Independent Evolution of Human Immunodeficiency Virus (HIV) Drug Resistance Mutations in Diverse Areas of the Brain in HIV-Infected Patients, with and without Dementia, on Antiretroviral Treatment

Abstract

AIDS dementia complex (ADC) in human immunodeficiency virus (HIV)-infected patients continues to be a problem in the era of highly active antiretroviral therapy (HAART). A better understanding of the drug resistance mutation patterns that emerge in the central nervous system (CNS) during HAART is of paramount importance as these differences in drug resistance mutations may explain underlying reasons for poor penetration of antiretroviral drugs into the CNS and suboptimal concentrations of the drugs that may reside in the brains of HIV-infected individuals during therapy. Thus, we provide a detailed analysis of HIV type 1 (HIV-1) protease and reverse transcriptase (RT) genes derived from different regions of the brains of 20 HIV-1-infected patients (5 without ADC, 2 with probable ADC, and 13 with various stages of ADC) on antiretroviral therapy. We show the compartmentalization and independent evolution of both primary and secondary drug resistance mutations to both RT and protease inhibitors in diverse regions of the CNS of HIV-infected patients, with and without dementia, on antiretroviral therapy. Our results suggest that the independent evolution of drug resistance mutations in diverse areas of the CNS may emerge as a consequence of incomplete suppression of HIV, probably related to suboptimal drug levels in the CNS and drug selection pressure. The emergence of resistant virus in the CNS may have considerable influence on the outcome of neurologic disease and also the reseeding of HIV in the systemic circulation upon failure of therapy.

AIDS dementia complex (ADC) is a neurologic complication of human immunodeficiency virus (HIV) infection affecting approximately 25% of immunocompromised HIV-positive individuals in the late stage of disease (20, 30). Only a minority of patients receiving antiretroviral therapies succeed in long-term suppression of viral replication (6, 9, 19). Various factors including toxicity, poor drug compliance, lack of potency, and drug interactions contribute to the limited benefit of therapy in clinical practice (5, 24). More importantly, the selection and accumulation of drug resistance mutations become inevitable in the absence of effective suppression of viral replication. However, little is known about the activity of antiretroviral drugs and the incidence and distribution of drug resistance in diverse areas of the central nervous system (CNS).

The CNS is an immunologically privileged site providing a sanctuary and reservoir for HIV (18, 22, 27). Several studies have shown that drug concentrations in vivo can vary considerably from one tissue type to another during therapy (14, 36). In addition, discordant changes in peripheral blood and cerebrospinal fluid (CSF) HIV type 1 (HIV-1) RNA levels have been reported in response to antiretroviral therapy (ART) (10, 31). Similar and discordant mutations in the reverse transcriptase (RT) and protease (PR) regions have been detected in resistant isolates from the blood or plasma compartment compared to those from the CSF or CNS (3, 4, 7, 34, 35).

Although the incidence of ADC has gone down with highly active ART (HAART), the continuation of ADC in the era of HAART raises further queries about the efficacy of these drugs in reaching the desired penetration and levels in the CNS. Only a small number of antiretroviral agents can efficiently penetrate the blood-brain barrier (BBB) and attain optimal concentrations in the CSF and CNS. There is considerable evidence showing that unique anatomical structures limit the distribution of antiretroviral drugs in the CNS: the BBB located between the blood and brain tissue and the blood-CSF barrier, primarily formed by the choroid plexus. High levels of plasma protein binding of PR inhibitors and the unidirectional efflux by P glycoprotein membrane proteins in the BBB limit the penetration and absorption of antiretroviral agents into the CNS (15, 26). Consequently, systemic treatment with antiretroviral drugs does not necessarily reduce the viral load or prevent viral replication in the CNS, likely as a result of inadequate drug penetration and hence inadequate concentrations in the CNS. Supporting this, anti-HIV drugs have been shown to accumulate differently within lymphoblastoid cell lines and peripheral blood mononuclear cells of virologically suppressed patients in vivo as follows: nelfinavir>saquinavir>lopinavir>ritonavir> indinavir (13). Also, drug concentrations in vivo have been shown to vary considerably between cell and tissue types during HAART (14). Further, recent studies from our laboratory and others have shown that there is compartmentalization of HIV-1 variants in different regions of the brain of patients with and without ADC (2, 25, 28) and also in diverse blood leukocytes of patients on HAART (23). But what remains to be determined is whether such populations in diverse areas of the brain segregate in genes (pol and PR) which are the targets of antiretroviral therapy. We therefore hypothesized that there was a possibility that drug resistance mutations in pol and PR gene regions, which develop in response to ART or HAART, may also be present within the CNS and may vary from one region to another. The present pilot study searched for evidence of independent evolution of drug resistance in diverse areas of the CNS during HAART by examining drug resistance mutations in HIV-1 PR and RT genes from HIV-infected patients with and without dementia. These analyses suggest that in addition to host selection pressure in vivo, the antiretroviral drugs and their distribution in diverse areas of the CNS may further influence viral evolution.

MATERIALS AND METHODS

Patient details, brain tissue collection, and therapy details.

Brain autopsy tissue samples were obtained, after informed family consent and institutional human ethics approval, from a variety of brain regions from 20 HIV-1-infected patients (5 without ADC, 2 with probable ADC, and 13 with various stages of ADC) (Table 1) on HAART. There were a few exceptions where no clear details of therapy were available (Table 1). Samples from patients #1509, #1791, #2057, #2020, #2434, and #2453 were obtained from the National Neurological Research Specimen Bank (Los Angeles, Calif.). Samples from patients #H0001GM, #H0007GA, #H0011DB, #H0038RR, #D0012KE, and #G0010MB were obtained from the Texas Repository for AIDS Neuropathogenesis Research (Galveston, Tex.). Samples from patients #10015, #10018, #10045, and #10064 were provided by the Manhattan Brain Bank (New York, N.Y.). Autopsy samples from patients #LD, #LB, and #LG were obtained via the Coroners Court, Glebe and St Vincent's Hospital, Sydney, Australia (Table 1).

TABLE 1.

