Derivatives of β-2′,3′-didehydro-2′,3′-dideoxy-5-fluorocytidine (d4FC) (Fig. ) are potent and selective inhibitors of both wild-type and nucleoside reverse transcriptase inhibitor-resistant human immunodeficiency virus type 1 (HIV-1) in clinical development (2
). A phase I-II study with HIV-infected patients having virus with the M184V resistance mutation showed that the substitution of l
-d4FC (elvucitabine or ACH-126,443) for lamivudine produced a 0.67 to 0.78 log10
decline in plasma HIV RNA (copies/ml) over a 28-day period, although dose-dependent hematological toxicity was evident (L. M. Dunkle, J. C. Gathe, D. E. Pedevillano, H. G. Robison, W. G. Rice, J. C. Pottage, Jr., and the ACH-006 Study Team, Abstr. XII Int. HIV Drug Resist. Workshop: Basic Princ. Clin. Implic., abstr. 2, 2003). A phase I-II study in which d
-d4FC was given once daily at 50 mg, 100 mg, or 200 mg for 10 days to 30 antiretroviral naïve, HIV-infected subjects showed marked reductions in plasma HIV-1 RNA averaging 1.7 log10
copies/ml without discernible toxicity (R. L. Murphy, D. Schürmann, A. Beard, L. Cartee, R. F. Schinazi, and M. J. Otto, Abstr. 11th Conf. Retrovir. Opport. Infect., abstr. 137, 2004). The potency and favorable safety profile of d
-d4FC have led to its accelerated clinical development. Because of the interest in both enantiomers of d4FC, the goal of the current study was to evaluate the influence of stereochemistry on resistance to d4FC resistance by performing in vitro selections.
Structures of 2′,3′-didehydro-2′,3′-dideoxy-5-fluorocytidine derivatives.
Selection experiments were conducted as previously described (1
). A total of 106
MT-2 cells (AIDS Research and Reference Reagent Program, National Institute of Allergy and Infectious Diseases, National Institutes of Health) were pretreated for 2 h with l
-d4FC or d
-d4FC (synthesized by R. Schinazi) (12
) and then infected with HIV-1LAI
at a multiplicity of infection of 0.1 50% tissue culture infective doses/cell. A 100-μl aliquot was used to initiate a new cycle of infection and the selective pressure (i.e., drug concentration) was generally doubled every two to three passages. The passaged virus was regularly monitored for a reduction in susceptibility to the compounds by determining the 50% inhibitory concentration (IC50
) of the passaged virus relative to that of the unpassaged HIV-1LAI
). Student's t
test was used to determine the statistical significance of differences between log-transformed IC50
Amino acid changes in passaged virus were identified by extracting RNA (TRIZOL reagent; Gibco BRL, Grand Island, NY) from pelleted virions (25,000 × g
for 1 h at 4°C). The near full-length coding region of reverse transcriptase (RT) (amino acids [aa] 1 to 511) was amplified by RT-PCR (1
), purified (Wizard PCR purification system; Promega, Madison, WI), and sequenced (aa 1 to 350) by using an ABI 377 sequencer (Applied Biosystems, Foster City, CA). To confirm the role of these changes, recombinant mutants were made by using oligonucleotide-directed mutagenesis (Altered Sites II; Promega) or by cassette cloning of the XmaI-to-XbaI RT fragment from passaged virus into the pxxHIV-1LAI
). Infectious recombinant HIV-1 was generated as previously described (14
In two independent experiments, virus resistant to l
-d4FC was isolated after 4 (selection no. 1) or 10 (selection no. 2) passages. Both selections were initiated at an l
-d4FC concentration of 0.5 μM (~5 times the IC50
of the wild type). The selective pressure could not be increased above 1.1 μM because of cytotoxicity for MT-2 cells. Phenotypic analyses of viruses from passages 4 and 10 showed >20-fold (selection no. 1) and >11-fold (selection no. 2) levels of resistance to l
-d4FC, respectively; and genotype analyses of these viruses identified mutations at codon 184 (M to I or V) (Table ). A recombinant HIV-1 encoding the M184V mutation showed >46-fold resistance to l
-d4FC. The M184V mutation was also selected in HIV-1LAI
passaged with l
-d4FC in human peripheral blood mononuclear cells (13
Selection of HIV-1 resistant to d- and l-d4FC
Viruses resistant to d-d4FC were isolated in three independent passage experiments. The first selection was initiated at a concentration of 0.75 μM (~6 times the IC50 of d-d4FC) and increased to a final concentration of 4.0 μM over the course of 37 passages. Phenotypic analysis of the virus from passage 37 showed a 14-fold decrease in susceptibility to d-d4FC (Table ). Genotype analysis of virus from this passage identified mutations at codons 70 (K to N), 90 (V to I), and 172 (R to K). All three mutations are novel and, with the exception of R172K, are in highly conserved regions of HIV-1 RT. Site-directed mutants encoding K70N or V90I/R172K mutations showed a 2.1-fold reduction in susceptibility. Both the site-directed mutant encoding K70N/V90I/R172K mutations and the p37 RT recombinant virus also showed significant reductions in susceptibility to d-d4FC, 3.9- and 2.4-fold, respectively (Table ).
