(Received for publication, October 9, 1996, and in revised form, December 20, 1996)
From the Department of Biochemistry,
§ Division of Infectious Diseases, and ¶ Center for
AIDS Research, Case Western Reserve University School of Medicine,
Cleveland, Ohio 44106-4984
Alterations to the highly conserved
Asp549 of the retroviral ribonuclease H (RNase H)
domain were evaluated in the heterodimeric (p66/p51) reverse
transcriptases of human immunodeficiency and equine infectious anemia
viruses. In addition to the polymerization-dependent and
-independent modes of template hydrolysis, mutants were evaluated via
their ability to select and extend the 3 polypurine tract (PPT)
primers of these two lentiviruses into (+) strand DNA. Concerted and
two-step reactions were designed to evaluate (+) strand priming, the
latter of which allows discrimination between selection end extension
events. In contrast to enzyme mutated at the highly conserved
Glu478, substitution of Asp549 with Asn or Ala
reduces, rather than completely eliminates, RNase H activity. When the
requirement for RNase H function becomes more stringent, differences in
activity are readily evident, most notably in the cleavage events
liberating the 5
terminus of the PPT primer. PPT selection thus
appears to represent a specialized form of RNase H activity that is
more sensitive to minor structural alterations within this domain and
may provide a novel therapeutic target.
Although nonspecific hydrolysis of the RNA-DNA replication
intermediate might be considered the primary function of the C-terminal ribonuclease H (RNase H)1 domain of
retroviral reverse transcriptase (RT), increasing attention is being
given to the extensive repertoire of highly specialized RNase
H-mediated events necessary to synthesize a double-stranded proviral
DNA from the viral RNA genome. For example, polymerization-independent RNase H activity (which hydrolyzes the RNA template from the point of
initial endonucleolytic cleavage to within eight nt of the primer
terminus) has been demonstrated essential for efficient transfer of
nascent () strand DNA between the 5
and 3
termini of the RNA genome
(1, 2). In addition, (+), or second-strand DNA synthesis requires
highly specific RNase H cleavage to release the 3
OH of the polypurine
tract (PPT) primer (3-5). Finally, the (+) and (
) strand RNA primers
(the PPT and a host-derived tRNA, respectively) must be precisely
removed from nascent DNA to preserve the integrity of the 5
and 3
long terminal repeat termini of the provirus for recognition by the
retroviral integration machinery (3, 6, 7). Each of these events can be
considered candidates for therapeutic intervention by antiviral agents
in the ongoing effort to stem the progression of human immunodeficiency virus (HIV) infection and acquired immunodeficiency syndrome. However,
the success of such ventures would benefit from a more detailed
analysis of RNase H-mediated events during retroviral replication.
Structural similarities between RNase H domains of retroviral and
bacterial origin (8) make a strong case for metal ion co-ordination by
the side chain carboxylates of Asp443, Glu478,
and Asp498 (Asp10, Glu48, and
Asp70 in Escherichia coli RNase H1). However, a
fourth residue, Asp549 (Asp134 in E. coli) is also conserved among these enzymes, the significance of
which is not immediately apparent. From the crystal structure of the
isolated RNase H domain of HIV-1 RT, Davies et al. (8) have
suggested that Asp549 could participate in binding a second
metal ion, thereby invoking a two-metal ion-catalyzed mode of
hydrolysis (9). Alternatively, hydrogen bonding between the backbone
carbonyl oxygen of Asp549 and the side chain hydroxyl of
Ser553 may have important architectural consequences for
structural elements that aid in accommodating the template primer
duplex within the C-terminal RNase H domain (-helix E
and the
5
-
E
connecting loop (10, 11)). The latter possibility was
initially supported by mutagenesis studies with HIV-1 RT, indicating
that removing Ser553 via an eight-residue C-terminal
deletion selectively eliminated polymerization-independent RNase H
activity, concomitant with which was a substantial reduction in DNA
strand transfer activity (12). However, when an equivalent enzyme from
EIAV was evaluated, RNase H activity was unaffected but could be
altered by removal of an additional two residues (13). In the enzyme
from E. coli, Haruki et al. (14) have suggested
that Asp134 is less likely to be involved in maintaining
overall architecture, and that it might indirectly contribute toward
catalysis by preserving active site geometry through negative charge
repulsion. However, because Asp134 of E. coli
RNase H forms a salt bridge with Arg138, an additional role
in stabilizing
-helix V (counterpart of
-helix E
in the HIV-1
enzyme) cannot be excluded. These observations prompted us to more
precisely evaluate the role of Asp549 in two structurally
related lentiviral enzymes (both HIV-1 and EIAV RTs are heterodimers of
66- and 51-kDa subunits) by site-directed mutagenesis.