Patient demographics

Patient	Sexa	Age at death (yr)	ADC Stageb	ARTc	Yr seropositive for HIV	Yr of death	Postmortem interval (h)
#LG	F	34	0	AZT, d4T, amprenavir, and nevirapine started 5 mo prior to death	2001	2002	23
#LB	M	48	0	∼3 yr prior to death started combivir, AZT, 3TC, ritinovir (1 mo), nelfinavir (1 mo) and then on d4T, 3TC, AZT, nelfinavir, and efavirenz (unknown duration)	1999	2002	Not recorded
#2020	M	64	0	Short period on AZT	1988	1992	21
#2434	M	62	0	AZT alternating with ddC (unknown duration): in yr of death on AZT and d4T	1988	1995	22.5
#H0011DB	M	35	0	No information available	1994	1999	11
#G0010MB	M	38	0.5	Multiple NRTIs and PIs plus efavirenz	1987	1999	5.5
#H0007GA	M	45	0.5	Multiple NRTIs and PIs given, but duration and combinations not provided	1991	1999	19
#GL	M	U	1	None 2 mo prior to death; no other details available	1996	2001	90
#H0001GM	M	U	1	Multiple NNRTIs NRTIs, and PIs given; duration of treatment and combinations not provided	1983	1999	20
#1791	M	51	2	Short period on AZT	1988	1991	36
#10045	F	31	2	d4T, 3TC, and nelfinavir started 20 mo prior to death, but duration not known	Not known	2000	9
#10064	M	49	2	d4T, 3TC, and nevirapine started ± 10 mo prior to death	1984	2000	12
#1509	M	55	3	∼12 mo on AZT; stopped ∼20 mo prior to death	1985	1989	4.4
#2057	M	45	3	∼6 mo on AZT; d4T (briefly)	1988	1992	7
#LD	F	44	3	∼22 mo on AZT; 12 mo on ddI	1988	1994	Not recorded
#2453	M	63	3	AZT 2-3 yr prior to death	1985	1995	13
#D0012KE	M	32	3	Combivir (d4T, 3TC, and nevirapine); duration of treatment not provided	1992	2000	15
#H0038RR	M	61	4	No information available	1992	2000	Not recorded
#10015	M	33	3	d4T, 3TC, and efavirenz started 8 yr prior to death; duration unknown	1991	1999	20
#10018	M	39	3	Last ART given ∼12 mo prior to death; no details provided	1994	1999	3

^aM, male; F, female; U, unknown.

^b0, 0.5, 1, 2, 3, and 4, stages of ADC.

^cAZT, zidovudine; d4T, stavudine; 3TC, 2′-deoxy-3′-thiacytidine; ddC, dideoxycytosine;

As the samples were collected from different tissue banks retrospectively to this study, only minimal details of ART were available. Many of the patients (e.g., #1509, #1791, and #2020) became HIV seropostive and/or died prior to 1992 and the era of protease inhibitors (PIs). The postmortem interval varied from 3 h (patient #10018) to 90 h (patient #GL), but for some patients this information was not recorded (patients #LD, #H0038RR, and #LB) (Table 1). All samples were collected postmortem and stored frozen at −70°C in separate sterile containers until used. The samples obtained for different brain regions varied between patients due to their availability at the time of autopsy. The CNS regions included the frontal, temporal, parietal, and occipital lobes, caudate body and nucleus, choroid plexus, putamen, cerebellum, periventricular white matter, anterior hippocampus, and corpus callosum (Table 2). Where possible, blood and/or CSF samples were also collected postmortem.

TABLE 2.

Patients and tissue samples available for PCR and sequencing

Patient	ADC stagea	Tissue sampleb
		Frontal	Temporal	Parietal	Occipital	Caudate nucleus	Choroid plexus	Putamen	Corpus callosum	Anterior hippocampus	Frontal gyrus	Deep frontal white matter	Cerebellum	Periventricular white matter	Brain lavage	CSF	Serum/plasma	Whole blood	Bone marrow	Spleen	Lung
#LG	0	PR													PR			A			PR
#LB	0	PR	A	PR	PR											PR	PR		A
#2020	0	A	A	A	A	A	A	A							A
#2434	0	A	A	A	A	A	A	A							A
#H0011DB	0										A	A	A		A					PR
#G0010MB	0.5										PR	PR	PR		PR					PR/RT
#H0007GA	0.5										A	A	A							PR/RT
#GL	1		PR/RT	PR/RT		A	PR/RT		PR/RT	PR/RT		PR/RT	PR/RT		PR/RT
#H0001GM	1										PR	PR	PR		PR
#1791	2	A	PR	A	A	A	A								A
#10045	2					RT					RT		RT	RT	A		A, P
#10064	2					RT					RT		RT	A	A		A, P
#1509	3	PR/RT	PR/RT	A	PR/RT										PR/RT	PR/RT
#2057	3	PR/RT	PR/RT	PR/RT	A	PR/RT	PR/RT	PR/RT							PR/RT
#LD	3	RT		PR	PR													PR
#2453	3		A	A	A	A	A	A							A	A	A, S
#D0012KE	3										PR/RT	PR/RT	PR/RT		A		A, P			PR/RT
#H0038RR	4										PR	PR	PR		PR					PR
#10015	3					PR					PR		PR	PR	A		A, P
#10018	3					RT					RT		RT	RT	RT		RT, P

^a0, 0.5, 1, 2, 3, and 4, stages of ADC.

^bA, availability of tissue specimen; PR, specimen for which PR gene was amplifiable and sequenced; RT, specimen for which RT gene was amplifiable and sequenced; PR/RT, specimen for which both PR and RT genes were amplifiable and sequenced; empty block, no tissue specimen available.

Proviral DNA extraction from brain tissues.

Frozen tissue sections (±3 mm in size) were finely cut up with individual scalpel blades, in individual sterile petri dishes to avoid cross contamination. The genomic DNA was extracted according to the tissue protocol of the QIAamp Mini Blood kit (QIAGEN, Chatsworth, Calif.). In addition, brain lavage samples were obtained by removing and pooling macroscopically visible blood vessels from the surface of as many of the brain tissue samples for each patient as possible. The lavage sample was collected as a representation of the viral variants circulating in the systemic blood at the time of the patient's death. DNA was extracted from the lavage samples by using the QIAamp Mini Blood kit.

RNA extraction from the CSF, plasma, and serum.

RNA was extracted from the CSF, plasma, or serum, where available, by using the QIAamp Viral RNA Isolation kit (QIAGEN). cDNA synthesis was performed by using the Reverse Transcription System (Promega, Madison, Wis.) per the manufacturer's instructions.

PCR amplification, cloning, and sequencing of HIV-1 variants.

