CD8⁺-Cell-Mediated Suppression of Virulent Simian Immunodeficiency Virus during Tenofovir Treatment

Abstract

The ability of tenofovir to suppress viremia in simian immunodeficiency virus (SIV)-infected macaques for years despite the presence of virulent viral mutants with reduced in vitro susceptibility is unprecedented in this animal model. In vivo cell depletion experiments demonstrate that tenofovir's ability to suppress viremia during acute and chronic infection is significantly dependent on the presence of CD8⁺ lymphocytes. Continuous tenofovir treatment was required to maintain low viremia. Although it is unclear whether this immune-mediated suppression of viremia is linked to tenofovir's direct antiviral efficacy or is due to independent immunomodulatory effects, these studies prove the concept that antiviral immune responses can play a crucial role in suppressing viremia during anti-human immunodeficiency virus drug therapy.

Although highly active antiretroviral therapy (HAART) regimens provide significant progress in the clinical management of human immunodeficiency virus (HIV) infection, their long-term use is often limited by problems such as incomplete virus suppression and drug resistance. To alleviate this problem, it is believed we need to invoke the help of the immune system to help limit virus replication and/or the virus-induced immune dysfunction.

Immune responses, especially cell-mediated immune responses (CD4⁺-T-helper cells and CD8⁺ T lymphocytes) are thought to play an important role in controlling virus replication and determining disease-free survival. Evidence for this comes from studies in untreated long-term nonprogressors (for a review, see reference 2). In addition, macaque data have shown conclusively that CD8⁺ lymphocytes play an important role in limiting lentivirus replication because in vivo CD8⁺-cell depletion of untreated macaques resulted in an increase in viremia (23, 34, 67).

A number of dilemmas exist, however, for patients receiving HAART. HAART initiated during acute HIV infection can often preserve or increase antiviral immune responses, but episodes of transient viremia may be required to maintain these responses (54, 63). In contrast, when HAART is initiated during chronic HIV infection, HIV-specific immune responses are more difficult to restore (for a review, see reference 2).

The goal of immunotherapeutic strategies is to augment these antiviral immune responses to prevent the emergence and/or the replication of drug-resistant mutants during HAART or to allow simplified regimens or periods without drug treatment. Although different approaches are investigated to stimulate antiviral immune responses (e.g., structured treatment interruptions, active immunizations, and cytokines), the results have been poor or success has been limited mainly to patients with acute infection with often transient results (2, 3, 25, 63). Progress in this area may depend on a better understanding of the contribution of antiviral immune responses to the reduction of viremia during HAART and of the complex in vivo interactions of viral replication, drug resistance, immune responses, and drug pharmacokinetics.

To gain further insights in the viral and host immune factors that determine successful therapy, animal models can be very useful because they allow controlled experimental approaches that are not feasible in humans. Infection of macaques with simian immunodeficiency virus (SIV) is a good animal model for such studies because it shares many similarities in disease pathogenesis with AIDS and has also been used to study the efficacy of antiviral compounds, including the emergence, virulence, and clinical implications of drug-resistant mutants. The reverse transcriptase (RT) inhibitor tenofovir {9-[2-(phosphonomethoxy)propyl]adenine} has been used extensively in this macaque model of AIDS because of its unprecedented efficacy, compared to other compounds, to prevent infection or suppress viremia (33, 55, 64, 69-71, 75-79, 80, 82, 85, 88). Most other commonly used antiviral drugs that have been tested in the macaque model were usually able to delay or reduce the peak of primary viremia if given early during infection, but in contrast to tenofovir, these other drugs were not very efficient in suppressing viremia once virus dissemination was already well established or the emergence of drug-resistant viral mutants inevitably led to increased virus replication and disease (22, 30, 35, 47, 70, 84, 87, 90, 102).

The reasons for this high efficacy of tenofovir in the macaque model have been unclear but warrant further investigation, as this could lead to the development of additional antiviral strategies. The experiments described here demonstrate that suppression of viremia during tenofovir treatment of SIV-infected macaques requires CD8⁺-cell-mediated antiviral immune responses. Our data provide proof of concept for a model of antiviral drug therapy in which the combined forces of antiviral immune responses and antiviral drugs are the key to success, especially in the presence of drug-resistant viral mutants. The available evidence suggests that the reason for tenofovir's success in this animal model is because tenofovir treatment allows the host to mount strong CD8⁺-cell-mediated immune responses that complement tenofovir's antiviral effects.

MATERIALS AND METHODS

Animals.

All rhesus macaques (Macaca mulatta) were from the type D-retrovirus-free and SIV-free colony at the California National Primate Research Center (CNPRC). The newborn and infant macaques were hand reared in a primate nursery and were housed in accordance with American Association for Accreditation of Laboratory Animal Care standards. We adhered to the standards outlined in the Guide for Care and Use of Laboratory Animals (50). For blood collections, animals were immobilized with 10 mg of intramuscular ketamine-HCl (Parke-Davis, Morris Plains, N.J.) /kg of body weight.

Virus inoculation.

Animals 29045 and 29003 were inoculated orally with wild-type uncloned SIVmac251 at birth and animal 29276 was inoculated at birth intravenously with the K65R mutant isolate SIVmac385; these animals were started on tenofovir treatment 3 weeks later, as described previously (80, 81). As indicated in Table 1, all other animals were juvenile animals (12 to 17 months of age) at the time of virus inoculation; they were inoculated orally twice (on 2 consecutive days) with 1 ml of uncloned SIVmac251 (with internal reference number 5/98). This virus was grown on rhesus peripheral blood mononuclear cells (PBMC), had a titer of 10⁵ 50% tissue culture infectious doses and 1.4 × 10⁹ RNA copies/ml (as measured by SIV branched DNA [bDNA] assay), and was pathogenic for infant and adult macaques (43, 83). Our SIVmac251 isolates are very difficult to control in vivo. Of the more than 180 rhesus macaques (including 54 previously vaccinated animals) that became persistently infected after inoculation with these SIVmac251 virus stocks over the past decade at CNPRC, none suppressed viremia to persistently low or undetectable levels (19, 43, 65, 79, 80, 82, 83, 85, 87, 88, 90-92).

TABLE 1.

Summary of tenofovir-treated SIV-infected macaques used for CD8⁺-cell depletion experiments

Animal no.	Presence of MHC I allelea:		Age (mo) at virus infectionb	Start of tenofovirc (wk after SIV infection)	CD8+-cell depletiond (time after SIV infection)	Temporary tenofovir interruptione
	MamuA*01	MamuB*01
29045	+	−	0 (birth)	3	7 yr	None
29276	−	−	0 (birth)	3	6 yr	None
32137	−	−	17	2	39 wk	None
32186	−	−	17	2	39 wk	Wk 64-71
32993	−	−	12	2	2 wk	None
33088	−	+/−	12	2	2 wk	Wk 32-39
33091	+/+	−	12	2	2 wk	Wk 32-39

^aThe presence of the MHC type I alleles MamuA*01 and MamuB*01 is indicated as + (present but unknown whether homozygous or heterozygous), +/+ (homozygous for the presence of the alleles), +/− (heterozygous based on known haplotypes of parents), and − (homozygous for the absence of the particular allele). The frequency of the MamuA*01 and Mamu B*01 alleles in the CNPRC rhesus macaque colony is approximately 25%.

