Coresistance to Zidovudine and Foscarnet Is Associated with Multiple Mutations in the Human Immunodeficiency Virus Type 1 Reverse Transcriptase

Abstract

Human immunodeficiency virus type 1 (HIV-1) isolates obtained from a patient with AIDS were assessed for coresistance to foscarnet and zidovudine. An HIV-1 strain (AP20) coresistant to foscarnet and zidovudine was isolated after 20 months of continuous combination therapy. The reverse transcriptase (RT) gene of AP20 had 41 substitutions which were different from the HXB2-D sequence and 9 that were different from the sequence of its foscarnet-sensitive, zidovudine-resistant progenitor virus (AP6). Six of these mutations were nonpolymorphic (T39A, V108I, K166R, K219R, K223Q, and L228R). Both strains had the conventional mutations mediating zidovudine resistance. In vivo selection may result in HIV-1 strains that are coresistant to foscarnet and zidovudine, but coresistance appears to require a complex evolutionary path and multiple RT mutations.

Foscarnet (PFA) is a broad-spectrum viral DNA polymerase inhibitor which also inhibits human immunodeficiency virus type 1 (HIV-1) (27). Despite its activity towards HIV-1 (4), PFA is used exclusively to treat opportunistic viral infections such as human cytomegalovirus (CMV) (28), acyclovir-resistant herpes simplex (2, 26), and varicella-zoster virus infections (25) in patients with immunodeficiency. PFA also inhibits Karposi’s sarcoma-associated herpesvirus in vitro (18) and may thereby decrease the risk of Kaposi’s sarcoma in patients with AIDS (5, 20).

PFA-resistant strains of HIV-1 have developed in patients with AIDS receiving long-term PFA therapy for CMV retinitis (19, 29). The reverse transcriptase (RT) substitutions W88G, W88S, Q161L, and H208Y were observed in these clinical isolates (19, 29). In vitro selection readily generates PFA-resistant strains of HIV-1 (19, 30) with single (E89K, L92I, or S156A) (30) or double (Q161L and H208Y) (19) amino acid substitutions in the RT region.

Zidovudine (AZT) is a thymidine analogue inhibitor of the HIV-1 RT which has been used extensively to treat HIV-1-infected individuals. Long-term AZT monotherapy is associated with the development of HIV-1 strains with reduced susceptibility to this drug (16). Resistance is mediated by the stepwise accumulation of up to six mutations in the HIV-1 RT including M41L, D67N, K70R, L210W, T215Y/F, and K219Q (7, 9, 11, 14).

Given that both AZT and PFA may occasionally be administered either sequentially or in combination to HIV-infected individuals, it was of interest to determine whether strains coresistant to these drugs would emerge in vivo. We have previously demonstrated that several mutations which confer PFA resistance (W88G, E89K, L92I, Q161L) will reverse phenotypic AZT resistance and that at least based on in vitro selection studies, PFA and AZT resistance appear to be mutually exclusive (31), suggesting reciprocal conformational changes between the PFA and AZT-triphosphate binding sites on the HIV-1 RT (31). Given these data, we have hypothesized that a complex evolutionary path involving multiple RT mutations would be required to generate a strain coresistant to AZT and PFA (31). Here we describe a phenotypic and genotypic analysis of two HIV-1 clinical isolates obtained from a patient with AIDS after 6 and 20 months of continuous AZT and PFA therapy; the latter strain was coresistant to these drugs. Consistent with our hypothesis, coresistance was associated with multiple RT mutations.

To assess whether HIV-1 strains coresistant to PFA and AZT could emerge in vivo, we studied HIV-1 strains from an AIDS patient who had received long-term combination therapy with AZT and PFA for the treatment of HIV-1 and CMV retinitis, respectively (Fig. 1). The patient presented with an HIV-1 seroconversion illness in June 1985. After 5 years of clinically stable HIV-1 infection, AZT monotherapy (300 to 500 mg/day) was initiated in June 1990 because of a drop in CD4 T-cell count from 440 to 110 cells/μl and the appearance of mucosal candidiasis (Fig. 1). PFA treatment was initiated in November 1993. AZT and PFA were administered concurrently for 20 months until AZT was discontinued in July 1995 because of pancytopaenia resulting in central venous catheter sepsis (Fig. 1). The clinical history of the patient, including the relationship between antiviral therapy, serum p24 antigen concentrations, and CD4 T-cell counts from the time of commencement of AZT monotherapy, is shown in Fig. 1.