All DNA extracts and cDNA were used in a triple-nested PCR to amplify a 376-bp region of PR and an 808-bp region of RT, separately. Although somewhat unconventional, a triple-nested PCR was required to obtain adequate amplification for further cloning and sequencing. Twelve microliters of genomic DNA was used in the first-round reactions with primers bpol1 (5′-GGGCTGTTGGAAATGTGGAA-3′) and pol1466 (5′-TTTATATGTCCATTGG-3′) in a final volume of 50 μl. The first-round reaction was performed with an initial denaturation at 95°C for 5 min followed by 35 cycles consisting of denaturation at 95°C for 30 s, annealing at 55°C for 30 s, and elongation at 72°C for 2 min. For the second-round PCR amplification, 12 μl from the first-round reaction was used as a template with primers bpol3 (5′-ATGTGGAAAGGAAGGACACCAAATGAA-3′) and MJ2 (5′-CTGTTAGTGCTTTGGTTCCCCT-3′). In a final volume of 100 μl, the same cycling conditions as those used for the first-round reaction were used for the second round, with a reduced extension cycle of 72°C for 1 min 30 s. The third-round PCR (if required) was run with 12 μl of second-round amplicon and either PR primers (PR-RT3 [5′-TGACAAGGAACTGTATCCTTTAGCTTC-3′] and pol 497R [5′-GCCATCCATTCCTGGCTTTAAT-3′]) or RT primers (MJ3 [5′-CTGTTGACTCAGATTGGTTGCACT-3′] and PR-RT4 [5′-TTCTGTATGTCATTGACAGTCCAGCT-3′]). PCR amplicons were detected on an agarose gel and visualized under UV light. The amplified PR or RT products were cloned into a pGEM-T vector (pGEM-T Easy Vector System II; Promega) per the manufacturer's instructions. Up to six clones per sample with the correct PR or RT gene insert were then sequenced on an ABI 377 automated sequencer (Applied Biosystems, Foster City, Calif.).

Sequence alignment, phylogeny, and interpretation of drug resistance.

The PR and RT sequences derived from different regions of the brain and, where available, from blood, lavage samples, CSF, or other organs and the cloned variants were aligned by using CLUSTAL-W (33) from the Genetics Computer Group package, with default settings, and improved manually. The Kimura two-parameter model in PHYLIP (16) was used to calculate the pairwise nucleotide distances. The phylogenetic trees were constructed by using the neighbor-joining algorithm integrated in the PHYLIP package (8). These phylogenetic reconstructions were further verified for tree topology and evolutionary relationships between HIV variants by using the maximum likelihood algorithm. Upon examination, if the tree topologies did not show significant differences between two algorithms, neighbor-joining trees were adopted in our analysis. These phylogenetic analyses were based on gap-stripped nucleotide sequences. The trees were bootstrapped with 100 replications to generate phylogenetically significant relationships between variants from different brain regions by using SEQBOOT. The drug resistance profiles were analyzed by using genotypic and phenotypic interpretations defined by the Stanford HIV RT and Protease Sequence Database (http://hivdb.stanford.edu). This program identifies primary and secondary resistance mutations at specific codons in the PR and RT regions and determines whether they confer resistance or sensitivity to certain nonnucleoside reverse transcriptase inhibitors (NNRTIs), nucleoside reverse transcriptase inhibitors (NRTIs), or PIs.

RESULTS

Independent evolution of primary and secondary drug resistance in diverse areas of the CNS during ART.

Although the PR and RT regions could be successfully amplified for most patients and their brain regions, the only notable exceptions from whom we could not amplify these regions, despite repeated attempts, were patients #2020 and #2434, without ADC, and patient #2453, with ADC (Table 2). Sequencing and cloning of PCR products from diverse areas of the CNS showed compartmentalization of primary and secondary drug resistance mutations to RT, PR, or both inhibitors in some of the patients from whom PR (e.g., patients #GL, #H0001GM, and #1509) (Table 3) or RT (e.g., patients #LB, #GL, #10045, and #10064) (Table 4) genes were successfully amplified. Our analyses showed that several primary drug resistance mutations to both PIs (patients #LG, #G0010MB, and #D0012KE) (Table 3) and RT inhibitors (patients #GL and #2057) (Table 4) were observed in several of the brain tissue samples and clonal variants derived from the same patient. The results for other patients with and without ADC were similar (Tables 3 and 4). Overall, the drug resistance mutations to PIs were more consistent in different brain tissue regions and clonal variants from the same patient (#LG, #G0010MB, and #D0012KE (Table 3). By contrast, the resistance mutations to the RT inhibitors were more heterogeneous within the different tissue regions and clonal variants from the same patient (#H0001GM, #10064, #GL, and #10018) (Tables 3 and 4). This study provides the first evidence of compartmentalization and possible independent evolution of primary and secondary drug resistance mutations in diverse areas of the CNS of HIV-infected patients. This is one of the notable features of our analysis, which was consistent also between patients with and without dementia.

TABLE 3.

Primary and secondary resistance mutations to PIs in patients with and without ADC