In addition to the polymerization-dependent and
-independent modes of RNase H-mediated hydrolysis, we elected to
investigate the consequences of amino acid alterations in the RNase H
domain on PPT: (a) selection; (b) extension into
(+) strand DNA; and (c) removal from nascent (+) strand DNA,
using RNA-DNA duplexes within which the PPTs of HIV and EIAV were
embedded. Recent reports have indicated that the r (purine)/d
(pyrimidine) configuration confers considerable thermal stability on
the duplex compared with its d (purine)/r (pyrimidine) counterpart
(15). Furthermore, model hybrids mimicking the PPT demonstrate an
unusual CD spectrum, suggestive of a novel conformation (16). Given
these potentially unique features of the PPT, subtle alterations to the
geometry of the RNase H domain may have significant impact on PPT
selection and/or its subsequent release from nascent (+) strand DNA. In addition to highlighting differences in the modes of hydrolysis of
HIV-1 and EIAV RT, we demonstrate in this communication that substituting Asp549 with Asn compromises selection of the
HIV-1 3 PPT and its release from (+) strand DNA. The ability of the
same mutant to extend a preselected PPT into (+) strand DNA as
efficiently as wild-type enzyme suggests that the mutation
Asp549
Asn does not have global structural
consequences.
The coding sequence for EIAV protease (17) was amplified from pEIAV (a gift of Dr. S. L. Payne) by the polymerase chain reaction and ligated into pDS56RBSII (18) as a BamHI/HindIII fragment. The resulting EIAV protease expression cassette was introduced to p6H EIAV RT (19) as an XhoI fragment, creating pEIAV PR-6H RT, which controls co-expression of p66 EIAV RT and protease. The analogous HIV vector, p6HRT-PROT, has been described (18). Mutation of HIV and EIAV RT genes at amino acid position 549 (D549N and D549A) was achieved using the USE mutagenesis kit (Pharmacia Biotech Inc.).
Purification of HIV RT has been described (18), whereas the EIAV enzyme
was prepared essentially as described by Rausch et al. (13),
with the exception that cultures were harvested and cooled to 4 °C
30 min after induction to prevent overdigestion of EIAV RT by protease.
Metal chelate (Ni2+-NTA-Sepharose) and Mono-S ion exchange
chromatography (Pharmacia Biotech Inc.) yielded highly pure enzymes
free of contaminating nucleases, with a 1:1 stoichiometry of p66 and
p51 subunits. Purified enzymes were stored at 20 °C in 50 mM Tris-HCl, pH 7.0, 1 mM EDTA, 25 mM NaCl, and 50% glycerol.
The
location of the HIV 3 PPT has been determined (20), whereas location
of the EIAV 3
PPT was based upon homology to HIV. Polymerase chain
reaction was used to generate fragments of the HIV and EIAV genomes
containing 3
sequences from pHxb2 (21) or pEIAV, respectively. To
standardize the positioning of PPTs within each fragment, polymerase
chain reaction products were designed such that the extreme 3
-G-A-
dinucleotide of the PPT was flanked by 50 and 70 bases of genomic
sequence at the 5
and 3
termini, respectively. These products were
introduced as XhoI/BglII fragments into pSP72
(Promega Corp.) to generate transcription vectors pH3 and pE3,
respectively. 126 nt (+) strand, PPT-containing transcripts (H3 and E3)
were obtained from EcoRV-cleaved pH3 and pE3 using the SP6
Megascript kit (Ambion).
DNA-dependent DNA polymerase activities were assessed on a 71-nt DNA template annealed to a 36-nt DNA primer as described by Rausch et al. (13). Enzyme (8 nM) and template/primer (8 nM) were incubated 1 min in buffer containing 10 mM Tris-HCl, pH 8.0, 6 mM MgCl2, 80 mM NaCl, and 5 mM dithiothreitol. DNA synthesis was initiated by adding dNTPs to a final concentration of 50 µM. Aliquots were removed at the times indicated in the text and mixed with urea-based, gel-loading buffer. RNA-dependent DNA synthesis was assessed in a similar manner on the H3 RNA template annealed to a 20-nt DNA primer, with the exception that DNA synthesis was terminated after 10 min and the enzyme:template-primer ratio was increased 2-fold. Single-round DNA synthesis conditions were achieved by adding heparin to a final concentration of 2 mg/ml, as described previously (13). Reaction products were fractionated by high-voltage gel electrophoresis through 10% (w/v) polyacrylamide gels containing 7 M urea in Tris borate/EDTA buffer. After drying, gels were subjected to autoradiography, using the DuPont Reflections system.