^bAll animals were inoculated orally with wild-type SIVmac251, with the exception of animal 29276, which was inoculated with a K65R SIV isolate (SIV mac385) derived from tenofovir-treated SIVmac251-infected animals, as described previously (80, 81). K65R mutants became dominant in animal 29045 at 14 weeks of age (80). Genotypic drug resistance data are summarized in Table 4.

^cTenofovir was initially given to all of these animals at a once-daily regimen of 30 mg/kg of body weight, administered subcutaneously. Tenofovir regimens for animals 29045 and 29276 had been reduced over time to a low maintenance regimen of 2.5 mg/kg at the time of the CD8⁺-cell depletion (79a). The tenofovir regimen for the other animals listed in this table was reduced from 30 to 20 and then to 10 mg/kg after 10 and 14 weeks of treatment, respectively. Animal 32186's dose was reduced further to 5 and then to 2.5 mg/kg at weeks 30 and 51, respectively. Thus, at the time of the CD8⁺-cell depletion, the 4 long-term-treated animals (29045, 29276, 32137, and 32186) (Fig. 2) had been on a stable tenofovir monotherapy maintenance regimen of 2.5, 2.5, 10, or 5 mg/kg (once daily, subcutaneously) with no changes in dosage for 7, 9, 5, or 2 months, respectively.

^dAnimals 32137 and 32186 were first used as undepleted tenofovir-treated animals for comparison with the CD8⁺-cell-depleted tenofovir-treated animals (32993, 33088, and 33091) for the acute viremia experiment (Fig. 1). Animals 32137 and 32186 were subsequently depleted of CD8⁺ cells during chronic infection (Fig. 2).

^eTenofovir treatment was interrupted for a 7-week period for 3 animals at the indicated times after infection (Fig. 3) and then restarted at the same dosage regimen. At that time, the maintenance dosages were 10 mg/kg for animals 33088 and 33091 and 2.5 mg/kg for animal 32186 (subcutaneously, once daily).

Preparation and administration of tenofovir.

Tenofovir (Gilead Sciences) was suspended in distilled water, dissolved by the addition of NaOH to a final pH of 7.0 at 60 mg/ml, filter sterilized (0.2-μm pore size; Nalgene), and stored at 4°C. Tenofovir was administered subcutaneously into the back of the animal. The dosage was adjusted weekly based on weight. Tenofovir was initially given according to a once-daily regimen of 30 mg/kg of body weight, with dosage reductions described in Table 1.

Administration of cM-T807.

CD8⁺ cells were depleted by using the previously described cM-T807 antibody (67, 68); a total of 20 mg/kg of body weight was administered in 3 doses: 10 mg/kg subcutaneously on day 0 and 5 mg/kg intravenously 3 and 7 days later. The specific role of CD8⁺ cells on viremia of SIV and other viral infections has been well demonstrated with this cM-T807 antibody (34, 57, 67). CD8⁺-cell depletion during acute SIV viremia did not affect peak levels of viremia but inhibited the subsequent reduction in viremia (67). Because it is theoretically possible that administration of cM-T807 could induce virus replication through immune activation, others have previously performed control experiments and found that administration of a nondepleting control antibody or a B-cell-depleting antibody had no effect on SIV viremia (34, 57, 59, 66, 67, 73). In addition, the CD8⁺-cell depletion does not affect CD4⁺-T-cell counts. Thus, the increase of viremia following cM-T807 antibody administration and the return of viremia to baseline levels which coincides temporally with the repopulation of CD8⁺ cells can be considered the best proof of CD8⁺-cell-mediated suppression of virus replication. In our studies, no allergic reactions occurred and no lymphadenopathy or splenomegaly was observed as a result of the cM-T807 administration.

Quantitation of plasma viral RNA.

Viral RNA in plasma was quantified by a bDNA signal amplification assay specific for SIV, versions 2.0, 3.0, and 4.0, which have lower quantitation limits of 1,500, 500 and 125 copies per ml, respectively (P. J. Dailey, M. Zamroud, R. Kelso, J. Kolberg, and M. Urdea, Abstr. 13th Ann. Symp. Nonhum. Primate Models AIDS, abstr. 99, 1995; J. Booth, L. Sawyer, E. McNelley, D. Tayama, C. Wingfield, D. Cox, and K. Leung, Abstr. 18th Ann. Symp. Nonhum. Primate Models AIDS, abstr. 129, 2000; C. Wingfield, J. Booth, P. Sheridan, J. Detmer, and J. Turczyn, Abstr. 20th Ann. Symp. Nonhum. Primate Models AIDS, abstr. 135, 2002). These assays are similar to the Quantiplex HIV RNA assay, except that target probes were designed to hybridize with the pol region of the SIVmac group of strains including SIVmac251 and SIVmac239.

Virus isolation.

Infectious virus was isolated in cultures of PBMC with CEMx174 cells and subsequent p27 core antigen measurement, according to methods previously described (86). Levels of infectious virus in PBMC were determined by a limiting dilution assay (86).

Drug susceptibility assays.

Phenotypic drug susceptibilities of SIVmac isolates were characterized by a previously described assay based on a dose-dependent reduction of viral infectivity. This assay is able to detect SIV mutants with decreased susceptibility to several antiviral drugs (80, 90).

Sequence analysis of SIV RT-encoding region.

Because many plasma samples had insufficient viral RNA to allow sequencing, proviral DNA obtained from CEMx174 infected with PBMC isolates was used. In a previous study, this method gave sequence results that were similar to those of plasma viral RNA (49). Infected CEMx174 and PBMC cocultures were harvested as soon as culture supernatants were positive by antigen-capture enzyme-linked immunosorbent assay. For the DNA sequence analyses of codons 0 to 320 of RT, genomic DNA was extracted and used for nested PCR according to methods and with primers described previously (87). Each round of the nested PCR was carried out under the following conditions: samples were incubated at 94°C for 45 s; followed by 30 cycles at 94°C for 1 min, 57°C for 40 s, 72°C for 120 s; and followed by an extension at 72°C for 5 min. Amplicons were sequenced by Davis Sequencing, Davis, Calif., with primers 239-2786 and SIV-RT3 (87). Data were compared to published sequences of SIVmac251 and SIVmac239 (accession numbers M19499 and M33262, respectively). This method can detect the presence of a 20% subpopulation. Thus, if only K65R was detected, this means that >80% of the proviral DNA population had this mutation.

Lymphocyte phenotyping.