FIG. 1.

Relationship between antiviral therapy, serum p24 antigen concentrations, and CD4 T-cell counts from the time of commencement of AZT therapy (June 1990) for the patient. The time of isolation of HIV-1 strains AP6 and AP20 from the patient’s PBMCs are indicated. A p24 serum antigen concentration of 0 indicates that the antigen was undetectable (i.e., <20 pg/ml) by the Coulter enzyme immunoassay. Dosages for the administered antiviral drugs were as follows: AZT, 300 to 500 mg/day; ddC, 1.25 mg/day; ddI, 400 mg/day; PFA, 90 to 120 mg/kg of body weight/day; acyclovir, 200 to 400 mg/day.

We isolated HIV-1 from the patient’s peripheral blood mononuclear cells (PBMCs) after 6 (strain AP6) and 20 (strain AP20) months of combined therapy with AZT and PFA (Fig. 1). Virus isolation was performed by cocultivation of the patient’s PBMCs with PBMCs from an HIV-1 seronegative donor as previously described (17). Pretreatment isolates were not available, as we had identified this patient in June 1994 and blood specimens suitable for virus isolation or direct sequencing had not been collected prior to this date.

To allow for the assessment of drug susceptibilities in the HT4LacZ-1 cell line (24), we made recombinant strains rAP6 and rAP20, which had the RT coding regions of isolates AP6 and AP20, respectively, inserted in an HXB2-D genetic background. Recombinant strains rAP6 and rAP20 were generated by cotransfection of MT-2 cells (6) with 5 μg of PCR amplified pol fragments derived from HIV-1 strains AP6 and AP20, respectively, with 5 μg of BstEII linearized pHIVΔRTBstEII (12). All MT-2 cell transfections in this study were performed using DOTAP (Boehringer Mannheim, Mannheim, Germany) as described previously (31). The RT regions of AP6 and AP20 were PCR amplified from purified genomic DNA obtained from phytohemagglutinin-stimulated PBMCs infected with AP6 and AP20. We used the Expand High Fidelity PCR system (Boehringer Mannheim) and we performed two rounds of PCR using nested primers. The 2.2-kb DNA product contained all of the RT coding region and pol flanking sequences (HIVHXB2-D coordinates 2033 to 4201). PCR primers used in first- and second-round amplifications were 5′V3 and 3′V2, and 5′V2 and 3′V1HindIII (ATATAAGCTTAGGGAATTCCAAATTCCTGCTTG; HIVHXB2-D coordinates 4180 to 4203) (21), respectively, as described previously (31). Reactions were performed as described in the manufacturer’s protocol by using 3.5 and 2.5 mM MgCl₂ for the first and second PCR rounds, respectively. Each round of amplification was 35 cycles. Drug susceptibility assays were performed in HT4LacZ-1 cells as previously described (31) with the exception that cells were seeded into 24-well plates at 1.8 × 10⁴ cells per well. PFA (Fluka Biochemika, Buchs, Switzerland) was prepared as a 33 mM stock in sterile water. AZT (Sigma Chemical Company, St. Louis, Mo.) was prepared as a 37 mM stock in dimethyl sulfoxide. Zalcitabine (ddC) (Sigma) and didanosine (ddI) (Sigma) were prepared at a concentration of 25 mM in sterile water. The statistical significance of differences between 50% inhibitory concentration (IC₅₀) values was determined by the Wilcoxon rank-sum test (1).

Drug susceptibility testing with HT4LacZ-1 cells showed that recombinant virus with the RT coding region of AP6 (rAP6) was highly AZT resistant but fully susceptible to PFA, ddI, and ddC (Table 1). By contrast, rAP20 was resistant to both PFA and AZT but remained fully susceptible to ddC and ddI (Table 1). Therefore, while 6 months of continuous PFA and AZT therapy failed to select for HIV-1 coresistant to AZT and PFA, such a strain was selected after 20 months of combination therapy.