Evaluation of RNase H Activity by High-resolution Gel ElectrophoresisA 5-32P end-labeled, heteropolymeric
90-nt RNA hybridized to a 36-nt DNA primer was used to evaluate RNase H
activity in the absence of DNA synthesis (22). Enzymes were incubated
with template-primer (final concentrations, 150 and 50 nM,
respectively) in the absence of Mg2+, under conditions
described previously (12, 21, 23). Hydrolysis was initiated by addition
of MgCl2 to a final concentration of 6 mM and
terminated at the times indicated in the text by the addition of Tris
borate/EDTA buffer containing 100 mM Tris, pH 8.5, 100 mM borate, 2 mM EDTA, and 7 M urea.
Alternatively, the final concentration of MgCl2 was varied
as described. RNase H activity was also examined on the same substrate,
the 3
terminus of which was end-labeled with [32P]Cp and
RNA ligase (NEB) under conditions recommended by the manufacturer. In
these experiments, final enzyme and template-primer concentrations were
43 and 12 nM, respectively. Hydrolysis products were
fractionated by denaturing high-voltage electophoresis and analyzed by
autoradiography. Product size was determined by co-electrophoresis of
partial RNase A and alkaline hydrolysates of the radiolabeled RNA
templates.
126-base pair
RNA-DNA hybrids, within which the 3 PPTs of HIV-1 and EIAV were
embedded, were used to evaluate PPT selection and (+) strand synthesis.
These were generated by incubating RNA transcripts containing 20-nt DNA
primers at their 3
termini with HIV-1 p66E478Q/p51 (an RT devoid of
RNase H activity)(24, 25) and 200 µM dNTPs in a buffer of
10 mM Tris-HCl, pH 8.0, 6 mM MgCl2, 80 mM NaCl, and 5 mM dithiothreitol for 60 min
at 37 °C. DNA synthesis was terminated by extraction with an equal
volume of 1:1 phenol:chloroform (v/v), and the products precipitated in
three volumes of ethanol, 0.1 volume of 3 M NaOAc, pH 5, and resuspended in nondenaturing gel-loading buffer (30% glycerol,
0.25% bromphenol blue, and 0.25% xylene cyanol). The 126-nt RNA/DNA
duplexes were purified from 10% nondenaturing polyacrylamide gels and
quantified by UV spectroscopy. To evaluate (+) strand DNA synthesis,
RNA-DNA hybrid (50 nM) was incubated with RT (85 nM), 50 µM dNTPs, and
[
-32P]dATP in reaction buffer for 60 min at 37 °C.
Reactions were terminated by heat denaturation (90 °C for 2 min),
after which one-half of each mixture was added to 0.3 volume 1 N NaOH and incubated at 65 °C for 20 min to achieve
complete RNA hydrolysis (the remaining portion was stored at 4 °C).
Reaction products were then precipitated (50% isopropanol and 3.5 M NH4OAc) and resuspended in a urea-based,
gel-loading buffer.
To distinguish between PPT selection from primer-extension, the RNA-DNA
hybrid containing the HIV-1 3 PPT (50 nM) was initially incubated with a "selecting" enzyme (85 nM each) in the
absence of nt for 30 min at 37 °C, thus preventing DNA synthesis
from occurring. The RNase H hydrolysis products were next extracted with phenol/chloroform, precipitated, resuspended in DNA synthesis buffer, and evaluated for their ability to support (+) strand synthesis
by the addition of dNTPs and [
-32P]dATP in the
presence of an "extending" enzyme. After 30 min at 37 °C, DNA
synthesis was terminated, and the products were treated as described
above. Reaction products were fractionated by high-voltage gel
electrophoresis through 10% (w/v) polyacrylamide gels containing 7 M urea in Tris borate/EDTA buffer. Sequencing profiles of
pH3 and pE3 were generated using the dsDNA Cycle Sequencing System
(Life Technologies, Inc.). These manipulations are formally equivalent
to the "two-step" DNA synthesis reactions described by Wöhrl
et al. (19) and Ghosh et al. (26).
p66/p51
heterodimers of HIV-1 and EIAV RT carrying the mutations
Asp549 Asn and Asp549
Ala were
evaluated in this study to assess structurally related enzymes. The
conservative substitution with asparagine introduces a residue with
similar polarity and volume; in contrast, substitution with alanine
introduces a neutral side chain, while minimizing unfavorable steric
contacts and avoiding imposition of new charge interactions or hydrogen
bonds. To eliminate the possibility that major structural changes
accompanied amino acid substitutions in the RNase H domain, the DNA
polymerase activities of all mutants were first assessed on two
heteropolymeric template-primer combinations, the results of which are
illustrated in Fig. 1.