Patient	ART	Tissue	Clone	Primary and secondary resistance mutations in the PR gene
#LC	Zidovudine, stavudine, amprenavir, nevirapine	Blood	6	K20I, M36I
			9	K20I, M36I
			15	K20I, M36I, I84M
		Brain	15	K20I, M36I
			2	K20I, M36I
			3	K20I, M36I, L63P
			5	K20I, M36I
			6	K20I, M36I, I84V
		Lavage	1	K20I, M36I
			2	K20I, M36I
			3	K20I, M36I
			5	K20I, M36I
		Right lung	3	K20I, M36I
			4	K20I, M36I
			5	K20I, M36I
		Left lung	1	K20I, M36I, L33S
			2	K20I, M36I
			3	K20I, M36I
			4	K20I, M36I
			5	K20I, M36I
		Brain culture	1	K20I, M36I
			2	K20I, M36I
		Lung culture	1	K20I, M36I, F53L
			2	K20I, M36I
			3	K20I, M36I
			4	K20I, M36I
			5	K20I, M36I
#H0011DB	No information available	Spleen	1	L63P, V77I
			2	L63P, V77I
			3	L63P, V77I
			5	L63P, V77I
#G0010MB	Indinavir, nelfinavir, zidovudine, didanosine, stavudine, lamivudine, abacavir, efavirenz	Spleen	3	L63P, I93L, I50V
			4	L63P, I93L
			5	L63P, I93L
			8	L63P, I93L, L90S
				L63P, I93L
		Frontal gyrus	1	L63P, I93L
			2	L63P, I93L
			3	L63P, I93L
			4	L63P, I93L, M46V
			5	L63P, I93L
			6	L63P, I93L, M46T
		Cerebellum	3	I84M
		Frontal white matter	1	L63P, I93L
			3	L63P, I93L
			5	L63P, I93L, V77I
		Lavage	1	L63P, I93L
			2	L63P, I93L
			3	L63P, I93L, A71T
			4	L63P, I93L
			5	L63P, I93L
			6	L63P, I93L
#H0007GA	Indinavir, nelfinavir, saquinavir, ribinavir, amprenavir, zidovudine, didanosine, zalcitabine, abacavir, preveon,	Spleen	1	L63H, I93I
			6	L63H
#GL	None for 2 mo prior to death	Parietal cortex	1	I63P, V77I, I93L
			3	V77I, I93L
			4	V77I, I93L
			5	M36V, V77I, I93L
				V77I, I93L
		Anterior hippocampus	1	V77I, I93L
			3	V77I, I93L
			7	V77I, I93L
				I93L
		Choroid plexus	1	V77I
			4	D30N, M36I, V77I, I93L,
			7	L63N, I50T
				V77I, I93L
		Temporal white matter	2	I93L
			3	D30N, L63P, I93L,
			4	L63N
			5	I93L
			6
		Corpus callosum	4	V77I
				L63P, I84M, N88S
		Posterior frontal	5	V77I, I93L
			6	M36MV, V77I, I93L
			7	V77I, I93L
		Lavage	2	V77I, I93L
				L63P, V77I, I93L
#H0001GM	Zidovudine, dideoxyinosine, zalcitibane, stavudine, lamivudine, abacavir, indinavir, ritinovir, nevirapine, efavirenz, saquinavir	Frontal gyrus	1	L63P, V77I, I93L
		Cerebellum	1	I50T, L63P, V77I, I93L
			2
			4	L63P, V77I, I93L
				L10I, K20R, M36I, F53L, I54V, L63P, A71V, V82F, L90M, I93L
		Frontal white matter	2	L10I, K20R, M36I, F53L, I54V, L63P, A71V, V82F, L90M, I93L
			3	L10I, K20R, M36I, F53L, I54V, L63P, A71V, V82F,
			4	L10I, K20R, M36I, F53L, I54V, L63P, A71V, V82F, L90M, I93L
			5	L10I, K20R, M36I, F53L, I54V, L63P, A71V, V82F, L90M, I93L
		Lavage	7	L63P, V77I, I93L
#1509	Zidovudine (stopped 2 yr prior to death)	Frontal		L63P, A71V, V77I, I93L
		Temporal		L10I, L63P, A71V, V77I, I93L,
		Occipital		L63A, A71L, V77I, D30F, I50R, V82P, V32I, F53Q
				I94N, M36E, I54W, N88T, I47D, G73K, L90V
		CSF		L10I, L63P, A71V, V77I, I93L,
		Lavage		L63P, V77G, I93L, K20R, V32A, V82G, N88I
#2057	Zidovudine, stavudine,	Frontal		V77I
		Caudate nucleus		L63P, V77I
		Putamen		V77I
		Temporal lobe		V77I
		Parietal lobe		V77I
		Choroid plexus		M36I, L63P, V77I
		Lavage		V77I
#LD	Zidovudine, dideoxyinosine	Blood 1	1	L63P
			2	L63P
			3
			6	L63P
		Blood 2	1	L63P
			2	L63P
			3	L63P
			5	L63P
			6	L63P
		Left occipital	3	V77I
			4	M36I, L63P, A71T
		Right occipital	1	L63P
			3	L63P
			6
			7	L63P
			1	M36V, I93L
			3	L33V, I93L
			4	L63P
			5	L63P
			7	M36I, I93L
			8	M36I, I93L
#D0012KE	Stavudine, lamivudine, nevirapine	Spleen	1	L63P, V77I, I93L
			3	L63P, V77I, I93L
			8	L63P, V77I, I93L
				L63P, V77I, I93L
		Frontal gyrus	1	L63P, V77I, I93L
			2	L63P, V77I, I93L, V82D
			3	L63P, V77I, I93L
			4	L63P, V77I, I93L, L90S
			6	L63P, V77I, I93L
				L63P, V77I, I93L, L90M
		Cerebellum	2	L63P, V77I, I93L
			4	L63P, V77I, I93L
			5	L63P, V77I, I93L
			9	L63P, V77I, I93L, I84M
#I0015	Stavudine, lamivudine, efavirenz	Midfrontal gyrus	1	L63S, A71V, V77I, I93L
			2	D30V, V77I
			3	I93F
			4	D30V, V77I
			5
			6	L63P, A71T, I93L
				L63P, V77I, I93L
		Periventric white matter	1	M46T, L63P, A71T, I93L
			3	L63P, V77I, I93L
			4	L63P, V77I, I93L
			6	L63P, V77I, I93L
			7	L63P, V77I, I93L
				L63P
		Cerebellum	1	L63P, V77I
			3	L63P, V77I
			5
		Caudate body	1
			3
			4	L63P, A71V, V77I, I93L
#H0038RR	No information available			L63P
		Spleen	1	L63P, V77I, I93L
			2	L63P
			3	L63P, K20R, M36I, F53L, A71V, V82F/V, L90M,
			4	L63P, M36T, I84V
			5	L10I, K20R, M36IM, F53L, I54V, A71V
		Frontal gyrus	1
			2	L63P, V77I, I93L
			3	L63P, V77I, I93L
			4	L63P, V77I, I93L
			5	L63P, V77I, I93L
			6	L63P, V77I, I93L
		Cerebellum	1	L63P, V77I, I93L
			2	M46T, N88S
			5	L63P, V77I, I93L
				L63P, V77I, I93L, K20R
		Frontal white matter	3	L63P, V77I, I93L
			4	L63P, V77I, I93L, V82A
			5	L63P, V77I, I93L
			6	L63P, V77I, I93L
			8	V77I, I93L
		Lavage	1	L63P, V77I, I93L
			4	V77I, I93L
			7	L63P, V77I, I93L

TABLE 4.

Primary and secondary resistance mutations to RT inhibitors in patients with and without ADC

