Mutation of Amino Acids in the Connection Domain of Human Immunodeficiency Virus Type 1 Reverse Transcriptase That Contact the Template-Primer Affects RNase H Activity

Abstract

The crystal structure of human immunodeficiency virus type 1 (HIV-1) reverse transcriptase in a complex with an RNA-DNA template-primer identified amino acids in the connection domain that make specific contacts with the nucleic acid. We analyzed the effects of mutations in these amino acids by using a one-round HIV-1 vector. Mutations in amino acids in the connection domain generally had small effects on virus titers. To determine whether the mutations affected the level of RNase H activity or the specificity of RNase H cleavage, we used the two-long-terminal-repeat circle junction as a surrogate for the ends of linear viral DNA; specific RNase H cleavages determine the ends of the viral DNA. Several of the mutations in the connection domain affected the frequency of the generation of viral DNAs with aberrant ends. The mutation H361A had the largest effect on the titer and on the generation of DNAs with aberrant ends. H361 contacts the phosphate backbone of the nucleic acid in the same location as amino acid Y501 in the RNase H primer grip. Mutations at Y501 have been shown to decrease the virus titer and affect the specificity of RNase H cleavage. H361A affected the frequency of the generation of linear viral DNAs with aberrant ends, but in general the connection domain mutations had subtle effects on the efficiency of RNase H cleavage. The results of this study suggest that, in addition to its primary role in linking the polymerase and RNase H domains, the connection subdomain has a modest role in binding and positioning the nucleic acid.

Human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT) is the virally encoded enzyme that converts the single-stranded RNA viral genome into double-stranded DNA. The conversion of the RNA genome into DNA is accomplished through the collaboration of a DNA polymerase that can use either RNA or DNA as a template for DNA synthesis and an RNase H that cleaves RNA if (and only if) it is present in an RNA-DNA duplex (3, 6, 7, 28). Both the polymerase and the RNase H are required for the conversion of the viral RNA genome into double-stranded DNA (17, 19, 23, 25). RNase H degrades the viral RNA genome and removes the tRNA^Lys3 used as the primer for minus-strand DNA synthesis (5, 15, 20, 27). RNase H also generates and removes the polypurine tract (PPT) primer used to initiate plus-strand DNA synthesis (14, 16). The removal of the RNA primers defines the ends of the viral DNA; these specific ends are the substrates for the integration of the viral DNA into the host cell genome, which is carried out by the virally encoded enzyme integrase (3).

The crystal structure of HIV-1 RT in a complex with an RNA-DNA primer template whose sequence was derived from the PPT has provided information on the interactions of RT and its template-primer (18). The complex of HIV-1 RT and an RNA-DNA duplex differs from the complex of HIV-1 RT and a DNA-DNA duplex; however, there are considerable similarities (8, 18). In a complex with RT, both DNA-DNA and RNA-DNA duplexes have a bend of about 40° 5 to 9 bp from the polymerase active site. In the DNA-DNA-containing complex, the bend in the nucleic acid is associated with a transition from A-form to B-form geometry; a similar transition has been observed in DNA-DNA substrates bound to other polymerases.The majority of the contacts between the enzyme and the nucleic acid are near the polymerase active site, where the nucleic acid geometry is closer to A-form. Farther from the polymerase active site, the geometry of the DNA-DNA duplex is closer to B-form. The RNA-DNA duplex adopts a geometry between A-form and B-form (H-form). During viral replication, the PPT must remain resistant to RNase H degradation long enough to serve as the primer for plus-strand DNA synthesis. In the RT-RNA-DNA complex, the 4 bp adjacent to the RNase H active site are properly base paired; however, there are two unpaired bases in the RNA-DNA duplex, one in the primer strand and the other in the template strand. These two unpaired bases shift the base pairing out of and then back to the normal register; this unusual structure may contribute to the PPT primer's resistance to RNase H degradation. In the RT-RNA-DNA complex, there are additional contacts between the nucleic acid and the protein relative to the RT-DNA-DNA complex. HIV-1 RT interacts with several of the 2′ hydroxyl groups of the RNA template. Near the RNase H active site, a network of amino acids interacts with the DNA primer strand. These amino acids form the RNase H primer grip, which plays a role in positioning the primer strand relative to the RNase H active site and helps determine cleavage specificity.