The 71-nt DNA template/36-nt DNA primer of our studies contains a short template hairpin 15 nt upstream from the primer terminus. This structure has the consequence of transiently stalling the replication machinery but is eventually overcome, resulting in synthesis of a full-length, 72-nt cDNA (10, 13). In the time course of Fig. 1A, this pattern is demonstrated for the wild-type HIV-1 (panel [i]) and EIAV enzymes (panel [iv]). Minor differences in processivity were observed when Asp549 was substituted with either Asn (panels [ii] and [v]) or Ala (panels [iii] and [vi]), although this most likely reflects experimental error. Taken together, the data of Fig. 1A suggest that DNA-dependent DNA polymerase activity of the HIV-1 and EIAV RT mutants was unaffected. This is strengthened by the data of Fig. 1B, which assesses the RNA-dependent DNA polymerase activity of each mutant on a 126-nt template derived from the HIV-1 genome. In this experiment, DNA synthesis was also evaluated in the presence of the competitor heparin, which restricts polymerization to a single binding event. For all enzymes tested, the addition of heparin highlighted dissociation of the replication complex within 10 nt of the primer terminus and subtle alterations in the stalling pattern further along the template. Collectively, the data of Fig. 1 thus indicates that the DNA polymerase catalytic center of all enzymes retained the appropriate structural integrity.
Analysis of RNase H Activity on a 3Initial characterization of RNase H activity of the
HIV-1 and EIAV mutants made use of a 90-nt RNA template/36-nt DNA
primer described previously (12, 13), with the exception that
radiolabel was introduced at the 3 end of the template. By doing so,
it was possible to distinguish products of the first endonucleolytic cut from those defined by polymerization-independent (27, 28) or
directional processing RNase H activity (29). Under these conditions,
the data of Fig. 2 indicate considerable differences in
the manner in which HIV-1 and EIAV RT hydrolyze the model RNA-DNA hybrid.
For HIV-1 RT, the primary hydrolysis products span template nt 16 to
21, with the most prominent events at positions
20 and
21 after
5 s. This hydrolysis profile indicates relaxed cleavage specificity of the HIV-1 enzymes but is essentially in keeping with
reports suggesting that the RNase H and DNA polymerase catalytic centers are separated by a distance of ~17-20 base pairs (27, 28,
30). Under the same conditions, template hydrolysis by the HIV-1 mutant
p66D549N/p51 was only slightly reduced, while activity with mutant
p66D549A/p51 was virtually undetectable. Under equivalent conditions, a
different hydrolysis profile was generated by the EIAV enzyme. In this
case, there was considerably more stringency in the position of
hydrolysis, which was restricted predominantly to template nt
16 and
17. Once again. the overall level of hydrolysis was reduced with EIAV
mutant p66D549N/p51 and absent in mutant p66D549A/p51. Because both the
HIV-1 and EIAV enzymes used in this study are heterodimers of 66- and
51-kDa subunits sharing ~50% amino acid homology, differences in
their RNase H hydrolysis profiles are unlikely to reflect major
structural differences between the two enzymes. Although features of
the two enzymes contributing to the different hydrolysis profiles will
be addressed later, the data of Fig. 2 suggest that substituting Asp549 of either retroviral enzyme with Asn appears to have
little effect on initial endonucleolytic cleavage of the template,
whereas introducing Ala at the same position has severe
consequences.
Upon prolonged incubation with wild-type HIV-1 and EIAV enzymes,
considerably shorter RNA fragments predominate, indicating RNase H
activity cleaving toward the radiolabeled template 3 terminus. Results
of this nature are consistent with recent reports that, under the
appropriate conditions, the RNA 5
terminus can direct the position of
RNase H hydrolysis (31, 32). This concept is outlined schematically in
the lower portion of Fig. 2B. Because RT binding is
initially controlled by the DNA primer 3
terminus, the combination of
synthesis-dependent and -independent RNase H activities
will have the effect of producing a "gapped" template. In doing so,
this creates an RNA-DNA hybrid with a readily available RNA 5
terminus, which can serve to re-direct RT to cleave further 3
,
i.e. toward the radiolabel. This activity is virtually
absent from HIV-1 and EIAV RT harboring the Asp549
Asn
mutation, reflecting either: (a) loss of RNA-5
directed RNase H activity; or (b) an inability of Asp549
Asn mutants to create a "gapped" duplex, effectively shielding the RNA 5
terminus. Experiments of the following section suggest that
the latter explanation is more likely. Equally important is the
observation that, despite almost complete template hydrolysis, the
hydrolysis profile of EIAV p66D549N/p51 RT is still distinguishable from its HIV-1 counterpart. The possibility that both the HIV and EIAV
enzymes work in a similar manner but that the EIAV enzyme is simply
slower can thus be discounted. Finally, although the extent of
hydrolysis is substantially reduced, both Asp549
Ala
mutants generate the same hydrolysis pattern as the Asp549
Asn enzymes.
