`1420-682X:00:101408-15 $ 1.50(cid:27)0.20:0
`© Birkha¨user Verlag, Basel, 2000
`
`Review
`
`Molecular mechanisms of HIV-1 resistance to nucleoside
`reverse transcriptase inhibitors (NRTIs)
`N. Sluis-Cremer, D. Arion and M. A. Parniak*
`
`Lady Davis Institute for Medical Research and McGill University AIDS Centre, Sir Mortimer B. Davis-Jewish
`General Hospital, 3755 Cote Ste-Catherine Road, Montreal, Quebec H3T 1E2 (Canada), Fax (cid:27)1514 340 7502,
`e-mail: mparniak@ldi.jgh.mcgill.ca
`
`Received 16 December 1999; received after revision 3 April 2000; accepted 3 April 2000
`
`Abstract. Nucleoside reverse transcriptase inhibitors
`typic mechanisms have been identified or proposed to
`such as 3%-azido-3%-deoxythymidine, 2%,3%-
`(NRTIs),
`account for HIV-1 RT resistance to NRTIs. These
`dideoxyinosine and 2%,3%-dideoxy-3%-thiacytidine, are ef-
`mechanisms include alterations of RT discrimination
`fective inhibitors of human immunodeficiency type 1
`between NRTIs and the analogous dNTP (direct effects
`(HIV-1)
`replication. NRTIs
`are deoxynucleoside
`on NRTI binding and:or incorporation), alterations in
`triphosphate analogs, but lack a free 3%-hydroxyl group. RT-template:primer interactions, which may influence
`Once NRTIs are incorporated into the nascent viral
`subsequent NRTI incorporation, and enhanced removal
`of the chain-terminating residue from the 3% end of the
`DNA, in reactions catalyzed by HIV-1 reverse tran-
`primer. These different resistance phenotypes seem to
`scriptase (RT), further viral DNA synthesis is effectively
`terminated. NRTIs should therefore represent the ideal
`correlate with different sets of mutations in RT. This
`review discusses the relationship between HIV-1 drug
`antiviral agent. Unfortunately, HIV-1 inevitably devel-
`resistance genotype and phenotype, in relation to our
`ops resistance to these inhibitors, and this resistance
`correlates with mutations in RT. To date, three pheno-
`current knowledge of HIV-1 RT structure.
`
`Key words. Human immunodeficiency virus type 1; reverse transcriptase; nucleoside reverse transcriptase in-
`hibitors; DNA polymerization; chain termination; antiviral drug resistance; phosphorolysis; pyrophosphorolysis.
`
`Introduction
`
`Retroviruses such as the human immunodeficiency virus
`(HIV) are RNA viruses that replicate through a double-
`strand DNA intermediate. This novel viral replication
`cycle requires that retroviruses carry a specific enzyme,
`reverse transcriptase (RT), since there are no cellular
`enzymes that can convert single-strand RNA into dou-
`ble-strand DNA. RT is a DNA polymerase that can
`copy both DNA templates (like cellular enzymes) and
`RNA templates (unlike cellular enzymes). HIV RT dif-
`
`* Corresponding author.
`
`fers from cellular DNA polymerases in two additional
`respects. First, HIV RT readily utilizes many chemically
`altered analogs of the normal deoxynucleoside triphos-
`phate (dNTP) DNA polymerase substrates. Second,
`HIV RT lacks a formal %proofreading% activity. These
`characteristics are important from a pharmaceutical
`focus, and direct the use of nucleoside analog inhibitors
`as anti-HIV pharmaceuticals. As of January 2000, six of
`the current FDA-approved anti-HIV drugs are nu-
`cleoside reverse transcriptase inhibitors (NRTIs) [1].
`NRTIs are analogs of the normal dNTP substrates of
`DNA polymerases, with important modifications (fig.