The connection domain of HIV-1 RT has not been well characterized. It plays a structural role in the protein, linking the polymerase and RNase H domains of the protein and aligning the two domains so that both active sites can simultaneously contact an RNA-DNA substrate. It is likely that retroviral RTs were derived by the fusion of a polymerase and an RNase H, which were originally physically separate enzymes; a structural role for the connection domain fits this evolutionary hypothesis (7). However, the crystal structure of HIV-1 RT in complex with an RNA-DNA template-primer derived from the PPT shows that there are amino acids in the connection domain that make specific contacts with the nucleic acid (18). This finding suggests that amino acids in the connection domain may also play a role in helping to bind and properly position the nucleic acid substrate in the vicinity of the RNase H active site. To test this hypothesis, we generated a series of mutations in the amino acid residues in the connection domain of HIV-1 RT that contact the DNA primer or the RNA template. There are nucleic acid contacts involving both the p66 and p51 subunits. Amino acids K395 and E396 in the p51 subunit and amino acids A360, H361, and T362 in the p66 subunit contact the DNA primer strand; K390 in the p51 subunit contacts the RNA template strand. We wanted to determine if mutations in any of these amino acids would affect the replication of the virus. In particular, we wanted to know if the mutations would alter the binding and proper positioning of the RNA strand, which might affect the efficiency and specificity of RNase H cleavage. The ends of the linear viral DNA are determined by specific RNase H cleavages; the cleavage specificity of RNase H was monitored in vivo using the two-long-terminal-repeat (2-LTR) circle junction sequence as a surrogate for the ends of the viral DNA. The 2-LTR circles arise from the joining of the ends of unintegrated viral DNA, presumably by host cell ligases (12). We used the sequence of the 2-LTR circle junctions to characterize the effects of mutations in the RNase H domain of HIV-1 RT that contact the RNA primer and DNA template in the vicinity of the RNase H active site. There are mutations in the RNase H domain that increase the frequency of aberrant viral DNA ends (10). We analyzed the 2-LTR circle junction sequences derived from infecting cells with wild-type virus and viruses containing mutations in amino acids in the connection domain that contact the RNA-DNA template-primer. The results indicate that most of the mutations in the connection domain have relatively small effects on the virus titer; however, one of the mutations (H361A) decreased the virus titer to about 25% of that of the wild-type virus. The sequence of the 2-LTR circle junctions indicates that most of the mutations that contact the primer strand have relatively subtle effects on RNase H activity and specificity. However, the mutation that had the largest effect on the titer (H361A) also had a significant effect on RNase H cleavage; this amino acid contacts the nucleic acid through the p66 subunit. A mutation in one of the amino acids that contacts the nucleic acid through the p51 subunit (E396A) also had a significant effect on RNase H cleavage, but this mutation had a smaller effect on the titer than did H361A.

Effects of mutations in the connection domain on virus titers.

Amino acids in the connection domain of HIV-1 RT that contact either the RNA template or the DNA primer strand in the vicinity of the RNase H active site were identified in the crystal structure of HIV-1 RT in a complex with an RNA-DNA duplex whose sequence was derived from the PPT (18) (Fig. 1). K390, K395, and E396 contact the nucleic acid through the p51 subunit; K395 and E396 contact the DNA primer strand, while K390 contacts the RNA template strand. G359, A360, H361, and T362 contact the DNA primer strand through the p66 subunit. As has been previously described, site-directed mutagenesis has been used to introduce amino acid substitutions in amino acids at positions in the RNase H domain that contact the DNA primer strand and RNA template strand in an HIV-1-based vector, pNLNgoMIVR⁻E⁻.HSA (Fig. 2A) (10). The vector was derived from the NL4-3 isolate of HIV-1; it does not express the viral env or vpr gene (9). The vector expresses the gene for a murine cell surface marker, the heat-stable antigen (HSA) gene (hsa) from the nef reading frame (26). 293 cells were cotransfected with the vectors and with pHCMV-g, a plasmid that expresses the vesicular stomatitis virus glycoprotein (1, 29). This cotransfection produced virus-containing supernatants that were harvested and used to infect HOS cells. Forty-eight hours after infection, the cells were labeled with antibody directed against HSA and the virus titer was determined by using a fluorescence-activated cell sorter (FACS). The effects of the mutations in the connection domain of RT on virus titers are shown in Fig. 2B. The T362A, K390A, and K395A mutations decreased the virus titers less than twofold. The G359S, A360D, and E396A mutations decreased the virus titers to about 50% of that of the wild-type virus. The mutation H361A decreased the virus titer to about 25% of that of the wild-type virus, suggesting that the contacts involving the amino acid at position 361 are more important for virus infectivity than the contacts involving other amino acids that we analyzed in the connection domain. H361 contacts the phosphate backbone of the DNA primer strand at position −5 relative to the RNase H cleavage site. Y501 in the RNase H domain of RT contacts the same phosphate as H361; it has previously been shown that the Y501A mutation affects the viral titer and RNase H cleavage specificity (10). The other contacts between amino acids in the connection domain and the DNA primer strand are located farther away from the RNase H active site than the contact made by H361.

