Highly active antiretroviral therapy (HAART) has significantly changed the morbidity and mortality associated with HIV-1 infection [1-3]. A non-nucleoside reverse transcriptase inhibitor (NNRTI) combined with two or three nucleoside reverse transcriptase inhibitors (NRTIs) is recommended in the current treatment guidelines for the treatment of antiretroviral naïve patients [4]. NNRTI-based regimens are frequently the preferred choice for many patients compared to protease inhibitor (PI)-based regimens because of lower pill burdens resulting in both better compliance [5] and better lipid profiles [6,7].

Among NNRTIs, efavirenz (EFV) has been included in the majority of treatment guidelines as the preferred first-line regimen for HIV-infected adults [4] and children>3 years old in the United States and developed countries. In contrast, nevirapine (NVP) is the most commonly used NNRTI in developing countries where it is also commonly used for the prevention of mother-to-child transmission (MTCT) [8,9,10,11,12].

Most patients experience an excellent response to NNRTI containing regimens; however, a subset of individuals responds poorly with slow virologic suppression and immunologic recovery, due to the rapid development of drug resistance [13]. Although there are other factors contributing to these different clinical responses, recent pharmacogenetic studies suggest that at least some of these variable responses to NNRTI-containing regimens are predictable based on variants in genes responsible for metabolizing and transporting antiretrovirals [14]. We previously reported that specific CYP2B6 genetic variants were determinants of NVP pharmacokinetics (PK) and affected the immunologic recovery of HIV-1 infected children who received NVP containing HAART [15]. Children with CYP2B6 variants had lower NVP oral clearance and demonstrated better immunologic recovery than those with the wild-type CYP2B6 genotype. In contrast, although children with CYP2B6 variants receiving an EFV containing HAART regimen achieved higher plasma concentrations of EFV, we failed to observe improved immunologic recovery [16].

Intracellular concentrations (ICs) of NNRTIs are determined by a net result of passive transport, active uptake, efflux from cells, and metabolization of enzymes such as CYP3A4 and CYP2B6 in peripheral blood mononuclear cells (PBMCs). The cellular accumulation of NNRTIs may vary among individuals and measurement of ICs of NNRTIs may adequately reflect treatment efficacy and drug toxicity than measurement of plasma drug concentrations. Therefore, measuring IC of antiretrovirals is important. We hypothesized that ICs of EFV and NVP are different and the difference may affect cell survival because EFV is known to induce caspase- and mitochondrion dependent apoptosis in PBMCs [17]. A few studies evaluated ICs of EFV and NVP in HIV-infected patients [18,19,20]; however, no study compared the differences in ICs of EFV and NVP using an in vitro model.

Peripheral blood mononuclear cells (PBMCs) isolation

PBMCs were obtained from healthy adult donors (n=5, multiple ethnicities and race) using a Ficoll-Paque (GE Healthcare) density gradient according to the manufacturer's instructions.

EFV and NVP treatment of PBMCs

Isolated PBMCs were counted and incubated in 3 mL of serum free media (AIM-V, Invitrogen, Carlsbad, CA) at a concentration of 1.0 x 10⁶/mL in 6-well plates (Corning, Corning, NY). EFV was purchased from Morevak Biochemicals (Brea, CA) and NVP was provided by Boehringer Ingelheim (New York, NY). Both compounds were reconstituted in 50% ethanol. The cells were treated with or without the mean peak steady-state levels (C_max) of EFV and NVP in human plasma during antiretroviral therapy multiplied by 0.5, 1.0, 2.0 and 4.0. The cells were also incubated with vehicle (50% ethanol) with the highest concentration of ethanol (0.78%) used to reconstitute EFV for each assay. The C_max of EFV and NVP were 12.4 µM and 17.0 µM, respectively. The cells were treated for 48 hr at 37°C at 5% CO₂. The assays contained cells treated with each treatment condition in duplicates and were repeated five times using different donor samples.

Measurement of intracellular concentrations of EFV and NVP in PBMCs

Treated and untreated cells were analyzed for ICs of EFV and NVP using liquid chromatoagraphy-ion trap/mass spectrometry (LCIT/ MS). Equal numbers of the cells in each treatment condition were collected from the same samples for flow cytometry and the cells were lysed in 100 µL of 50% methanol and were stored at -70°C until the measurement was performed. ICs of EFV and NVP were measured by positive ion electrospray ionization mass spectrometry, after separation with high pressure liquid chromatography. The response of protonated molecular ions (m/z = 267 and 316, for NVP and EFV, respectively) in PBMC were compared to the response of NVP and EFV calibrators prepared in artificial intracellular fluid [21], after adjustment for processing efficiency via internal standardization (Figure 3). The ICs were expressed µM per 10⁶ PBMC.

