Main

The gold compound auranofin (AF) is a candidate drug for curing HIV infection, showing the potential to hit crucial therapeutic targets that are not accessible to the antiretroviral drugs (ARVs) currently in use.1, 2, 3, 4, 5 Although ARVs are able to clear viremia and improve the immunological condition of HIV-infected individuals for prolonged time,6 the virus rebounds to levels comparable to those observed before treatment initiation shortly after treatment is withdrawn.6 AF is the first drug shown to induce a partially selective killing of central and transitional memory T cells (henceforth TCM and TTM, respectively7), which are pivotal carriers of HIV during conventional therapies.8 TCM and TTM constitute one main viral reservoir that is long-lived and harbors dormant proviral DNA copies integrated in the genome that cannot be targeted by either the immune system or drug-based therapies. However, different stimuli can reactivate proviral DNA to produce new infectious viral particles.9, 10

Also in vivo AF showed the potential to target the viral reservoir, given its ability to induce cell death in the memory T-cell compartment (recently reviewed in Badley et al.1). In this regard, we recently showed in the macaque AIDS model (i.e., rhesus macaques chronically infected with the HIV simian homolog, SIVmac251) that treatment with a drug combination consisting of ARVs and AF is able to induce a long-term reduction in the post-therapy viral set point following therapy suspension.7, 11 Importantly, the effects of AF on the memory cell subset were not associated with any detectable immune impairment in vivo. Moreover, we recently proved that the effects of AF can be enhanced in the context of a combined anti-reservoir therapy, inducing a drug-free remission of the disease in chronically SIVmac251-infected macaques.12

Despite the extensive study of gold salts conducted throughout the centuries,13, 14, 15 a comprehensive picture of the effects and mechanism of action of AF on the immune system is still lacking, particularly in the CD4+ T-cell compartment. It is known that AF impairs the proliferative capacity of T-lymphocytes in vitro16, 17 and decreases production of pro-inflammatory cytokines in macrophages and T cells.18, 19 Moreover, AF has been shown to induce intracellular oxidative stress by compromising the antioxidant defense due to the inhibition of thioredoxin reductase (TrxR).20, 21 However, the molecular pathways associated with these effects have been characterized in detail only in tumor cells, never in primary lymphocytes.22, 23

In the present study, we show that the main antioxidant defenses of the cell decrease paralleling the lymphocyte differentiation stage. This deficiency of antioxidant defense in differentiated memory CD4+ T cells is associated with increased susceptibility to a signaling cascade sparked by AF and leading to cell death. These findings suggest the molecular bases upon which the anti-reservoir effects of AF are grounded.

Results

AF exerts cytocidal and pro-differentiating effects on different lymphocyte memory T-cell compartments in vitro

In a previous study, we analyzed the in vivo and in vitro effects of AF7 on CD4+ T-cell subpopulations in peripheral blood of rhesus macaques infected with SIVmac251 and treated with antiretroviral therapy (ART) plus AF. We showed that AF induced a significant reduction in the frequency of the long-lived TCM/TTM cells.7 We first aimed at confirming these effects of AF on sorted CD4+ T-cell subpopulations isolated from a cohort of uninfected human donors. For this purpose, we measured, by flow cytometry, the expression of CD27, that is, a marker for long-lived phenotypes, and the frequency of Annexin V+ cells, that is, a predictive marker for apoptosis. The results confirmed that AF induced CD27 downmodulation with a concomitant increase in the frequency of Annexin V+ cells (Figure 1a). Annexin V staining was more pronounced in the memory compartment, including TCM and TTM lymphocytes, that is, the cell types that encompass the HIV-1 reservoirs (P<0.05, repeated measures ANOVA followed by the Newman–Keuls post-hoc test, five donors). These results confirm and extend those previously obtained in vitro in human CD4+ T cells and in vivo in SIVmac251-infected macaques.7

Figure 1
figure 1

Dot plots showing anti-CD27 and Annexin V staining after 48 h of treatment with AF (gate on live cells) in (a) CD4+ and (b) CD8+ T-cell subpopulations. CD4+ and CD8+ TN, TCM, TTM and TEM T cells were separated by sorting and treated with AF at 250 nM or left untreated (gated on live CD4+ T cells on the basis of the physical parameters). The dot plots show data from one representative experiment out of five (a) and out of three (b) with similar results

As during the progression of HIV infection the TCM and TTM CD8+ T cells become activated24 and this activation correlates with disease progression,25 we analyzed whether AF might also shorten the lifespan of the TCM and TTM compartments of CD8+ T cells. Experiments conducted in sorted CD8+ T-cell subpopulations showed that, similar to what was observed in CD4+ T cells, CD8+ TCM, TTM and effector memory T (TEM) lymphocytes succumbed more readily than the naive (TN) subset to AF treatment (P<0.05, repeated measures ANOVA followed by the Newman–Keuls post-hoc test, three donors). Moreover, downmodulation of CD27 was evident in all subsets (Figure 1b). We concluded that AF exerts a pro-differentiating effect and shortens the lifespan of memory T cells independent of their CD4+ or CD8+ lineage.