To monitor the 3-5
processing RNase H
activity, mutants were next investigated on the same RNA-DNA hybrid
within which the RNA component was uniquely end-labeled at the 5
terminus (Fig. 3A). In keeping with the data
of Fig. 2, a broader distribution of endonucleolytic cleavage products
were derived by wild-type p66/p51 HIV-1 RT, suggesting a more relaxed
specificity than its EIAV counterpart (Fig. 3B). In
contrast, both enzymes catalyze qualitatively identical
polymerization-independent cleavage, because the hybrid is hydrolyzed
as far as template nt
8. Differences in endonuclease activity are
subtle but reproducible, possibly reflecting minor differences in the
manner in which the two enzymes contact the template-primer duplex
previously alluded to by enzymatic footprinting experiments (10,
13).
Fig. 3C suggests that substituting Asp549 with
Asn has a differential effect on RNase H activity of both the HIV-1 and
EIAV enzymes. Although the rate of initial endonucleolytic cleavage
does not appear significantly compromised, the products of this
reaction were found to accumulate, apparently serving as poor
substrates for polymerization-independent hydrolysis to template nt
8. These observations can be likened to data recently presented on
HIV-1 RT mutants containing a deletion of either their p66 (p66
8/p51 (12)) or p51 subunit (p66/p51
13 (33)), although in the present case
we do observe limited
8 template cleavage. Previous studies on RNase
H activity in the presence of the competitor heparin indicate
predominant cleavage at template nt
17 (12), suggesting the same or a
second enzyme must re-bind this "nicked" intermediate to achieve
cleavage as far as template nt
8. Invoking this scenario, the data of
Fig. 3C suggest that subtle alterations to the geometry of
the RNase H domain may have serious consequences for the manner in
which the retroviral polymerase accommodates the more relaxed structure
of a nicked RNA-DNA hybrid. If under these conditions the RNA 5
terminus is not accessible to RT, this would account for loss of RNA
5
-directed hydrolysis in Fig. 2.
The hydrolysis profiles of Fig. 3D indicate that although
the highly conserved Asp549 is clearly important for RNase
H function, it is not absolutely critical, because low-level hydrolysis
could still be achieved with both p66D549A/p51 mutants. Although
slightly prolonged autoradiographic exposure was necessary to highlight
the hydrolysis products from each enzyme, retention of function can be
contrasted with an HIV-1 RT mutant containing the substitution
Glu478Gln, which completely abolishes
Mg2+-dependent RNase H activity (24, 25). A
second feature of both Asp549 Ala mutants is again
relaxed specificity of endonucleolytic cleavage. With p66D549A/p51
HIV-1 RT, hydrolysis products reflecting cleavage between positions
16 and
24 are evident, whereas its EIAV counterpart hydrolyzes
between positions
16 and
22. Reasons for this relaxed specificity
are not immediately apparent, although data of a later section suggest
this may reflect a hitherto unnoticed feature of the parental
enzyme.
If Asp549 were involved in Mg2+ co-ordination, this predicts that replacement with Asn or Ala would influence affinity for the divalent cation. To test this experimentally, the RNA-DNA hybrid of Fig. 2A was incubated with wild-type, p66D549N/p51, and p66D549A/p51 HIV-1 RT in the presence of increasing concentrations of Mg2+. The results of Fig. 3E indicate that both Asp549 mutants are unaffected at elevated Mg2+ concentrations. In fact, the increased ionic strength of the buffer appears to have a slight inhibitory effect on the wild-type enzyme. Other experiments indicate that the RNase H activity of p66D549A/p51 HIV RT, in contrast to that of p66D549A/p51 (23), is not restored in the presence of Mn2+ (data not shown). Hence, Asp549 does not appear to be directly involved in metal ion coordination.
PPT Selection by RNase H MutantsInitiation of (+) strand
synthesis in retroviruses represents an event requiring considerably
more precision of the RNase H domain to: (a) select the PPT
3 OH; (b) initiate DNA synthesis; and (c) remove
PPT sequences from nascent (+) strand DNA. Using purified RNA-DNA
hybrids, within which the 3
PPTs of HIV and EIAV were embedded, these
multiple events were evaluated via incorporation of radiolabel into (+)
strand DNA. The results of this analysis are presented in Fig.
4. (+) strand DNA products were analyzed prior to and
following alkali treatment, the latter of which liberates newly
synthesized DNA from the RNA primer.