`1). The 2%,3%-dideoxynucleosides such as ddC and ddI
`
`Columbia Ex. 2080
`Illumina, Inc. v. The Trustees
`of Columbia University in the
`City of New York
`IPR2020-00988, -01065,
`-01177, -01125, -01323
`
`
`
`CMLS, Cell. Mol. Life Sci. 57 (2000) 1408–1422
`1420-682X:00:101408-15 $ 1.50(cid:27)0.20:0
`© Birkha¨user Verlag, Basel, 2000
`
`Review
`
`Molecular mechanisms of HIV-1 resistance to nucleoside
`reverse transcriptase inhibitors (NRTIs)
`N. Sluis-Cremer, D. Arion and M. A. Parniak*
`
`Lady Davis Institute for Medical Research and McGill University AIDS Centre, Sir Mortimer B. Davis-Jewish
`General Hospital, 3755 Cote Ste-Catherine Road, Montreal, Quebec H3T 1E2 (Canada), Fax (cid:27)1514 340 7502,
`e-mail: mparniak@ldi.jgh.mcgill.ca
`
`Received 16 December 1999; received after revision 3 April 2000; accepted 3 April 2000
`
`Abstract. Nucleoside reverse transcriptase inhibitors
`typic mechanisms have been identified or proposed to
`such as 3%-azido-3%-deoxythymidine, 2%,3%-
`(NRTIs),
`account for HIV-1 RT resistance to NRTIs. These
`dideoxyinosine and 2%,3%-dideoxy-3%-thiacytidine, are ef-
`mechanisms include alterations of RT discrimination
`fective inhibitors of human immunodeficiency type 1
`between NRTIs and the analogous dNTP (direct effects
`(HIV-1)
`replication. NRTIs
`are deoxynucleoside
`on NRTI binding and:or incorporation), alterations in
`triphosphate analogs, but lack a free 3%-hydroxyl group. RT-template:primer interactions, which may influence
`Once NRTIs are incorporated into the nascent viral
`subsequent NRTI incorporation, and enhanced removal
`of the chain-terminating residue from the 3% end of the
`DNA, in reactions catalyzed by HIV-1 reverse tran-
`primer. These different resistance phenotypes seem to
`scriptase (RT), further viral DNA synthesis is effectively
`terminated. NRTIs should therefore represent the ideal
`correlate with different sets of mutations in RT. This
`review discusses the relationship between HIV-1 drug
`antiviral agent. Unfortunately, HIV-1 inevitably devel-
`resistance genotype and phenotype, in relation to our
`ops resistance to these inhibitors, and this resistance
`correlates with mutations in RT. To date, three pheno-
`current knowledge of HIV-1 RT structure.
`
`Key words. Human immunodeficiency virus type 1; reverse transcriptase; nucleoside reverse transcriptase in-
`hibitors; DNA polymerization; chain termination; antiviral drug resistance; phosphorolysis; pyrophosphorolysis.
`
`Introduction
`
`Retroviruses such as the human immunodeficiency virus
`(HIV) are RNA viruses that replicate through a double-
`strand DNA intermediate. This novel viral replication
`cycle requires that retroviruses carry a specific enzyme,
`reverse transcriptase (RT), since there are no cellular
`enzymes that can convert single-strand RNA into dou-
`ble-strand DNA. RT is a DNA polymerase that can
`copy both DNA templates (like cellular enzymes) and
`RNA templates (unlike cellular enzymes). HIV RT dif-
`
`* Corresponding author.
`
`fers from cellular DNA polymerases in two additional
`respects. First, HIV RT readily utilizes many chemically
`altered analogs of the normal deoxynucleoside triphos-
`phate (dNTP) DNA polymerase substrates. Second,
`HIV RT lacks a formal %proofreading% activity. These
`characteristics are important from a pharmaceutical
`focus, and direct the use of nucleoside analog inhibitors
`as anti-HIV pharmaceuticals. As of January 2000, six of
`the current FDA-approved anti-HIV drugs are nu-
`cleoside reverse transcriptase inhibitors (NRTIs) [1].
`NRTIs are analogs of the normal dNTP substrates of
`DNA polymerases, with important modifications (fig.
`1). The 2%,3%-dideoxynucleosides such as ddC and ddI
`
`
`
`CMLS, Cell. Mol. Life Sci. Vol. 57, 2000
`
`Review Article
`
`1409
`
`lack a 3%-OH on the sugar, whereas other analogs such
`as 3%-azido-3%-deoxythymidine (AZT) have the 3%-OH
`replaced by other functional groups that do not allow
`primer extension. NRTIs require intracellular metabolic
`transformation for antiviral activity, namely conversion
`to the triphosphate, a process catalyzed by cellular
`kinases [2].
`After conversion to the active triphosphate, NRTIs
`must compete with the natural dNTPs both for recogni-
`tion by RT as a substrate (binding) and for incorpora-
`tion into the nascent viral DNA chain (catalysis).