To confirm that susceptibility to AF-induced cell death was associated with the stage of lymphocyte differentiation, we tested the effects of AF in stem cells (CD34+ cells) purified from human cord blood. We stained stem cells with Annexin V after 24 and 48 h of treatment with AF. The results showed that AF had no effect on the frequency of Annexin V+ cells (Supplementary Figure S1; note that CD27 is not expressed by stem cells). We conclude that the cell-death-promoting effect of AF increases in parallel to the stage of lymphocyte differentiation.

The cytocidal and pro-differentiating effects of AF are associated with the baseline oxidative status of CD4+ T cells

As the pro-oxidant effects of AF are well known in the literature,26 we analyzed, in sorted CD4+ T-cell subpopulations, the baseline levels of the major marker of the intracellular redox state, that is, glutathione (GSH).27 The results showed that GSH levels were lower in the memory cell subpopulations than in the TN compartment (P<0.001; Figure 2a). Repletion of GSH, however, did not inhibit the cytocidal and pro-differentiating effect of AF (data not shown), indicating that GSH is unlikely to be a direct target of AF in CD4+ T cells.

Figure 2
figure 2

Baseline antioxidant levels in sorted CD4+ T-cell subpopulations reflect their susceptibility to AF. (a) Total GSH levels in sorted CD4+ T-cell subpopulations. CD4+ T-cell subpopulations were separated by sorting and baseline GSH levels were measured. Data points represent the means from three independent experiments normalized to TN cells. Data are shown as median±interquartile range (n=3; P<0.001; repeated measures ANOVA). (b) Total thiol content in sorted CD4+ T-cell subpopulations. The thiol content was measured, in all CD4+ T-cell subpopulations, by FACS after labeling cells with ThiolTracker violet dye. The data displayed in both the dot plot and the x, y graph are expressed as the median of fluorescence intensity. On the ‘x’ axis of the x, y graph, the level of CD4+ T-cell differentiation was associated, for clarity purposes, to an increasing score ranging from 0 to 3 (0=TN; 1=TCM; 2=TTM; 3=TEM; P<0.05; Spearman’s correlation test). (c) Replication of the effects of AF on CD4+ T cells by the TrxR inhibitor arsenic trioxide (As2O3). The dot plots show the effect of AF and As2O3 on total CD4+ T cells after 48 h of treatment, analyzed by CD27 (y axis) and Annexin V (x axis) staining. The dot plot shows data from one representative experiment out of three with similar results

To test whether this result could be extendable to the bulk of intracellular reactive thiols, we analyzed sorted CD4+ T-cell subpopulations by flow cytometry, using the ThiolTracker Violet (Invitrogen Molecular Probes, Milan, Italy) (Figure 2b). We found that the total thiol content decreased in parallel to the level of lymphocyte differentiation (P<0.05; Spearman’s correlation test, see Figure 2b).

The thiol/selenium proteins TrxRs represent another major intracellular defense against oxidative stress. AF is a well-known inhibitor of TrxRs and, in the present study, it was used at TrxR-inhibitory concentrations.28 To test whether inhibition of these enzymes in an AF-independent manner might replicate the effect of AF, we incubated primary CD4+ T cells with arsenic trioxide (As2O3), another well-known inhibitor of TrxRs. Results showed that As2O3 replicates the effect of AF at TrxR-inhibitory concentrations (i.e., 2 μM; see Figure 2c). As2O3 did not exert this effect at a concentration of 250 nM, that is, lower than its EC50 on purified TrxRs (Lin et al.28 data not shown). We conclude that compounds that inhibit TrxRs are able to induce both pro-differentiating and cytocidal effects in human primary CD4+ T cells.

Involvement of p38 MAPK activation in AF-mediated cell death

Some of the most important signaling molecules in lymphocyte cell death and differentiation are mitogen-activated protein kinases (MAPK), a family of serine/threonine kinases.29 The MAPK superfamily consists of different groups, one of which includes p38 MAPK,30 the activation of which represents one of the earliest sensors of TrxR inhibition. Incubation of CD4+ T cells with AF increased the phosphorylation of p38 MAPK in Tyr182 (Figure 3a). This phosphorylation could be detected at 1 h after treatment with AF, and, to a lesser extent, at 6 and 24 h (Figure 3a).