d-d4FC Susceptibility of recombinant viruses encoding mutations selected by serial passage in d-d4FC
The second and third selections for d
-d4FC-resistant virus were initiated at a lower concentration, 0.20 μM (~1.5 times the IC50
). The selective pressure was increased to a concentration of 4.8 μM over the course of 20 passages (selection no. 2) and to 1.6 μM over the course of 16 passages (selection no. 3). Passage 20 virus from selection no. 2 showed an 8.1-fold decrease in susceptibility to d
-d4FC, and genotypic analysis identified mutations at codons 65 (K to R) and 179 (V to D) in RT (Table ). In site-directed mutants, the K65R mutation alone decreased d
-d4FC susceptibility 6.2-fold; the addition of the V179D mutation did not increase the level of resistance (5.5-fold) (Table ), nor did it affect the replicative fitness of virus with the K65R mutation as determined in competition experiments in MT-2 cells (not shown). These results are consistent with findings from Geleziunas et al., who recently reported that K65R is the sole mutation after in vitro selection with d
In the third selection, virus with the I63L mutation was identified at passage 12 (13-fold resistance) and further selection resulted in virus having greater resistance (16-fold) and the K70E mutation (Table ). A site-directed mutant with the I63L mutation was not significantly resistant to d-d4FC (1.5-fold; P = 0.25), but the K70E mutation significantly decreased susceptibility to d-d4FC by 2.6-fold (Table ). We were unable to select d-d4FC-resistant viruses in human peripheral blood mononuclear cells after more than 20 passages (data not shown).
The site-directed mutant encoding the M184V substitution showed no reduction in susceptibility to d-d4FC (0.9-fold) (data not shown). Conversely, the K65R mutant showed 41-fold resistance to l-d4FC (not shown).
Selection of the M184I/V mutation by l
-d4FC is consistent with previous studies of lamivudine. The crystal structure of HIV-1 RT with a bound deoxynucleotide triphosphate predicts that replacing M184 with a β-branched amino acid, such as I or V, will create a “steric clash” between residue 184 and the pseudosugar ring of nucleoside analogs in the l
configuration, including l
-d4FC selected several different mutation patterns: (i) a combination of three novel mutations (K70N, V90I, and R172K); (ii) a mutation known to confer broad nucleoside reverse transcriptase inhibitor resistance (K65R) (1
); and (iii) two other mutations in the fingers subdomain of RT (I63L and K70E). Although the contribution of the I63L mutation to d
-d4FC resistance is uncertain, the K70E mutation, which replaced the I63L mutation at higher drug concentrations, is located near the incoming deoxynucleotide triphosphate and could affect the position of the bound analog triphosphate (3
). It is likely that the mutation confers some level of discrimination against incorporation of d
-d4FC triphosphate, although the specific molecular interactions that underlie this are not yet defined.
The selection of several different genotypes by d
-d4FC indicates that there is more than one genetic route to d
-d4FC resistance. It is notable, however, that each genotype conferred only low-level resistance (≤6.2-fold) to d
-D4FC in site-directed mutants. The reason different genotypes were selected may have been related to the variation in starting drug concentration (4
) or the stochastic nature of in vitro resistance selection (i.e., chance). Although our results suggest multiple genetic mechanisms of viral escape from d
-d4FC, clinical studies are needed to determine which, if any, of these occur in vivo.
Our finding of multiple mutations in the fingers subdomain of RT suggests that resistance to d
-d4FC is due to discrimination rather than excision. This concept is supported by the observation that d
-d4FC retains potency against virus containing the zidovudine resistance mutations D67N, K70R, T215Y, and K219Q, which confer excision competence (8
). Thus, d
-d4FC may be a potent inhibitor for patients with virus having zidovudine resistance mutations, although clinical studies are needed to confirm this.
In summary, the results of this study should prove useful in directing resistance testing of viruses from patients treated with l- or d-d4FC and in choosing combinations of nucleoside analogs for treatment and prevention of drug-resistant HIV-1.