`NRTIs thus inhibit RT-catalyzed proviral DNA synthe-
`sis by two mechanisms [3]. First, they are competitive
`inhibitors for binding and:or catalytic incorporation
`with respect to the analogous dNTP substrate. Second,
`they terminate further viral DNA synthesis, due to lack
`of a 3%-OH group. Chain termination is the principal
`mechanism of NRTI antiviral action.
`NRTIs should be the ‘ideal’ anti-HIV therapeutics.
`Each HIV virion carries only two copies of genomic
`RNA. There are about 20,000 nucleotide incorporation
`
`events catalyzed by RT during the synthesis of complete
`viral DNA, thus providing about 5000 chances for
`chain termination by any given NRTI. Since HIV-1 RT
`lacks a formal proofreading activity (i.e. some formal
`mechanism to identify and excise inappropriate nucle-
`otide incorporation), a single NRTI incorporation event
`should suffice to quell viral DNA synthesis. In reality,
`NRTIs are less potent inhibitors of HIV replication
`than might be expected; reasons for this will be dis-
`cussed later. In addition, although NRTI therapy is
`initially quite effective in reducing viral load in HIV-1-
`infected individuals,
`the viral burden inevitably re-
`bounds despite
`continued therapy, due
`to the
`appearance of drug-resistant strains of HIV. Numerous
`mutations in HIV-1 RT have been identified in NRTI-
`resistant HIV strains (table 1) [4, 5].
`The simplest mechanism for resistance would be one of
`discrimination, i.e. some mechanism for RT to exclude
`the NRTI, while retaining the ability to recognize the
`analogous natural dNTP substrate. However, this dis-
`crimination is actually somewhat of a problem, since in
`
`Figure 1. Structures of current clinically-used nucleoside reverse transcriptase inhibitors (NRTI). AZT (1), ddI (2), ddC (3), d4T (4),
`3TC (5), abacavir (6).
`
`
`
`1410
`
`N. Sluis-Cremer et al.
`
`HIV resistance to NRTI
`
`Table 1. Mutations in HIV-1 RT correlated with resistance to NRTI.
`
`ddN
`
`RT residue
`
`41
`
`65
`
`67
`
`69
`
`70
`
`74
`
`75
`
`151*
`
`184
`
`210
`
`215
`
`219
`
`AZT†
`ddC
`3TC
`ddI (ddA)
`d4T
`Abacavir
`
`M41L
`
`D67N
`
`K70R
`
`K70E
`
`T69D
`
`68¡¡70‡
`
`K65R
`K65R
`K65R
`
`K65R
`
`V75T
`
`V75T
`V75T
`
`L74V
`
`L74V
`
`M184V
`M184V
`
`M184V
`
`L210W T215Y:F
`
`K219Q
`
`* The mutation Q151M appears in patients with resistance to multiple ddN.
`† High-level AZT resistance requires the presence of two or more mutations.
`‡ Two-amino acid insertion mutants are not specific for d4T resistance, but were first identified in patients with d4T resistance. These
`insertion mutations are found in multidrug resistant HIV-1, generally over a background of AZT-resistance mutations.
`
`Table 2. Resistance phenotypes associated with NRTI-resistance mutations in HIV-1 RT.
`
`Resistance conferred
`
`NRTI-resistance phenotype
`
`Mutation
`
`K65R
`T69-S-S:G-K70*
`
`L74V
`V75T
`E89G
`Q151M
`M184I:V
`
`ddC, ddI, 3TC, PMEA
`multidrug resistance
`
`ddI, ddC
`ddC, d4T
`ddG
`multidrug resistance
`3TC
`
`discrimination
`uncertain (may be combination of discrimination and
`phosphorolysis)
`T:P repositioning
`uncertain
`T:P repositioning
`discrimination
`discrimination (also negative effect on phosphorolysis);T:P
`repositioning may also play a role
`phosphorolysis
`phosphorolysis†
`uncertain
`
`D67N:K70R:T215F(Y):Q219K
`M41L:T215Y
`L210W
`
`AZT
`AZT
`AZT
`
`* AZT-resistance mutations are required in addition to insertion mutations to provide multi-drug resistance.
`† D. Arion, N. Sluis-Cremer, M. A. Parniak, (unpublished).
`
`many cases this requires that RT must selectively ignore
`a structurally less rich compound (NRTI, lacking the
`3%-OH) in favor of the structurally more complex dNTP
`analog. In addition, our understanding of the molecular
`aspects of NRTI resistance has been complicated by the
`complex patterns of mutations required for resistance to
`some NRTIs such as AZT [6].