Figure 3
figure 3

Role of p38 MAPK in the pro-differentiating and pro-apoptotic effects of AF. (a) Activation of p38 MAPK by AF treatment. CD4+ T cells were incubated with 250 nM AF for 0–48 h. Protein extracts were prepared at the indicated time points and analyzed on western blot probed with specific antibody against the phosphorylated form of p38 MAPK (p-p38; Tyr182). Loading was controlled with an antibody against total p38 MAPK on the according blot. Respective bands were quantified using Quantity One software and relative p38 MAPK activation was calculated and normalized to the loading control. The blot is shown from one representative experiment out of the two performed. Data in densitometric analysis are graphed as means±s.d. of the two experiments performed (donors: n=2; *P<0.05; ***P<0.001; one-tailed Student’s t-test). (b) Effect of AF and the p38 MAPK inhibitor SB203580 on CD27 and Annexin V staining in human CD4+ T cells after 48 h of treatment. The dot plot shows data from one representative experiment out of three (similar results). (c) Percentages of CD27/Annexin V+ CD4+ T cells after 48 h of treatment with AF and SB203580. Data are shown as means±s.d. (n of donors=3; *P<0.05; **P <0.01; ***P<0.001; Newman–Keuls post-hoc test following repeated measures ANOVA and LOGIT transformation). (d) Percentages of CD27/Annexin V T cells after 48 h of treatment. Data are shown as means±s.d. (n of donors=3; *P<0.05; **P <0.01; ***P<0.001; Newman–Keuls post-hoc test following repeated measures ANOVA and LOGIT transformation)

To further confirm that p38 MAPK is involved in AF-induced lymphocyte cell death and differentiation, we tested a specific inhibitor of this protein (the pyridyl-imidazole inhibitor SB203580). Results showed that SB203580 inhibited the Annexin V stainability of T cells induced by AF and, to a lesser extent, their differentiation (Figures 3b–d). Instead, chemical inhibition of Erk1/2, another member of the MAPK superfamily, using the PD98059 inhibitor, enhanced AF-mediated cell death (Supplementary Figure S2). We conclude that p38 MAPK is an important mediator of AF-induced cell death and differentiation, whereas the Erk1/2 MAPK family member delivers a survival signal in primary CD4+ T cells, in line with previous evidence from AF-untreated cells.31

AF decreases the mitochondrial membrane potential

To further dissect the mechanism of AF-induced apoptosis, we examined the mitochondrial membrane potential (MMP, Δψm), a decrease of which may follow p38 MAPK activation. Changes in MMP were assessed by the fluorescence transition of the JC-1 probe, with a change from red to green fluorescence indicating a decrease in MMP (typically observed in cells undergoing apoptosis). As shown in Figure 4, treatment with AF resulted in a decrease in MMP of CD4+ T cells, indicating that AF-induced apoptosis involves the mitochondrial pathway.

Figure 4
figure 4

Analysis of MMP (Δψm) in CD4+ T cells after AF treatment. Cells were stimulated with AF for 12–48 h and stained with the JC-1 probe. After staining with JC-1, live cells display red fluorescence (high MMP), whereas apoptotic cells display intermediate or green fluorescence (low MMP). Note that the relatively high rates of cells displaying low MMP observed in the controls are in line with a spontaneous trend to undergo apoptosis in the memory compartment of non-stimulated CD4+ T cells (see Figure 1). (a) Dot plots of red and green fluorescence at 12, 24 and 48 h. (b) Histograms showing the decrease in mean red fluorescence intensity induced by auranofin at 48 h. Data are shown as mean±s.d. (n=4; *P<0.05; two-tailed paired Student’s t-test)

TN and TCM/TTM CD4+ lymphocytes respond to AF administration activating both common and different downstream signals involved in regulating cell survival, proliferation and differentiation

To explore the correlates of susceptibility to or protection from AF-induced cell death, we analyzed, in sorted CD4+ T-cell subpopulations, the AF-induced changes of an array of proteins involved in antioxidant and pro-/anti-apoptotic pathways. Protein array analysis showed that AF prompted dramatic changes in protein expression, which, in some cases, were peculiar to the different subsets analyzed (Figures 5a–c). In all long-lived CD4+ subsets, that is, TN, TCM and TTM, AF increased the levels of the pro-apoptotic factors Ask1 (S38 phosphorylated), p38 (Tyr180 phosphorylated, Figure 5b) and Bax/Bak (total, Figure 5c), but decreased the expression of the apoptosis-regulating factor p53 (Figure 5c). Moreover, AF increased the levels of the lymphocyte activation and differentiation factor phospho-ZAP-70(Tyr319)/SyK(Tyr352) (Figure 5c). In line with the well-known positive feedback effects induced by the presence of a specific inhibitor,32 TrxR1 and TrxR2 were also increased in all CD4+ subsets (Figure 5a). On the other hand, the mitochondrial peroxide generator, Mn2+ superoxide dismutase (MnSOD), was upregulated by AF only in the memory T-cell subsets (Figure 5a). No peculiar change was instead observed for the AF-treated TN subset, apart from extremely high levels of the antioxidant and anti-apoptotic enzyme TrxR2 (data not shown). As a support of the concept that susceptibility to AF-induced apoptosis increases in parallel to lymphocyte differentiation, the level of the effector caspases 3 and 9 was increased in the most differentiated subsets (Figure 5c).

