Effects of IFN-α as a signal-3 cytokine on human naïve and antigen-experienced CD8+ T cells

Authors


Abstract

IFN-α/β link innate and adaptive immune responses by directly acting on naïve CD8+ T cells. This concept unveiled in mice remains unexplored in humans. To investigate that, human CD8+CD45RO cells were stimulated with beads coated with anti-CD3 and anti-CD28 mAb, mimicking Ag (type-1) and co-stimulatory (type-2) signals, in the presence or absence of IFN-α and their transcriptional profiles were defined by cDNA-microarrays. We show that IFN-α provides a strong third signal directly to human CD8+ T cells resulting in regulation of critical genes for their overall activation. This transcriptional effect was substantiated at the protein level and verified by functional assays. Interestingly, the biological effects derived from this stimulation vary depending on the CD8+ T-cell population. Thus, whereas IFN-α increases the proliferative capacity of naïve CD8+ T cells, it inhibits or does not affect the proliferation of Ag-experienced cells, such as memory and effector CTL, including CMV-specific lymphocytes. Cytolysis and IFN-γ-secretion of all these populations are enhanced by IFN-α-derived signals, which are critical in naïve CD8+ T cells for acquisition of effector functions. Our findings in human CD8+ T cells are informative to understand and improve IFN-α-based therapies for viral and malignant diseases.

Introduction

Type I IFN (IFN-I) comprises a cytokine family that in humans includes 13 IFN-α subtypes and single proteins for IFN-β, IFN-ε, IFN-κ and IFN-ω 1. IFN-α/β are produced in response to viruses and are critical for viral defense. IFN-I signals through a common receptor (IFNAR) composed of two subunits, IFNAR1 and IFNAR2 2. The JAK-STAT pathway is critical for IFNAR signaling 3. It activates the transcription factors STAT1 and STAT2 4, 5, which control the transcription of hundreds of genes 3.

The human pharmacopeia includes IFN-I 6. Direct effects on malignant or virus-infected cells have been considered the main mechanism for the efficacy of IFN-I in therapy. However, IFN-I directly regulates many immune system cells such as NK cells, DC and B- and T-lymphocytes 7.

In mice, IFN-α/β are important enhancers of CD8+ T-cell responses 8. One contributing factor is DC stimulation 9. However, direct effects of IFN-I on DC seem to be insufficient for CD8+ T-cell priming 8, 10. IFN-I also exerts direct effects on murine CD8+ T cells 4, 10–13. The most definitive report came from Kolumam et al.12 who showed that IFN-I directly targets anti-viral CD8+ T cells in vivo allowing their clonal expansion and differentiation into memory cells.

Elegant experiments in mice by the group of Mescher 11 have shown that, in addition to signals via TCR (signal-1) and CD28 (signal-2), naïve CD8+ T cells require a third signal. Signal-3 delivered by IL-12 or IFN-α is required for expansion, acquisition of effector functions and memory differentiation. cDNA microarray analyses show that IFN-α as a signal-3 regulates critical genes involved in CTL functions 14, providing evidences that IFN-α promotes activation and differentiation of CD8+ T cells by sustaining the expression of key genes through chromatin remodeling.

There is very scanty information about the effects of IFN-I on human CD8+ T cells and how IFN-I may alter the response of different CD8+ T-cell subsets. Since IFN-α is frequently prescribed to patients with a variety of medical conditions, it is of great importance to determine whether mouse and human CD8+ T cells respond in the same way to this bio-therapeutic agent. Using good manufacturing practice (GMP)-grade recombinant IFN-α and Beads coated with anti-human CD3 and CD28 mAb to mimic type-1 and type-2 signals, we show that IFN-α provides a strong type-3 signal directly to human CD8+ T cells supporting the acquisition of effector functions. Intriguing distinct IFN-α effects on the expansion of human naïve and Ag-experienced CD8+ T cells are described.

Results

IFN-α-induced gene expression profiles in human CD8+ T cells

Magnetically sorted untouched CD8+CD45RO cells were stimulated (7 h) with GMP-grade recombinant IFN-α2b or IFN-α5 and their transcriptional profiles were defined by cDNA microarrays (Series GSE17299, deposited in the Gene Expression Omnibus (GEO) database, accession number GSE17302). In total, 195 genes changed at least two-fold by either IFN-α2b or IFN-α5 and 161 genes were regulated in common. Supporting Information Table 1 groups genes by functional pathways. The regulation of several transcripts involved in cell-mediated cytotoxicity [TNFSF10 (also known as TNF-related apoptosis-inducing ligand (TRAIL), FASLG and PRF1], chemotaxis (CXCL10 and CXCL11) and T-cell homeostatic proliferation (IL15RA) were confirmed by quantitative RT-PCR (Table 1A).