FIG. 1.

Diagram of the structure of HIV-1 RT in complex with an RNA-DNA template-primer whose sequence is derived from the PPT. The figure shows the portion of the structure near the RNase H active site; the contacts between the nucleic acid and the RNase H domain and the connection domain are indicated. The RNA template strand is shown in turquoise; the DNA primer strand is shown in purple. The scissile phosphate is designated with a green arrow pointing to the phosphate. The RNase H domain is shown in orange; the connection subdomain is shown in yellow. The thumb is in green. The p51 subunit is in gray. Amino acids in the p51 subunit that contact the nucleic acids are marked with an asterisk next to the position number. Amino acids contacting the nucleic acid are labeled. Contacts between individual amino acids and the nucleic acid are shown in red. The RNA strand is numbered relative to the site of cleavage.

FIG. 2.

Mutations in the connection domain affect the viral titer. (A) The retroviral vector pNLNgoMIVR-E-.HSA has been previously described (9). Briefly, the vector was derived from the NL4-3 isolate of HIV-1, except for the portion of RT from amino acid 42 to amino acid 427, which was derived from the BH10 strain of HIV-1. The env and vpr genes are inactivated in the vector; consequently, the vector undergoes only a single cycle of replication. The murine hsa gene is expressed in the nef reading frame. HSA is a cell surface marker that allows for the determination of virus titers by labeling infected cells with anti-HSA antibody and then subjecting the cells to FACS analysis. (B) Effects of mutations in the connection domain on the virus titers. Virus titers, normalized to the amount of p24 antigen, are shown relative to that of the wild-type virus (WT) on the y axis; the RT mutations are shown along the x axis. Transfection, infections, and the determination of the virus titers were performed as previously described (9). These results are the averages of the results of three independent experiments.

Effects of mutations in the connection domain on the generation of the ends of the viral DNA.

The linear viral DNA that is the product of reverse transcription can be joined by host cell ligases to form a 2-LTR circle junction (12). To determine the effects of the mutations in the connection domain on the synthesis of the ends of the linear viral DNA, we used the 2-LTR circle junction as a surrogate. 2-LTR circle junctions were amplified from cells infected with viruses containing wild-type RT, with viruses containing reduced RNase H activity (viruses generated by cotransfecting 293 cells with a mixture in which 10% of the plasmids encoded wild-type RT and 90% of the plasmids encoded RT with the active-site mutation E478Q in RNase H), and with the connection domain mutants. The frequency of defects in the 2-LTR circle junction was determined by sequencing approximately 100 circle junction clones from cells infected with the wild-type virus, the virus with reduced RNase H activity, and each of the connection domain mutants. A summary of the sequence analysis is shown (Fig. 3). The data were analyzed by log linear categorical analysis, contingency table analysis, and related methods. I-by-J circle junction-by-mutant tables were decomposed through the use of likelihood ratio chi-square statistics into independent partitions to show associations between circle junction and/or mutant groupings and categories. Subsets of pertinent groupings were further studied with traditional two-by-two chi-square analyses and Fisher's exact tests. The proportion of 2-LTR circle junctions having a consensus sequence is shown in the top row (Fig. 3). For the wild-type virus and for virions with reduced RNase H activity, the results of two independent experiments are shown; the data for the first experiment are taken from previously published data (10). The fraction of consensus 2-LTR circle junction sequences was 0.60 from samples containing wild-type RT in each of two independent experiments. Statistical analysis of the two independent experiments indicated that the populations are not different. The 2-LTR circles derived from infecting cells with the virions containing reduced RNase H activity had consensus 2-LTR circle junctions at fractions of 0.54 and 0.48. Statistical analysis of the two experiments with virions with reduced RNase H activity indicated that these populations are not different. The mutations H361A and E396A (a position that contacts the nucleic acid through the p51 subunit) decreased the fractions of consensus circle junctions to 0.37 and 0.33, respectively. These decreases in the proportion of consensus 2-LTR circle junctions are highly significant (chi-square analysis; P < 0.001). The other mutations did not have a statistically significant effect on the proportion of 2-LTR circle junctions that had a consensus sequence.