Figure 5
figure 5

Protein array after AF treatment shows differential modulation of pathways involved in antioxidant responses and cell survival. (a) Antioxidant pathways and cytokines; (b) transcription factors and kinases; (c) cell death and survival pathways, and other proteins. The table was obtained by comparing the intensity level of proteins from each subpopulation of CD4+ T cells (TN, TCM, TTM and TEM), treated with AF for 6 h or left untreated. Endpoints examined are clustered on the horizontal axis. Within the heat maps, red color represents higher levels of relative expression, black represents same levels and green represents lower levels of relative expression. The proteins are clustered by similarity of their expression profiles

That AF also has a potential impact on reducing lymphocyte proliferation (and, hence, reservoir expansion) is shown by the fact that the cell cycle associated protein Ki67 was decreased in all three long-lived subsets (Figure 5c). Whereas TCM and TTM cells reacted to AF administration also downregulating CD25, that is, the α-chain of the receptor for interleukin-2 (IL-2), a well-known mitogenic and anti-apoptotic cytokine, TN cells compensated the AF-induced Ki67 downregulation by upregulating CD25. These results point to different mechanisms by which the TN and TCM/TTM subsets react to the AF-induced stress and that may be responsible for the different in vivo susceptibility of these cell compartments to AF-induced apoptosis.7

AF induces a burst in intracellular peroxide levels within CD4+ T cells

Given the ability of AF to increase the levels of MnSOD, we analyzed reactive oxygen species (ROS) following treatment with AF. To this aim, we measured, by flow cytometry, the levels of dihydrorhodamine (DHR), a compound that becomes fluorescent upon oxidation by intracellular peroxides such as hydrogen peroxide, one well-known cell-death-inducing agent.33, 34 As shown in Figure 6a, AF only moderately increased intracellular peroxides levels until 48 h of treatment, at which a major burst was observed (Figures 6a and b).

Figure 6
figure 6

The role of ROS in AF-induced cell death and differentiation. (a) ROS production in CD4+ T cells. ROS content was measured by flow cytometry after labeling cells with 1,2,3 DHR. The dot plot shows ROS fluorescence in total CD4+ T cells at different time points. (b) Histograms showing median intensity of the DHR fluorescence of CD4+ T cells from three independent experiments. Data are shown as mean±s.d. (**P<0.01; Bonferroni’s post-hoc test following two-way ANOVA). (c) Effect of pyruvate, a scavenger of peroxide, on AF-induced cell death and differentiation. Dot plots show CD27 (y axis) and Annexin V (x axis) staining of human CD4+ T cells after 48 h of treatment. (d) Histogram showing the percentages of CD27/Annexin V+ T cells after 48 h of treatment from three independent experiments. Data are shown as means±s.d. (*P<0.05; **P <0.01; ***P<0.001; Newman–Keuls post-hoc test following repeated measures ANOVA and LOGIT transformation). (e) Percentages of CD27/Annexin V T cells after 48 h of treatment from three independent experiments. Data are shown as mean±s.d. (*P<0.05; **P<0.01; ***P<0.001; Newman–Keuls post-hoc test following repeated measures ANOVA and LOGIT transformation)

It is well known that the cytocidal effects of hydrogen peroxide are mediated by ROS following its degradation through the iron-catalyzed Fenton reaction.35 In line with this view, the iron-containing drug ferroquine greatly enhanced AF-mediated cell death, although the iron compound was per se unable to carry out AF-like effects (data not shown).

To further test whether the cell-death-promoting effects of AF might be mediated by the production of hydrogen peroxide, we treated the CD4+ T cells with pyruvate, that is, the principal scavenger of hydrogen peroxide (Figure 6c). Pyruvate significantly prevented AF-induced Annexin V stainability but did not inhibit downregulation of CD27 (Figures 6d and e). Interestingly, apart from showing the involvement of hydrogen peroxide in the AF-mediated apoptosis of CD4+ T cells, these results show that the pro-apoptotic and pro-differentiating effects of AF can be de-coupled.

AF displays pro-differentiating effects that are not confined to CD27 downregulation

Among the panel of antigens that we used to define the stage of T-cell differentiation, AF induced downmodulation of CD27 but not of CCR7 (data not shown). CD27 CCR7+ cells have been described previously, although their role in vivo is not yet known.8 To elucidate whether the pro-differentiating effect of AF was confined to downmodulation of CD27, we verified the extent of downregulation of CD28, another common marker of the long-lived phenotypes TN, TCM and TTM. We found that AF downregulated CD28 expression similarly to CD27, suggesting that AF exerts a true pro-differentiating effect in human CD4+ T cells (Supplementary Figure S3).

AF induces phenotype changes in CD4+ T cells from HIV-infected patients

The pro-differentiating effects of AF were eventually confirmed in CD4+ T cells isolated from three HIV+ patients under ART. Treatment of these cells with AF for 2 days induced a significant reduction in the frequency of the long-lived TCM cells accompanied by a relative increase in the frequency of TEM cells (Figure 7). AF, however, did not induce viral emergence from latency in CD4+ T cells isolated from HIV-1+ individuals following 3 days of treatment (data not shown).

