Mutating conserved residues in the ribonuclease H domain of Ty3 reverse transcriptase affects specialized cleavage events.

The reverse transcriptase-associated ribonuclease H (RT/RNase H) domains from the gypsy group of retrotransposons, of which Ty3 is a member, share considerable sequence homology with their retroviral counterparts. However, the gypsy elements have a conserved tyrosine (position 459 in Ty3 RT) instead of the conserved histidine in the catalytic center of retroviral RTs such as at position 539 of HIV-1. In addition, the gypsy group shows conservation of histidine adjacent to the third of the metal-chelating carboxylate residues, which is Asp-426 of Ty3 RT. The role of these and additional catalytic residues was assessed with purified recombinant enzymes and through the ability of Ty3 mutants to support transposition in Saccaromyces cerevisiae. Although all mutations had minimal impact on DNA polymerase function, amidation of Asp-358, Glu-401, and Asp-426 eliminated Mg(2+)- and Mn(2+)-dependent RNase H function. Replacing His-427 and Tyr-459 with Ala and Asp-469 with Asn resulted in reduced RNase H activity in the presence of Mg(2+), whereas in the presence of Mn(2+) these mutants displayed a lack of turnover. Despite this, mutations at all positions were lethal for transposition. To reconcile these apparently contradictory findings, the efficiency of specialized RNase H-mediated events was examined for each enzyme. Mutants retaining RNase H activity on a heteropolymeric RNA.DNA hybrid failed to support DNA strand transfer and release of the (+) strand polypurine tract primer from (+) RNA, suggesting that interrupting one or both of these events might account for the transposition defect.

primers (6,7). In an initial step towards dissecting these complex events at the molecular level, we reported the purification of recombinant p55 Ty3 RT and preliminary characterization of its DNA polymerase and ribonuclease H (RNase H) activities (8). More recently, an active form of Ty1 RT has also been described by Wilhelm and co-workers (9,10). Although the Ty3 enzyme would recapitulate precise selection, extension and excision of its (+) strand PPT primer, we were unsuccessful in replacing Mg ++ in the Ty3 RNase H domain with Fe ++ to support hydroxyl radical-mediated cleavage of duplex DNA, a feature common to the RTs of human and feline immunodeficiency viruses (8,11). Such a result suggested that the mode of metal ion coordination in the Ty3 RNase H domain might differ from the extensively characterized retroviral enzymes.
Sequence alignment shows that a conserved histidine and a cluster of four carboxylate residues constitute the -D-E-D-H-D-motif common to the catalytic site of both retroviral and prokaryotic RNases H (12)(13)(14). In the proposed general acid-base model of catalysis (15,16) D134 of E.coli RNase H positions the attacking water molecule to donate a proton to H124, E48 to catalysis has been proposed (15,21,22), but it was not immediately clear how this function might be fulfilled by tyrosine in the retrotransposon enzyme without invoking its activation by a nearby acidic residue. However, a compilation of RNase H sequences from the Gypsy group of retrotransposons (8) suggested a conserved histidine immediately adjacent to one of the catalytic carboxylate residues (D426) might be implicated in catalysis, leading to our proposal of a -D-E-DH-Y-D-motif ( Fig. 1[A]). Interestingly however, a recent phylogenetic compilation of RNase H sequences from LTR-and non-LTR-containing elements suggests retrotransposon enzymes may lack the flexible "His loop" of retroviral and bacterial enzymes (13) (Fig. 1  outside primers. Mutants D358N, Y459A and D469N were generated using the appropriate 5' and 3' primers containing the desired mutated codon. The resulting fragments were cleaved and subcloned into p6HTy3RT. The final constructs were completely sequenced in the region derived by PCR amplification. These same mutations were introduced into plasmid pEGTy3-1 (24). 3' primers homologous to the mutation site were used for site-directed mutagenesis as described (25).
Expression and Purification of Ty3 RT mutants -Ty3 RT variants were purified from logarithmically grown and IPTG-induced E.coli cultures by a combination of metal chelate (Ni ++ -NTA Sepharose, Qiagen) and size exclusion chromatography (Superdex 200, Pharmacia).
Purified enzymes were free of contaminating nucleases and stored at -20 o C in a 50% glycerolcontaining buffer (50mM NaH 2 PO 4 /Na 2 HPO 4 (pH 7.8), 0.7 M NaCl). Under these conditions we observed minimal loss of activity over several months. Ten µl aliquots were removed at times indicated and processed as above. µg/ml of leupeptin and pepstatin, and 1mM PMSF. Cells were lysed by vortexing in presence of glass beads; the lysate was clarified by centrifugation and protein concentration was determined using the micro BCA assay kit (Pierce).