TABLE 1.

Drug susceptibility profile of strains rAP6 and rAP20

Strain	Mean IC50 ± SD (μM) (fold resistanceb) fora:
	PFA	AZT	ddI	ddC
HX	41 ± 10	0.024 ± 0.02	6.7 ± 2.1	0.8 ± 0.1
rAP6c	39 ± 10 (0.95)	>2 (>83)	5.1 ± 2.6 (0.8)	0.5 ± 0.5 (0.6)
rAP20c	123 ± 36 (3)	1.33 ± 1.0 (55)	10.5 ± 4.2 (1.6)	1.4 ± 1.0 (1.8)

^aIC₅₀s and standard deviations were determined in drug susceptibility assays performed in HT4LacZ-1 cells and were calculated from three independent assays. The differences in IC₅₀s of AZT for HX compared to those for rAP6 and rAP20 were significant (P = 0.05). The difference in PFA IC₅₀s for HX and rAP20 was also significant (P = 0.05). No significant differences in IC₅₀s of ddI or ddC were observed for HX compared to those for rAP6 and rAP20. 

^bFold resistance is defined as the IC₅₀ for the mutant strain divided by the IC₅₀ for wild-type HX. PFA resistance is defined as HIV-1 with a greater than twofold increase in IC₅₀ compared with HX. AZT resistance was as defined previously (31). 

^cRecombinant strains with pol fragment (2.2 kb) of clinical isolates AP6 or AP20. 

To determine whether the evolution of coresistance to PFA and AZT was associated with the appearance of multiple mutations in the RT coding region, we performed nucleotide sequence analysis of the RT gene of strains AP6 and AP20. The sequence was determined by both population-based DNA sequencing and sequencing of individual molecular clones. Molecular clones were prepared by PCR amplification of 2.2-kb pol fragments as described above. The inner primer pairs 5′V2 and 3′V1HindIII contained BamHI and HindIII sites, respectively, allowing cloning into the BamHI-HindIII sites of pT7T319U (AMRAD Pharmacia Biotech, Boronia, Australia). The nucleotide sequence of the entire RT coding region in recombinant phagemids was determined by automated sequencing using the PRISM Ready reaction DyeDeoxy Terminator Cycle Sequencing kit with Amplitaq FS (Perkin-Elmer, Foster City, Calif.) as previously described (30). The five and seven molecular clones derived from strains AP6 and AP20, respectively, were designated pAP6(1) to pAP6(5) and pAP20(1) to pAP20(7). The population-based DNA sequences for the RT genes of AP6 and AP20 were determined by the direct sequencing of amplimers by using automated dye primer sequencing as previously described (31). These amplimers were prepared by two rounds of PCR. The first round used outer primers 5′V3 and 3′V2 and was followed by one of two separate second-round amplifications using either the M13 forward- and reverse-primer pairs M13 5′V2 and M13Rcomb3 (to amplify codons 1 to 244) or M13 5′V4 and M13R 3′V6 (to amplify codons 218 to 511) as previously published (31).

Nucleotide sequence accession numbers.

The nucleotide sequence data reported in this article have been deposited in the GenBank database under accession no. AF011754 [AP6(4)] and AF011755 [AP20(6)].

Sequence analysis of the RT region of the five molecular clones derived from strain AP6 [pAP6(1) to pAP6(5)] showed 34 mutations common to four of the five clones which differed from HXB2-D (Table 2). Four of these mutations were known to confer AZT resistance (M41L, D67N, L210W, and T215Y). All clones also had the mutations W88C and H208Y. The H208Y mutation is known to confer low-level PFA resistance when present in a wild-type genetic background (19). However, this is the first report of a W→C mutation at codon 88 instead of the W88G and W88S mutations commonly observed in PFA-resistant strains (19, 30, 31). The T69D substitution, known to confer ddC resistance (3), was also observed (Table 2). However, the phenotype of rAP6 was ddC susceptible (Table 1), suggesting that the RT genetic background of strain rAP6 has had a modulatory effect on T69D. Other nonpolymorphic substitutions at codons 44, 104, 118, and 283 were common to the majority of clones (Table 2). In addition, two of the five clones [pAP6(4) and pAP6(5)] had the G190R mutation in the pol domain. Twenty of the remaining 23 substitutions in the RT coding region of AP6 are previously reported polymorphic changes (21). While codons 334, 480, and 489 are also in polymorphic regions, the specific amino acid changes at these codons have not been reported in other HIV-1 strains (21). Other changes at codons 50, 114, 166, 195, 200, 206, 220, 268, 376, 384, and 401 were noted in individual clones; these mutations may have been real or PCR amplification artifacts. Population-based DNA sequence analysis of the RT coding region of AP6 showed the mutations common to all clones (codons 1 to 480) in addition to a T376A substitution not observed in the molecular clones.