Patient	ART	Tissue	Clone	Primary and secondary resistance mutations to RT inhibitors
#LB	Stavudine, lamivudine, nelfinavir, zidovudine	Left parietal		V75I, F77I, Y115F, Q151M, M184V, T215F
		Left frontal		T215P
		Left occipital
		Right parietal
		Right frontal
		CSF		M184V, T215P, K219Q
		Plasma
#G0010MB	Indinavir, nelfinavir, zidovudine, didanosine, stavudine, lamivudine, abacavir, efavirenz	Spleen	1	K70M, L74S
			2
			4	A62V
			5	A62V
			6	M184I
			7
			8
			10	F77S, K219Q, M230V
#H0007GA	Indinavir, nelfinavir, saquinavir, ribinavir, amprenavir, zidovudine, didanosine, zalcitabine; no amplification of RT from frontal gyrus, cerebelleum, or frontal white matter	Spleen	1	M41L, E44D, K103N, Y181CY, L210W, T215S, F227L
			6	I62V, K70E
#GL	Not known, but none for 2 mo prior to death	Parietal cortex	1	D67G, Y181I, T215D, K219Q, Q151R
			2	D67G, Y181I, T215D, K219Q
			3	D67G, Y181I, T215D, K219Q, F227S
			4	D67G, Y181I, T215D, K219Q
			5	D67G, Y181I, T215D, K219Q
		Anterior hippocampus	1	D67G, K70R, T215V, K219Q
			3	D67G, K70R, T215V, K219Q
			7	D67G, K70R, T215V, K219Q
				D67G, K70R, T215V, K219Q
		Choroid plexus	1	D67G, K70R, K219Q
			4	D67G, K70R, K219Q
			7	D67G, K70R, K219Q, Y188H
		Temporal white matter	2	M41L, E44D, V118I, L210W, T215Y, T69A
			3	M41L, E44D, V118I, L210W, T215Y
			4	M41L, E44D, K70N, V118I, L210W, T215Y
			5	D67G, K70R, T215V, K219Q
			6	M41L, E44D, V118I, L210W, T215Y
				M41L, E44D, V118V, L210V, T215D/Y, K219Q
		Corpus collosum	4	M41L
		Posterior frontal	5	D67G, K70R, T215V, K219Q
			6	D67G, K70R, T215V, K219Q
			7	D67G, K70R, T215V, K219Q
		Lavage
#10045	Stavudine, lamivudine nelfinavir; last given 24 mo prior to death			K219Q
		Midfrontal gyrus	1	L210V, F77L
			6	L234S
		Periventric white matter	1
			2	L210V, K219Q, L234P
			5	P236H
				K219Q, K103R
		Cerebellum	1
			3	K219Q, G190V
			7	Q151R, F227V
			8	K219Q, M41V
		Caudate body		K219Q
#10064	Stavudine, lamivudine, nelfinavir; last given 12 mo prior to death	Midfrontal gyrus	1	K103E, V118A, K219P
			2	L210V
			3	L210V
			6	L210V
		Cerebellum	1	L210V
			2	M41I
			4	L210V, Y115F
			5	F116L
			6	L210V, K219P
		Caudate body	12
			13
#2057	Zidovudine, stavudine,	Caudate body		D67N, K70R
		Putamen		D67N, K70R, T69P
		Temporal		D67N, K70R
		Parietal		D67N, K70R
		Choroid plexus		D67N, K70R
		Lavage		D67N, K70R
#LD	Zidovudine, didanosine	Right frontal	1	M230V
			2
			3
			4	V118A
			5	V118A
			6
#D0012KE	Stavudine, lamivudine, nevirapine	Spleen	1
			3	Q151R, V179A
			8
		Frontal gyrus	1
			2
			3
			4	V179D
			6
		Cerebellum	2
			4
			5
			9	L74S
		Frontal white matter	1
			2
			3
			4	M230V
			5
#10018	No details; last given 12 mo prior to death	Midfrontal gyrus	1	K219Q
			3	K219Q
			8	D67E, L210V
		Periventric white matter	7
			11
		Cerebellum	2	K219Q, L210V
			3	K219Q
			4	K65R
			5	K219Q
			6	K219Q, V108E
		Caudate body	4
			9	Y181H
		Plasma		T69P, A98S

Noncanonical mutations evolving during ART in diverse areas of the CNS.

In addition to resistance mutations in RT and PR genes, other mutations, termed noncanonical, which do not confer resistance, were also analyzed in order to delineate genotypic differences between HIV strains from patients with and without dementia. We observed a pattern of compartmentalization between different regions of the brain, which was more pronounced than the pattern of segregation of primary and secondary resistance mutations (Table 5). These genotypic differences were apparent between diverse regions of the CNS and also between clones derived from each of the regions from which we obtained positive PCR and sequencing results. In Table 5, we attempted to present a cumulative picture of interregional segregation of mutations in different patients; therefore, the individual clones have not been shown. These clones are identified in Tables 3 and 4. At this time the actual biological significance of these noncanonical mutations in the development of drug resistance remains unknown.

TABLE 5.