Up until 1997, two-color flow cytometry was used as described previously to detect CD4⁺ and CD8⁺ cells (80). From 1997 to 2001, three-color flow cytometry techniques were used to detect CD3, CD4, CD8, and CD20 with fluorochrome-conjugated antibodies described previously (82, 87) Starting in 2002, four-color flow cytometry was used, consisting of a single tube containing peridinin chlorophyll protein (PerCP)-conjugated anti-human CD8 (clone SK1; Becton Dickinson Immunocytometry Inc., San Jose, Calif.), fluorescein isothiocyanate-conjugated anti-human CD3 (clone SP34; Pharmingen), phycoerythrin-conjugated anti-human CD4 (clone M-T477; Pharmingen), and allophycocyanin-conjugated anti-human CD20 (clone L27; Becton Dickinson). Red blood cells were lysed, and the samples were fixed in paraformaldehyde by the Coulter Q-prep system (Coulter Corporation, Hialeah, Fla.). Flow cytometry was performed on a FACSCalibur flow cytometer (Becton Dickinson). Lymphocytes were gated by forward and side light scatter and were then analyzed with Cellquest software (Becton Dickinson). CD4⁺ T lymphocytes and CD8⁺ T lymphocytes were defined as CD3⁺CD4⁺ and CD3⁺CD8⁺ lymphocyte populations, respectively. B lymphocytes were defined as CD3⁻CD20⁺ lymphocytes. NK cells were defined as CD3⁻CD8⁺ lymphocytes. For the CD8⁺-cell-depletion experiment, as suggested in the cM-T807 protocol (67), the anti-CD8 antibody was replaced by the DK25 clone (DAKO, Carpinteria, Calif.) conjugated to fluorescein isothiocyanate (and combined with anti-CD3-PerCP, anti-CD4-phycoerythrin, and anti-CD20-allophycocyanin).

ELISPOT assay for SIV-specific IFN-γ-secreting cells.

The number of antigen-specific gamma interferon (IFN-γ)-producing cells in rhesus monkeys was measured with an enzyme-linked immunospot (ELISPOT) assay. This assay was performed as previously described (83), with the exception that a pool of 15-mer peptides with 10 amino acids overlapping of the entire p24 gag region of SIVmac239 was used (56). Results were considered positive if the number of spot-forming cells (SFC) for 2 × 10⁵ cells was ≥10 per well and greater than the average of the negative-control (medium only) wells plus 2 standard deviations.

Genetic assessment of MHC class I alleles.

DNA extracted from lymphoid cells (with QIAamp DNA mini kit; Qiagen, Valencia, Calif.) was used to screen for the presence of the major histocompatibility complex (MHC) class I alleles MamuA*01 and MamuB*01 by a PCR-based technique (15, 28). The frequency of MamuA*01 and Mamu B*01 alleles in the CNPRC rhesus macaque colony is approximately 25%.

Statistical analysis.

Statistical analyses were performed with Prism 3 and Instat 3 (GraphPad Software, Inc., San Diego, Calif.).

RESULTS

Role of CD8⁺-cell-mediated immune responses during suppression of acute viremia by tenofovir.

The acute stage of infection consists of a fierce battle between the replicating virus and emerging antiviral immune responses, and relatively small disturbances can have a great impact. Cell depletion experiments demonstrated previously that the reduction of acute SIV viremia is determined mainly by CD8⁺-cell-mediated immune responses instead of humoral immunity (66, 67).

To test the hypothesis of whether tenofovir alone is sufficient to suppress acute viremia, we designed an experiment in which CD8⁺ cells were depleted simultaneously with the start of tenofovir treatment. The week-2 time point of infection was selected for these interventions because this is approximately the time (i) of peak viremia, (ii) when antiviral CD8⁺-cell-mediated responses can be detected in blood and mucosal tissues of SIV-infected macaques (29, 60, 66, 93, 98), and (iii) when the number and activity of NK cells in the blood have been reported to be high (18). Others have already demonstrated that CD8⁺-cell depletion of untreated animals during acute SIV viremia did not affect peak levels of viremia but inhibited the subsequent reduction in viremia so that viremia remained relatively unchanged (67). In our experiments, 11 juvenile macaques (∼1 to 2 years of age) were divided into three groups and infected orally with highly virulent wild-type SIVmac251. At 1 and 2 weeks after infection, virus levels in plasma were not different among the 3 groups (P ≥ 0.4 by analysis of variance) (Fig. 1; Tables 2 and 3), and virus had wild-type susceptibility to tenofovir. Six animals remained untreated and maintained high viremia (virus set point > 6 to 7 log RNA copies per ml), confirming the high virulence of this SIVmac251 stock (Fig. 1). The other 5 animals were started on tenofovir treatment (30 mg/kg subcutaneously once daily) 2 weeks after infection; 3 of these 5 animals were simultaneously depleted of their CD8⁺ lymphocytes through administration of the well-characterized monoclonal antibody cM-T807 (67).

FIG. 1.

Effect of CD8⁺-cell depletion on efficacy of tenofovir during acute SIV viremia. Eleven animals were inoculated orally with wild-type SIVmac251. Six animals were untreated, and 5 animals were started on tenofovir treatment at 2 weeks of infection (arrow). Three of these tenofovir-treated animals were simultaneously depleted of CD8⁺ cells by 3 injections of cM-T807 (on days 14, 17, and 21 after SIVmac251 infection). Data presented are averages (± standard errors of the means) for each group (left side) and individual values (right side). (A and B) Viral RNA levels; (C and D) CD8⁺CD3⁺ T lymphocytes; (E and F) absolute CD8⁺ CD3⁻ NK cell counts. The CD8⁺-cell depletion had no effect on CD4⁺-T-lymphocyte counts (data not shown). Five of the 6 untreated SIVmac251-infected animals had to be euthanized at 14, 22, 25, 32, and 36 weeks. In panel B, the first time points at which virus isolated from PBMC predominantly had the K65R mutation in RT are indicated by squares. More-detailed RT sequence data are given in Table 4.

TABLE 2.

Effect of CD8⁺-cell depletion on ability of tenofovir to suppress acute viremia^a

Animal group (n), treatment	Mean (SD) log-transformed viral RNA copies/ml of plasma at wk (after virus inoculation):			Mean (SD) difference in log-transformed viral RNA copies/ml of plasma between wk (after virus inoculation):
	2	3	4	3 and 2	4 and 3	4 and 2
A (2), tenofovir	7.39 (0.14)	6.06 (0.23)	4.90 (0.91)	−1.34 (0.37)	−1.16 (0.68)	−2.50 (1.05)
B (3), tenofovir, anti-CD8	7.38 (0.42)	6.98 (0.35)	5.75 (0.39)	−0.40 (0.10)	−1.23 (0.26)	−1.63 (0.20)
C (6), none	7.56 (0.35)	6.88 (0.38)	6.72 (0.61)	−0.68 (0.29)	−0.15 (0.38)	−0.83 (0.36)

^aAll animals were inoculated with SIVmac251. Tenofovir treatment and CD8⁺-cell depletion were started 2 weeks after infection (Fig. 1).

Table 3.

Statistical comparison of viral RNA levels of the three animal groups of ^a

Test	Groups	P value at wk:			P value for difference between wk:
		2	3	4	3 and 2	4 and 3	4 and 2
One-way ANOVA	A, B, C	0.72	0.045	0.015	0.015	0.009	0.007
Posttest	A, B	ND	<0.05	>0.05	<0.01	>0.05	>0.05
	A, C	ND	<0.05	<0.05	<0.05	<0.05	<0.01
	B, C	ND	>0.05	>0.05b	>0.05	<0.01	<0.05

^aThe posttest comparison was the Bonferroni test for selected pairs. Groups are defined based on treatment and are as given in Table 2. ANOVA, analysis of variance; ND, not done.

^bThe P value was <0.05 with the two-tailed unpaired t test.