Virus-like particle (VLP) preparation -One liter cultures of yTM443 cells transformed with
pEGTy3 derivatives were grown to late log phase in SG medium to induce Ty3 expression. A mock VLP preparation was made with non transformed yTM443 cells. VLPs were partially purified from whole-cell extract as previously described (29). Briefly, the cells were harvested, washed in buffer, digested with zymolyase and lysed by vortexing with glass beads. Whole cell extract was fractionated over a 70%, 30% and 20% (5, 5 and 15 ml, respectively) sucrose step gradient by centrifugation in a Surespin 630/36 ml rotor (Sorvall) at 22,000 rpm for 3 h at 4 o C.  (23). Secondary antibodies to rabbit IgG were detected by chemiluminescence, using the ECL system as described by the manufacturer (Amersham).  Equivalent results were obtained on several different substrates (data not shown). Thus, to a first approximation, mutating conserved residues of the Ty3 RNase H domain had minimal effects on the structure of the DNA polymerase catalytic center.

Preliminary Characterization of Ty3 RNase H Mutants
In a second experiment, relocating radiolabel to the template 5' terminus allowed RNase H activity to be evaluated during RNA-dependent DNA synthesis. Under such conditions, the primary hydrolysis products generated by wild type Ty3 RT are 21 and 12 nt (region "a"), which would be the expected products from a template on which the primer had been fully extended.
Despite prolonged incubation, these products are absent for mutants D358N, E401Q and D426N ( Fig. 2 evident. The larger of these products (region "b") is only slightly smaller than the intact template, suggesting longer residency of these mutants on the initial template/primer duplex and low-level template hydrolysis prior to polymerization. Further downstream, a second region of the RNA template is susceptible to hydrolysis. Since the DNA synthesis profiles of these mutants shows no major termination products in this region, this might reflect (i), transient pausing at a region of template secondary structure (ii), resolution of this structure via RNaseH-mediated hydrolysis and (iii), continued polymerization. Such a mechanism has in fact been proposed for HIV-1 RT (30)(31)(32) and RNA folding programs indicate a stable stem-loop structure in this region (data not shown). Thus, while presenting a more complex hydrolysis profile, the cumulative data with mutants H427A, Y459A and D469N indicates they retain significant RNase H activity on a heteropolymeric RNA/DNA hybrid.

Divalent Cation Requirement of Ty3 RT Mutants -Previous work indicated that the HIV-1
RNase H mutant p66 E478Q /p51 (33) recovered polymerization-dependent RNase H activity in Mn ++ on both a random heteropolymeric RNA/DNA hybrid and a second substrate mimicking release of the tRNA primer (33,34). We therefore determined whether such a phenotype could be reproduced with Ty3 RT, and if this was restricted to the equivalent catalytic residue, E401. RT to a single binding event), the more established pattern of hydrolysis emerges: polymerasedependent (-21) and polymerase-independent (-13) cleavages represent 65 and 35% of the total product, respectively (data not shown). Though speculative, it is possible that when Ty3 RT cleaves at position -21, this is immediately followed by cleavage of the same substrate at position -13 when multiple binding events are permitted. This subsequent cleavage, however, is suppressed in the presence of heparin because re-binding of Ty3 RT cannot occur.
Experiments to address this possibility are being considered.

3[B] and [D]
) had several consequences for the latter mutants. Firstly, in contrast to HIV-1 RT, Mn ++ failed to stimulate any mutant of the D358/E401/D426 triad (Fig. 3[B], Lanes 4-6, respectively). Based on data from E.coli (37)  and Site 2 via D358/D469. Since both potential sites share D358, amidation of E401 or D426 appears to affect D358 geometry such that the occupancy by either divalent cation is affected.
Secondly, the Mn ++ -dependent activity of mutants H427A, Y459A and D469N exceeded that observed in Mg ++ (Fig. 3[D], Panels [ii] -[iv]). Retention of Mn ++ -dependent activity again suggests these residues are less critical for catalysis. Thirdly, wild type Ty3 RT exhibited relaxed RNase H specificity in Mn ++ , hydrolyzing the template at almost every position between nucleotides -24 and -6 (Fig 3[C], Lane 2). Although indirect, lack of Mn ++ -dependent activity with mutants of the D358/E401/D426 triad (Fig 3 [B]) ruled out the trivial possibility E.coli RNase H contamination in our enzyme preparations. Finally, the data of Fig. 3 Fig. 3[D] with that of Fig. 3[C]). This suggests slower turnover, i.e. the enzyme, having cleaved the substrate, fails to dissociate from the product and, as a result, additional cleavage is not observed. Enhanced Mn ++ -dependent activity, for mutants H427A, Y459A, and D469N, was directed primarily on -13 cleavage, yielding hydrolysis profiles qualitatively similar to that observed with wild type enzyme in Mg ++ . Stated differently, if the H427/Y459/D469 triad is implicated in Site 2 metal binding, abrogating this event results in catalysis mediated by metal bound exclusively at the biologically relevant site. The results of this assay are presented in Figure 4 hydrolysis level required in vivo, and that the Mn ++ -dependent activity observed in vitro is not biologically relevant. Alternatively, while Ty3 RT mutants H427A, Y459A and D469N could process random, heteropolymeric substrates, the same might not hold true for precise RNase H-mediated events required during replication. This notion was investigated in the next sections.

