We first sought to establish the overall effect Nef has on virus replication and depletion of CD4⁺ T lymphocytes in ex vivo-cultures of human tonsil tissue from HIV-negative donors. Such cultures can be established as tissue blocks or in suspension as aggregates (human lymphoid aggregate cultures, HLAC) 12. Initial parallel testing of both experimental systems gave identical results for the comparison of wt and nef-negative HIV-1 (HIV-1Δnef) (data not shown). We also compared normalization of virus input based on amounts of viral antigen (p24) or virus infectivity (TCID₅₀) and again did not observe significant differences (data not shown). We therefore employed HLAC and virus inoculum normalization by p24 ELISA for the remainder of this study. HLAC were infected one day after cell preparation with wt and Δnef HIV-1 corresponding to 3 ng p24 per 2 × 10⁶ cells. 24 h later the virus input was washed out and the HLAC maintained for 11 more days. On day 4, 8, and 12 p.i., cell culture supernatant was analyzed by p24 ELISA to quantify HIV-1 production and CD4⁺ T cell depletion was determined by flow cytometry. Quadruplicate parallel infections were harvested for each time point.

Fig. 1A presents the results of such an analysis 12 days p.i.: viable lymphocytes were identified in the FSC/SSC (gate R1) and analyzed for expression of CD3 and CD8. Direct staining of CD4 was avoided due to the reduction of CD4 surface exposure in HIV-1 infected cells, but a control staining for mock infected cells reveals that virtually all cells in this gate were positive for CD4 (see Additional file 1). A pronounced population of CD3⁺/CD8^- cells that represent CD4⁺ T lymphocytes was found in mock-infected cultures (53.8% of all lymphocytes). CD8⁺ T lymphocytes were less abundant (7.4%) and approximately 40% of all cells in the lymphocyte gate did not carry these T cell markers. In the HIV-1 wt infected culture, the CD4⁺ population was markedly reduced to 14.6%, reflecting the strong and specific depletion of CD4⁺ T lymphocytes due to HIV-1 replication. This CD4⁺ T lymphocyte depletion was significantly less pronounced in the culture infected with the HIV-1Δnef virus that maintained 34.9% of viable CD4⁺ T lymphocytes. To normalize for variations in cell populations between different donors/experiments, we employed a well-established strategy 111736, which determines the relative abundance of CD4⁺ T cells by calculating the ratio of CD4⁺ and CD8⁺ T lymphocytes with values for mock-infected cultures set to 100%. Accordingly, average CD4⁺ T lymphocyte depletion of 83.8 ± 5.1% and 54.2 ± 1.0% was observed in parallel quadruplicate infections of the experiment shown in Fig. 1A for HIV-1 wt- or Δnef- infected HLAC, respectively (Fig. 1B). Control analyses confirmed that virtually identical degrees of CD4⁺ T lymphocyte depletion were obtained by using the total amount of CD4⁺ lymphocytes rather than the CD4⁺ to CD8⁺ ratio for evaluation (data not shown). Quantification of viral p24 antigen over the time course of infection showed that wt HIV-1 replicated more rapidly and to higher levels than HIV-1Δnef (Fig. 1C). In HLAC from a few donors, HIV-1 replication was overall accelerated, resulting in late peak virus production for HIV-1Δnef that was comparable to that observed early in HIV-1wt infected cultures, indicating that the lack of Nef delays but not generally abrogates HIV-1 replication in HLAC (data not shown). To facilitate assessment of overall virus production during the time course of infection, the integral area under the curve (AUC) was calculated from the p24 replication kinetics plot (e.g. Fig. 1C). This evaluation accounts best for changes in p24 concentration in the cell culture medium over time 13. Plotting of the mean AUC of the independent quadruplicates analyzed in parallel revealed that over 3-times more p24 were produced in the wt HIV-1-infected culture when compared to HIV-1Δnef (Fig. 1D). These results recapitulate the previously described phenotype of Nef on HIV-1 replication and CD4⁺ T lymphocyte depletion in ex vivo-cultures of human tonsil tissue 836 and provide the experimental framework to perform standardized multi-donor HLAC analyses and map Nef determinants that are critical for these activities.