Noncanonical mutations in diverse areas of the CNS^a

Patient and brain region	Noncanonical mutations in PR	Noncanonical mutations in RT
#LG
Blood	W6R, I15T, Q18R, I84M	−
Frontal brain tissue	Q6R, L63P, I84V	−
Lavage	I62T, R57G	−
#HOO11DB
Frontal gyrus	N37C, G49E, L89P	−
Frontal white matter	N37C, G49E, L89P	−
Cerebellum	N37C, G49E, L89P	−
Brain lavage	N37C, G49E, L89P	−
#G0010MB
Frontal gyrus	M48V/T, P39L, K43R, R8F	N136D/T, K11R, D121G, G51R, T216P, E6G, K220G/E, L149P, W212G, D218R, P14S, Q197H, T200P, R206E, E6G
Cerebellum	I15M, P9T	−
Frontal white matter	I84M, Q2H, W6R, K14R, K45R, E34G, V77I	−
Lavage	I62V, L5P, G51R, R8L, Q58R, C95R	−
#GL
Parietal cortex	K55E, R8L, I13K, E34K, L89P, M36V	Q151R, F227S
Anterior hippocampus	N83D, G68R, N83S, V32A	T7A, E233G, D218G, K220R
Choroid plexus	R41K, C67R, T74A, R8Q, D25N, G51E, K70E, D60N, R87K, R57G, M36I	Y188H, H198P, V60I, P170S, M16I, N81S, Q242R, Q222R
Temp white matter	R8L, Q18L, R8P, T31I, P18S, D30N, K14K/R, H69R	T69A, K13R, T131I, Y146H, R211K, K201R, K30E, F124L
Corpus collosum	I15V, L38S, R57K, I72L, N88S, L89P	R172K, R211K, C38W, L80P
Postfrontal	I13T, G51A/G, N83Y	D110V, N175I, P140L, L149F, I195M, K102R, K154E, R206S
Lavage		None
#H0001GM
Frontal gyrus	None	−
Cerebellum	None	−
Frontal white matter	None	−
Lavage	None	−
#1791
Temporal	+, not analyzed	−
Frontal	−	−
Parietal	−	−
Occipital	−	−
Caudate nucleus	−	−
Choroid plexus	−	−
Brain lavage	−	−
#10045
Midfrontal gyrus	−	F77L, L234S, E169K, R172K, R206G, T216P, Q222P, T216P, T27A, K22N, R78G, F171Y, I244L, Q174H, R72G
Perivascular white matter	−	S3R, R72G, I180V, H221P, Q222P, I47T, R78G, Y144C, Q145R, Q197K, H235L
Cerebellum	−	K20N, V111G, E169K, Q91L, L187W, E194K, H198L, E204D, P25L, L228P, I132M, I195L, S117P, V90A, K20R
	−	M16V, K20N, G51E, V148G, R172K, E194K, H198L, H208P, K22R, Q23L, E194G, P217T, R206S, F214S
#1509
Frontal	−	None
Temporal	L10I	L10V
Occipital	D30F, I50R, V32I, F53Q, M36E, G73K	Y59S
CSF	L10I	L10V
Lavage	R8P, G94D, K70R, I64T, L89V, T31A, I93L, R87K	K14M, L19I, A28S, P39G, G40R, V56G, I64V, I72V, N83P, Q92K
#LD
Blood	V11A, W42R, I64V, T96S, D29H	−
Left occipital	Q2R, R8P, E34K, V77I, M36I, G86R	−
Right occipital	W6R, I64V, G94D, K20R	−
Right parietal	T26A, T31A, C95R, I93L, M36V/I	−
Right frontal	−	D113N, F171L, L12S, E53G, V148A, K73N, Q242R, V21I
#2453	No amplification	No amplification
#D0012KE
Frontal gyrus	V82D, L90S, C95Y, G78R, L89P, I62T, T74I, Q92P, R87K,	T7A, K173N, P4S, I180T, R206G, I47T, W88R, Q91R
Cerebellum	K43R, K70R, L23P, K45R, F99S, I64T, L89V, L90M, I84M,	V148G/V, E169K, R172K/R, H198P, E204V, L260Y, K49R, S68G, N136D, V90I
Frontal white matter	I62T, K45E, L90S, V77I	Y56H, E169D, W212G, G45R, T240A, E28D/E, E224G, V111M, T27A, P226Q
#H0038RR
Frontal gyrus	K45R, I72T, C95R, I15V, F99L
Cerebellum	K45R, I72T, C95G
Frontal white matter	D25Y, K43R, R8L, T31A, R8L, K45R
Lavage	R8L, K45R
#10015
Midfrontal gyrus	D30V, I15V, I64T, Q18R, T12A, E34G, Q18H, C95Y, Q7R, L19Q, L89P, L97I
Perivascular white matter	V11A, A22P, K43R, E21D, I15V, T31A, T96S, R8T
Cerebellum	G27R, R57K, I62M, I64V, M36I, L10I, R57K, T96I, W6R, V11A, K70E, V75I
Caudate body
#10018
Midfrontal gyrus	−	K49R, S68G, R72G, K104N, W212G, N57D, Q197R, V8E, S68G, P170S, M16I, V21A/V
Perivascular white matter	−	M16I, K82M, R143G, N57D, Q91R, T139A, E233G, V21A
Cerebellum	−	E42Q, I132L, L246P, P247Q, R72S, Y183H, L246P, R72S, Y183H, P247Q, T27S, C38R, K166E, I195V, Q242P
Caudate body	−	D177E, I178L, Q222R, I244V, E138G, D177E, I178L, G196E, V241I, I244V
#LB
Left parietal	−	F171Y, Q174R, K11R, T200I, I202V, F214L
Left frontal	−	K166R
Left occipital	−	K166R
Right parietal	−	S162C
Right frontal	−	F214L
CSF	−	Y183S, E204D, D218P, V245E, I178L
Plasma	−
#2020	Only 1 PCR positive, not included
#2434	Only 1 PCR positive, not included
#2057
Frontal	None	None
Caudate nucleus	None	None
Putamen	None	None
Temporal	None	None
Parietal	None	None
Choroid plexus	None	None
Lavage	None	None
#H0001GA
Frontal gyrus	L63H	−
Cerebellum	L63H	−
Frontal white matter	L63H	−
#10064
Midfrontal gyrus	−	K103A, V118A, K219P, V10L/R, K22I, W24R, R172K, E194K, V245E, V8G, P9S, T240A, V10E, S68G, I132L, R143S, V245K
Cerebellum	−	R125G, L193F, E204G, K220E, P226H, Q23R, E42G, K82R, Y146H, W212R, I195V, L149P, M41I, F116L, K219P
Caudate body	−	Y56S, V10A, K13N, T165A, K43N, P55L, Y56S, T216A

^a−, negative for both RT and PR gene amplification by PCR. Where positive for RT and PR at least five clones were analyzed for each region. The table represents a cumulative analysis of five clones for a given region. None, no noncanonical mutations in RT or PR region; only primary and secondary resistance mutations were observed +, PCR positive but could not be analyzed, as other samples from the patient were PCR negative. No amplification, could not be amplified by PCR.

Naturally occurring polymorphisms in the CNS with drug resistance genotypes.

As therapy details were available from most patients, we also analyzed the resistance mutations that were harbored by patients naturally. The fact that some of the primary and secondary resistance mutations, such as L63P (amino acid change from L to P at residue 63), V77I, I93L, and E35D (Table 3, were so consistently present in many of the patients in this study, irrespective of their ART regimen, suggests that some of these resistance mutations may not have developed as a result of drug pressure. It is also likely that these mutations may rather be naturally occurring polymorphisms (17) of the virus variants not only within the CNS but also in the systemic circulation, as the same mutations were frequently present in blood, CSF, spleen, and lavage samples from the same patients (Table 3). Interestingly, PR resistance mutations were also detected in patients not receiving PIs (patients #LG, #GL, #1509, #2057, #LD, #D0012KE, #10015, and #H0038RR (Table 3), further supporting the suggestion that these mutations may not be all drug induced but may be natural polymorphisms.

Phylogenetic analysis supporting independent evolution.

Viral variants in both RT and PR genes were analyzed from diverse areas of the brain in order to determine phylogenetic relationships between various genotypes and also to assess the extent of intrapatient diversity in drug resistance in the CNS. Although, the phylogenetic relationships of the PR (Fig. 1) and RT regions (Fig. 2) support both intra- and interregional viral variation in diverse areas of the CNS, no specific segregation pattern of viral variants between patients with and without dementia was apparent. The most notable features of phylogenetic reconstructions were the compartmentalization and independent evolution of drug resistance mutations to PR and RT inhibitors in different regions of the brain (Fig. 1 and 2) and a good regional concordance between viral strains and clonal variants derived from a given brain region. This remains consistent with our previous studies of the Env region from a smaller subset of patients (28). This compartmentalization was evident in patients both with and without ADC. Most of the tissue and clonal variants arose from two or three major branches of the trees. However, in most patients, the variants present in at least one region of brain tissue (an outlier) appeared to be distinct from the other variants present in other brain regions from the same patient in both RT and PR regions (Fig. 1a, LL.2; Fig. 1b, C3; Fig. 1d, C2; Fig. 1f, parietal; Fig. 1g, LO.3; and Fig. 2a, right parietal). This suggests that infection of the CNS may have occurred as two distinct events and at different times. It is possible that the more outlying variant is archival and gained entry into the CNS during the early stages of HIV-1 infection while the other variants arose more recently from certain quasispecies circulating in the blood and/or CSF of the patient.

FIG. 1.