In contrast to the untreated animals, the SIVmac251-infected animals that received tenofovir had a 12- to 39-fold decrease in viral RNA levels after 1 week of tenofovir treatment (P < 0.05) (Fig. 1; Tables 2 and 3), indicating a half-life of productively infected cells of ∼1.3 to 2 days. The magnitude of this reduction of viremia in the tenofovir-treated animals is consistent with many previous observations of a rapid reduction of viral RNA and infectious titers after the start of tenofovir treatment (∼20- to 40-fold reduction in 1 week; half-life, ∼1.3 ± 0.2 days) (52, 76, 80, 82). In contrast, the three CD8⁺-cell-depleted animals (which had no detectable CD8⁺ cells in the blood at days 3 and 7 after the first injection of cM-T807) had only a two- to threefold decrease in viral RNA levels after 1 week of tenofovir treatment (Fig. 1), indicating that in the absence of CD8⁺ cells, productively infected cells have a half-life of 4 to 6 days. Thus, for these CD8⁺-cell-depleted tenofovir-treated animals, the change in viral RNA between weeks 2 and 3 of infection was significantly different from those of the tenofovir-treated undepleted animals (P < 0.01) but indistinguishable from those of the untreated SIVmac251-infected animals (Tables 2 and 3). In these tenofovir-treated CD8⁺-cell-depleted animals, by week 4 of infection (i.e., 2 weeks after the start of tenofovir treatment and CD8⁺-cell depletion), CD8⁺ cells had become detectable again and this return of CD8⁺ cells was associated with a 10- to 30-fold decrease in viral RNA between weeks 3 and 4 (indicating a half-life of productively infected cells of 1.4 to 2 days). This sudden decrease in virus levels was significantly different from the relatively stable viremia of the untreated SIVmac251-infected animals (P < 0.01) but was indistinguishable from the decline rate of the tenofovir-treated undepleted animals (Fig. 1; Tables 2 and 3). The CD4⁺-T-cell counts did not differ significantly among the different groups.

Virus isolated from PBMC of the tenofovir-treated undepleted animals at 4 weeks of infection (i.e., 2 weeks after the start of tenofovir treatment) was wild type, but by 6 weeks of infection, viral mutants with ∼5-fold-reduced in vitro susceptibility to tenofovir and a lysine-to-arginine mutation at codon 65 in RT (K65R) were the only detectable virus populations (Table 4). This rapid emergence of K65R mutants after only 4 weeks of treatment demonstrates that tenofovir is a highly effective inhibitor of reverse transcription of wild-type SIV and exerts strong selection pressure for the K65R mutation (51). Despite the initial immunological disturbance induced by the CD8⁺-cell depletion and the emergence of K65R mutants, two of the three tenofovir-treated depleted animals obtained undetectable viremia by 32 weeks after infection, similar to one of the tenofovir-treated undepleted animals. The two remaining tenofovir-treated animals had a virus set point of ∼10⁴ to 10⁵ RNA copies per ml of plasma but were clinically healthy at ≥19 months of infection. In contrast, the 6 untreated SIVmac251-infected animals maintained high viremia (∼10⁶ to 10⁷ copies/ml) and 5 animals had to be euthanized due to severe immunodeficiency at 14 to 36 weeks after infection. All of the tenofovir-treated animals had detectable levels of SIV-specific IFN-γ-producing cells in the blood, although no correlation was found between their absolute level and the plasma virus RNA set point (data not shown). In summary, this CD8⁺-cell depletion experiment demonstrated that tenofovir's efficacy to suppress acute viremia of wild-type virus was significantly reduced in the absence of CD8⁺ cells.

TABLE 4.

Mutations in RT of SIV isolates from tenofovir-treated macaques^a

Animal no.	Time point	Mutation(s) in RT sequence
29045	9 mo P-SIV	K65R, N69S, I118V
	3 yr P-SIV	K65R, N69S, I118V
	7 yr P-SIV, day 35 P-CD8	K65R, N69S, I118V.
29276	9 mo P-SIV	K64R, K65R, N69S, Y115F, I118V, F171Y, S211N
	4 yr P-SIV	K65R, N69S, S211N
	6 yr P-SIV; wk 1, 3, and 6 P-CD8	K65R, N69S, S211N
32137	2 and 4 wk P-SIV	WT
	6 wk P-SIV	K65R
	8 wk P-SIV	K65R, S211N
	12 wk P-SIV	K65R, I118V, S211N
	28 wk P-SIV	K65R, N69S, I118V, S211N
	40 wk P-SIV, 1 wk P-CD8	K65R, N69S, I118V, S211N
	55 wk P-SIV	K64R, K65R, N69S, I118V, I145V, S211N
32186	2 and 4 wk P-SIV	WT
	6 wk P-SIVI	K65R
	12 wk P-SIV	K65R, I118V
	28 wk P-SIV	K65R, N69S, I118V
	40 wk P-SIV, 1 wk P-CD8	K65R, N69S, I118V
	64 wk P-SIV	L21V, K65R, N69S, I118V
	71 wk P-SIV, 7 wk P-TI	L21V, K65R, N69S, I118V
32993	2 and 4 wk P-SIV	WT
	8 wk P-SIV	T88A
	10 wk P-SIV	K65R, R82K
	12 wk P-SIV	K65R, N69S, S211N
	32 wk P-SIV	K65R, N69S, I118V, S211N
	43 wk P-SIV	K65R, N69S, R82G, Y115F, S211N
33088	2 wk P-SIV	WT
	4 wk P-SIV	S211N
	8 wk P-SIV	K65R, S211N
	10 and 12 wk P-SIV	K65R
	35 wk P-SIV, 3 wk P-TI	K65R, N69S, R82K, S211N
	39 wk P-SIV, 7 wk P-TI	K65R, N69S
	42 wk P-SIV	K65R, N69S
33091	2, 4, and 8 wk P-SIV	WT
	12 wk P-SIV	K65R
	32 wk P-SIV	K65R, N69S
	39 wk P-SIV, 7 wk P-TI	K65R, N69S, S211N
	43 wk P-SIV	K65R, N69S, I118V, S211N

^aDNA from CEMx174 cells infected with SIV isolated from PBMC of SIV-infected animals was used for sequence analysis of RT (codons 1 to 320). The RT amino acid sequence was compared with that of our uncloned SIVmac251 isolates, which is similar to that of the molecular clone SIVmac251, with the exception that at codon 11 alanine is found instead of threonine. For each animal, selected isolates with the wild-type and K65R genotypes were tested for phenotypic drug susceptibility and found to have wild-type and approximately 5-fold-reduced susceptibility to tenofovir, respectively. RT sequences of early time points for animals 29045 and 29276 have been reported previously (80, 81). For more details on the timing and virus levels during the CD8⁺-cell depletion experiments and the 7-week tenofovir treatment interruption experiment, see Table 1 and Fig. 1, 2, and 3. Although we did not sequence viral RNA in plasma, another study found that when SIVmac251-infected juvenile macaques (inoculated with the same SIVmac251 stock as used here) were started on tenofovir treatment (30 mg/kg per day, subcutaneously), the K65R mutation could often be detected in viral RNA in plasma after 1 week of treatment by using a high-sensivity heteroduplex tracking assay; at 2 weeks of treatment, all viral RNA could already have the K65R mutation (40). Abbreviations: WT, wild type; P-SIV, post SIV infection; P-CD8, after the start of the CD8⁺-cell depletion experiment; P-TI, after tenofovir treatment interruption.