Table 1. List of immunologically relevant genes regulated by IFN-α in human CD8+CD45RO cells determined by RT-PCRa)
AFold induction of stimulated versus unstimulated cells
Genes§IFN-α2bIFN-α5
  • a)

    a) (A and B) Untouched CD8+CD45RO cells negatively selected from seven individuals were left unstimulated or stimulated with the indicated stimuli () for 7 h (A) or 48 h (B). After this time, RNA was isolated and the expression of the indicated genes (§) were determined by quantitative RT-PCR. These tables show the quantitative RT-PCR increases of stimulated sample over unstimulated controls represented as fold change (FC)±SEM. (*) Statistical comparison of FC (Beads versus Beads+IFN-α2b or Beads versus Beads+IFN-α5) (p≤0.05) (Wilcoxon signed-rank test). (†) Statistical comparison of FC (IFN-α2bversus Beads+IFN-α2b) (p≤0.05) (Wilcoxon signed-rank test).

CXCL101085±2581 677±1607 
CXCL11415±594 183±313 
TRAIL7.1±3.6 5.8±4.4 
IL15RA3.9±2.3 4.0±2.3 
PERF12.1±0.4 1.9±0.2 
FASLG2.1±0.7 1.8±1.0 
BFold induction of stimulated versus unstimulated cells
Genes§IFN-α2bBeadsBeads+IFN-α2bBeads+IFN-α5
IFNG0.91±0.705.3×102±5.7×1021.2×103±1.9×103* †9.0×102±5.9×103*
GZB1.70±0.4432.48±39.8856.75±63.04*51.82±67.30*
FASL2.85±4.722.03±2.4311.76±19.26*8.39±9.54*
TRAIL4.99±1.860.96±0.546.40±3.68*5.74±3.68*
CXCL106.3×102±1.4×1037.4×104±9.1×1045.1×105±7.7×105*3.6×105±4.8×105*
CXCL1136.86±60.684.5×103±7.2×1034.0×104±8.0×104*2.0×104±3.2×104*
CCL204.50±5.7815.02±27.3855.91±86.82*42.31±40.80*
CCL31.17±0.2717.45±15.6748.90±29.62*40.41±21.99*
CD384.56±2.883.82±1.1218.28±12.34*12.22±8.83*
IL224.22±62.773.0×102±6.3×1026.0×102±1.0×103*6.7×102±1.3×103*

A second set of microarrays (Series GSE17301) analyzed the effects of IFN-α on human CD8+ T cells co-stimulated by Beads coated with anti-human CD3 and CD28 mAb (henceforth Beads). CD8+CD45RO cells were left unstimulated or stimulated (48 h) with IFN-α2b, or with Beads alone or together with IFN-α2b or IFN-α5. As a signal-3 cytokine, IFN-α2b and IFN-α5 regulated in common 74 genes (Supporting Information Table 2). IFN-α-derived type-3 signals on human CD8+ T cells induced transcripts involved in effector functions (IFNG, GZB, FASLG and TRAIL) and T-cell immune responses (CD38 and IL2) that were confirmed by quantitative RT-PCR (Table 1B). Genes involved in chemoattraction were also regulated by IFN-α-derived type-3 signals (Table 1B and Supporting Information Table 2).

No substantial differences were found between IFN-α2b and IFN-α5 either when acting as single agents or in combination with Beads (Table 1).

IFN-α provides a strong stimulating signal directly to human CD8+CD45RO cells

CD3/CD28-triggering induced blastic transformation on CD8+CD45RO cells, as depicted by forward versus side scatter changes (Fig. 1A and C). IFN-α-derived signals by themselves did not induce blast transformation, but strongly enhanced the CD3/CD28-induced pro-blastic effects. Moreover, IFN-α by itself was unable to increase the expression of CD25 or CD38 (Fig. 1B and D) and barely induced a marginal up-regulation of CD69 (Supporting Information Fig. 1). However, in combination with CD3/CD28-signaling IFN-α markedly enhanced the surface expression of these three molecules (Fig. 1B and D and Supporting Information Fig. 1).