Using the experimental set-up described above, we first compared virus replication and CD4⁺ T lymphocyte depletion of wt and Δnef HIV-1 using tonsils from 14 donors in 35 independent experiments, e.g. with independent virus stocks, with quadruplicate parallel infections for each time point analyzed. These studies demonstrated that CD4⁺ T lymphocyte depletion was consistently less severe for the nef-deficient virus compared to the isogenic wt HIV-1 (wt: 84.4 ± 1.4% vs Δnef: 53.8 ± 1.4%; p < 0.0001) (Fig. 2A, C). Similarly, p24 production and thus virus replication was also significantly reduced in the absence of Nef (wt: (mean) 251.1 ± 9.8 vs Δnef: (mean) 134.1 ± 4.6; p = 0.0001) (Fig. 2B, D), suggesting, for this cross-donor analysis, a correlation between virus replication and loss of CD4⁺ T lymphocytes. Upon closer examination of the results for HLAC from individual donors, we noted that while HIV-1Δnef depleted CD4⁺ T lymphocytes less vigorously than an average infection with wt HIV-1 in essentially all experiments (Fig. 2A, C), p24 production by HIV-1Δnef reached levels in the range of the highest ones obtained with wt HIV-1, in some, but not all experiments (Fig. 2B). To explore this further, we stratified results obtained for individual donor HLAC according to p24 production into two groups: the first, for which AUC values for Δnef HIV-1 were statistically lower than those of wt (p < 0.05, 18 experiments; different replication; Fig. 3A), and the second, for which the AUC values did not reveal statistically significant differences between wt and Δnef HIV-1 (p ≥ 0.05, 17 experiments; similar replication; Fig. 3B). Expectedly, both p24 production and CD4⁺ T lymphocyte depletion were markedly decreased in HIV-1 Δnef-infected relative to wt HIV-1-infected cultures in the "different replication" cohort (production: wt: 248.9 ± 13.4 vs Δnef: 79.3 ± 5.7 (p < 0.0001); depletion: wt 84.5 ± 1.9% vs. Δnef 50.7 ± 2.1% (p < 0.0001); mean values) (Fig. 3A). In the "similar replication" cohort p24 production was, expectedly, statistically indistinguishable between both viruses (wt: 253.0 ± 14.7 vs Δnef: 202.1 ± 7.0; p = 0.26; mean values) (Fig. 3B), but wt HIV-1 infection still resulted in a more pronounced CD4⁺ T lymphocyte depletion than HIV-1Δnef infection (wt: 84.4 ± 2.2% vs Δnef: 57.2 ± 2.0%, mean values), with a high statistical significance (p < 0.0001). In line with these findings, no statistically significant correlation between HIV-1 replication and CD4⁺ T lymphocyte depletion was observed (data not shown). These multi-donor studies demonstrate that Nef boosts HIV-1 replication and depletion of CD4⁺ T lymphocytes in HLAC, and suggests that Nef's effect on the loss of CD4⁺ T lymphocytes is not strictly coupled to the elevation of virus spread.