Long-term suppression of viremia in tenofovir-treated SIV-infected macaques.

In the experiment described above and in previously published studies, when macaques were inoculated with the highly virulent SIVmac251 isolate, initiation of tenofovir treatment near the time of peak viremia rapidly suppressed viremia (80). Prolonged treatment led to the emergence of viral mutants with a K65R mutation in RT (followed by other, presumably compensatory, mutations). Remarkably, however, the emergence of these K65R mutants was for most animals not associated with an increase in viremia (80). Approximately half of the animals showed a progressive decline of viral RNA levels in plasma to undetectable levels (<125 to 500 RNA copies/ml) after ∼6 to 12 months of continued tenofovir treatment. A similar gradual decline in viremia was also observed in an animal inoculated at birth with a virulent K65R mutant and started on tenofovir treatment 3 weeks later: virus levels, initially high (>6 log RNA copies/ml), declined gradually to undetectable levels after approximately 4 years of tenofovir treatment (Fig. 2, animal 29276). For these animals, viral RNA levels remained undetectable, with the exception of an occasional transient episode of low-level viremia (125 to 600 RNA copies/ml), throughout the observation period (up to 7 years), and the frequency of infected PBMC (as determined by virus isolation) decreased to less than 1 infected cell per 10⁷ to >10⁸ PBMC. Immune activation through booster immunizations of these animals with tetanus toxoid did not result in detectable increases in viremia. This tenofovir-induced long-term suppression of viremia was observed in animals regardless of the presence of MamuA*01 and Mamu B*01 alleles.

FIG. 2.

CD8⁺-cell depletion during chronic tenofovir treatment. Four animals with chronic SIV infection and prolonged tenofovir treatment (started 2 or 3 weeks after infection) (Table 1) were depleted of CD8⁺ cells through administration of the cM-T807 antibody (3 doses). All animals had developed K65R SIV mutants. No viral RNA data were available for animal 29045 for the first 6 weeks of infection, but infectious PBMC titers were available for the whole period. At the time of the CD8⁺-cell depletion, these 4 long-term-treated animals were on a stable tenofovir monotherapy maintenance regimen. Please note the discontinuous x axes. Abbreviations: TCID₅₀, 50% tissue culture infective dose; m, months; y, years.

Those tenofovir-treated animals with low or undetectable viremia had detectable cell-mediated immune responses, as measured by gag-specific IFN-γ ELISPOT assay (Table 5). Thus, we decided to perform 2 sets of experiments to detect whether the low viremia in these tenofovir-treated animals was due to CD8⁺-cell-mediated antiviral immune responses (by depleting CD8⁺ cells) and/or direct suppression of K65R virus replication by tenofovir (by removing tenofovir).

TABLE 5.

IFN-γ ELISPOT data in tenofovir-treated animals with chronic SIV infection undergoing CD8⁺-cell depletion^a

Animal no.	Time point	Tissue	SFC/million cells
29045	3 mo pre-CD8	PBMC	290
	2 mo pre-CD8	PBMC	350
	1 wk post-CD8	PBMC	<50
	5.5 wk post-CD8	PBMC	<50
		Axillary lymph node	145
29276	3 mo pre-CD8	PBMC	140
	1 mo pre-CD8	PBMC	80
	3 wk post-CD8	PBMC	55
	8 wk post-CD8	PBMC	65
	10 wk post-CD8	PBMC	385
32186	2 mo pre-CD8	PBMC	60
	1 mo pre-CD8	PBMC	175
	1 wk post-CD8	PBMC	<50
	3 wk post-CD8	PBMC	65
	12 wk post-CD8	PBMC	180
32137	1 mo pre-CD8	PBMC	80
	1 wk post-CD8	PBMC	<50
	3 wk post-CD8	PBMC	57
	12 wk post-CD8	PBMC	210

^aData presented in this table are from the 4 tenofovir-treated macaques presented in Fig. 2, which had been chronically infected but had low or undetectable viremia at the time of the CD8⁺-cell depletion experiment. Enumeration of virus-specific IFN-γ-producing cells was performed by ELISPOT assay. The cutoff value was 50 SFC per million cells. Prior to the in vivo CD8⁺-cell depletion experiment, PBMC collected from these animals were also enriched for CD4⁺ and CD8⁺ cells in vitro (by negative selection with anti-CD8 and anti-CD4 Dynabeads). It was found that for animals 29276 and 32186, the IFN-γ-producing activity was approximately equally divided among CD4⁺ and CD8⁺ enriched populations, whereas for animals 29045 and 32137, the IFN-γ production resided almost exclusively in the CD8⁺-cell population. Similar results were obtained by intracellular IFN-γ flow cytometry assays (56). For animal 29045, IFN-γ-producing cells were detected readily in lymphocytes derived from an axillary lymph node at a time when no SFC could be detected in PBMC, suggesting that PBMC data may not reflect the antiviral activity that occurs in lymphoid tissues and that SIV-specific IFN-γ-producing antiviral CD8⁺ cells may accumulate in lymphoid tissues rather than in PBMC (similar to observations with humans) (4). This finding explains why the timing of reappearance of IFN-γ-producing cells in PBMC did not necessarily coincide exactly with the clearance of viremia (Fig. 2). pre- and post-CD8, before and after, respectively, the administration of the first dose of the CD8⁺-cell-depleting cM-T807 antibody.

CD8⁺-cell-mediated suppression of viremia during prolonged tenofovir treatment.

To determine the contribution of cell-mediated immune responses to the low viremia, 4 tenofovir-treated animals with chronic SIV infection were depleted of CD8⁺ cells (Fig. 2). All 4 animals had been started on tenofovir treatment 2 to 3 weeks after virus inoculation, had K65R SIV mutants, and had been on continuous tenofovir treatment for 9 months (2 animals) to 6 to 7 years (2 animals) (Table 1). Three of these animals had undetectable viremia (<125 RNA copies/ml of plasma), and one had moderate viremia (animal 32137; ∼29,000 copies RNA/ml of plasma). In all 4 animals, CD8⁺ cells were undetectable in peripheral blood within 3 days after the first cM-T807 antibody dose (Fig. 2). Viral RNA levels increased rapidly and reached peak levels 3 days to 3 weeks after the first cM-T807 dose. This increase in viral RNA levels in plasma was accompanied by an increase in infectious titer in PBMC (Fig. 2). The increase in viremia was highest (∼3,000- to 5,000-fold) for the two oldest animals, which had been on tenofovir for 6 to 7 years and had undetectable viral RNA levels for ≥20 months (Fig. 2, animals 29045 and 29276). The increase in viral RNA was relatively smaller (∼20- to 210-fold) for the ∼2-year-old animals, which had only been infected and on tenofovir treatment for 9 months (Fig. 2, animals 32137 and 32186). Virus isolated during this resurging viremia had the expected genotypic and phenotypic resistance to tenofovir that had previously been detected in these animals (Table 4). Because a stable tenofovir treatment regimen was continued throughout the CD8⁺-cell depletion experiment, this sudden increase in viremia demonstrates that K65R SIV mutants can replicate well in the presence of tenofovir. Viremia was reduced to baseline values as soon as the CD8⁺ T lymphocytes and NK cells became detectable again. The steepest interval of the viral RNA decay curve (Fig. 2) suggests a half-life of productively infected cells of ∼1 day, similar to the shortest half-life reported for cytotoxic-T-lymphocyte (CTL)-mediated clearance of acute viremia in untreated macaques (72). Following the return of CD8⁺ cells, the number of SIV-specific IFN-γ-producing PBMC (by ELISPOT assay) increased again to near baseline values (i.e., prior to the CD8⁺ cell depletion) (Table 5). In summary, these CD8⁺-cell depletion studies in chronically infected animals indicate that the sustained suppression of K65R SIV replication during prolonged tenofovir treatment required CD8⁺-cell-mediated immune responses, and tenofovir alone was not sufficient.