TABLE 2.

Amino acid sequences of RT coding region derived from strains AP6 and AP20

Strain	Amino acid at RT codona:
	pol domain
	35	39	41	44	50	67	69	88	98	104	108	114	118	122	123	147	166	177	190	195	196	200	206	208	210	211	214	215	220	219	223	228
HX	V	T	M	E	I	D	T	W	A	K	V	A	V	E	D	N	K	D	G	I	G	T	R	H	L	R	L	T	K	K	K	L
AP6b	–	–	L	D	–	N	D	C	–	N	–	–	I	K	E	–	–	E	–	–	–	–	–	Y	W	K	F	Y	–	–	–	–
Molecular clones derived from AP6
pAP6(1)	–	–	L	D	–	N	D	C	–	N	–	–	I	K	E	–	R	E	–	M	–	–	–	Y	W	K	F	Y	–	–	–	–
pAP6(2)	–	–	L	D	V	N	D	C	–	N	–	V	I	K	E	–	–	E	–	–	–	A	–	Y	W	K	F	Y	–	–	–	–
pAP6(3)	–	–	L	D	–	N	D	C	–	N	–	–	I	K	E	–	–	E	–	–	–	–	–	Y	W	K	F	(framec)
pAP6(4)	–	–	L	D	–	N	D	C	–	N	–	–	I	K	E	–	–	E	R	–	–	–	–	Y	W	K	F	Y	–	–	–	–
pAP6(5)	–	–	L	D	–	N	D	C	–	N	–	–	I	K	E	–	–	E	R	–	–	–	G	Y	W	K	F	Y	R	–	–	–
AP20b	I	A	L	D	–	N	D	C	–	N	I	–	I	K	E	–	R	E	–	–	E	–	–	Y	W	K	F	Y	–	R	Q	R
Molecular clones derived from AP20
pAP20(1)	I	A	L	D	–	N	D	C	V	N	I	–	I	K	E	–	R	E	–	–	E	–	–	Y	W	K	F	Y	–	R	Q	R
pAP20(2)	I	A	L	D	–	N	D	C	–	N	I	–	I	K	E	S	R	E	–	–	E	–	–	Y	W	K	F	Y	–	R	Q	R
pAP20(3)	I	A	L	D	–	N	D	C	–	N	I	–	I	K	E	–	R	E	–	–	E	–	–	Y	W	K	F	Y	–	R	Q	R
pAP20(4)	I	A	L	D	–	N	D	C	–	N	I	–	I	K	E	–	R	E	–	–	E	–	–	Y	W	K	F	Y	–	R	Q	R
pAP20(5)	I	A	L	D	–	N	D	C	–	N	I	–	I	K	E	–	R	E	–	–	E	–	–	Y	W	K	F	Y	–	R	Q	R
pAP20(6)	I	A	L	D	–	N	D	C	–	N	I	–	I	K	E	–	R	E	–	–	E	–	–	Y	W	K	F	Y	–	R	Q	R
pAP20(7)	I	A	L	D	–	N	D	C	–	N	I	–	I	K	E	–	R	E	–	–	E	–	–	Y	W	K	F	Y	–	R	Q	R