FIG. 3.

Sequences at the 2-LTR circle junctions from wild-type and mutant viruses. The top of the figure shows the linear viral DNA that is the product of reverse transcription. The PBS is indicated by a white box, the leader sequence downstream of the PBS is shown with gray dots, the PPT is a box with black horizontal bars, the U-tract is a gray box, and the sequences immediately upstream of the U-tract are shown by the box with diagonal bars. This linear viral DNA can be ligated by cellular enzymes in the nucleus of an infected cell to form a 2-LTR circle. A consensus circle junction is shown. The underlined “T” is derived from the riboA located at the end of the minus strand; it is derived from the tRNA^Lys3 primer used by HIV-1 RT to initiate minus-strand DNA synthesis. The 2-LTR circle junctions from cells infected with wild-type virus, with virions containing reduced RNase H activity, or with strains with mutations in the connection domain were amplified by using PCR, cloned into a cloning vector, and sequenced as previously described (10). Approximately 100 circle junctions were analyzed for each of the mutants. The boxes show the fraction of 2-LTR circles in the category specified at the left of the figure. Statistically significant increases, compared to levels with the wild-type virus, are designated by an asterisk.

If the PPT is properly generated but is not removed (or is incompletely removed) by RNase H, then a 2-LTR circle junction containing a simple PPT insertion is generated. Simple PPT insertions can also be generated if an improper primer is used to initiate plus-strand DNA synthesis. There was one simple PPT insertion in the 2-LTR circle junctions obtained with the wild-type vector and one in the circle junctions from the virions that contained reduced RNase H activity (Fig. 3). This finding shows that reducing RNase H activity approximately 10-fold does not lead to a measurable increase in PPT insertions. The 2-LTR circle junctions derived from infecting cells with the A360D mutant did not contain any simple PPT insertions; however, the fraction of 2-LTR circle junctions containing a simple PPT was increased with the other connection domain mutants. The largest effects were with the K390A and K395A mutants. For both of these mutants, the fraction of simple PPT insertions was 0.06; however, this increase is not statistically different than that with the wild-type virus (P > 0.05). The G359S, H361A, and E396A viruses gave rise to 2-LTR circles that contained simple PPT insertions at frequencies of 0.05, 0.05, and 0.04, respectively. For the T362A mutation, the fraction was 0.02. None of these increases in the proportion of simple PPT insertions are statistically significant.

2-LTR circle junctions containing PPTs with short flanking sequences could arise by the improper generation of the PPT. If the aberrant PPT is then retained as a result of a failure of RNase H to remove the primer, a PPT plus a short flank (the U-tract) is found at the 2-LTR circle junction. This type of insertion can also arise from plus-strand priming by an aberrant primer. PPTs with short flanking sequences were not observed from 2-LTR circle junctions derived from infecting cells with virions containing wild-type RT or virions with reduced RNase H activity. Aberrant 2-LTR circle junctions containing short flanking sequences were observed at a low frequency (0.01 in each case) after cells were infected with the G359S, A360D, K395A, and E396A mutants. The low frequency of 2-LTR circle junctions with PPTs and flanking sequence observed with the connection domain mutants differs from the larger increases that we found for the RNase H primer grip mutants, suggesting that the contacts with amino acids in the connection domain amino acids are less important than the contacts with amino acids in the RNase H domain in determining RNase H cleavage specificity.