`To date, three mechanisms have been proposed to ac-
`count for the molecular basis of the NRTI resistance
`phenotype. These mechanisms apply to different stages
`of NRTI inhibition, and include (i) Selective alterations
`in NRTI binding and:or incorporation (i.e. discrimina-
`tion), (ii) template:primer (T:P) repositioning, which
`then influences NRTI incorporation and (iii) phospho-
`rolytic removal of an incorporated chain-terminating
`NRTI residue from the 3%-end of the nascent viral
`DNA. Correlations of these phenotypes with specific
`mutations in RT and the NRTIs affected are summa-
`rized in table 2.
`
`HIV-1 RT structure and function
`
`The HIV-1 RT gene encodes a 66-kDa protein; how-
`ever, the presumed biologically relevant form of HIV-1
`RT is a heterodimer comprising of subunits of 66 and
`51 kDa (termed p66 and p51) [7]. The p51 subunit is
`produced during viral assembly and maturation via
`HIV-1 protease-mediated cleavage of the C-terminal
`domain of a p66 subunit. The structure of the HIV-1
`RT heterodimer is illustrated in fig. 2.
`The overall shape of the p66 subunit has been likened to
`that of a ‘right hand’ [8], with the major subdomains of
`the polymerase domain of p66 appropriately termed
`fingers (residues 1–85, 118–155), palm (86–117, 156–
`237) and thumb (238–318) (fig. 2). The DNA poly-
`merase catalytic aspartate residues (D110, D185, D186)
`are in the palm subdomain. In addition, the p66 has two
`additional major subdomains, the ‘connection’ (residues
`319–426) and the C-terminal ribonuclease H (RNase
`
`
`
`CMLS, Cell. Mol. Life Sci. Vol. 57, 2000
`
`Review Article
`
`1411
`
`H) (427–565) domains. The latter subdomain is missing
`in the p51 subunit. Whereas the overall folding of the
`subdomains is similar in both p66 and p51 subunits, the
`spatial arrangement of the subdomains differs markedly
`[8, 9]. The p66 subunit adopts an ‘open’, catalytically
`competent conformation that can ‘grasp’ a nucleic acid
`template, whereas the p51 subunit is in a ‘closed’ confor-
`mation. The p51 subunit is considered to play a largely
`structural role, although it may also be important in
`interacting with the transfer RNA (tRNA)Lys3 primer
`used for the initiation of HIV-1 DNA synthesis [8].
`Mutations associated with NRTI resistance occur pri-
`marily in the fingers and the palm subdomains of RT
`(table 1, fig. 2). Because of the nature of HIV-1 RT
`
`heterodimer formation, NRTI resistance mutations ob-
`viously occur in both p66 and p51 subunits. However,
`only those in the p66 subunit are generally considered to
`have phenotypic consequences. Whereas one report has
`hypothesized that mutations in the p51 subunit may also
`contribute to the resistance phenotype [10], there are no
`biochemical data to support this conjecture.
`The conversion of HIV-1 genomic RNA into double-
`strand viral DNA is a complex process, yet all chemical
`steps are catalyzed by RT. This requires RT to be
`multifunctional, with two types of DNA polymerase
`activity, RNA-dependent DNA polymerase (RDDP) to
`synthesize a DNA strand copy of the viral genomic
`RNA template and DNA-dependent DNA polymerase
`
`Figure 2. Structure of the HIV-1 RT heterodimer showing locations of residues mutated in NRTI resistance. The crystal coordinates
`used to generate this figure are Brookhaven Protein Data Bank (PDB) 2HMI [19].
`
`
`
`1412
`
`N. Sluis-Cremer et al.
`
`HIV resistance to NRTI
`
`Figure 3. Reaction mechanism for RT-catalyzed DNA synthesis, showing the multiple mechanistic forms of RT involved, and the
`relationship between forward reaction DNA synthesis and reverse reaction pyrophosphorolysis.
`
`(DDDP) to complete the synthesis of double-strand
`viral DNA. As well, RT possesses an intrinsic RNase H
`activity, to degrade the genomic RNA component of
`
`the DNA:RNA duplex intermediate formed during RT
`RDDP synthesis. NRTIs are directed against RT
`RDDP and DDDP.
`
`
`
`CMLS, Cell. Mol. Life Sci. Vol. 57, 2000
`
`Review Article
`
`1413
`
`HIV-1 RT DNA synthesis follows an ordered ‘bi bi’
`mechanism [11–14] involving several RT mechanistic
`species. Free RT first binds the template:primer (T:P) to
`form a tight RT-T:Pn binary complex (fig. 3, step 1).