Figure 7
figure 7

AF-induced phenotype changes in CD4+ T cells from HIV-infected individuals. Shown are the proportions of CD4+ T-cell subpopulations of human HIV+ subjects, under ART, after 2 days of treatment with auranofin. CD4+ T-naive (TN; i.e., CD45RA+ CD27+ CCR7+), CD4+ T-central memory (TCM; CD45RA CD27+ CCR7+), CD4+ T-transitional memory (TTM; CD45RA CD27+ CCR7) and CD4+ T-effector memory (TEM; CD45RA CD27 CCR7), CCR7+ CD27. The P-values are shown, as obtained by paired t-test analysis following a LOGIT transformation to restore normality

Discussion

We here show that AF promotes differentiation and apoptosis of the memory CD4+ T-cell subsets that encompass the HIV-1 reservoir. This potential anti-reservoir mechanism is novel and different from the currently investigated ‘shock and kill’ strategies for HIV reservoir elimination, in that the latter are aimed at inducing viral replication in latent viral reservoirs (which cannot be recognized by drugs or the immune system), whereas AF can shorten the lifespan of the viral reservoirs, favouring their elimination. In this regard, AF-based strategies do not present the risk of enhancing viral replication in HIV+ patients. The pro-apoptotic and pro-differentiating effects of AF were also observed in another cell subpopulation, the central memory CD8+ T cells, the expansion of which is associated with progression to AIDS.25 High proportions of memory CD8+ T cells are also associated with anergy in SIV-infected macaques, and, therefore, their elimination by AF may explain the improved immune responses associated with control of viral load that we previously observed in SIV-infected macaques following suspension of AF containing therapies.11, 12 Alternatively, it is also possible that only the pro-differentiating effect of AF contributed to the strong CD8-based immunity, which we observed after treatment interruption, and that death of CD8+ T cells is a so-far inevitable by-product of this strategy.

A major finding of the present study is that the viral reservoir-harbouring CD4+ TCM/TTM cells display an overall antioxidant defense lower than that of TN cells, thus uncovering an Achilles’ heel exploitable for the elimination of the HIV reservoir. TrxRs are well studied intracellular targets of AF and, in the present study, the pro-apoptotic and pro-differentiating effects of AF could be replicated by As2O3, another TrxR inhibitor. We can therefore deduce that these enzymes represent important targets for viral reservoir elimination by drugs. Inhibition of these enzymes is in line with the activation of downstream molecular machineries that we observed in the present study. These include phosphorylation of Ask1 and MAP kinases, which are essential components of the oxidative stress signal transduction pathways, and have a central role in cell growth, differentiation and programmed cell death.34, 35, 36, 37, 38, 39

Our data show that activation of the redox-sensitive protein p38 MAPK is associated with the pro-apoptotic activity of AF. The late increase in ROS that we observed here is also consistent with mitochondrial depolarization, shown in other studies to occur following MAPK activation.40, 41, 42 It is well known that mitochondria have an increased content of ROS that, following mitochondrial membrane depolarization, are released into the cytosol. One likely explanation behind the partially selective ‘antimemory’ effect of AF is that, unlike the memory subsets, TN cells did not show AF-induced upregulation of Mn2+ SOD, an enzyme catalyzing conversion of superoxide into hydrogen peroxide. Although Mn2+ SOD may provide an initial antioxidant benefit by limiting the pools of the superoxide ion, the accumulation of its reaction product hydrogen peroxide can be deleterious in the longer term.43 In this regard, a marked pro-apoptotic effect of hydrogen peroxide on the memory T-cell pool has been previously demonstrated.44

A possible model accounting for the pro-apoptotic effect of AF in TCM and TTM cells can now be reconstructed, as shown in Figures 8a and b, respectively. In this model, the downstream effects are the results of a balance of pro-apoptotic and pro-survival signals. AF-mediated inhibition of TrxRs would represent a sparking event in the activation of the apoptotic cascade in human primary CD4+ T cells. This event subsequently activates p38 MAPK, which becomes phosphorylated. It is known that phosphorylation of p38 MAPK is associated with a decrease in the anti-apoptotic potential of the Bcl-2 protein.45, 46 If this reconstruction is correct, p38 MAPK activation would in turn induce mitochondrial depolarization and caspase activation leading to cell death. The eventual cell death observed suggests that this pathway is able to overcome a likely AF-induced pro-survival signal emerging from our protein array analysis. This signal involves Erk1/2 and likely results in the phosphorylation of Bad in S136/S155, hampering its dimerization with BclXL (Figure 5c).