A PPT with a long flanking sequence can arise if RT uses (and removes) a primer upstream of the PPT to initiate plus-strand DNA synthesis or if an aberrant primer is used but not removed. The resulting 2-LTR circle junction contains the PPT with a considerable amount of flanking sequence (typically >20 nucleotides upstream of the PPT). This type of aberrant 2-LTR circle junction was observed at a frequency of 0.02 when RNase H activity in the virions was reduced, at 0.02 with the A360D mutant, and at 0.01 for each of the G359S, K390A, and K395A mutants. These differences are not statistically significant.

If RNase H fails to correctly recognize the 3′ ends of the PPT and cleaves within U3 to generate the 3′ end of the PPT primer and then the PPT primer is removed, there will be a deletion in the 5′ end of U3. We analyzed the frequency of small deletions in U3 or U5 to determine if the connection domain mutants incorrectly generate (and remove) the U3 primer (Table 1). It has previously been shown that the RNase H primer grip mutants caused a dramatic increase in the frequency of 5-bp deletions in the U3 region (10); however, no similar increases in 5-bp deletions were observed with the connection domain mutants. Circle junctions derived from infections with the K390A mutant showed an increase in the frequency of smaller deletions (2 to 3 bp) compared to the frequency with circle junctions from infections with either the wild type or mutants with reduced RNase H activity. However, the differences are small, and it appears that the K390A mutation does not cause significant cleavage within U3.

TABLE 1.

Frequency of small deletions in U5 or U3

Size of deletion (bp)	Frequency of deletion in U5 or U3 in:
	WTa		Reduced-RNase H activity		G359S mutant		A360D mutant		H361A mutant		T362A mutant		K390A mutant		K395A mutant		E396A mutant
	U5	U3	U5	U3	U5	U3	U5	U3	U5	U3	U5	U3	U5	U3	U5	U3	U5	U3
1	0.07	0.01	0.01	0.01	0.06	0.01	0.03		0.05		0.02		0.04		0.02	0.01	0.04	0.01
2	0.02	0.01	0.01		0.01	0.01	0.02		0.06			0.02	0.04	0.04	0.06	0.01	0.04	0.04
3		0.01		0.02		0.01				0.02		0.01		0.04		0.01	0.01
4				0.01										0.01
5

^aWT, wild-type virus.

When RNase H activity is reduced in the virion, the number of 2-LTR circle junctions containing a tRNA-derived insertion at the circle junction is dramatically increased (about 10-fold) (Fig. 3). All of the connection domain mutants tested showed increases in the fraction of 2-LTR circle junctions containing tRNA-derived insertions. Except with the G359S and K390A mutants, these increases were statistically significant (P < 0.05). Circle junctions obtained from infections with virions with reduced RNase H activity also showed a statistically significant increase in tRNA insertions. Of the mutations tested, the most dramatic increases were observed with H361A and E396A; these mutations had larger effects on the virus titer than did the other mutations in the connection domain. These data suggest that mutations in the connection domain decrease the levels of RNase H activity, which would explain the failure to remove the tRNA primer. The A360D and T362A mutations did not cause a corresponding increase in the fraction of 2-LTR circle junctions containing a PPT insertion. These data are consistent with the notion that tRNA removal is more difficult than PPT removal in vivo; a similar result was obtained with the RNase H primer grip mutants (10).

We also observed 2-LTR circle junctions containing insertions of tRNAs with the leader sequences immediately adjacent to the primer binding site (PBS). These insertions were observed at a frequency of 0.01 for the wild type and for the G359S, A360D, H361A, and E396A mutants. The frequency was increased to 0.04 for the K390A mutant; however, this increase is not statistically significant. We also observed aberrant junctions of this type with the RNase H primer grip Q475A mutant and the N474A + Q475A double mutant. A circle junction with tRNA sequences and the leader could be generated by the retention of the tRNA on the end of the minus-strand DNA (Fig. 4). The retained tRNA can be copied a second time, and the resulting plus-strand DNA can undergo strand invasion to the PBS of the viral DNA in an inter- or intramolecular fashion, which allows RT to copy the sequences adjacent to the PBS (Fig. 4).