Amino acid at RT codona:
pol domain																							RNase H domain
240	268	272	283	293	321	326	334	341	356	358	371	375	376	384	390	394	395	400	401	403	428	437	460	480	483	487	489	512	517	519	523	550
T	S	P	L	I	P	I	Q	I	R	R	A	I	T	G	K	Q	K	T	W	T	Q	A	N	Q	Y	Q	S	Q	L	N	E	K
–	–	A	I	V	–	V	L	–	K	K	V	V	A	–	R	–	–	A	–	M	–	–	D	E
–	I	A	I	V	–	V	L	–	K	K	V	V	–	–	R	–	–	A	–	M	–	–	D	E	H	–	A	K	I	S	–	–
–	–	A	I	V	–	V	L	–	K	K	V	V	S	E	R	–	–	A	R	M	–	–	D	E	H	–	A	K	I	S	–	–
–	–	A	I	V	–	V	L	–	K	K	V	V	–	–	R	–	–	A	–	M	–	–	D	E	H	–	A	K	I	S	–	–
–	–	A	I	V	–	V	L	–	K	K	V	V	–	–	R	–	–	A	–	M	(amberd)	–	D	E	H	–	A	K	I	S	–	–
–	–	A	I	V	–	V	L	F	K	K	V	V	A	–	R	–	–	A	–	M	–	G	D	–	H		–
–	–	A	I	V	–	V	L	–	K	K	V	V	A	–	R	–	–	A	–	M	–	–	D	–	H	–	–	K	I	S	–	–
–	–	A	I	V	T	V	L	–	K	K	V	V	A	–	R	–	–	A	–	M	–	–	D	–	H	–	–	K	I	S	(ambere)	–
–	–	A	I	V	–	V	L	–	K	K	V	V	A	–	R	–	–	A	–	M	–	–	D	–	H	–	–	K	I	S	–	E
I	–	A	I	V	–	V	L	–	K	K	V	V	A	–	R	P	–	A	–	M	–	–	D	–	H	–	–	K	I	S	–	–
–	–	A	I	V	–	V	L	–	K	K	V	V	A	–	R	–	–	A	–	M	–	–	D	–	H	–	–	K	I	S	–	–
–	–	A	I	V	–	V	L	–	K	K	V	V	A	–	R	–	R	A	–	M	–	–	D	–	H	–	–	K	I	S	–	–
–	–	A	I	V	–	V	L	–	K	K	V	V	A	–	R	–	–	A	–	M	–	–	D	–	H	H	–	K	I	S	–	–

^aRT amino acid residues shown are numbered as for HX (HXB2-D sequence) and are those that differ from isolate HX, as predicted from the observed nucleotide sequence. Codons that are underlined or in boldface type are those associated with PFA and AZT resistance, respectively. Previously reported polymorphic substitutions are in italics while codons both italicized and underlined may represent previously unreported polymorphic changes. Dashes (–) denote no change from HX. 

^bPopulation sequence of the RT coding region of strains AP6 (codons 1 to 480) and AP20 (codons 1 to 483). 

^cA single nucleotide deletion was noted at nucleotide 2738 (RT codon 215) resulting in a −1 frameshift. 

^dTAG sequence (stop codon) present at RT codon 428. 

^eTAG sequence (stop codon) present at codon 523. 

Sequence analysis of seven clones derived from AP20 [pAP20(1) to pAP20(7)] revealed 41 substitutions common to the majority of clones in both the pol and the RNase H domains which differed from the HXB2-D sequence (Table 2). Compared to mutations present in the RT coding region of AP6, an additional nine substitutions were acquired by AP20, including V35I, T39A, V108I, K166R, G196E, K219R, K223Q, L228R, and T376A. Three of these, V35I, G196E, and T376A, are previously reported polymorphic substitutions (21). The mutation conferring ddC resistance, T69D, was also present in strain AP20. However, as with strain AP6, AP20 remained ddC susceptible (Table 1). Mutations at codons 98, 147, 240, 321, 394, 395, 487, and 550 were observed in individual clones, which may have been genuine or errors introduced by the PCR amplification procedure. Population-based DNA sequence analysis of the RT coding region of AP20 showed the same mutations common to all clones (codons 1 to 483) with the exception of the I341F and A437G substitutions, which were not observed in the molecular clones.