`This is followed by the binding of dNTP, to form the
`RT-T:Pn-dNTP ternary complex (step 2, fig. 3). Binding
`of dNTP appears to be a two-stage process [15]. The
`initial interaction may be nonselective, with each of the
`four dNTPs binding with relatively similar affinities.
`The second stage is more selective, and involves the
`correct positioning of only that dNTP complementary
`to the template base. The selectivity involved in posi-
`tioning and incorporating the correct nucleotide is pri-
`marily controlled by the free energy of base pairing with
`the complementary template nucleotide base. Since NR-
`TIs and the analogous dNTPs have identical base struc-
`tures, in most cases RT has little opportunity to readily
`effect discrimination between natural substrates and
`inhibitors at this stage.
`Formation of the initial RT-T:Pn-dNTP induces a rate-
`limiting change in protein conformation to form a very
`tight ternary complex, RT*-T:Pn-dNTP (step 3, fig. 3).
`The formation of this ternary complex allows the criti-
`cal transition state to be reached, enabling nucleophilic
`attack by the 3%-OH primer terminus on the a-phos-
`phate of the bound dNTP. This results in phosphodi-
`ester-bond formation and extension of the viral DNA
`strand by one nucleotide, along with formation of the
`RT-T:Pn (cid:27) 1-PPi ternary pyrophosphate complex (step
`4). The PPi product then dissociates, leaving the RT-T:
`Pn (cid:27) 1 binary complex, which is kinetically indistinguish-
`able from the RT-T:Pn binary complex. RT can then
`either dissociate from the newly elongated T:Pn (cid:27) 1 (dis-
`tributive mode of polymerization) or translocate along
`the bound template, bind the next complementary
`dNTP and continue DNA synthesis (processive mode of
`polymerization). The key kinetic steps that define the
`incorporation of dNTP (and the analogous NRTIs) are
`the nucleotide binding event (step 2), the rate-limiting
`conformational change (step 3) and the efficiency of
`incorporation (step 4).
`Crystal structures for three of the RT mechanistic forms
`have been solved. These include substrate-free enzyme
`(RT) [16, 17], the RT-T:P binary complex [18, 19] and
`an RT*-T:P-dNTP ternary complex [20]. The ‘missing’
`structures in terms of the RT reaction pathway include
`the RT-T:P-PPi product ternary complex. Also missing
`is that complex resulting from the initial binding of the
`dNTP (RT-T:P-dNTP). However, such a structure has
`been solved for another DNA polymerase [21]. These
`structural data have provided considerable insight into
`the conformational changes associated with DNA poly-
`merization, the molecular structure of the RT active
`site, and the relative position of certain critical residues
`during catalysis, all of which are important for under-
`
`standing the molecular basis of HIV-1 RT resistance to
`NRTIs.
`Comparison of the structures of free RT with the RT-T:
`P binary complex shows that interaction of RT with the
`nucleic acid induces a conformational rotation of the
`p66 thumb subdomain [18, 19]. Comparison of the
`structures of the RT-T:P binary complex with the RT*-
`T:P-dNTP ternary complex reveals that dNTP binding
`induces further conformational changes in RT [20]. In
`particular, parts of the fingers subdomain rotate in-
`wards toward the palm subdomain and the polymerase
`active site, effectively ‘clamping’ the dNTP into the
`active site region. Two fingers residues make important
`contacts with the dNTP substrate. The o-amino of K65
`and the guanidinium group of R72 interact with the g-
`and a-phosphates of the bound dNTP substrate, respec-
`tively (fig. 4). As seen in tables 1 and 2, the K65R
`mutation arises in HIV-1 resistance to a number of
`different NRTIs. The dNTP triphosphate moeity also
`interacts with the main-chain amide-NH of residues
`D113 and A114, and with the Mg2 (cid:27) divalent metal ion
`cofactor bound to the polymerase active site aspartates
`(D110, D185, D186).
`The divalent metal ions facilitate the nucleophilic attack
`by the 3%-OH of the primer terminus on the a-phospho-
`rous of the incoming dNTP, and stabilize the transition
`state [22]. Only two of the active site aspartates (D110,
`D185) are coordinated to Mg2 (cid:27) in the HIV-1 RT
`ternary complex [20]. The third aspartate (D186) may
`be involved in positioning the primer terminus by inter-
`acting with the 3%-OH of the primer terminal nucleotide
`(this functional group is absent in the ternary complex).