FIG. 4.

Model for the capture of tRNA insertions with leader sequences. (i) The dotted gray line represents degraded viral RNA. The solid gray line represents the PPT primer. The solid black line represents viral DNA. The arrowhead represents the RNase H cleavage that removes the PPT primer. The white arrows represent the tRNA primer. (ii) Plus-strand DNA transfer occurs by using the complementary sequences at the PBS. If the tRNA primer is not removed by RNase H during the synthesis of strong-stop plus-strand DNA, it can be copied a second time when the plus strand is extended. (iii) The removal of the tRNA by RNase H generates a linear viral DNA with a plus-strand overhang (shown in red). (iv) The plus-strand overhang is complementary to the tRNA sequence. (v) This plus-strand DNA can undergo strand invasion of the PBS sequence shown in blue. As drawn, it is assumed that DNA synthesis occurs from both viral RNAs; intramolecular strand invasion may also occur. RT extends the DNA that has invaded. Resolution of the products generates a linear viral DNA containing the tRNA and a large leader sequence.

Together these results indicate that the mutations in amino acids in the connection domain that make specific contacts with the nucleic acid reduce the level of RNase H activity. The simple explanation is that the mutant RTs bind the nucleic acid substrate more weakly than does the wild-type enzyme; this weak binding leads to a reduction in RNase H cleavage. It appears, in general, that there is a larger effect on the level of RNase H activity than on cleavage specificity. Two of the mutations, H361A and E396A, generated a higher percentage of aberrant 2-LTR circle junctions and specifically increased PPT insertions than did other mutations; however, this increase was not statistically significant. The relatively small effects seen with the individual mutations suggest that the binding and proper positioning of the nucleic acid substrate depends on the cooperative effects of RT-nucleic acid contacts in the connection domain rather than on only one or a few specific interactions. The data also demonstrate a direct role for the p51 subunit in binding nucleic acid. K395 and E396 contact the nucleic acid through the p51 subunit; both the K395A and E396A mutations appear to affect the extent of RNase H cleavage.

As has previously been mentioned, retroviral RTs likely evolved from the fusion of two different proteins, one containing a polymerase and the other containing an RNase H (7). The fusion of the two enzymes into a single protein allowed RNase H to make the highly specific cleavages necessary for retroviral reverse transcription. Retroviral RNase H domains are structurally and functionally related to those of Escherichia coli RNase H. The E. coli RNase H has a structural element called the basic loop, which is necessary for enzymatic activity. This basic loop helps the enzyme bind its nucleic acid substrate (11, 22). Murine leukemia virus (MLV) RNase H contains a related basic loop that is essential for virus replication (13, 24). In in vitro assays, recombinant MLV RT lacking the basic loop was deficient in both polymerase and RNase H activities, presumably because it binds the nucleic acid weakly (2). Because it retains the basic loop, the separately expressed MLV RNase H domain retains enzymatic activity. Unlike MLV RT, the RNase H domain of HIV-1 RT has lost the basic loop. If the RNase H domain of HIV-1 RT is separately expressed, it is properly folded but it lacks enzymatic activity. This lack of enzymatic activity is the result of poor substrate binding; activity can be restored to the isolated RNase H domain of HIV-1 RT by the insertion of a basic loop (11, 22). In addition, expressing RNase H domains containing increasing amounts of the N-terminal extensions that included the connection and thumb domains increased the activity of the RNase H domain (21). This finding suggests, in the context of HIV-1 RT, that the lack of a basic loop in the RNase H domain is compensated for, at least in part, by contacts in the polymerase domain. Examination of the crystal structure shows that the majority of nucleic acid contacts involve the fingers, thumb, and palm of the p66 subunit (4, 8, 18). However, there are contacts in the connection domains of both p51 and p66; these connection domain contacts are in positions similar to those that involve the basic loop. We suggest that in HIV-1 RT some of the connection domain contacts may substitute for the loss of the basic loop contacts.