The nucleotide sequence of the RT coding regions of strain AP20 contained 41 mutations which were common to the majority of clones analyzed. Therefore, it was unlikely that the observed phenotype for rAP20 was due to a mixture of strains displaying AZT or PFA resistance. To confirm this hypothesis, a recombinant strain was generated, rAP20(6), containing the RT coding region of pAP20(6) in the HXB2-D genetic background. rAP20(6) was prepared by cotransfection of MT-2 cells with BamHI-HindIII linearized pAP20(6) and BstEII linearized pHIVΔRTBstEII. The pAP20(6) clone contained all 41 mutations common to the seven clones, in addition to a nonpolymorphic substitution at codon 395 (Table 2). Strain rAP20(6) was coresistant to AZT and PFA (P = 0.018) (Table 3), indicating that the mutations in the RT gene of rAP20(6) were sufficient for conferring this phenotype.

TABLE 3.

Drug susceptibility of recombinant HIV-1 strains

Strain	Susceptibility to:
	PFA		AZT
	Mean IC50 ± SD (μM)a	Resistance (fold)b	Mean IC50 ± SD (μM)	Resistance (fold)
HX	29.0 ± 4.0	1	0.014 ± 0.006	1
HX88Cc	49.5 ± 9.9	1.7	0.0097 ± 0.0006	0.7
rAP6(4)d	46.2 ± 5.7	1.6	>2	>143
rAP6(4)G190e	42.6 ± 9.3	1.5	1.5 ± 0.3	106
rAP20(6)f	208 ± 32	7.2	>2	>143

^aIC₅₀s and standard deviations were determined in drug susceptibility assays performed in HT4LacZ-1 cells and were calculated from at least three independent assays. The differences in IC₅₀s of AZT for HX and rAP6(4) or rAP6(4)G190 were significant (P = 0.018). The differences in IC₅₀s of PFA and AZT for HX and rAP20(6) were also significant (P = 0.018). 

^bIC₅₀ for strain divided by IC₅₀ for wild-type HX. 

^cRecombinant strain with W88C introduced in the RT coding region. 

^dRecombinant strain with the pol fragment from clone pAP6(4). 

^eRecombinant strain with the pol fragment from clone pAP6(4) with wild-type sequence at codon 190. 

^fRecombinant strain with the pol fragment of clone pAP20(6). 

Two of the five molecular clones containing the RT coding region from strain AP6 had the G190R substitution. This change has previously been reported to confer resistance to the nonnucleoside RT inhibitor, S-2720 (13). To assess the influence of G190R on AZT susceptibility, we constructed mutant strains which had the consensus RT sequence of AP6 in an HXB2-D genetic background with [rAP6(4)] and without [rAP6(4)G190] the G190R mutation. pAP6(4) was used as the template to change the G190R mutation to the wild type. The nucleotide substitution AGA to GGA (G190) was introduced by a mutagenic oligonucleotide complementary to the sense strand to generate the construct pAP6(4)G190. Mutagenesis was performed with the Transformer site-directed mutagenesis kit (Clontech Laboratories, Inc., Palo Alto, Calif.) with modifications previously described (33). The presence of the desired nucleotide change was verified by nucleotide sequence analysis. Strains rAP6(4) and rAP6(4)G190 were recovered by cotransformation of MT-2 cells with BamHI-HindIII linearized pAP6(4) and pAP6(4)G190, respectively with BstEII linearized pHIVΔRTBstEII. Strains rAP6(4) and rAP6(4)G190 were both highly resistant to AZT and susceptible to PFA (Table 3), indicating that both genotypes could confer this phenotype.

As mutations W88G and W88S have been shown to confer 7.7- and 2.3-fold increases in resistance to PFA, respectively (31), we wished to assess the capacity of the W88C substitution observed in strains AP6 and AP20 to confer PFA resistance. The phagemid clone pHX/HOM was used to introduce the PFA resistance mutation W88C in a wild-type genetic background. pHX/HOM contains a 4.3-kb HindIII fragment of HXB2-D encompassing the complete pol gene (nucleotides 1258 to 5578) cloned into the HindIII site of pTZ19U (Bio-Rad Laboratories, Inc., North Ryde, Australia) (9). The nucleotide substitution TGG to TGT (W88C) was introduced by site-directed mutagenesis (pHX88C). The recombinant virus (HX88C) was recovered by cotransfection of MT-2 cells with HindIII linearized pHX88C and BstEII linearized pHIVΔRTBstEII. Susceptibility testing in HT4LacZ-1 cells showed that the recombinant strain HX88C was susceptible to PFA and AZT (Table 3).

