotransposons appear to lack this residue. This `deletion' has gone unremarked in previous reports as most of these analyses focused on the alignment of RNH domains from E. coli (and other Eubacteria) with those of HIV-1 and other vertebrate retroviruses (Johnson et al. 1986; Doolittle et al. 1989; McClure 1991).

We have presented a simplified topological diagram of the three-dimensional structure of E. coli RNH (Structure 1RDD) in Figure 2A, highlighting the four catalytic residues and the histidine residue believed to play a direct role in the RNH catalytic mechanism. RNH domains have been characterized as -helix/-sheet/-helix with the mixed -sheet consisting of five strands in the order 3-2-1-4-5 with strand 2 antiparallel to the rest and an -helix between strands 4 and 5. The proposed active site of rnhA is shown in Figure 2B. An alanine substitution for this histidine residue in E. coli resulted in a large drop in k_cat/K_m (Kanaya et al. 1990). The deletion of this histidine-bearing subdomain in the LTR retrotransposons would suggest altered (perhaps weaker) enzymatic ability for the LTR retrotransposon-borne RNH domains.

View larger version (24K): [in this window] [in a new window]   Figure 2   (A) A simplified topological diagram of the Escherichia coli Ribonuclease HI (RNH) domain, indicating the active site residues (see Fig. 1). -strands are indicated by arrows, and -helices are shown by boxes. The four carboxylates and single histidine residue are shown. (B) A schematic of the proposed RNH catalytic mechanism is shown (modified with permission from Kanaya et al. 1996). The carboxylate triad typical of other endonucleases with an RNH fold (Yang and Steitz 1995; Rice et al. 1996) is indicated by the dotted triangle.

Vertebrate Retroviruses "Reacquired" Their RNH Domain

One of the most surprising aspects of the RNH sequence alignment in Figure 1 is the suggestion that the vertebrate retroviruses' RNH domain is enzymatically more similar to those of eubacterial and eukaryotic genomes, and non-LTR retrotransposons, than it is to other LTR retrotransposons. We performed a phylogenetic analysis of the RNH domains based on the multiple alignment in Figure 1 to test whether the origin of the RNH domains in vertebrate retroviruses was distinct from other LTR-retrotransposons. Figure 3 presents the neighbor-joining tree obtained from this comparison. To test the effect on the phylogeny arising simply from the fact that the subdomain containing H124 is missing from the LTR retrotransposons, we have also excluded this region in a separate analysis. The same phylogeny was obtained as that with the full RNH sequences (data not shown).

View larger version (50K): [in this window] [in a new window]   Figure 3   Phylogeny of the Ribonuclease HI (RNH) domains. A Neighbor-Joining (NJ) tree of the various RNH domains was performed based on the alignment of ~140 amino acid residues in Figure 1. Bootstrap analysis was performed and nodes were collapsed to a 50% consensus. Bootstrap support (percentage from 1000 trials) for the various nodes is shown above the nodes. Maximum parsimony (MP) analysis of the RNH sequences agreed with the NJ analysis but showed lower bootstrap values for most nodes. Bootstrap values from the MP analysis for the major groupings are shown in italics if greater than 50%. The phylogeny is rooted using the eubacterial RNH domains as the outgroup. Note that the phylogenetic position of the vertebrate retroviruses is in conflict with that shown in Figure 4. All retroelement sequences are readily accessible from GenBank and previous reports (Bowen and McDonald 1999; Malik and Eickbush 1999; Malik et al. 2000).

For comparison with the RNH phylogeny, we present in Figure 4 a phylogeny of representative LTR retrotransposons based on the RT domain using non-LTR retrotransposons as an outgroup. This phylogeny is in general agreement with those presented earlier with the only significant uncertainty being the relative position of the Ty1/copia and hepadnaviral groups (Xiong and Eickbush 1990; Bowen and McDonald 1999; Malik et al. 2000). Both the RT and RNH phylogeny reveal four distinct lineages of LTR retrotransposons: the Ty1/Copia, BEL, DIRS1 and Ty3/gypsy groups, as well as three classes of viruses: the retroviruses, hepadnaviruses, and caulimoviruses. The phylogenetic relationship within and among these groups is virtually the same using these two sets of data, with one striking exception. In the RNH phylogeny, retroviruses are located distal to the four retrotransposon groups as well as to caulimoviruses and hepadnaviruses, whereas in the RT phylogeny, retroviruses are a sister group to the Ty3/ gypsy elements and caulimoviruses.