Neighbor-joining phylogenetic tree based on a 297-bp region of the PR gene for patients #LG (a), #G0010MB (b), #GL (c), #H0001GM (d), #1509 (e), #2057 (f), #LD (g), #D0012KE (h), #10015 (i), and #H0038RR (j). The trees were bootstrapped with 100 replications. The distance bar at the bottom of the tree indicates the degree of variation on the horizontal plane. Boldface indicates analysis of sequences obtained directly from PCR amplification of DNA extracted from the tissue. Numbers are for individual clones and correspond to the tissue regions from which they were derived. Sequences for HXB2 and JRCSF, subtype B prototype strains, were derived from GenBank. All trees were rooted with HXB2 as the outlier. FG, midfrontal gyrus; FWM, frontal white matter; C, cerebellum; S, spleen; TWM, temporal white matter; PFWM, posterior frontal white matter; AFWM, anterior frontal white matter; CC, corpus collosum; CP, choroid plexus; HIP, anterior hippocampus; LAV, lavage; CSF, cerebrospinal fluid; PC, parietal cortex; LO, left occipital; RO, right occipital; RP, right parietal; RF, right frontal; B1, blood 1; B2, blood 2.

FIG. 2.

Neighbor-joining phylogenetic trees based on a 741-bp region of the RT for patients #LB (a), G0010MB (b), #GL (c), #10045 (d), #10064 (e), #LD (f), #D0012KE (g), and #10018 (h). The trees were bootstrapped with 100 replications. The distance bar at the bottom of the tree indicates the degree of variation on the horizontal plane. Boldface indicates analysis of sequences obtained directly from PCR amplification of DNA extracted from the tissue. Numbers are for individual clones and correspond to the tissue regions from which they were derived. Sequences for HXB2 and JRCSF, subtype B prototype strains, were derived from GenBank. All trees were rooted with HXB2 as the outlier. FG, midfrontal gyrus; FWM, frontal white matter; C, cerebellum; S, spleen; TWM, temporal white matter; PFWM, posterior frontal white matter; AFWM, anterior frontal white matter; CC, corpus collosum; CP, choroid plexus; HIP, anterior hippocampus; LAV, lavage; CSF, cerebrospinal fluid; PC, parietal cortex; LO, left occipital; RO, right occipital; RP, right parietal; RF, right frontal; B1, blood 1; B2, blood 2.

Distinct nature of brain lavage-derived variants.

As we did not have blood from a number of patients, the information on blood variants for each patient was derived by obtaining brain lavage samples by removing and pooling macroscopically visible blood vessels from the surfaces of as many of the brain tissue samples as possible. These strains showed only partial genetic relationships with the tissue PR region variants from which the lavage sample was obtained. Thus, apart from being the first demonstration of compartmentalization of drug resistance mutations within different regions of the CNS, another notable feature of our analysis was the distinct molecular nature of HIV-1 variants present in the lavage sample obtained from blood present in macrovisible vessels in the brain. This was seen in patients #LG, G0010MB, H0001GM, 2057, 1509, and #H0038RR (Fig. 1 a, b, d, e, f, and j, respectively). Overall, the phylogenetic analysis (based on 297 bp in the PR gene region) (Fig. 1) of these variants from the lavage samples showed distinct and also intermixed clustering with the tissue-derived HIV-1 variants. Nonetheless, the lavage variants consistently were genetically related in each case to the respective tissues from which they were derived. The only difference was the heterogeneous viral quasispecies present in lavage samples as opposed to tissue-derived variants, suggesting that only a fraction of variants enter the brain tissue successfully. Where blood was available, there was a distinct difference between the variants present in the blood or lavage and most of the variants in different brain regions. In addition to compartmentalization of viral variants, there was some intermingling of CSF- and lavage-derived HIV variants with the brain tissue-derived variants (Fig. 1a to f and j and Fig. 2a), which further supports selective trafficking of drug-resistant and other HIV-1 variants between the CNS and blood.

Detection of PR resistance mutations in PR inhibitor-naive patients.

Interestingly, PR resistance mutations were also detected in patients not receiving PIs (Table 1) (patients #LG, #GL, #1509, #2057, #LD, #D0012KE, #10015, and #H0038RR [Table 3]), further supporting the suggestion that these mutations may not be all drug induced but may be natural polymorphisms. However, the sexual transmission of these mutations from drug-experienced patients cannot be ruled out.

DISCUSSION

The introduction of HAART, which includes PR and RT inhibitors, has been a major step toward the control of HIV-1 infection and also in the treatment of CNS infection by HIV. However, only a small number of anti-HIV drugs pass the BBB and reach effective levels in the CNS. Thus, it is likely that suboptimal concentrations of drugs reside in the brain and lead to HIV- and drug-resistant variants being concealed in sanctuaries created in diverse areas of the CNS. Thus, drug resistance in the CNS may become a consequence of incomplete suppression of HIV. As in the era of HAART, AIDS dementia continues to be a problem; it is possible that this may be related to poor penetration and/or suboptimal concentrations of drugs in the brain. Whether the independent evolution of drug-resistant variants occurs in the brain during HAART and whether the emergence of resistant virus has some relationship to the continued problem of ADC in some HIV patients were the subjects of this study. In this study, we analyzed antiretroviral drug resistance mutation profiles in RT and PR regions of HIV-1 in diverse areas of the CNS of patients with and without dementia during HAART.

The most notable feature of our analyses is the demonstration that primary and secondary drug resistance mutations to both RT and PR inhibitors remain compartmentalized in diverse regions of the brain during ART or HAART. A wide range of resistance and nonresistance mutations shown here stands as a testimony to the genetic plasticity of HIV. Overall, the phylogenetic relationships of the PR and RT regions support both intra- and interregional viral variation within the CNS. Further, this is consistent with the regional compartmentalization and independent evolution of the env gene variants in diverse areas of the CNS (28). It further indicates that systemic treatment does not necessarily reduce the viral load or prevent viral replication in the CNS. In fact, differential penetration along with suboptimal drug concentrations may permit independent development and evolution of drug resistance mutations within the CNS. Supporting this, the CNS is already a recognized reservoir and an immunologically privileged site for HIV infection and replication. Therefore, these data are particularly significant, as both primary and secondary resistance mutations, which actually confer resistance to antiretroviral drugs, were regionally compartmentalized in diverse areas of the CNS during therapy. Although the clinical and biological significance of this regional segregation of drug-resistant variants remains unclear, it is plausible to hypothesize that the drug-resistant mutants in the CNS may reseed systemic circulation, resulting in failure of therapy.