To determine the molecular determinants that govern Nef's activity in this ex vivo-model, tonsil aggregate cultures were challenged with a panel of isogenic HIV-1 NL4-3 viruses coding for characterized SF2 Nef variants 27. Fig. 4 summarizes the results from up to 12 individual experiments. HIV-1 wt and Δnef served as reference controls and showed the expected and statistically highly significant difference in CD4⁺ T cell depletion and p24 production. While all HIV-1 infections with the analyzed Nef mutants displayed apparently intermediate p24 production (Fig. 4A), this difference only reached low statistical significance for NefLLAA and NefΔ12–39 (Δnef p = 0.0017, G2A p = 0.16, E4A4 p = 0.19, AxxA p = 0.12, LLAA p = 0.03, KKAA p = 0.33, Δ12–39 p = 0.026). When compared among them, virus production between these various Nef mutant viruses was statistically indistinguishable. Despite these comparable intermediate levels of virus production, specific Nef mutants significantly differed in their ability to deplete CD4⁺ T lymphocytes (Fig. 4B). Four of the analyzed mutants were statistically indistinguishable in their depletion activity from wt HIV-1. This included the G2A and KKAA mutants that lack N-terminal myristoylation or membrane microdomain targeting signals, respectively, and thus display reduced membrane binding (G2A) or lack membrane microdomain incorporation (KKAA) (G2A: 79.5%, p = 0.93; KKAA: 79.9%, p = 1.0) 37. The two other mutants, E4A4 and AxxA, are not disturbed in their subcellular localization 27, but lack protein interaction motifs for the phosphofurin acidic cluster sorting protein (PACS) sorting adaptor (E4A4) or SH3 domains (AxxA), and are deficient in modulating MHC class I cell surface levels and cell activation, respectively (E4A4: 79.1%, p = 0.899; AxxA: 80.9%, p = 0.93) (reviewed in 2425). The lack of requirement for these motifs suggested that these activities are dispensable for Nef-mediated CD4⁺ T lymphocyte depletion in HLAC infections. In contrast, the two other Nef mutants LLAA and Δ12–39 were significantly impaired in CD4⁺ T lymphocyte depletion, displaying only partial activity relative to wt (LLAA: 58.6%, p = 0.0002; Δ12–39: 55.6%, p = 0.0005). NefLLAA fails to interact with the endocytic machinery responsible to internalize CD4 and is thus defective in downregulating cell surface CD4 (reviewed in 38). The Δ12–39 Nef variant, in contrast, is fully active in CD4 downregulation but lacks the interaction surface for the Nef-associated kinase complex (NAKC), a multiprotein complex that facilitates transcription of the HIV-1 genome 3940. These results reveal that Nef requires a complex set of molecular determinants to boost HIV-1 spread in HIV-1-infected HLAC and identify two independent protein interaction motifs in Nef that facilitate CD4⁺ T lymphocyte depletion independently of their effect on HIV-1 replication.