Apoptosis and active mitochondrial membrane potential assays by flow cytometry

Treated and untreated cells were evaluated for the degree of induction of apoptosis and mitochondrial membrane potential (Δψm) by flow cytometry. The following markers were used; Annexin-V (Benton Dickinson, Franklin Lakes, NJ), propidium iodide (PI) (Benton Dickinson), and DiIC₁(5) (Invitrogen). After 48-hr, the treated and untreated cells were counted and 3x10⁵ cells per each treatment condition were assessed by flow cytometry. Briefly, cells were incubated with DiIC₁(5) for 30 min with appropriate positive controls. Subsequently, the cells were stained with Annexin-V, followed by PI according to the manufactures' protocols. Stained cells were analyzed by FACS at the FACS Core Laboratory at the Veterans Affair Medical Center, San Diego. Viable cells were defined based on side and forward scatter and the cells positive for Annexin-V were defined as apoptotic cells. The cells positive for PI and Annexin-V were defined as late apoptotic cells and the proportion of late-apoptotic cells to earlyapoptotic and viable cells was calculated. The cells positive for DiIC₁(5) and negative for PI were defined as active Δψm cells and the proportion of the cells with active Δψm to viable cells was calculated.

Statistical analyses

Statistical analyses were performed using the SPSS 13.0 software (Chicago, IL). The Student T-test was used for comparison of numerical variables in two independent groups. The analysis of variance (ANOVA) test was used for the comparison of numerical variables in =3 independent groups. All P-values calculated were two-sided and a P-value of < 0.05 was considered to be statistically significant.

Intracellular concentrations of EFV and NVP

The mean log ICs of NVP (mean ± S.D) at x0.5 C_max (1.62 ± 0.26 µM) were significantly lower than those of EFV at x0.5 C_max (2.87 ± 0.13 µM) (P < 0.01). The mean log ICs of NVP (mean ± S.D) at x1.0 C_max and x2.0 C_max (2.00 ± 0.23 µM and 1.99 ± 0.39 µM, respectively) were significantly lower than those of EFV at x1.0 C_max and x2.0 C_max (2.95 ± 0.22 µM and 3.11 ± 0.21 µM, respectively) (P < 0.01, P < 0.01, respectively) (Figure 1). These results demonstrate ICs in PBMCs treated with NVP were approximately one log lower compared to the cells treated with EFV.

Figure 1: Intracellular concentrations of NNRTI in PBMCs treated with nevirapine and efavirenz. Open and closed bars indicate the log intracellular concentrations of nevirapine and efavirenz in PBMCs treated with different concentrations for 48 hrs, respectively. Significant differences were observed when the intracellular concentrations were compared with the same Cmax (P < 0.01). Cmax: the mean peak steady-state levels in human plasma during antiretroviral therapy.

Survival of PBMCs treated with NNRTIs

After the 48-hr incubation period, viable PBMCs in untreated groups decreased to 60% (range: 51-68%). Compared to untreated cells (1.00), the cell-survival ratios (mean ± S.D.) in PBMCs treated with 0.5x, 1.0x, 2.0x, and 4.0x C_max of EFV were decreased in a dose dependent manner: 0.88 ± 0.18 (P = 0.15), 0.94 ± 0.16 (P = 0.24), 0.71 ± 0.28 (P = 0.06), and 0.03 ± 0.04 (P < 0.01), respectively. In contrast, the cell-survival ratios in PBMCs treated with 0.5x, 1.0x, 2.0x, and 4.0x C_max of NVP did not change significantly: 0.93 ± 0.05 (P = 0.14), 0.92 ± 0.12 (P = 0.17), 0.84 ± 0.13 (P = 0.09), and 0.84 ± 0.13 (P = 0.09), respectively. The cell-survival ratio in the group treated with vehicle used for the maximum concentration of EFV (x2.0 C_max) was 0.91 ± 0.12 (P = 0.87).