This is the first report of an HIV-1 strain coresistant to AZT and PFA. AZT and PFA coresistant strains of HIV-1 could not be selected by passage in vitro (31) and required >6 months of combined AZT and PFA treatment for selection in vivo. The transition from an AZT-resistant, PFA-susceptible strain (AP6) to the coresistant strain, AP20, required six nonpolymorphic mutations in the HIV-1 RT polymerase domain (T39A, V108I, K166R, K219R, K223Q, and L228R); none of these mutations has been previously described for PFA-resistant strains of HIV-1 selected in PFA alone (19, 30). It is highly likely that at least some of these mutations confer coresistance to AZT and PFA and/or compensate for the adverse effects on enzyme function which result from drug-resistance mutations (8, 10).

Taken together, these data illustrate the extraordinary capacity of HIV-1 to adapt by mutation to almost any in vivo chemotherapeutic environment. These data also support our previous hypothesis, based on observations of a reciprocal relationship between AZT and PFA resistance, that development of an HIV-1 strain coresistant to AZT and PFA would require a complex evolutionary path involving multiple RT mutations (31).

Mutations known to confer PFA resistance (and hypersusceptibility to AZT), such as W88G and Q161L, were not observed in strain AP20. This is consistent with our hypothesis that coresistance to AZT and PFA would require RT mutations that do not cause reciprocal effects on AZT and PFA resistance (29, 31). Instead, the W88C and H208Y mutations developed in both AP6 and AP20. While mutations at codon 88 conferring PFA resistance have been described (19, 31), the cysteine substitution at this site has not been reported previously. Our data show that W88C alone does not impart PFA resistance. By contrast, the H208Y mutation has been observed in several PFA-resistant HIV-1 clinical isolates and has been shown to confer twofold increases in PFA resistance (19). H208Y can also potentiate PFA resistance when combined with other mutations conferring PFA resistance (19). However, strain AP6 was PFA-susceptible even though it had both W88C and H208Y mutations, suggesting that RT genotype modulates PFA resistance due to these mutations.

The presence of the K219R mutation in AP20 but not AP6 is notable as it also appears in the PFA-resistant, AZT-susceptible strain, PFA330AZT0.2p25, generated by in vitro selection of a wild-type strain in the presence of both AZT and PFA (31), and it has also been found in several PFA-resistant HIV-1 clinical isolates (19, 29). This mutation differs from that classically associated with AZT resistance, K219Q. Studies of the effect of PFA resistance mutations on AZT resistance in the LAI MC/Y genetic background (which has the RT mutations D67N, K70R, T215Y, and K219Q), suggest that K219Q prevents the concomitant increase in PFA susceptibility which is observed with other AZT resistance genotypes (31). Although we have previously hypothesized that mutations at this codon may permit HIV-1 to become coresistant to AZT and PFA (31), mutagenesis studies will be required for proof. It will also be of interest to determine whether the mutations mediating AZT and PFA coresistance also impair the fitness of such strains (32).

The virologic data from this patient suggest that the development of AZT and PFA coresistance follows an evolutionary path much more complicated than that of resistance to either AZT or PFA alone (14, 19, 30, 31) or to other antiretroviral combinations, such as AZT and nevirapine (23). The impediment to developing AZT and PFA coresistance may be attributable in part to the reciprocal relationship between the RT mutations mediating AZT and PFA resistance (31) and to the phenotypic reversal of AZT resistance by mutations conferring resistance to PFA (31). A recent report has shown that coresistance to AZT and lamivudine also requires multiple RT mutations (22), and drug-resistant mutations to these individual drugs show phenotypic reversal (15). While additional strains from other patients given prolonged AZT and PFA treatment will be needed to fully validate these conclusions, our observations may be exploited in the design of improved antiretroviral agents and treatment strategies.