View larger version (51K): [in this window] [in a new window]   Figure 4   Phylogeny of the long-term repeat (LTR) retrotransposons based on their reverse transcriptase (RT) domains. The phylogram is a 50% consensus tree of the elements' RT domains (~240 amino-acid residues) based on the neighbor-joining (NJ) method, and is rooted using non-LTR retrotransposon RTs as an outgroup (not shown). Bootstrap values are shown associated with corresponding nodes. This tree is in agreement with prior analyses (Bowen and McDonald 1999; Malik et al. 2000) except for the relative position of the Ty1/copia and hepadnaviral groups and the additional DIRS1-like sequences from sea urchin and two teleosts (accession nos. AZ181274, AL305423, and AF112374, respectively).

Which of these two analyses is an accurate reconstruction of LTR retrotransposon evolution? The RT phylogeny is consistent with the more generally held view of LTR elements. In particular, the integrase (IN) domains of retroviruses is clearly most similar in sequence and domain structure to that of the Ty3/gypsy group of elements (Capy et al. 1996; Malik and Eickbush 1999). Other features of these elements, including the order of the different enzymatic domains of the pol gene are similar between retroviruses and the Ty3/gypsy group. Thus, parsimony suggests that it is the RNH phylogeny that is at odds with the evolution of LTR retrotransposons. This discrepancy could be reconciled if we propose that the ancestral vertebrate retrovirus "replaced" its preexisting RNH domain with another RNH domain from a source outside the LTR retrotransposon group.

Is there any evidence for this ancient replacement of the retroviral RNH domain? In all members of the LTR retrotransposon lineage, the RNH domain is found immediately adjacent to the RT domain. Examination of the relative positions of the RT and RNH domains in retroviruses clearly reveals "paleontological" evidence for an RNH replacement. Retroviruses have an additional domain separating the RT and RNH domains. This additional domain has been referred to as the "tether" or "connection" domain of the retroviral RT-RNH structure (Kohlstaedt et al. 1992). There is little primary sequence similarity to suggest that this retroviral `tether' was the remnant of a previous RNH domain. However, as presented in Figure 5C, the three-dimensional structure of the HIV-1 tether region from the HIV-1 RT-RNH crystal structure reveals a remarkable structural similarity between it and the functional RNH domains of HIV-1 (Davies et al. 1991; Kohlstaedt et al. 1992), E. coli (Yang et al. 1990) and Thermus thermophilus (Ishikawa et al. 1993).

View larger version (53K): [in this window] [in a new window]   Figure 5   Schematic three-dimensional diagrams of the RNH domains from Escherichia coli (PDB structure 1RDD), Thermus thermophilus (1RIL) and HIV-1 (1RVT) are shown along with the tether domain of HIV-1 (1RVT). -strands and -helices are represented by arrows and cylinders, respectively, using the Cn3D viewer software (version 3.0). Note that the tether (connection) domain has the same fold (also see Artymiuk et al. 1993) as the enzymatically active ribonuclease HI domains.

The tether domain of HIV-1 has the same organization as the enzymatically functional RNH domains (Fig. 5C), except that it lacks the carboxy-terminal motif and possesses none of the conserved catalytic residues. This similarity of the HIV tether and RNH domains has been previously noted using three-dimensional searching techniques and was suggested to have been the result of an RNH gene duplication event (Artymiuk et al. 1993). These authors found a RMS error of only 1.77 A over 48 core C--atoms on superposition of the proposed equivalent five  strands and single  helix (Fig. 5C). Our phylogenetic analysis suggests that this domain was not the result of a duplication, but was rather the acquisition of a new domain from a source outside the LTR retrotransposons. The "new" RNH domain acquired by vertebrate retroviruses may have been more proficient than the "old" one by virtue of the conserved histidine residue (H124 in Fig. 1) involved in the suggested catalytic mechanism (Fig. 2B).