RNase H-Proficient RT mutants Fail to Support DNA Strand Transfer -DNA strand
transfer, i.e. relocation of nascent DNA to an acceptor template (35), is a specialized event in Ty3 replication requiring RNase H activity. Although this has been studied in retroviruses (35,40,41), model Ty3 systems to investigate the mechanism and its dependence on RNase H function have not been reported. The features of our DNA strand-transfer system (Fig. 6 [A]) are similar to those described by Peliska  transfer to the acceptor template, followed by resumption of DNA synthesis, yields a 60-nt strand transfer product. In this model system, efficient strand transfer requires that polymerization-independent RNase H activity reduces the donor RNA template to a size permitting its dissociation and relocation of nascent DNA onto the acceptor (33,35,42,43). However, while strand transfer product accumulates with wild type Ty3 RT, it is barely detectable with mutants H427A, Y459A and D469N. Figure 6[C] follows the same reaction using a 5' end-labeled donor RNA template rather than radiolabeled primer, which allowed us to monitor RNase H function prior to and concomitant with DNA strand transfer. The accumulation of a 30 nt hydrolysis product early in the time course (Fig. 6 [B], panel [I]) correlates with transient pausing shortly after DNA synthesis is initiated. Thereafter the primary hydrolysis products are 18 nt and shorter, each of which results from an enzyme which has completed DNA synthesis to the 5' terminus of the donor template. A comparison of the hydrolysis and polymerization products gives insight into the size to which the donor template must be reduced to allow strand transfer. For example, although we observe rapid accumulation of an 18 nt hydrolysis product, strand transfer and continued synthesis is not evident. With time, the 18 nt RNA diminishes and is replaced with fragments of 11 and 10 nt. As these accumulate, there is a parallel rise in strand transfer activity, suggesting the donor RNA template must be reduced tõ 10 nt to permit its displacement and relocation of the growing point onto the acceptor template. termini, respectively. On the non-PPT template, wild type Ty3 RT produces a major 21 nt fragment, suggesting the DNA polymerase catalytic center is positioned at the template 3' OH.
A second series of hydrolysis products corresponding to cleavage between -11 and -14 is also evident (Fig. 7 [B Fig. 7 [B], Panel [ii] indicates removal of this RNA would yield an 11 nt radiolabeled fragment. While this is achieved by wild type RT, mutant enzymes are virtually inactive. Precise processing of the PPT flanked by 11 ribonucleotides at its 3' terminus was predicted to yield a 13 nt radiolabeled fragment (Fig. 7 [A], Panel [iii]). 13-and 20-nt fragments are produced by wild type Ty3 RT (Fig. 7 [B], Panel [iii]), the latter most likely arising from positioning of its polymerase domain over the 5' terminus of the PPT-containing primer (44)(45)(46). Interestingly, although the three mutants hydrolyze the non-PPT portion of the primer, cleavage at the PPT/U3 (+) RNA junction is impaired. Finally, release of the PPT 3' OH on substrate PPT/5'-3' is predicted to generate an 18 nt radiolabeled fragment. This is the primary product observed with wild type RT (Fig. 7 [B

DISCUSSION
The availability of recombinant RT from the S. cerevisiae retrotransposons Ty1 (9,49,50) and Ty3 (1,8) Fig. 1 [B]). Thus, the catalytic mechanism for Ty3 RNase H need not be strictly reconciled with other more extensively studied bacterial and retroviral enzymes, perhaps exemplified by recent hydroxyl radical footprinting efforts (8) and in this report by the activity of wild type and mutant enzymes as a function of divalent cation requirement.
One unequivocal feature of our data is complete loss of Mg ++ -and Mn ++ -dependent RNase H activity when any residue of the D358/E401/D426 triad is altered, while substitutions in the H427/Y459/D469 triad are only partially inhibitory. Thus, in keeping with bacterial and retroviral RNases H, we propose that D358/E401/D426 constitute the primary metal binding site of Ty3 RNase H, which presumably is occupied with Mg ++ in vivo. Relaxed RNase H specificity and reduced enzyme turnover with wild type Ty3 RT in the presence of Mn ++ was surprising, but i.e. tighter binding that slow down product release, with the consequence of decreasing enzyme turnover and inducing relaxed specificity. Indeed, for EcoRV, Mn ++ was shown to accelerate the chemical reaction and stabilize the enzyme-substrate complex. Under these conditions, extended residency at a non-cognate site might be predicted to enhance hydrolysis, thus accounting for relaxed specificity (51). Invoking this argument, data of Fig. 3 [D] suggests the relaxed specificity of wild type Ty3 RT in Mn ++ reflects enhanced affinity for the substrate, Le Grice, unpublished observations), although the extent of distortion may be less severe. We therefore suggest that, similar to the RNase H primer grip of HIV-1 RT (59) H427, Y459 and D469 constitute a subset of residues whose interaction with an RNA/DNA hybrid induces the appropriate trajectory of the RNA template into the RNase H catalytic center for hydrolysis.
Although speculative, Y459 of Ty3 RT, a residue conserved in many LTR-containing retrotransposons, may be the counterpart of Y501 of the HIV-1 RNase H primer grip, alteration of which to alanine yields an enzyme exhibiting an abnormal PPT processing phenotype (60).
Experiments to better understand the role of this Ty3 residue are currently underway.