Figure 8
figure 8

(a and b) Proposed model of the molecular mechanisms underlying AF-induced apoptosis of human primary CD4+ central and transitional memory T cells

Our data also show that the pro-differentiating effect of AF can be de-coupled from the pro-apoptotic effect. Differently from the pro-apoptotic effect, the pro-differentiating effect of AF is not hydrogen peroxide dependent, as evidenced by the lack of inhibition by pyruvate of the AF-induced CD27 downmodulation. The pro-differentiating pathway is likely to diverge from the pro-apoptotic pathway downstream of p38 MAPK activation. In this regard, activation of ZAP70/Syk, a pivotal mediator of lymphocyte activation and differentiation, represents a possible explanation for the pro-differentiating effects of AF.47 Ingenuity Pathway Analysis results support several possibilities for ZAP70/Syk activation as a consequence of pathways activated by AF (Supplementary Figure S4). Unfortunately, our experiments using the ZAP70/Syk inhibitor (Piceatannol, Sigma-Aldrich, St Louis, MO, USA) were hampered by the extremely high toxicity of the compound starting from 24 h of incubation (data not shown). The pro-differentiating effect of AF may have an important role in viral reservoir restriction in vivo, limiting the ‘stem-cell-ness’ of the TCM and TTM pools, and turning these cells into short-lived lymphocytes with limited in vivo persistence of the associated viral reservoir. Further investigation of these mechanisms, for example, through RNA macro-array analysis, might open new avenues for novel therapeutic strategies aimed at broader and more ambitious targets, such as complete eradication of lentiviral infections.

Materials and Methods

Cell sorting and flow cytometric analysis

Low-density mononuclear cells (less than 1,077 g/mL) were isolated by Ficoll-Hypaque density-gradient centrifugation (Sigma-Aldrich). Human peripheral blood mononuclear cells (PBMCs) were isolated from whole blood by Ficoll centrifugation and resuspended in RPMI-10% fetal bovine serum (FBS). CD4+ T cells were isolated by positive selection (MACS Separation Columns, Miltenyi Biotec, Teterow, Germany). Isolated CD4+ T cells (purity: >90%) were stained for 20 min at 4 °C in MACS buffer (PBS-1,25%FBS, 2 mM EDTA) with the following combination of antibodies: CD45RO-FITC-conjugated antibody (Becton Dickinson (BD), Franklin Lakes, NJ, USA), CD27-PE-conjugated antibody (Miltenyi Biotec, Bergisch Gladbach, Germany) and CCR7-PE Cy7-conjugated antibody (BD). Cells were washed in MACS buffer before being resuspended at 30 × 106 cells per mL in the same buffer for sorting. Cell subpopulations (TN, TCM, TTM and TEM; purity: >98%) were sorted on a BD fluorescence-activated cell sorting (FACS) Aria, (Becton-Dickinson, San Jose, CA, USA). Debris and cell doublets were gated out on the basis of physical parameters. CD45RO+ and CD45RO lymphocytes were then selected after analyzing single-cell events in a FITC versus FSC plot. The gated CD45RO+ T-cell events were then plotted for CD27 versus CCR7 expression, which allows the selection of the TCM, TTM and TEM subpopulations, whereas the gated CD45RO T-cell events were plotted for CD27 versus CCR7 expression and the cells positive for both markers were sorted as TN. Purity of the sorted populations was assessed by flow cytometry with a FACSCanto flow cytometer (Becton-Dickinson) and analyzed with the FACS Diva software (Becton-Dickinson).

Total CD4+ T cells and isolated subsets were cultured at 1 × 106 cells/mL in RPMI supplemented with 10% FBS and IL-2. Cells were incubated with AF (250 nM) and with or without different compounds aimed at enhancing/inhibiting or mimicking the effect of AF: p38 MAPK inhibitor (20 μ M; SB203580 Calbiochem, La Jolla, CA, USA), Erk1/2 inhibitor (20 μ M, PD98059 Calbiochem), pyruvate (1 mM), As2O3 (2 μ M) and ZAP70/Syk inhibitor (piceatannol, P0453 Sigma-Aldrich). After 48 h of culture, cells were collected and stained for memory and activation markers using the following combination of antibodies and reagents: CD45RO-FITC (BD) and CD27-PE (BD) in combination with Annexin V staining (Sigma-Aldrich). Cells were stained for 20 min at 4 °C in MACS buffer and washed in the same buffer before acquisition. 200 000 events were collected during flow cytometric analysis on a FACSCalibur flow cytometer (Becton-Dickinson) and analyzed using the CellQuest Pro software (BD).

Measurement of GSH

Intracellular GSH was measured with the Glutathione assay kit (Sigma-Aldrich, Milan, Italy) according to manufacturer’s instructions. The samples were first deproteinized with the 5% 5-sulfosalicylic acid solution and their GSH content was then measured using a kinetic assay. Briefly, in this assay, catalytic amounts of GSH cause a continuous reduction of 5,5′-dithiobis-(2-nitrobenzoic) acid (DTNB) to TNB. The quantity of TNB produced in the reaction is measured at 412 nm, although the oxidized GS-TNB product arising from the reaction is recycled to GSH by GSH-reductase and NADPH. Values were expressed as nanomoles of GSH per milligram of protein in the original cell extract.

Measurement of thiol content

The thiol content was measured using a ThiolTracker Violet dye (10 μ M; T10095, Invitrogen Molecular Probes). The staining was performed by applying the dye directly to live cells (500 000 cells) in thiol-free buffer for 30 min in the dark at 37 °C. 200 000 events were subsequently analyzed using a flow cytometer at 405 nm excitation as described by the manufacturer.