RNH folds are typical of other endonucleases, including the retroviral and DNA-mediated transposases/INs, the RuvC resolvases as well as exonuclease domains of DNA polymerases (Dyda et al. 1994; Yang and Steitz 1995; Rice et al. 1996). In each of these enzymes, three catalytic carboxylates are similarly arranged (Fig. 2A), whereas RNH and RuvC resolvases have an additional fourth conserved carboxylate (D134). Thus, although an RNH fold by itself is not an absolute indicator that the tether was originally an RNH domain, parsimony argues against the likelihood that the tether was derived from any of the other endonucleases; this would invoke not only the loss of the ancestral RNH domain, but the subsequent acquisition and degeneration of another endonuclease.

Non-LTR Retrotransposons Arose Earlier Than LTR Retrotransposons

The phylogeny in Figure 3 is rooted on the various eubacterial representatives of RNH (the Archaea have no rnhA homolog). Using this rooting, non-LTR retrotransposon and the LTR-retrotransposon RNH domains group together, indicating a common evolutionary origin. The phylogenetic proximity of the retrotransposon lineages to the eukaryotic RNH sequences suggests that the origin of this RNH domain was an early eukaryote. Indeed, the diplomonad Giardia lamblia appears to be the outgroup not only to other eukaryotic but also to all retroelement-encoded RNH domains. Is this (acquisition of an RNH domain from an early eukaryote) yet another example of replacement of a preexisting RNH domains, as hypothesized for retroviruses? Or does this phylogeny reflect the original acquisition of this enzymatic domain by both non-LTR and LTR retrotransposons?

We have addressed this issue previously, in the non-LTR retrotransposon lineage. We and others have postulated that the most likely origin of the non-LTR elements are the group-II introns found in eubacteria and the organelles of fungi and plants (Zimmerly et al. 1995; Cousineau et al. 1998; Malik et al. 1999; Lambowitz et al. 1999). This model is based on both the phylogenetic relationship of their RT domains (Xiong and Eickbush 1990; Malik et al. 1999) and the similarity of their target primed reverse transcription mechanisms used for insertion (Luan et al. 1993; Zimmerly et al. 1995). When the group-II introns are used to root the non-LTR retrotransposon phylogeny, it suggests that the original non-LTR elements were elements which encoded a single open reading frame (ORF) and contained an endonuclease domain with an active site similar to certain restriction enzymes (Malik et al. 1999 Yang et al. 1999). Evolving from these original non-LTR retrotransposons were elements that acquired a gag-like first ORF and replaced the original restriction-like endonuclease with an apurinic-like endonuclease (APE). This nonspecific APE domain enabled the non-LTR elements to insert more widely throughout the genome resulting in the diversification of a number of different lineages. One of these new lineages acquired an RNH domain giving rise to the present day lineages we have termed the I, R1 and Tad clades (see Malik et al. 1999). Thus, our previous analysis of all non-LTR retrotransposon sequences suggests that the phylogeny in Figure 3 reflects the original acquisition of an RNH domain from a eukaryotic host. The phylogeny of the RNH domains is unable to resolve branching order of the three extant non-LTR clades that are derived from this lineage (Malik et al. 1999).