`The sugar 3%-OH of the dNTP projects into a small
`pocket formed by the side chains of D113, A114, Y115,
`F116 and Q151. Interestingly, the 3%-OH of the nucle-
`otide forms a hydrogen bond with the main-chain -NH
`of Y115. Since all NRTIs are missing the 3%-OH, this
`interaction may be of significance in understanding the
`kinetic efficiency of NRTI incorporation into the grow-
`ing viral DNA chain.
`
`Mechanisms of HIV-1 resistance to NRTIs
`
`Discrimination-resistance due to impaired NRTI binding
`and:or incorporation
`A large number of mutations in HIV-1 RT have been
`identified in resistance to NRTIs (tables 1 and 2). In a
`few cases, such as the AZT resistance and certain multi-
`drug resistance phenotypes, multiple mutations are
`needed to observe high phenotypic resistance. For the
`most part, however, single point mutations suffice to
`confer resistance to individual NRTI.
`Point mutations such as K65R, T69D, Q151M and
`M184V:I lead to alterations in affinity of RT for spe-
`
`
`
`1414
`
`N. Sluis-Cremer et al.
`
`HIV resistance to NRTI
`
`cific NRTIs with little or no change in affinity for the
`corresponding dNTP substrate. These mutations are
`located in, or close to, the dNTP substrate binding site
`[20], and may therefore affect the initial binding and:or
`the subsequent positioning of the NRTI for catalysis in
`a manner such that the mutant RT is able to discrimi-
`nate between the NRTI and the analogous dNTP sub-
`strate. As mentioned previously, discrimination between
`NRTIs and analogous dNTPs requires that RT must
`selectively ignore the structurally less rich NRTI in
`favor of the structurally more complex dNTP analog. It
`is important to note that wild-type RT is generally less
`efficient at catalyzing the incorporation of bound
`ddNTP into the nascent DNA compared with the corre-
`sponding dNTP substrate [13, 23]. Two examples will
`
`be used to illustrate how RT is able to carry out this
`discrimination.
`K65R. The first example concerns the K65R mutation.
`As seen in table 1, this mutation confers some level of
`resistance to a variety of NRTIs, but is primarily recog-
`nized for viral cross-resistance to 3TC, ddC and ddI
`(ddA), as well as to PMEA [24–27]. In vitro studies
`with purified recombinant RT show that the K65R
`HIV-1 RT exhibits an eight-fold decrease in the affinity
`for ddCTP and ddATP, compared with only a twofold
`decrease in KM for dCTP and dATP [24, 28]. K65 is in
`the highly flexible b3– b4 loop in the fingers domain of
`the p66 subunit of RT, and the o-amino of K65 inter-
`acts with the g-phosphate of the bound dNTP substrate
`[20]. Molecular modeling studies indicate that the K65R
`
`Figure 4. View of the dNTP binding pocket in the RT*-T:P-dNTP ternary complex. The figure was created on Sybyl 6.5 using PDB
`coordinates 1RTD [20]. The RT residues shown are those that interact with the dNTP substrate (K65, R72, G112, D113, A114, Y115,
`Q151), the primer 3%-terminal nucleotide (Y183, M184), the complementary template nucleotide (L74, V75) and the catalytic aspartate
`triad (D110, D185, D186) with coordinated metals.
`
`
`
`CMLS, Cell. Mol. Life Sci. Vol. 57, 2000
`
`Review Article
`
`1415
`
`substitution induces packing rearrangements of the nu-
`cleotide substrate, leading to subtle differences in orien-
`tation of the phosphate backbone [28]. This leads to
`displacement of the base component of the nucleotide,
`which may hinder the positioning of the base for hydro-
`gen bonding to the complementary template base. But
`why is this effect more pronounced for the NRTI than
`the analogous dNTP? As previously discussed, the 3%-
`OH of the dNTP substrate makes significant interac-
`tions with the ‘3%-OH’ binding pocket in RT, and in
`particular to residue Q151. These interactions may facil-
`itate proper positioning of the dNTP base to allow base
`pairing with the template. NRTI lack a 3%-OH, and are
`therefore unable to contact the ‘3%-OH’ pocket on the
`enzyme. The absence of this positioning interaction
`makes it more difficult for the NRTI base to hydrogen-
`bond to the template, and the RT-bound NRTI is
`therefore more likely to adopt an orientation in the
`polymerase active site that is unfavorable for catalysis.