Detection of MMP (Δψm)

MMP was estimated by flow cytometry after staining with the JC-1 fluorescent dye (M34152, MitoProbe JC-1 Assay Kit, Life Technologies, Grand Island, NY, USA). 500 000 cells were treated with AF for 12, 24 and 48 h, harvested and then washed with cold PBS, and incubated with JC-1 solution (2.5 μg/mL) for 30 min in the dark at 37 °C. Following incubation, cells were washed twice with cold PBS and resuspended in 200 μL of cold PBS. Immediately after resuspension, the fluorescence emitted by JC-1 was analyzed by flow cytometry (FACSCalibur). 200 000 events were collected during flow cytometric analysis and analyzed using the CellQuest Pro software (BD). High MMP characterizes cells in the normal (live) state and favours the formation of JC-1 aggregates that predominantly yield red fluorescence (590 nm). Reduced MMP characterizes cells in the apoptotic state and favours the maintenance of monomeric JC-1 that yields green fluorescence (529 nm). Thus, a red to green switch in fluorescence indicates a decrease in the MMP.

Western blot analysis

Cell lysates extracted from 1 × 106 cells were resuspended in reducing SDS sample buffer, separated on SDS–polyacrylamide gels, and blotted onto nitrocellulose membranes. The membranes were blocked with 10% nonfat dry milk in TBS-tween for 1 h at room temperature. The activation of p38 MAPK was detected with phospho-specific rabbit monoclonal antibody (1/500, Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA). After stripping of bound antibody, loading was controlled with an antibody against p38 MAPK using rabbit polyclonal antisera (1/500, Santa Cruz Biotechnology, Inc.) on the according blot, as described in ref. 48. Anti-rabbit IgG horseradish-peroxidase–conjugated antibody (1/5000, Jackson ImmunoResearch Laboratories, West Grove, PA, USA) was applied for visualization. Blots were developed with an enhanced chemiluminescence system (GE Healthcare, Milan, Italy). Respective bands were quantified using Quantity One software (Bio Rad Laboratories, Hercules, CA, USA), and relative p38 MAPK activation was calculated and normalized to the loading control.

Reverse phase protein microarray analysis

Samples were printed on glass-backed nitrocellulose array slides using an Aushon 2470 arrayer (Aushon BioSystems, Burlington, MA, USA) equipped with 350 μm pins as previously described.49 Sample lysates were printed on 50 nitrocellulose slides and subjected to reverse phase protein microarray analysis (RPMA). Each sample was printed in triplicate. Slides 15 and 25 were then stained with Sypro Ruby Protein Blot Stain (Molecular Probes, Eugene, OR, USA) and visualized with NovaRay Image Acquisition Software (Alpha Innotech, San Leandro, CA, USA) to determine the total protein concentration. Before antibody staining, the slides were treated with Reblot antibody stripping solution (Chemicon, Temecula, CA, USA) and incubated for 5 h at room temperature in blocking solution with constant shaking. Blocked slides were stained with 43 different antibodies targeting total and phosphorylated proteins using an automated stainer (Dako Cytomation, Carpinteria, CA, USA). Each antibody was subjected to rigorous validation for single band specificity at the correct molecular weight (MW) by western blotting along with the use of appropriate ligand-induction controls for phospho-specific antibodies. Catalyzed Signal Amplification System kit (Dako Cytomation) and fluorescent IRDye 680 Streptavidin (LI-COR, Lincoln, NE, USA) were used as detection system. Stained slides were scanned with NovaRay Image Acquisition Software (Alpha Innotech).

Acquired images of each slide were analyzed using MicroVigene software (Vigenetech, Carlisle, MA, USA), which finds spots, performs local background subtraction, averages replicates and normalizes each sample for the total protein value. Normalization between slides was performed using control cell lysate printed on each slide as a bridging case (Supplementary Figure S5).

Systems biology

Ingenuity Pathway Analysis (IPA7.0, Ingenuity System, http://www.ingenuity.com/) was used to establish physical and/or functional interactions among proteins found either upregulated or downregulated in AF-induced CD4+ T-cell subpopulations (TN, TCM, TTM and TEM).

Cell separation and isolation of CD4+ T-cell subsets from HIV+ subjects by cell sorting

PBMCs from HIV+ subjects were isolated from whole blood by density gradient centrifugation (Ficoll) and resuspended in RPMI/10% FBS. Total CD4+ T cells were isolated by negative selection on a Robosep (Stemcell–Human CD4+ T cell enrichment kit). Isolated CD4+ T cells (more than 90% pure, as determined by flow cytometry) from HIV+ subjects were then sorted on an ARIA II (BD Biosciences) into TN (CD45RA+CD27+CCR7+), TCM (CD45RACD27+CCR7+), TTM (CD45RACD27+CCR7) and TEM (CD45RA+CD27CCR7) using the following combination of antibodies: CD3-PB, CD4-A700, CD45RA-APC Cy7, CD27-PE, CCR7-PE Cy7 and Vivid-AmCyan (Invitrogen Life Technologies, Carlsbad, CA, USA). Sorted subsets were more than 98% pure.