To further analyze how Nef accelerates depletion of CD4⁺ T lymphocytes in HLAC HIV-1 infection, we sought to determine whether Nef primarily affects the killing of productively infected or uninfected (bystander) CD4⁺ T cells. Since reproducible analyses of specific markers for apoptosis such as active caspase 3 or cell surface annexinV in combination with the fixation method and intracellular stain for p24 proved difficult in our experimental system (data not shown), we analyzed cell death by staining with 7AAD, a nucleic acid dye that binds to genomic DNA of dead, necrotic and late apoptotic cells because of their increased membrane permeability. For analysis, we gated on CD4⁺ T lymphocytes to determine their 7AAD incorporation and intracellular p24 expression (see Methods for details). Control experiments ensured that the 7AAD staining was stable during the fixation and permeabilization procedures (data not shown). The infection with wt HIV-1 was compared to Δnef HIV-1 as well as HIV-1 coding for the Nef LLAA or Δ12–39 variants, the two mutants that displayed defects in CD4⁺ T lymphocyte depletion in the above analyses. Fig. 5 shows primary data of such an analysis for one donor. As before, CD4⁺ T lymphocytes were identified as CD3⁺/CD8^- population, carefully avoiding autofluorescent cells in the center of the dot plot (Fig. 5A). When these CD4⁺ T lymphocytes were analyzed for the expression of p24 and staining for 7AAD (Fig. 5B), the mock-infected culture, expectedly, only showed background staining for p24. On day 8 p.i., 27.1% of all CD4⁺ T lymphocytes stained positive for 7AAD and were thus considered dead. In wt HIV-1-infected cultures, a population of p24-positive viable CD4⁺ T lymphocytes was readily detectable (10.9%, upper left quadrant). A small fraction of p24-positive cells also stained positive for 7AAD (1.7% of all CD4⁺ T lymphocytes; 13.7% of all p24⁺ CD4⁺T lymphocytes; upper right quadrant) (Fig. 5D). Notably, wt HIV-1 infection increased the number of 7AAD⁺/p24^- cells (33.2% of all CD4⁺ T lymphocytes; 38.0% of all p24^- CD4⁺ T lymphocytes; lower right quadrant), reflecting bystander killing in response to virus infection (Fig. 5E). This increase in death of p24^- bystander cells was less pronounced in cultures infected with HIV-1Δnef (19.8% of all CD4⁺ T lymphocytes; 21.1% of all p24^- CD4⁺ T lymphocytes). Plotting of the total number of productively infected CD4⁺ T lymphocytes revealed slightly fewer infected cells in cultures infected with the Δnef relative to the wt virus (Fig. 5C). Interpretation of these results was complicated by the varying degree of HIV-unrelated cell death between HLAC replicates, but also HLAC cultures from different donors. To account for this, analysis of quadruplicate infections of HLAC from five different donors was performed to quantify and statistically evaluate this effect (Fig. 6). First, total amounts of p24⁺ cells present in these cultures (irrespective of their sensitivity to staining with 7AAD) (Fig. 6A) were significantly increased in the presence of Nef (wt: 17.7 ± 0.8% vs Δnef: 8.6 ± 0.7%, p = 0.0079). Nef thus expands the pool of productively infected cells in HIV HLAC infections. Although the 7AAD staining appeared slightly increased in cells infected with wt HIV-1 relative to Δnef or the two Nef mutants (Fig. 6B, data expressed as percentage relative to wt that was arbitrarily set to 100%), these differences were not statistically significant (wt: 100 ± 15.2%; Δnef: 65.1 ± 11.7%; LLAA: 81.6 ± 6.3%; Δ12–39: 74.9 ± 11.0%). This indicated that Nef does not have major pro- or anti-apoptotic effects on infected CD4⁺ T lymphocytes in HIV-infected HLAC. However, since HIV infection clearly reduces the overall life span of CD4⁺ T lymphocytes 41, the fact that Nef expands the target cell population will indirectly contribute to T cell depletion in HLAC. In contrast, significantly more killing of bystander CD4⁺ T lymphocytes (Fig. 6C and 6D) was observed in HIV-1 wt-infected cultures when compared to infections with the Δnef or the LLAA and Δ12–39 Nef mutant viruses [wt: 154.6 ± 34.2% (p = 0.0079 compared to mock); Δnef: 92.8 ± 6.5% p = 0.0317; LLAA: 107 ± 5.3% p = 0.0556; Δ12–39: 99.7 ± 5.9% p = 0.0159, data expressed as percentage relative to mock that was arbitrarily set to 100%]. These results identify the interaction motifs for endocytic machinery and the NAKC signalosome as key determinants for Nef-mediated depletion of CD4⁺ T lymphocytes in HIV-1 HLAC infections and suggest that Nef selectively augments the efficacy of bystander killing.

All proviral plasmids used here were previously described and the expressed Nef variants characterized 2737. Briefly, nef genes and thus also the overlapping U3 enhancer/promoter region from the HIV-1_SF2 (wild-type or carrying the indicated mutations) were inserted into the HIV-1_NL4-3 proviral DNA and compared to a nef-deleted variant entirely lacking Nef expression. All Nef variants analyzed herein express to comparable levels in HIV-1 infected T lymphocytes 2737. Virus stocks were generated by transfection of proviral HIV plasmids into 293 T cells as described 65. Two days after transfection, culture supernatants were harvested. The HIV-1 p24 antigen concentration of concentrated stocks was determined by a p24 antigen enzyme-linked immunosorbent assay (ELISA) 66.