`Thus, the absence of the 3%-OH positioning interaction
`in NRTI binding, coupled with the subtle displacement
`of the NRTI base arising from the K65R substitution,
`can account for the selective decrease in RT affinity for
`ddCTP and ddATP relative to the dNTP substrates.
`The slightly decreased affinity of the K65R mutant for
`dATP and dCTP further increases the already existing
`bias of RT against NRTI (in this case ddCTP and
`ddATP) incorporation into the nascent viral DNA,
`resulting in a readily observable resistance phenotype.
`The K65R mutant RT possesses an increased processiv-
`ity in DNA synthesis relative to wild-type enzyme [29].
`This may compensate for the modest loss of affinity for
`dCTP and dATP exhibited by the mutant enzyme,
`thereby enabling normal replication kinetics.
`It is intriguing that the K65R mutation appears to
`confer selective resistance to some, but not all, NRTIs.
`For example, we demonstrated that whereas the K65R
`mutation provides cross-resistance to 3TC, ddC and ddI
`(ddA), the mutant is not resistant to either ddG or ddT.
`While the precise mechanism for this selective resistance
`is not yet known, a clue is obtained from crystal studies
`of the dNTP complexes with another DNA polymerase,
`the Klenow fragment of Thermus aquaticus DNA poly-
`merase I [30]. The binding of all the dNTP is due in part
`to interaction of the g- and a-phosphates of the dNTP
`with two basic amino acid residues (analogous to K65
`and R72 of HIV-1 RT). However, the base components
`of the dNTPs occupy different positions in the bound
`structures. The base rings of dCTP and dATP occupy
`similar positions in the bound structure, and have few
`contacts with protein residues. The base rings of dGTP
`and dTTP also occupy similar regions of space as well,
`but these differ from the regions occupied by the bases
`of dCTP and dATP. The base components of dGTP
`and dTTP make significant additional contacts with
`
`surrounding enzyme residues, which may stabilize the
`positioning of these dNTPs for incorporation. We pro-
`pose that a similar situation may exist for RT. The
`selective resistance to ddCTP and ddATP conferred by
`the K65R mutation results from (i) the loss of the
`3%-OH pocket interaction, and (ii) the lack of additional
`base-stabilizing contacts with surrounding RT residues.
`This results in an unacceptable flexibility in ddCTP and
`ddATP binding such that correct positioning for cata-
`lytic incorporation is impaired. In contrast, the flexibil-
`ity of ddGTP and ddTTP binding is reduced due to
`additional contacts of the base ring with RT residues.
`These NRTIs are more readily positioned for incorpo-
`ration into the growing DNA chain.
`We have recently compared the properties of recombi-
`nant RTs with a variety of amino acid substitutions at
`K65 [28]. Whereas the K65R mutation confers resis-
`tance only to specific NRTIs, other mutations such as
`K65A and K65Q result in broad resistance to all NR-
`TIs (i.e. multidrug resistance). So why does HIV not
`generate these mutations? These mutant RTs have
`severe impairments in the incorporation of the natural
`dNTP substrates as well. Our modeling experiments [28]
`suggest that only enzymes with K or R at position 65
`are able to make significant contacts with the g-phos-
`phate of the dNTP. The loss of this contact in the other
`position 65 mutants prevents the proper positioning of
`the dNTP for catalytic incorporation. Thus, HIV-1 is
`severely curtailed in the choice of mutations at this
`position.
`M184V (and M184I). The second example of discrimi-
`nation as a resistance phenotype concerns the point
`mutation M184V (sometimes M184I) that confers high-
`level resistance to 3TC [31, 32]. M184 is located in the
`YMDD motif, which is highly conserved among retro-
`viral RTs. In HIV-1 RT, this motif contains two of the
`three catalytic aspartates of the DNA polymerase active
`site [16–20]. The M184V RT is about 140-fold less
`efficient in incorporating 3TCTP into the nascent viral
`DNA compared with the wild-type enzyme [33]. This
`decreased incorporation efficiency results from steric
`hindrance that leads to decreased binding and:or inap-
`propriate positioning of the 3TCTP for catalysis. In the
`RT*-T:P-dNTP ternary complex of wild-type RT [20],
`the side chain of M184 contacts the sugar and the base
`of the 3% nucleotide in the primer. Molecular modeling
`experiments [20, 34] show that substitution of M184
`with an amino acid with a b-branched side chain
`(isoleucine or valine) results in an additional contact
`with the dNTP sugar moiety (fig. 5). The side chain of
`a b-branched amino acid at position 184 makes inap-
`propriate contact with the sulfur of the sugar oxathi-
`olane ring of 3TCTP, preventing proper positioning of
`3TCTP for catalysis, consistent with the data of Feng
`and Anderson [33]. The analogous substrate dCTP does
`
`
`
`1416
`
`N. Sluis-Cremer et al.