Cell culture

Total CD4+ T cells or isolated subsets, obtained from HIV+ subjects, were cultured from 1 to 5 × 106 cells/mL in RPMI completed with 10% FBS for 3 to 6 days in the presence or absence of increasing concentrations of AF ranging from 0 to 500 nM.

Reactivation assays and HIV-1 viral RNA quantification

The virus was pelleted by centrifugation for 60 min at 17 000 r.p.m. at 4 °C. To generate the standard curve, a sample of titer-known ACH2 virus was pelleted in the same run. After the centrifugation, total viral RNA was extracted with the QIAamp viral RNA extraction kit from Qiagen (Crawley, UK) according to the manufacturer’s instructions. Briefly, samples were lysed under highly denaturing conditions to inactivate RNases. Alcohol was added and lysates loaded onto the QIAamp spin column. Wash buffers were used to remove impurities and pure, ready-to-use RNA was then eluted in 60 μL of low-salt buffer (supplied with the kit). The RNA solution was stored at −80 °C until use.

The purified RNA was then used as a matrix for a two-step quantitative real-time reverse transcription-PCR (RT-PCR followed by qRT-PCR). For each sample, a minimum of two independent replicates (separate wells) were performed, including the ACH2 RNA sample as a standard ranging from 300 000 copies to 3 copies.

Reverse transcription (RT-PCR)

Total viral RNA (17 μL) was first treated with 1 U DNase in DNase I reaction buffer 1 × for 10 min at 25 °C. The DNase was inactivated with 1 μL 25 mM EDTA for 10 min at 65 °C. Total viral RNA was then reverse-transcribed into complementary DNA for qPCR analysis. RT-PCR was performed in 50 μL of solution containing 22 μL of DNase-treated RNA, 0.5 μL of each gag gene-specific primers (50 μ M each), LM667 (5′-ATG CCA CGT AAG CGA AAC TCT GGC TAA CTA GGG AAC CCA CTG-3′) and GagR (5′-AGC TCC CTG CTT GCC CAT A-3′), 2 μL of Superscript III RT/Platinum Taq mix and 1 × Reaction mix (RT-PCR One Step kit – Invitrogen) in a final 50 μL volume. No-template samples were used as negative controls. The running conditions were as follows: reverse transcription 30 min at 50 °C, denaturation 2 min at 94 °C followed by 20 cycles of 94 °C for 15 sec (denaturation), 62 °C for 30 s (annealing) and 68 °C for 1 min (extension). The reaction was achieved by a final elongation at 68 °C for 5 min before cooling gradually to 4 °C. The cDNAs were diluted 10-fold with DNase–RNase-free water, then subjected to quantitative real-time PCR analysis.

Quantitative real-time PCR

Experiments were performed with a LightCycler Carousel-based system (Roche, Penzberg, Germany). H2O was included as a no-template control. All reactions were carried out in 20 μL of reaction mixtures containing 6.4 μL of cDNAs, 0.3 μL of Taq DNA polymerase (Invitrogen), 1X Jumpstart mix (Sigma), 1.8 μL MgCl2 25 mM, 0.25 μL of each gag gene-specific primers (100 μ M each), Lambda T (5′-ATG CCA CGT AAG CGA AAC T-3′) and A55M (5′-GCT AGA GAT TTT CCA CAC TGA CTA A-3′), 0.5 μL each hybridization probes (8 μ M each) LTR-LC (LCred640-5′-CAC TCA AGG CAA GCT TTA TTG AGG C-3′-Phosphate) and LTR-FL (5′-CAC AAC AGA CGG GCA CAC ACT ACT TGA-3′-Fluorescein). The running conditions were as follows: 4 min at 95 °C, followed by 50 cycles of 95 °C for 10 sec (denaturation), 60 °C for 10 sec (annealing) and 72 °C for 9 sec (extension). Following the PCR reaction, melting curve analysis was performed to control amplification specificity by measuring the fluorescence intensity across the temperature interval from 45 °C to 95 °C. The absence of nonspecific products or primer dimers was indicated by observation of a single melting peak in melting curve analysis.

Statistical analyses

Statistical analyses were conducted using GraphPad Prism, v5.0 (GraphPad software, La Jolla, CA, USA), unless otherwise specified. Correlation between two variables was analyzed using Spearman’s correlation coefficients (r). Differences between variables were analyzed using parametric tests such as t-tests, or, in case of multiple comparisons, one- or two-way analysis of variance (ANOVA) followed by an appropriate post-hoc test. Paired or repeated measures tests were adopted for matched observational data points. An appropriate transformation such as the LOGIT transformation of percentage values was done to restore normality, where necessary. A P-value <0.05 was considered to be statistically significant.