In keeping with previous reports (5, 20), HIV-1 RT selects the
appropriate 3 PPT primer to initiate (+) strand synthesis immediately
downstream of a contiguous stretch of G residues (G6). The
presence of correctly sized (+) strand DNA in the absence of alkali
treatment also indicates that this enzyme efficiently removes the PPT
primer. Under the same conditions, several differences were noted with
mutant p66D549N/p51: (a) the overall amount of (+) strand
product is considerably reduced, despite near wild-type levels of
DNA-dependent DNA polymerase activity on the model
substrate of Fig. 1; (b) although (+) strand synthesis
initiates from the correct position, the primer selected contains the
entire PPT and an additional five U residues 5
to this; and
(c) alkali treatment is required to free the extended PPT
from (+) strand DNA. Subtle alterations to RNase H activity
demonstrated in Fig. 3 thus appear to have more serious implications
for 3
PPT selection. Wild-type EIAV RT was found to initiate (+)
strand synthesis at the equivalent position as its HIV-1 counterpart,
whereas the activity of mutant p66D549N/p51 was also substantially
reduced. For both lentiviral enzymes, (+) strand DNA could not be
detected in reactions supported by p66D549A/p51 mutants.
In contrast to the HIV-1 3 PPT, wild-type HIV-1 and EIAV RT respond
differently to the EIAV 3
PPT. The EIAV enzyme displays the expected
precision to initiate exactly at the 3
end of the PPT, whereas its
HIV-1 counterpart initiates at several positions within the 3
G6 sequence. A slight reduction in (+) strand DNA was
evident in reactions containing p66D549N/p51 EIAV RT; surprisingly, (+)
strand products were virtually undetectable in reactions supported by
p66D549N/p51 HIV-1 RT. Finally, in keeping with results from the HIV-1
3
PPT, enzymes containing the Asp549
Ala mutation
failed to generate (+) strand product.
Experiments of Fig. 4 cannot exclude the possibility
that PPT selection is unimpaired in our HIV-1 and EIAV RNase H mutants, but DNA synthesis was affected by unique structural features of the (+)
strand initiation complex, where an RNA-DNA hybrid is gradually
replaced in the nucleic acid binding cleft by duplex DNA. This notion
is not without precedent, evidenced by recent documentation that
intermolecular base pairing between genomic RNA sequences outside the
primer binding site and the tRNA primer influences initiation of ()
strand synthesis in lentiviruses and the transition from initiation to
elongation (34, 35). We, therefore, sought to develop a strategy
capable of independently assessing PPT primer selection and (+) strand
synthesis. The approach we adopted is presented in Fig.
5A, and results with the HIV-1 and EIAV
enzymes are presented in Fig. 5B.
The strategy of Fig. 5A takes advantage of the HIV-1 RNase H
mutant p66E478Q/p51 (24, 25), which is completely devoid of
Mg2+-dependent RNase H activity (Fig.
3D) but retains full DNA polymerase activity. Initially,
wild-type HIV-1 RT is permitted to select the PPT primer (in this
experiment only, the HIV-1 enzymes and 3 PPT were used), after which
the nucleic acid substrate is recovered by phenol extraction and
ethanol precipitation. Precleaved substrate is subsequently offered to
each RNase H mutant in the presence of dNTPs, and (+) strand synthesis
was determined. In the converse experiment, the HIV enzymes are
required to select the 3
PPT, which is recovered and offered as
substrate to p66E478Q/p51 RT. The RNase deficiency in this mutant
ensures that it cannot contribute to PPT selection.
Fig. 5B, panel [i], assesses the ability of each mutant to
initiate (+) strand synthesis from a PPT selected by wild-type HIV-1
RT. Following alkali treatment, all enzymes showed identical levels of
DNA synthesis, which is in keeping with the data of Fig. 1,
illustrating that neither substitution affects initiation from the PPT.
However, when p66E478Q/p51 RT is required to extend (+) strand primers
selected by Asp549 mutants, major differences become
apparent (Fig. 5B, panel [ii]). As expected, PPT extension
into (+) strand DNA is highly efficient with wild-type p66/p51. In
contrast, the efficiency of synthesis drops approximately 10-fold when
p66E478Q/p51 RT is required to extend the PPT selected by mutant
p66D549N/p51, and no product was found when p66D549A/p51 RT was
required to select the PPT. Fig. 5C, panels [i] and
[ii], represent equivalent reactions as Fig.