`
`HIV resistance to NRTI
`
`not have a sulfur in its sugar ring, and no steric hin-
`drance of dCTP binding is noted. In contrast, Krebs et
`al. [23], while demonstrating that the M184V mutant RT
`shows a markedly reduced efficiency of incorporation of
`3TCMP into viral DNA, found no differences in binding
`affinity of the wild-type and M184V mutant enzymes for
`3TCTP. These observations differ from those of Feng
`and Anderson, and may support the T:P repositioning
`model (see below). Thus, the precise mechanism by
`which the M184V:I mutations confer resistance to 3TC
`is still not completely certain. However, we feel that
`steric hindrance preventing correct positioning of
`3TCTP in the RT active site likely plays a major role in
`this resistance mechanism.
`
`Template:primer repositioning may, however, play a
`role in the decreased DNA synthesis processivity associ-
`ated with the L74V mutation for ddI resistance [36].
`However, this altered processivity does not necessarily
`contribute to the resistance phenotype. Whereas L74
`interacts with the template nucleotide that is base-
`paired to the incoming dNTP molecule, it is also proxi-
`mal
`to Q151 and R72,
`two residues
`that make
`significant contacts with the bound dNTP [20]. Thus the
`L74V mutation may alter ddNTP binding and:or catal-
`ysis by altering the packing rearrangements of ddNTP
`through its
`interactions with the aforementioned
`residues.
`
`Template:primer repositioning
`Boyer et al. [35] showed that wild-type RT was inhibited
`by ddITP when the template extension was four nucle-
`otides or more beyond the 3% end of the primer, but was
`resistant to ddITP when the template extension was
`three nucleotides or fewer [35]. In contrast, ddITP resis-
`tance shown by the L74V and E89G mutant enzymes
`was independent of template extension length. The RT-
`T:P binary complex crystal structure [18] suggested that
`residues 74 and 89 contacted the DNA T:P; the recently
`solved RT*-T:P-dNTP ternary complex [20] shows that
`L74 interacts with the template nucleotide that is base-
`paired to the incoming dNTP. Boyer et al. [35] therefore
`proposed that the L74V and E89G mutations (and
`perhaps others) confer resistance to NRTI as a result of
`repositioning of the T:P in the RT-T:P complex, which
`in turn alters the ability of the enzyme to select for, or
`reject an incoming dNTP. While there is no structural
`evidence for T:P repositioning in the L74V or E89G
`mutants, comparison of the M184I RT-T:P binary com-
`plex with the wild type RT-T:P binary complex reveals
`some alterations in the position of the T:P in the M184I
`mutant RT structure [34]. As discussed above, the data
`of Krebs et al. [23] provide circumstantial support for
`the T:P repositioning model in 3TC resistance. Nonethe-
`less, the relevance of T:P repositioning in phenotypic
`resistance to NRTI is uncertain. How could this reposi-
`tioning selectively alter incorporation of ddNTP, with-
`out also dramatically affecting incorporation of the
`corresponding dNTP substrate? In addition, it is not
`certain whether the shift in T:P position noted in the
`M184I RT-T:P binary complex would be retained in the
`subsequent RT-T:P-dNTP ternary complex. Finally, as
`discussed above, the phenotype for 3TC resistance is
`satisfactorily explained by the steric conflict between the
`oxathiolane ring of 3TCTP and the side chain of the
`branched b-amino acid. Therefore, there does not ap-
`pear to be a need to invoke T:P repositioning to explain
`this resistance.
`
`Phosphorolytic removal of incorporated
`chain-terminating NRTI—the novel mechanism of AZT
`resistance
`The efforts of investigators to elucidate the biochemical
`phenotype for HIV-1 resistance to AZT was compli-
`cated by the absence of any measurable effect on inhibi-
`tion of RT by AZTTP in vitro, despite the more than
`200-fold resistance to AZT demonstrated in cell culture
`[37, 38]. AZT resistance correlates with multiple muta-
`tions in RT, including M41L, D67N, K70R, L210W,
`T215Y:F and Q219K [39–42] (table 1). The extent of
`AZT resi