Tonsil tissue was removed during routine tonsillectomy from HIV-, HBV-, HCV-negative patients with informed consent. To prepare HLAC, tonsil tissue was mechanically dispersed by cutting tissue in 2- to 3-mm blocks and passing them through 40-μm cell strainers (BD Falcon, Belgium). Cells were washed in PBS, and 2 × 10⁶ cells were plated in 96-well V-bottom plates (Corning Incorporated, New York, NY) in a final volume of 200 μl. Culture medium (RPMI 1640 containing 15% fetal bovine serum, 1% L-glutamine, 1% fungizone, 1% gentamycin (all from GIBCO), 0.25% ampicillin (Roth, Karlsruhe, Germany), 1% non-essential amino acids, and 1% sodium pyruvat (both from Invitrogen)). Detailed cultivation methods have been reported 1217. One day after tonsil preparation, the HLAC was inoculated with HIV-1 (3–6 ng p24 per 2 × 10⁶ cells per well). Following overnight infection cells were washed and the culture medium was subsequently changed every three days without dispersing the pellet. At the same time intervals supernatant samples were harvested and stored at -20°C for subsequent analysis by p24 ELISA.

All flow cytometry was performed with a FACS Calibur with BD CellQuest Pro 4.0.2 Software (BD Pharmingen). For the analysis of CD4⁺ T lymphocyte depletion, cells were stained with the following antibodies without prior permeabilization: anti-human CD3 FITC (clone HIT3a, BD Pharmingen), anti-human CD8 APC (clone RPA-T8, BD Pharmingen) and subsequently fixed (1.5 hrs, 2% paraformaldehyde). Using standard gating procedures 17, CD4⁺ T lymphocytes were then defined as the CD3⁺/CD8^- lymphocyte population. To determine the death of HIV-infected and non-infected CD4⁺ T lymphocytes, cells were first stained for anti-human CD3 APC (clone HIT3a, BD Pharmingen) and anti-human CD8 PE (clone RPA-T8, BD Pharmingen) and 7AAD (2.5 μg/ml) (BD Pharmingen) as described above. Following subsequent fixation cells were permeabilized in PBS with 0.1% Triton X-100 and simultaneously stained for intracellular p24 using anti-p24 antibody KC57 FITC (Beckman Coulter, Fullerton, CA) 20 min at RT.

Analysis of HLAC infections was carried out in independent quadruplicate infections for each time point investigated. Mean values and standard deviation were calculated from these quadruplicates. For comparison of individual HLAC infections, mean values of each data set were averaged with the indicated standard error of the mean. For evaluation of statistical significance, data sets were first analyzed for normal distribution by the Kolmogorov-Smirnow test. As this test for Normality failed for some populations, the non-parametric Mann-Whitney-U analysis was employed throughout (***, P ≤ 0.0005; **, P < 0.005; *, P < 0.05). GraphPad Prism software (GraphPad Software Inc.) was used for all statistical analyses.

For the analysis of CD4⁺ T cell depletion, the ratio of CD4⁺ to CD8⁺ T lymphocytes was determined and expressed relative to the uninfected control whose ratio was arbitrarily set to 100%. Depletion efficiencies were calculated relative to this value with the CD4⁺ to CD8⁺ ratio of uninfected cultures considered as 0% depletion. For quantification of virus replication, p24 concentrations of individual infections were plotted over the course of the experiment. To account best for rapid changes in p24 concentration, the integral area under the curve (AUC) was determined and used as measure of overall p24 production. To determine the percentage of killed infected and non-infected bystander CD4⁺ T cells, four populations of CD4⁺ T lymphocytes were defined and quantified: uninfected, living cells (p24^-/7AAD^-), uninfected, dead cells (p24^-/7AAD⁺), infected, living cells (p24⁺/7AAD^-) and infected, dead cells (p24⁺/7AAD⁺).