5B, with the exception that alkali treatment was eliminated, allowing visualization of the PPT/(+) strand DNA chimera and its removal by the retroviral enzyme. In Fig. 5C, panel [i],
wild-type RT is clearly efficient in PPT removal, with almost 90% of
the (+) strand product represented by (+) strand DNA. These proportions are reversed with mutant p66D549N/p51, i.e. the (+) strand
product is predominantly the PPT-DNA chimera, whereas (+) strand DNA
synthesized by p66D549A/p51 RT is exclusively an RNA/DNA chimera. Data
of Fig. 5C, panel [i], also indicates that the (+) strand
primer selected by wild-type RT is restricted almost exclusively to PPT sequences, i.e. through precise RNase H-mediated hydrolysis
at both its 5 and 3
extremities. This is not the case in Fig.
5C, panel [ii], where p66E478Q/p51 RT extends PPT
sequences preselected by Asp549 mutants. Under these
circumstances, a unique PPT primer is selected by p66D549N/p51 RT but
is some 6-7 nt longer at its 5
terminus, corresponding to a stretch
of contiguous Us immediately 5
to the PPT (Fig. 5D).
Finally, it is interesting to note that the difference between this
novel cleavage site and the PPT 5 terminus is seven to eight nt, which
is consistent with the extremity of the "directional processing"
RNase H activity relative to the position of initial endonucleolytic
cleavage (23, 27, 28). Furthermore, wild-type RT cleaves at several
positions immediately 5
to the PPT (Fig. 5C, [ii]) which
cannot be resolved by Asp549 mutants deficient in
directional processing activity (Fig. 5C, [i], and Fig.
3). These observations suggest that PPT selection may be subject to
fine control by either sequences or structural elements of the RNA/DNA
replication intermediate in the immediate 5
vicinity. This model would
envisage 5
RNA-directed RNase H activity of RT (31, 32) as a mechanism
allowing the replication machinery to "walk" along and hydrolyze
the fragmented RNA genome by a series of endonucleolytic and
directional processing steps until it encounters the 3
PPT. Once the
last of these combined hydrolysis events is accomplished, the free PPT
5
terminus directs a final RNase cleavage event to liberate the 3
terminus for extension into (+) strand DNA. Such a model remains
speculative and is currently under evaluation.
An improved understanding of RNase H-mediated events is critical
to defining novel therapeutic targets to combat HIV infection and
acquired immunodeficiency syndrome. However, despite the wealth of data
available on the DNA polymerase domain of HIV RT, the RNase H domain
has received surprisingly little attention, despite documentation of
its absolute requirement for replication (36, 37). One contributing
factor may have been a lack of defined heteropolymeric substrates,
reflecting critical steps in retroviral replication that demand
considerable precision of the RNase H domain. This problem has largely
been resolved with the advent of chemically synthesized
oligonucleotides, which have provided model systems to evaluate events
such as DNA strand transfer, PPT selection, and tRNA primer removal. In
combination with a program of site-directed mutagenesis, these
strategies should provide a fine dissection of RNase H-mediated events
and their potential for therapeutic intervention. In this study, we
elected to evaluate the consequences of altering a highly conserved
residue of the retroviral RNase H domain, Asp549, on
overall catalysis, as well as selection and utilization of the 3 PPT
primer. Analyzing two structurally related lentiviral RTs has also
provided insights into subtle differences in RNase H-mediated
hydrolysis.
A clear difference when comparing p66/p51 heterodimers of HIV-1 and
EIAV RT is the stringency of RNase H cleavage (Figs. 2 and 3). This is
best exemplified by use of a 3-labeled substrate, which reveals that
HIV-1 RT cleaves the RNA-DNA hybrid at multiple sites centered around
20/
21, whereas its EIAV counterpart cleaves with considerably
greater precision at positions defined by the spatial separation of the
DNA polymerase and RNase H catalytic centers (30). One explanation for
these differences might lie in observations of Palaniappan et
al. (32) that when presented with an RNA-DNA hybrid within which
the DNA 3
end is unannealed, the position of RT (and hence the RNase H
domain) is defined by the first hybrid base pair, which is upstream
from the primer terminus. Partial hybridization or "breathing" of
the duplex at the primer terminus could therefore lead to variability
in enzyme localization, thereby inducing a corresponding change in
RNase H cleavage specificity. In contrast, the EIAV enzyme may bind and
stabilize the unannealed primer terminus, resulting in predominant cleavage around positions
16/
17. Alternatively, both enzymes may
locate themselves in a similar manner at the DNA polymerase catalytic
center but differ in the manner in which the RNA-DNA hybrid is
accommodated within the RNase H domain.