Continued tenofovir treatment is required to sustain low viremia.

Because the long-term efficacy of tenofovir to suppress viremia is determined by the induction of strong antiviral immune responses, an important question is whether continued tenofovir treatment is required to maintain low viremia once these antiviral immune responses exist. Four SIVmac251-infected macaques, including three animals from the CD8⁺-cell depletion experiment discussed above, had K65R virus and eventually reached undetectable viremia (<125 RNA copies/ml). When treatment was interrupted, a slow increase in viremia was observed. Upon reinitiation of tenofovir treatment, viral RNA levels returned to undetectable levels (Fig. 3). The absence of a rapid rebound of viremia following tenofovir withdrawal suggests that effective antiviral immune responses initially limited virus replication in vivo, but continued tenofovir treatment was required to maintain optimal suppression of K65R SIV viremia.

FIG. 3.

Effect of tenofovir treatment interruption on viremia. (A) As previously described (80), animal 29003 had been inoculated at birth with wild-type SIVmac251 and started 3 weeks later on chronic tenofovir treatment (30 mg/kg once daily subcutaneously). Even after the emergence of K65R viral mutants, virus levels decreased gradually. At 22 months of age (time zero, indicated by an arrow), when virus levels were undetectable (<500 copies RNA/ml) (81), tenofovir treatment was withdrawn. Tenofovir therapy was restarted 18 months later (large arrow and shaded area). The dotted line represents the 1,500-copies/ml cutoff value of the bDNA assay, version 2.0. (B) Some animals which had previously been CD8⁺-cell depleted had subsequently reached undetectable virus levels (<125 RNA copies/ml; cutoff value of bDNA, version 4.0) upon return of the CD8⁺ cells (Fig. 1, 2; Table 1). For 1 animal (no. 29276), tenofovir was continued and virus levels re mained undetectable throughout the observation period (>12 months after the CD8⁺-cell depletion). For 3 other animals that had reached undetectable viremia (no. 32186, 33088, and 33091), tenofovir treatment was interrupted and restarted 7 weeks later (shaded area) at the same dosage regimen. Virus isolated during the interruption had the expected genotypic (K65R) and phenotypic resistance to tenofovir that had been detected previously in isolates of these 3 animals (Table 4). The reason for the transient spike of viral RNA for animals 33088 and 33091 at 2 and 3 weeks after restarting tenofovir (which was associated with an increase in infectious titer in PBMC) is not clear but may represent transient replication of an immune escape mutant which was subsequently controlled. No significant changes in CD8⁺ or CD4⁺ lymphocyte counts were observed during this period.

DISCUSSION

While currently available models of viral kinetics during HAART focus mainly on direct inhibition of virus replication by the antiviral drugs, the contribution of antiviral immune responses to the success of HAART is less understood. Our data support viewing antiviral drug therapy and immunology from a perspective in which there is synergistic interaction. Thus, we propose that models that describe viral kinetics during drug therapy must incorporate an important variable, namely the strength of antiviral immune responses. In other words, antiviral drugs require immune responses to reach full effectiveness. This way of thinking may seem provocative at first, but it helps to explain why tenofovir has been more effective than many other antiviral drugs in the SIV-macaque model.

Tenofovir has potent direct antiviral effects against SIV in vivo, as demonstrated by the rapid selection for the K65R mutation in RT. The controlled animal experiments described here provide further insights into the role of immunological events during successful tenofovir therapy. In particular, CD8⁺-cell-mediated antiviral immune responses were important in reducing viremia (i) of wild-type virus at the start of tenofovir therapy and (ii) of K65R mutants during prolonged tenofovir treatment. In other words, the ability of tenofovir to suppress viremia relied on the presence of antiviral CD8⁺-cell-mediated immune responses. Both these antiviral effects of CD8⁺ cells and tenofovir treatment complement each other and are required, as neither tenofovir nor CD8⁺-cell-mediated immune responses alone were sufficient to induce and maintain low or undetectable viremia. As predicted by mathematical models (11), strong antiviral immune responses can lower the viral burst and reduce the basic reproductive ratio, so that even in the presence of drug-resistant mutants, a low virus set point can be achieved.

Tenofovir's efficacy in suppressing acute viremia of wild-type virus was reduced significantly in the absence of CD8⁺ lymphocytes (P < 0.01). This observation is consistent with a mathematical model that proposed that the initial phase of rapid viral RNA decline (with a half-life of 1 to 2 days) following initiation of HAART in HIV-infected individuals may be due to immune-mediated killing or inhibition of productively infected cells (e.g., CTL, apoptosis, and noncytotoxic inhibition) (7). In the absence of immune-mediated killing or inhibition of productively infected cells, the decline of viral RNA induced by a completely effective drug regimen would represent the natural death rate of productively infected cells (as determined by the cytopathogenicity of the virus) (7). Although some evidence suggested a role of antiviral immune responses in the virological response to HAART, other studies did not confirm this (37, 58, 61). This question has been difficult to answer conclusively in human studies due to many uncontrollable variables (12, 53). In contrast, the SIV-macaque model allows us to better distinguish whether the decay of productively infected cells after initiation of drug therapy is due to host immune responses or to viral cytopathogenicity. In our animal study, we controlled many variables (the virulence of the virus strain and the timing of drug treatment) and we performed the experiment that was proposed by mathematicians to answer this question (7): we used a very drastic approach (in vivo CD8⁺-cell depletion), which is not feasible in humans, to maximally eliminate CD8⁺-cell-mediated immune responses. The very slow decline in viral RNA levels in the CD8⁺-cell-depleted SIV-infected macaques during the first week of tenofovir treatment suggests that, in the absence of CD8⁺ cells, productively infected cells during acute viremia had a longer half-life (∼4 to 6 days), suggesting that virulent SIVmac251, during concomitant tenofovir treatment, may not be as cytopathic as expected.