EFV treated PBMCs undergo greater apoptosis than NVP treated PBMCs

PBMCs treated with EFV demonstrated a higher proportion of late-apoptotic cells in a dose dependent manner compared to untreated cells (Figure 2A). The percentages (mean ± S.D.) of late-apoptotic cells in PBMCs treated with x0.5, x1.0, and x2.0 C_max of EFV were 3.0 ± 4.1 %, 7.4 ± 3.5%, and 24.9 ± 12.9%, respectively. The apoptosis assay could not be performed when a concentration of x4.0 C_max of EFV because at this concentration and duration of treatment no viable cells remained. NVP also exhibited a dose-dependent increase in populations of late-apoptotic cells; however, the proportion of late-apoptotic cells to early-apoptotic and viable cells was substantially less under NVP treatment when compared to EFV treatment. The percentages (mean ± S.D.) of late-apoptotic cells in PBMCs treated with x0.5, x1.0, x2.0, x4.0 C_max of NVP were 0.8 ± 4.5%, 3.3 ± 5.0%, 5.9 ± 5.7%, and 12.5 ± 5.8%, respectively. When equal concentrations of EFV and NVP were compared, no significant differences were observed at x0.5 C_max (P = 0.55) and x1.0 C_max (P = 0.36), but a significant difference between EFV and NVP was observed at x2.0 C_max (P = 0.02).

Figure 2: Induction of late apoptotic cells and mitochondrial membrane potentials in peripheral blood mononuclear cells (PBMCs) treated with nevirapine and efavirenz using different concentrations. The changes in late apoptotic cells (A) and mitochondrial membrane potential (B) from untreated cells in each treatment condition are shown with open bars (nevirapine) and closed bars (efavirenz). Significant differences were observed when the cells were treated with x2.0 Cmax of efavirenz. No data are shown in cells treated with x4.0 Cmax of efavirenz because of the low number of viable cells after 48 hr incubation. Cmax: the mean peak steady-state levels in human plasma during antiretroviral therapy. Actual FACS figures for late apoptotic cells (C) and cells with active mitochondrial membrance potential (D) are shown. (C): The viable cells were gated (Panel C-A) and the proportion of late apoptotic cells which were defined as positive for both Annexin-V and propidium iodine (PI) are indicated at the right upper corner of the panel for late apoptotic cells: untreated cells (Panel C-B), vehicle (50% ethanol concentration used for efavirenz x2.0 Cmax, Panel C-C), nevirapine treated cells with different concentrations (Panels C-D, E, F), and efavirenz treated cells with different concentrations (Panels C-G, H, I). (D): The viable cells were gated (Panel D-A) and the proportion of cells with mitochondrial membrane potential were defined as positive for DiIC1(5) and negative for PI are indicated at the left upper corner of the panel for cells with mitochondrial membrane potential: untreated cells (Panel D-B), vehicle (50% ethanol concentration used for efavirenz x2.0 Cmax, Panel D-C), nevirapine treated cells with different concentrations (Panels D-D, E, F), and efavirenz treated cells with different concentrations (Panels D-G, H, I). Cmax: the mean peak steady-state levels in human plasma during antiretroviral therapy. PI: propidium iodine.

The active mitochondrial membrane potential of PBMCs is reduced to a greater extent with exposure to EFV compared to NVP

Active Δψm exhibited trends parallel to the data on apoptosis. PBMCs treated with x0.5, x1.0, and x2.0 C_max of EFV all showed significant decreases in populations of cells with active Δψm in a dose dependent manner (Figure 2B). No data were obtained from cells treated with x4.0 C_max of EFV as no viable cells remained. A similar dose-dependent decrease in cells with uncompromised Δψm was observed in cells treated with NVP, although to a far lesser extent compared those treated with EFV. No differences were observed when the cells were treated with x0.5 C_max of EFV and NVP (P = 0.84), but significant differences were observed when the cells were treated with x1.0 C_max (P < 0.01) and x2.0 C_max of EFV and NVP (P < 0.01).

Figure 3: Mass spectrometry (MS)/MS response of peripheral blood mononuclear cell extract with 50 ng nevirapine and efavirenz, including their respective internal standards: The two left most peaks illustrate the MS/MS response of nevirapine (m/z = 267.4 > 267.4, Rt = 2.23, C) and its internal standard (BIRH-414) (m/z = 255 > 255, Rt = 3.00, A). The right most peaks represent the response of efavirenz (m/z = 316 > 272, Rt = 7.56, B) and its internal standard (d4-EFV) (m/z = 320 > 276, Rt = 7.47, D).