In the case of the LTR retrotransposons (excluding the vertebrate retroviruses), acquisition of the original RNH domain again appears to be monophyletic. This single lineage is not clearly resolved from the multiple extant non-LTR retrotransposon RNH lineages. Because the phylogeny derived from this RNH domain is the same as the RT phylogeny (with the exception of the retroviruses describe above), the entire lineage of LTR retrotransposable elements thus appears to be no older than one of the younger lineages of non-LTR retrotransposons. Consistent with the proposal that the LTR retrotransposons arose later in the eukaryotic lineage than the non-LTR retrotransposons is their phylogenetic distribution. Although non-LTR retrotransposons have been found in the oldest eukaryotes, the diplomonad Giardia lamblia (Arkhipova and Meselson 2000; Burke et al., in prep.) and trypanosomes (Kimmel et al. 1987; Teng et al. 1995), LTR-retrotransposons have not been found in these lineages. This phylogeny thus suggests that the original LTR-retrotransposon RNH domain was acquired from a non-LTR retrotransposon. This event may have been repeated when retroviruses replaced their RNH domain. However, the poor resolution of the non-LTR retrotransposon and vertebrate retroviral lineages does not allow us to rule out alternate possibilities for the source of this acquisition.

What was the structure of the precursor LTR element that acquired this RNH domain? In the following section we present arguments for what we believe is the most likely origin of the LTR retrotransposons, the fusion of a DNA-mediated transposon and a non-LTR retrotransposon.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

Vertebrate Retroviruses: RNH Connections

In this report we have presented phylogenetic analyses that indicate the vertebrate retroviral lineage has replaced its RNH domain. This event must have occurred early in the evolution of retroviruses because all known retroviral lineages contain this new RNH domain. Our analysis indicates that the retroviruses probably obtained their RNH domain from a non-LTR retrotransposon (Fig. 3). This close relationship of the RNH domain from retroviruses and non-LTR retrotransposons can be observed in the first comparisons of RNH domains in different types of retroelements (Doolittle et al. 1989).

The presence of a connection domain represents the most dramatic difference between retroviral RTs and the RTs of LTR retrotransposons. Because of the advantages of reducing its genome size, we would have expected the preexisting RNH domain to have been rapidly lost after the retroviral lineage gained a new domain. The fact that the connection domain still exists suggests that this "fossilized" RNH domain is performing another important role in the lifecycle of the virus. What is this present-day function? In the case of the HIV protein, the best studied retroviral reverse transcriptase, the active enzyme is a heterodimer composed of a p66 subunit containing an RT, a connection and an RNH domain and a p51 subunit containing only the RT and connection domains. Several studies have remarked on the structural and possible functional role of the connection domain in the formation of this heterodimer (Wang et al. 1994; Divita et al. 1994; Debyser and De Clercq 1996). For example, it has been shown to be crucial in mediating the conformational changes required of the p66/p51 heterodimer for reverse transcription (Bahar et al. 1999) and for RNH activity (Smith et al. 1994). Indeed, contacts by the connection domains make up one-third of the total contacts between the two subunits, and the connection domain in the p51 subunit makes close contact with the tRNA primer annealed to the viral RNA template (Kohlstaedt et al. 1992). Finally, the connection domain may even play a role in the incorporation of protein in the virus particle. Mutations in the connection domain prevent the efficient packaging of HIV viral particles (Mak et al. 1997). It appears likely that, although the RNH domain in other LTR-retrotransposons may carry out both enzymatic and structural roles, the presence of a connection domain in retroviruses has allowed subfunctionalization (Lynch and Force 2000). Thus, although the newly acquired RNH domain is enzymatically active, the connection domain may still carry out its ancestral structural function.

The Chimeric Origin of LTR Retrotransposons

Perhaps the most interesting aspect of the RNH phylogeny described in this report is its implication for the origin of eukaryotic retrotransposons. The RNH domains of the I, R1, and TAD clades of non-LTR elements and the original RNH domain in LTR elements (i.e., before the reacquisition of the RNH domain by retroviruses) appear to have a common origin. These acquisitions appear to have arisen sometime after the origin of eukaryotes. Note that the branch containing these retrotransposon sequences is more closely related to the RNH from the crown group of eukaryotes than is the RNH domain of G. lamblia (Fig. 3). Based on the phylogeny of the RT, APE, and RNH domains of the non-LTR retrotransposons, we had previously concluded that the acquisition of the RNH domain was a monophyletic event occurring late in the evolution of these elements (Malik et al. 1999).