The initial rapid decline of SIV RNA levels in plasma and of infectious titers of PBMC (shown in previous studies) (80, 82) with a half-life of 1 to 2 days, which is generally observed after the start of tenofovir treatment, suggests that during tenofovir treatment strong CD8⁺-cell-mediated antiviral immune responses rapidly destroy or inhibit productively infected cells. A model to fit these data is proposed in Fig. 4. In contrast, the inability of most other antiviral drugs to rapidly suppress established viremia of highly virulent virus in this animal model suggests that even though these drugs may have strong direct antiviral effects (as indicated by the rapid detection of drug-resistant mutants) (87), the antiviral immune responses were probably not sufficient to kill productively infected cells. Because untreated macaques infected with virulent SIV isolates such as SIVmac251 have higher viremia, lower CTL responses, and more rapid disease than HIV-infected humans (for a review, see reference 44), this model could explain why most antiviral drugs that are effective in HIV-infected people have been less effective in SIV-infected macaques, with the exception of tenofovir, which must somehow allow or rescue the development of such antiviral immune responses. In the absence of CD8⁺ cells, tenofovir's effects on viremia resembled those of these other drugs. Accordingly, further research is needed to confirm whether this positive interaction we observed between tenofovir and the development of CD8⁺-cell-mediated immune responses is unique or more pronounced for tenofovir than for other drugs. Repeating the present CD8⁺-cell depletion experiments with other antiviral drugs, however, is unlikely to provide further insights, because these drugs are not very effective in suppressing established viremia of very virulent SIV isolates, so the effect of CD8⁺-cell depletion on drug efficacy would be difficult to detect unless attenuated virus isolates were used.

FIG. 4.

Proposed model of tenofovir's effects on virus replication. (A) Without drug treatment, virulent virus can replicate to high titers because of high infection rates of CD4⁺-T-helper cells and antigen-presenting cells and insufficient CD8⁺-cell-mediated immune responses. (B) Tenofovir acts (i) as an inhibitor of RT, reducing the number of CD4⁺-T-helper cells and antigen-presenting cells that become newly infected and (ii) possibly also as an immunomodulator. Potent CD8⁺-cell-mediated immune responses help to reduce the half-life (T_1/2) and thus the burst of viral progeny for those cells that became infected. Both tenofovir and antiviral CD8⁺ cells are required to induce and maintain low viremia. The role of the CD8⁺ cells becomes even more pronounced after the emergence of drug-resistant K65R viral mutants. (C) During artificial CD8⁺-cell depletion, productively infected cells survive longer and produce more progeny virus, resulting in an increase in viremia.

The ability of chronic tenofovir treatment to persistently suppress SIVmac251 viremia to low or undetectable levels in some animals for more than 6 to 7 years despite the emergence of drug-resistant viral mutants is remarkable and has not been achieved with other antiviral drugs (see the introduction). With pathogenic simian-HIV isolates, short-term drug- or vaccine-induced suppression of acute viremia is often sufficient to promote strong antiviral immune responses that induce a low virus set point and provide protection against rapid disease, although the emergence of CTL escape mutants in some animals still leads to disease (9, 16, 32). In contrast, with highly virulent isolates such as SIVmac251, sustained control of viremia and disease has rarely been achieved (16, 32). In this regard, SIVmac251 infection closely mimics HIV infection but is more virulent. Although there is variation in the virulence of different SIVmac251 stocks, the high viremia induced by the SIVmac251 stocks used in the experiments described here has been very difficult to control (see Materials and Methods).

Tenofovir-treated animals can show a progressive and sustained suppression of viremia even after the emergence of K65R viral mutants. At first thought, potential explanations for this observation would include the possibility of strongly reduced fitness and virulence of K65R mutants and drug pharmacokinetics resulting in continued suppression of K65R mutants which have fivefold-reduced levels of in vitro susceptibility to tenofovir. However, the available evidence, including the results of the present experiments, points against each of these explanations as the sole factor. K65R SIV mutants also accumulate other mutations in RT, which are believed to be compensatory and improve replication fitness in vivo (80, 81). These K65R SIV mutants, when inoculated in infant or juvenile macaques in the absence of tenofovir treatment, were found to replicate to high levels and be as virulent as wild-type virus with similar time to disease without reversion to the wild type (46, 81, 89). Persistent suppression of virulent K65R SIV mutants has only been observed in tenofovir-treated animals. The gradual decline in virus levels in these animals could not be explained by progressive accumulation of tenofovir levels because the dosage had been reduced from a high induction regimen (30 mg/kg) to a low maintenance regimen (2.5 to 10 mg/kg) and pharmacokinetic studies confirmed that plasma levels at these low regimens were within the expected range (79a). The CD8⁺-cell depletion experiments reported here confirmed that the tenofovir regimen by itself was insufficient to suppress K65R viremia in the absence of CD8⁺ cells (Fig. 2). Instead, while continuous tenofovir was still required, CD8⁺-cell-mediated immune responses played a major role in suppressing K65R viremia (Fig. 4). In other words, the emergence of K65R viral mutants is not associated with virological failure of tenofovir therapy as long as there are good antiviral immune responses.

When animals were started on tenofovir treatment at peak SIVmac251 viremia (∼2 to 3 weeks after infection), approximately half of them were able to gradually suppress viremia to undetectable levels. Persistent suppression of virulent SIVmac251 viremia has been more difficult to achieve when animals were started on tenofovir monotherapy during intermediate-to-late-stage infection, and the emergence of K65R viral mutants is then usually associated with a gradual increase in viremia (92a). Because virulent SIV infection induces immune dysfunction at many levels, the reduced ability of tenofovir to persistently suppress viremia in such immunocompromised animals could be due to a defect in any events that make it more difficult to rescue the generation, maturation, and maintenance of strong antiviral CD8⁺-lymphocyte responses to complement tenofovir's direct antiviral effects (20, 26, 27, 39, 101). Without proper T-helper-cell function and antigen presentation, CD8⁺-cell-mediated immune responses may not remain active at low levels of antigen (24, 31, 45) and are thus not able to suppress viremia to low or undetectable levels, especially once drug-resistant mutants emerge. However, chronic viremia was suppressed more efficiently during tenofovir treatment when less-virulent SIV isolates were used (52, 76, 80) or when tenofovir treatment was combined with other antiviral drugs and/or immunotherapy strategies (17, 21, 36, 95).

Because the cM-T807 antibody depletes both the CD3⁺CD8⁺ T lymphocytes as well as CD3⁻CD8⁺ NK cells, the relative contribution of these two cell populations to the immune-mediated suppression of viremia could not be determined in the present experiments. CD8⁺ T and NK lymphocytes inhibit virus replication in vitro through a variety of mechanisms, including cytotoxicity and noncytolytic pathways (secretion of antiviral factors) (31, 45). In macaques, most NK T cells (believed to be important immunoregulators) are also CD8⁺ (48). Thus, further studies are needed to identify which CD8⁺ cells and antiviral mechanisms are most important in suppressing viremia. In another study, the number of expanding CD4⁺- and CD8⁺-T-cell clonotypes (as determined by Vβ chain CDR3 region DNA heteroduplex tracking assay) in SIVmac251-infected macaques did not correlate with their response to tenofovir treatment (9a). As suggested by data from both human and macaque studies (1, 10, 34), the present in vitro assays, especially when PBMC are used, may not sufficiently capture the strength and breadth of the cell-mediated immune responses that are active in vivo, especially in lymphoid tissues where most of the virus replication occurs. In our long-term tenofovir-treated macaques with undetectable viremia, the cell-mediated immune responses must be unusually strong and/or broad, as immune-escape mutants have not emerged in these animals even after 6 to 7 years of tenofovir treatment (Fig. 2).