In this study, we also analyzed such mutations in diverse areas of the CNS because the role of mutations, other than drug resistance (noncanonical), remains unclear. Interestingly, a similar degree of compartmentalization was also apparent at the level of noncanonical mutations. The role of these polymorphisms in conferring drug resistance remains undetermined, but their role in conferring cross-resistance to antiretroviral drugs cannot be ruled out. In our study, most of the tissue and clonal variants, at both the level of resistance and nonresistance mutations, arose from two or three major branches of the trees. However, in most patients, the variants present in at least one region of brain tissue (an outlier) appeared to be distinct from the other variants present in other brain regions from the same patient. This suggests that infection of the CNS may have occurred as two distinct events and at different times within the same patient. It is possible that the more outlying variant is archival and gained entry into the CNS during the early stages of HIV-1 infection while the other variants arose more recently from certain quasispecies circulating in the blood and/or CSF of the patient. Further, a critical examination of drug-resistant mutants and noncanonical mutations showed no genotypic distinction between patients with and without dementia. To some extent, this diversity was expected because of the genetic differences between infecting strains in different patients. Moreover, distinct changes that may segregate patients with and without dementia may lie in other regions of the HIV genome (28).

A serious concern is that in the era of HAART, ADC continues to be a problem. Thus, the degree of regional compartmentalization of drug-resistant viral variants during HAART suggests that poor and perhaps differential penetration of antiretroviral drugs may occur in the CNS. This may encourage the independent development of HIV quasispecies in regions of the brain with characteristic resistance profiles through a milieu of subtherapeutic drug concentrations. This phenomenon may further be accentuated by the varied tropism of HIV variants arising as a consequence of selection pressures imposed on HIV in each of the local areas and also by cellular differences in the CNS that may lead to differential permeability to antiretroviral drugs. Further, a great concern is that mutations conferring resistance to multiple antiretroviral drugs may predominate in brain regions where drug levels are suboptimal (4). Hence, the CNS may become a source and a reservoir for multiple drug-resistant viral strains that may emerge systemically at the failure of therapy. Further, we also reiterate that, in addition to resistance mutations, there was also a marked absence of drug-resistant mutants in certain areas of the CNS in almost all patients. Presently, there is no clear explanation for the notable absence of resistance in certain regions, but several studies have shown that drug concentrations in vivo can vary considerably from one tissue type to another, or one organ to another, during therapy (14, 36). In addition, some compartments, including the CSF (12), genital secretions (21), and lymphoid tissue (29), have been shown to be poorly accessible to different antiretroviral drugs. In rhesus monkeys a dramatic difference in the levels and concentration-time profiles of 2′-deoxy-3′-thiacytidine (3TC) between lumbar and ventricular CSF was observed (1). Therefore, the suboptimal therapeutic drug levels in the CNS and poor penetration of these drugs in various regions of the CNS may be more likely explanations for the independent evolution of drug-resistant variants in diverse areas of the CNS. Cumulatively, the spectrum of primary and secondary resistance mutations in diverse areas of the CNS, which develop as a consequence of the administration of ART or HAART, may significantly influence the outcome of therapy both in the CNS and systemic circulation.

We observed a significant degree of variation in the mutations present in clonal variants derived from the brain tissue. This indicates the importance of cloning in assessing the true drug resistance profiles from HIV-infected patients and also predicting the role of minority populations in influencing the outcome of antiretroviral therapy. Furthermore, given that the resistance is more likely to be clinically significant (32), it remains to be determined which drug resistance mutations and polymorphisms may be important clinically. Further, whether drug-resistant variants possess altered neurotropism as a consequence of changes occurring in tandem in other regions (such as env) implicated in viral tropism remains unclear. Further, these changes can certainly impact the outcome of neurologic disease during therapy. Previously (28), our group showed that HIV-1 strains from diverse areas of the CNS showed a higher propensity to grow in macrophages of patients with ADC, as opposed to patients without ADC. The key determinants of viral tropism lie in the envelope gene. Therefore, the changes in RT and PR regions occurring under drug pressure are bound to impact changes in the envelope gene. These changes may have a direct effect in altering tropism. Thus, a better understanding of the drug resistance mutations that emerge in the CNS during HAART is needed to minimize the possibility that they may “seed” the systemic circulation (11). In addition, it is important to dissect mutations, which may occur in drug target genes and the envelope gene in tandem, as these changes may clarify molecular issues related to tropism and neurovirulence of drug-resistant strains.

In this study we also observed that some of the primary and secondary mutations such as L63P, V77I, I93L, and E35D were consistently present irrespective of the ART regimen. This suggests that these mutations may be naturally occurring polymorphisms (17) of the virus variants not only within the CNS but also in the systemic circulation, as these mutations were frequently present in blood, CSF, spleen, and lavage samples of the same patients (Table 3). Interestingly, PR resistance mutations were also detected in PR inhibitor naive patients (e.g., #LG, #GL, #1509, #2057, #LD, #D0012KE, #10015, and #H0038RR) (Table 3), further supporting the suggestion that these mutations may not all be drug induced but may be natural polymorphisms. Further, it cannot be ruled out that some of these naturally occurring mutations may have been sexually acquired from drug-experienced partners. Recent unpublished data from our group, obtained from a survey of Australian patients on HAART, have shown that in 15% of HIV-positive homosexuals, both RT and PR mutations could be sexually transmitted (D. E. Dwyer and N. K. Saksena, Bristol Myers Squibb Study, 2000-2001). Nonetheless, although these mutations (L63P, V77I, I93L, and E35D) on their own do not confer resistance to PIs, they may predispose an individual to developing resistance and in combination with other mutations may enhance the risk of development of primary drug resistance and eventual treatment failure. In a recent study by Sune et al. the L63P polymorphism, while not conferring inhibitor resistance on its own, did however provide significant replication benefit to certain mutant viruses (32). Presently, there is no clear explanation of whether drug resistance mutations in viral variants may provide more CNS affinity to viral strains or cause altered tropism of HIV strains. But with data in hand, we believe that drug penetration, drug pharmacokinetics, and viral load within the brain tissue may have a role in defining the differential distribution and possible independent evolution of viral strains in diverse areas of the CNS. Moreover, it is likely that the evolution of HIV drug target genes may occur in concert with the envelope gene. These changes, occurring in tandem during HAART, may have a critical role in defining the therapy outcome and also the neurotropism and selection of viruses fit to infect the brain during therapy.

This study is the first to provide a detailed analysis of drug-resistant HIV-1 genotypes regionally compartmentalized in diverse regions of the CNS during antiretroviral therapy. Our data clarify that both primary and secondary resistance mutations are regionally distributed in diverse areas of the CNS, which may be significantly important in a clinical context. It remains unknown which cell types in the CNS may harbor resistant viruses and permit their active replication and whether poor penetration of drugs and suboptimal drug concentrations have some role in encouraging viral replication in diverse areas of the CNS. Nonetheless, further clarification of these aspects may have important implications for future design of antiretroviral treatment strategies for treating CNS infection.