In contrast to the non-LTR retrotransposons, few models have been proposed for the origin of the LTR retrotransposons. First, no prokaryotic elements have been found that could be regarded as likely progenitors of the present-day LTR retrotransposons. Second, the oldest lineage of extant LTR retrotransposons, the Ty1/copia lineage (Xiong and Eickbush 1990; Fig. 4) contains all the components of a complete LTR retrotransposon (a gag-like ORF1, and a pol gene with protease, RT, RNH, and IN domains). The only difference between the Ty1/copia group of elements and the other groups of LTR retrotransposons is the position of the IN domain. It is found upstream of the RT/RNH domains in the Ty1/copia group but is downstream from the RT/RNH in the BEL, Ty3/gypsy and retroviral clades. Unlike the gradual addition and replacement of domains in the non-LTR retrotransposons, the only dramatic changes that have occurred since the evolution of LTR retrotransposons were the addition in several lineages of env-like domains (Malik et al. 2000) and the loss of the IN domain in the DIRS group (Cappello et al. 1985).

We propose that the origin of the LTR retrotransposons was the fusion of a DNA-mediated transposon and a non-LTR retrotransposon. Although this model is highly speculative, it is the only simple model that can explain the sudden origin of the two-step mechanism used by LTR retrotransposons in the absence of obvious eubacterial precursors. Based on the similarity of the IN of LTR retrotransposons and the transposases of DNA transposons, Capy et al. (1998) have also recently postulated that one likely origin of the LTR retrotransposons was by a DNA-mediated transposon acquiring RT activity.

Transposition Mechanism

IN

RT

RNH

gag-Like ORF

View larger version (35K): [in this window] [in a new window]   Figure 6   The role of the eukaryotic nucleus in the evolution of long-term repeat (LTR) retrotransposons. In eubacteria, transcription and translation are coupled; thus the encoded transposase (here labeled IN for integrase) of a DNA-mediated element can immediately bind the donor element for transposition. For simplicity, we have only shown the DNA-mediated transposition reaction as cut-and-paste, but in Eubacteria, a replicative form of transposition can also occur (Mizuuchi 1992). In the case of the RNA-mediated reaction, the reverse transcriptase (RT) can immediately bind its own transcript and initiate target-primed reverse transcription. This eubacterial precursor of the eukaryotic retrotransposons is assumed to be a mobile group II intron (Cousineau et al. 1998). The situation differs for mobile elements in eukaryotes where transcription and translation are uncoupled. Synthesis of IN in the cytoplasm means that this enzyme must enter the nucleus and find a donor for transposition. This can result in the transposition of defective copies that only retain the correct terminal repeats. In the case of the non-LTR retrotransposons, the RT must drag its RNA template back into the nucleus (cis preference) or find a new RNA template. Only the former will insure the production of active copies (Wei et al. 2001). This need to stabilize the template for entry back into the nucleus (or to wait for the breakdown of the nuclear membrane during cell division) is postulated to be the selective force that enabled the evolution of the LTR retrotransposons. LTR retrotransposons utilize both RT and IN activities. First, the RNA template is reverse transcribed into a double stranded DNA template. Second, an integrase complex shuttles this complex to a target site for integration either through the nuclear membrane or during nuclear breakdown at cell division.

	METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

RNH sequences were obtained from GenBank via PSI-BLAST and TBLASTN (Altschul et al. 1997) searches against the nonredundant database. The sequences were then aligned using CLUSTALX (Thompson et al. 1997) and manually refined using PSI-BLAST alignments as a template. RNH structures were obtained from the PDB database (http://www.rcsb.org/pdb). Schematic diagrams of the various PDB files were made using the Cn3D viewer software version 3.0 (http://www.ncbi.nlm.nih.gov/Structure/CN3D/cn3d.shtml). Phylogenetic analyses were performed according to the neighbor-joining method (Saitou and Nei 1987) using PAUP* (Swofford 1999). Maximum parsimony analysis was also carried out using PAUP* and the heuristic search option with the number of trees retained at each step limited to 10. Although trees obtained by both methods are in strict agreement, the bootstrap support is generally lower using maximum parsimony.