Why does tenofovir behave differently from other antiviral drugs in the SIV macaque model? Several features of tenofovir that may be dependent or independent of its direct antiviral effects may contribute to the beneficial immunological events observed in the macaque studies. It is possible that due to the intrinsic antiviral potency and favorable pharmacokinetics in comparison to other drugs, tenofovir treatment provides better preservation of virus-specific CD4⁺-T-helper-cell function in vivo, which is expected to lead to improved antiviral CD8⁺-cell-mediated immune responses (24). Early tenofovir treatment was also found to reduce CD8⁺-lymphocyte apoptosis during acute SIVmac239 viremia (71). In vitro studies have demonstrated that tenofovir is more active to inhibit wild-type HIV type 1 (HIV-1) replication in monocytes/macrophages, dendritic cells, and Langerhans cells than in lymphocytes or lymphocytoid cell lines because of (i) a more efficient conversion to its active diphosphate metabolite, (ii) a longer intracellular half-life of these metabolites, and (iii) the lower dATP levels in these cells (6, 8, 62). Tenofovir may thus be more effective in inhibiting replication, including that of K65R SIV mutants, in vivo in these antigen-presenting cells than in lymphocytes. Relatively subtle improvements in the early steps of an antiviral immune response are likely to be amplified in later steps, as suggested by the beneficial effects of administration of exogenously pulsed dendritic cells to SIVmac251-infected macaques (38).

Tenofovir may have immunopreserving or immunomodulatory effects for macaques that are independent of its direct antiviral effects. Immunomodulatory effects of tenofovir and related compounds have been extensively demonstrated in murine models in vitro and in vivo, including enhancement of NK cell activity and secretion of cytokines and chemokines (13, 14, 94, 99, 100). During in vitro experiments with PBMC of uninfected rhesus macaques, tenofovir did not directly stimulate NK cell activity or cytokine production but primed cells for more rapid and enhanced interleukin-12 (IL-12) production following exposure to bacterial antigens (K. K. A. Van Rompay, M. L. Marthas, and N. Bischofberger, submitted for publication). Considering the many biological effects of IL-12 (i.e., activation of NK cells and CTLs, stimulation of IFN-γ production and Th1 immune responses, and antiapoptotic effects) (for a review, see reference 74), further research is needed to determine its potential role in the observations with tenofovir in the macaque model. Interestingly, studies performed with exogenous IL-12 administration to SIV-infected macaques during acute and chronic SIV infection showed effects on viremia remarkably similar to those of tenofovir therapy during these respective stages (5, 96, 97).

Other observations provide additional evidence for immune-preserving or immunomodulatory effects of tenofovir therapy in SIV-infected macaques. Tenofovir therapy initiated during late-stage SIVmac251 infection significantly improved disease-free survival even in the presence of high viremia and suppressed acquired immune responses, which has not been observed with other drugs in this animal model (81, 92a).

Because CD8⁺-cell depletion experiments are not feasible in humans, clinical studies are needed to investigate whether similar immune events occur also in drug-treated HIV-infected humans and are preferentially seen with tenofovir-containing regimens. Two pieces of evidence from tenofovir-treated HIV-1 infected patients are consistent with our proposed synergistic effects of CD8⁺ cells and tenofovir in SIV-infected macaques. First, short-term studies comparing the early decay kinetics of viral RNA following the initiation of therapy have demonstrated that a tenofovir-containing regimen induced a faster decline in viral RNA levels than other commonly used combination regimens without tenofovir (37, 42). This can be interpreted as a reflection of the higher efficacy of the tenofovir-containing regimen than the standard regimen to directly inhibit virus replication. However, our experiments offer an alternative explanation: when different regimens are equally effective in preventing infection of new cells, the decay rate of virus-producing cells is determined mainly by the potency of CD8⁺-cell-mediated antiviral immune responses. Thus, the faster decay rate during a tenofovir-containing regimen could be consistent with enhanced CD8⁺-cell-mediated inhibition of virus replication. The second observation is that K65R mutants of HIV-1 are generally infrequently (∼2%) detected after long-term tenofovir treatment as part of most HAART regimens, and their emergence is generally not associated with a rebound in viremia (41). These clinical observations with HIV-1 are consistent with our observations of reduced viremia of K65R SIV mutants in macaques due to improved antiviral immune responses during tenofovir treatment. Of note, effects of improved antiviral immune responses are not mutually exclusive of effects of decreased replication capacity of the K65R mutant virus and/or residual drug activity of tenofovir. In particular, even a relatively minor decrease in intrinsic replication fitness of K65R viral mutants during tenofovir treatment can have a major impact on viremia if it provides more opportunity for antiviral CD8⁺ cells to kill productively infected cells prior to the major viral burst. Thus, multiple factors are likely to contribute to reduced viremia in tenofovir-treated humans who develop K65R mutants of HIV-1.

Recently, it has been reported that when treatment-naive patients were started on a combination of tenofovir, lamivudine, and abacavir used once daily, approximately half of them showed less-than-desired (i.e., <2 log) suppression of viremia by week 8 or had evidence of an early viral rebound, which was associated with the detection of M184V mutants in all failure patients and K65R mutants in approximately 50% of failure patients (J. Gallant, Abstr. 43rd Intersci. Conf. Antimicrob. Agents Chemother., late-breaker abstract 1722a, 2003; C. Farthing et al., Prog. Abstr. 2nd Int. AIDS Soc. Conf. Pathog. Treat., abstr. 43, 2003). While further investigations are currently being done to explain these findings, an overall poor potency of this investigational triple-nucleoside regimen with abacavir once daily is a likely explanation, and as a result, resistance emerged to one or more drugs in the regimen. The findings of this human study are consistent with our macaque data, confirming the high selection pressure for the emergence of K65R mutants during tenofovir treatment and the impact of these K65R mutants on reducing tenofovir's direct efficacy in suppressing viremia. Our macaque studies demonstrated that even after a relatively modest early virological response to tenofovir or partial viral rebound due to the emergence of K65R viral mutants, continued tenofovir treatment leads in some animals to a gradual suppression of viremia to low or undetectable levels over the course of months to years, due to a strengthening of antiviral CD8⁺-cell-mediated immune responses. The human patients with a poor early response to this tenofovir-containing regimen were switched to a different regimen, so the long-term response was not determined.

If these interactions between antiviral immune responses and antiviral drugs such as tenofovir are confirmed in humans, the clinical implications are multiple. First, early drug treatment is more likely to promote the induction and maintenance of strong antiviral immune responses (63). Second, the detection of drug resistance mutations such as K65R does not necessarily indicate that therapy is failing and is by itself not a valid reason to stop tenofovir treatment unless better treatment options are available. Finally, elucidation of the CD8⁺-cell-mediated antiviral immune responses that are elicited during successful tenofovir treatment in macaques and that are unusually strong in suppressing virus replication for years may aid the development of novel immunotherapeutic strategies. As mentioned above, macaque studies have suggested that the efficacy of tenofovir during chronic infection can be improved further by combining it with additional immunotherapeutic strategies. Accordingly, further research in this area should be explored, as this may result in finding long-term sustainable and affordable strategies that combine the strengths of antiviral drugs and the immune system to indefinitely delay disease progression.