The experiment of Fig. 2, which restricts RNase H activity to
visualization of the primary site of endonuclease activity, indicates
little difference between wild-type RT and enzymes carrying the
Asp549 Asn mutation, which would be in keeping with a
recent observation of Hostomsky et
al.2 However, we have documented
several HIV-1 mutants that support efficient endonuclease or
synthesis-dependent RNase H activity but are severely
compromised for the subsequent stepwise template hydrolysis as far as
nt
8 (12, 29, 34). This scenario is again apparent following extended
hydrolysis of a 3
-labeled hybrid or through use of a 5
-labeled RNA
template, the latter of which monitors events subsequent to initial
endonucleolytic cleavage. Markedly slower hydrolysis kinetics of
p66D549A/p51 HIV-1 RT also highlight the reduced stringency of
cleavage, where products corresponding to cleavage between positions
16 and
24 are clearly evident. The highly conserved
Asp549 thus appears to make an important contribution to
RNase H function, although retention of low-level activity in
Asp549
Ala mutants of HIV-1 and EIAV RT suggests that,
in contrast to Glu478, it is not absolutely required for
metal ion co-ordination (24, 25). A plausible alternative might be
assisting in maintaining the integrity of
-helix E
of the RNase H
domain, which spans Gly544-Gly555 in the HIV-1
enzyme (30). Structural data from the isolated RNase H polypeptide (8)
indicates hydrogen bonding between side chain residues of
Asp549 and Ser553 in HIV-1 RT. Loss of hydrogen
bonding function may conceivably alter the geometry of
-helix E
,
which, together with
-strand 5
, is proposed to constitute a
"floor and wall" to accommodate RNA-DNA hybrids in the RNase H
domain (10, 11). Interestingly, when the equivalent residue of the duck
hepatitis B virus polymerase gene (Asp755) is altered to
Glu, a substantial difference in DNA synthesis was observed (38),
suggesting that alterations to this RNase H domain mutant can also
impact upon the DNA polymerase catalytic center.
The importance of the synthesis-independent or directional processing
RNase H activity has been illustrated by the inability of mutants
lacking this property to support DNA strand transfer (12, 29). A second
feature of impaired processing function can be highlighted when another
key step in retroviral replication is evaluated, i.e.
initiation of (+) strand synthesis from the 3 PPT primer. Data of Fig.
5 indicate p66D549N/p51 and p66D549A/p51 HIV-1 RT catalyze equivalent
levels of (+) strand synthesis as wild-type enzyme from a preselected
PPT. However, when these mutants are required to select the PPT for
extension by a second enzyme, major reductions in (+) strand product
are evident. Furthermore the HIV-1 mutants are virtually unable to
remove the PPT RNA primer from nascent (+) strand DNA, an event the
parental enzyme accomplishes efficiently. Recent studies have indicated
that the RNA-DNA PPT-containing hybrid has unusual structural features
that may contribute to its resistance to RNase H-mediated hydrolysis.
Given the importance for accurate initiation of (+) strand synthesis in
defining sequences at the 5
long terminal repeat terminus to be
recognized by the retroviral integration machinery, it is not
unreasonable that RT has adapted to sequence and/or structural features
of the PPT. Should the manner in which the PPT RNA/(
) strand DNA
hybrid is accommodated be affected by perturbations to the geometry of
the RNase H domain, this would be magnified in a PPT
selection/extension assay. Stated differently, PPT selection/extension
assays of Figs. 4 and 5 provide an additional example of the necessity
for highly selective assay systems to finely dissect multiple functions
of the retroviral polymerase.
Data with the HIV-1 mutant p66D549N/p51 may also provide clues to the
mechanism of PPT selection. Fig. 5C, [ii], indicates that
the major 3 PPT primer selected by wild-type HIV-1 RT is ~14 nt, the
5
terminus of which lies within the -A-A-A-A- sequence at the 5
end
of the PPT. The presence of additional (+) strand products with
identical 3
termini places the 5
terminus of these primers within one
to seven nt of the PPT. Invoking the model of DeStefano et
al. (31), RNA 5
-directed RNase H cleavage might be a means of
directing the replication machinery to within seven nt 5
to the PPT.
At this stage, although RT located over the RNA 5
terminus is unable
to cleave within the PPT, the spatial distance between its catalytic
centers would permit cleavage at the PPT 3
terminus (a distance of
~20 nt). A conformational rearrangement, possibly promoted by unusual
structural features of the PPT RNA-DNA duplex, would relocate RT with
its polymerase catalytic center over the PPT 3
OH. Finally, the
directional processing activity of this enzyme would remove extraneous
sequences 5
to the PBS. Mutant p66D549N/p51 would follow the same
route but simply make a series of endonucleolytic cuts. However, when
this mutant selects the PPT 3
OH and undergoes a conformational
rearrangement, a lack of processing activity would leave these
additional seven nt on the PPT, as we observed experimentally.