In the current study, we have shown that ICs of NVP in PBMCs were approximately one-log lower than ICs of EFV when the PBMCs were treated with different concentrations of NVP and EFV based on the mean peak steady-state levels. Furthermore, the induction of apoptosis of PBMCs is lower in PBMCs treated with concentrations of NVP above the C_max as compared to those treated with comparable concentrations of EFV. These results suggest that differences in ICs of NVP and EFV may contribute to different cell survival in PBMCs treated with NVP and EFV.

Similar IC differences between EFV and NVP have also been demonstrated in vivo [20]. Rotger et al. [20] demonstrated that the IC of EFV was higher in than IC of NVP in HIV-infected adults and was affected by CYP2B6-G516T genotype; however, no direct comparison of the IC between EFV and NVP was performed. A number of potential mechanisms may account for the observation of a difference of ICs between NVP and EFV. First, although a clear mechanism is unknown, the degree of passive transport and active uptake may be different between NVP and EFV. Second, a role of the ATP-binding cassette, sub-family B, member 1 (ABCB1) needs to be considered. ABCB1 is a trans-membrane protein that actively transports many substrates from the cytoplasm to the extracellular space and is therefore a principle determinant of the pharmacokinetics of many drugs, including protease inhibitors [22]. Genetic variants in the ABCB1 gene have been reported to be responsible for variable protease inhibitors pharmacokinetics in children [23] and adults [24]. Although neither NVP or EFV are known substrates of ABCB1 [25], several studies have shown clinical associations between ABCB1 genotypes and NNRTIs pharmacokinetics (including EFV) [24], virologic outcomes in patients receiving EFV containing regimens,[26] and the risk of NVP-induced hepatotoxicity [27,28]. Furthermore, ICs of NVP inversely correlate with ABCB1 expression on PBMCs in subjects receiving NVP [19]. In contrast, Janneh and colleagues found that ABCB1 expression has no effect on the accumulation of EFV and NVP, and suggest that some an important role for other factors including MRP1/2 expression, lipophilicity and SLCO-like transporters [29] .

In the current studies, we used the mean peak steady-state levels (C_max) of NVP and EFV in human plasma during antiretroviral therapy multiplied by 0.5, 1.0, 2.0 and 4.0 C_max to reflect the actual concentrations of EFV and NVP in vivo. The actual x1.0 C_max of EFV (12.4 µM) was lower than x1.0 C_max of NVP (17.0 µM), demonstrating that the differences observed in the current study did not result simply from the differences of concentrations of both drugs.

Although our data demonstrate that ICs of NVP are lower than that of EFV and induce less apoptotic cells compared to EFV in PBMC when the cells are treated with close to the mean-peak steady state concentrations, no clinical data, to our knowledge, have identified differences in immunologic recovery between NVP and EFV regimens [30,31]; therefore, the clinical relevance of our findings remains uncertain. Other factors that may alter the ICs of NVP and EFV will need to be considered including hepatic CYP2B6 genotypes [32] and metabolization of each drug by CYP2B6 and CYP3A4 in PBMC [33]. Because we were not able to find a direct correlation between ICs of NNRTIs and cell survival or induction of apoptosis, these factors may contribute our findings.

In addition, although our observations showed significant differences in late apoptotic cells between NVP and EFV during the 48-hr incubation period, in vivo data demonstrated that no differences were observed in the extent of apoptosis or in ΔΨ between EFV or NVP treatments in lymphocytes from HIV-1 infected patients [34]. In this study, the duration of the NVP regimen correlated with apoptosis and Δψm in lymphocytes from HIV-1 infected patients. Therefore, the cumulative effects of NVP on lymphocytes may account for the discrepancy between in vitro and in vivo studies.

In conclusion, PBMCs treated with NVP induced significantly less apoptosis than EFV when treated with higher than the mean peak steady-state levels of NNRTI in human plasma during antiretroviral therapy. The clinical relevance of this observation remains to be elucidated.

This work was supported by Grants from the National Institute of Allergy and Infectious Diseases (5K23AI-56931) and The International Maternal Pediatric Adolescent AIDS Clinical Trials (IMPAACT) Network (AI-069536). Authors acknowledge Dr. Celsa Spina, Neal Sekiya, Judy Nordberg, Peggy O'Keefe, Dale Disharoon at the Center for AIDS Research Flow Cytometry Core at the Veterans Affair Medical Center, and Carol Mundy and Joe Sanding at the University of California, San Diego for their assistance.

Supported by Grants from the National Institute of Allergy and Infectious Diseases (5K23AI-56931) and The International Maternal Pediatric Adolescent AIDS Clinical Trials (IMPAACT) Network (AI-069536).