Figure 1.

IFN-α markedly enhances blast transformation and surface expression of CD38 and CD25 on CD3/CD28-triggered cells. Purified CD8+CD45RO cells were unstimulated or stimulated with IFN-α2b, anti-CD3/CD28-Beads alone or together with IFN-α2b or IFN-α5. (A and C) Forward (FSC) and side (SSC) scatter parameters (72 h). (A) Numbers in brackets refer to the FSC and SSC mean values (FSC, SSC) corresponding to the R1 cell population. (C) Average data (Mean±SEM) (n=10) of the FSC×SSC values (FSC×SSC=product of FSC by SSC values). (B and D) Surface expression of CD38 and CD25 on CFSE-labeled cells at 48 h (CD38) or 72 h (CD25) of culture. Numbers in brackets refer to percentage of cells expressing that molecule. (D) Percentage of cells expressing CD38 or CD25. Each line represents an individual subject. (C and D) Numbers in italics indicate p values obtained by Wilcoxon signed-rank (C) or paired T (D) test. One subject representative of ten (A) or three (B) tested.

IFN-α significantly enhanced CD3/CD28-induced cell number expansion of CD8+CD45RO cells (Fig. 2A). Cell division as assessed by CFSE dilution required CD3/CD28-triggering and was not detected until 72 h of culture (Supporting Information Fig. 2A). In some individuals (5/12) we observed that at day 4 of culture Beads+IFN-α-stimulated cells displayed a slightly higher CFSE intensity than cells stimulated only with Beads, indicating fewer divisions (Supporting Information Fig. 2B). However, from day 5, the content of CFSE was always lower in those cells receiving CD3/CD28/IFNAR-derived signals, and this higher level of division is accompanied of a higher percentage of divided cells (in 12/12 individuals) (Fig. 2B and C and Supporting Information Fig. 2). Figure 2D and E show that cell death mediated by CD3/CD28-triggering was reduced in the presence of IFN-α. Of note, IFN-α did not protect against cell death in the absence of CD3/CD28-stimulation.

Figure 2.

IFN-α-derived type-3 signals enhance expansion of human CD8+ T cells and protect them against activation-induced cell death. CFSE-labeled purified CD8+CD45RO cells were cultured as in Fig. 1. (A) Fold expansion (day 5). (B and C) Extent of proliferation assessed by CFSE dilution (day 5). Percentages of divided cells are indicated. “B” stands for Beads. (A–C) Cells were gated as in Fig. 1. (D and E) Percentage of dead cells assessed by TOPRO-3 staining at the indicated time points (D) or at day 5 (E). Analyses were performed in the total cell population, including those cells with high (R1, as detailed in Fig. 1A) and low FSC values. Numbers in brackets (D) and “Y” values (E) indicate the percentages of divided cells that were TOPRO-3+. (A, C and E) Average data (Mean±SEM) from 12 (A and C) and six (E) different subjects. Numbers in italics indicate p values obtained by Wilcoxon signed-rank test. One subject representative of 12 (B) or six (D) tested.

Importantly, IFN-α acts on CD3/CD28-triggered cells to increase the expression of IFN-γ, Granzyme-B and TRAIL (Fig. 3A). No other further in vitro stimulation step (most usually stimulation with PMA/ionomycin) was used to detect these three effector molecules. In other words, Fig. 3A is the confirmation at the protein level of the effects of IFN-α on IFNG, GZB, and TRAIL transcripts. Although the production of IFN-γ, as measured by intracellular staining, was marginal (Fig. 3A), the levels of secreted IFN-γ determined by ELISA confirmed the IFN-α-mediated enhanced production of IFN-γ (Fig. 3B). IFN-α in combination with Beads also markedly increased the release of CXCL10 (Fig. 3C). IFN-α2b and IFN-α5 effects were almost identical over the broad range of concentrations tested (Supporting Information Fig. 3). The necessary role of IFNAR was revealed by neutralizing anti-human IFNAR2 mAb (Supporting Information Fig. 4).

Figure 3.

IFN-α provides a strong type-3 signal to human CD8+CD45RO cells that promotes acquisition of effector functions. Purified CD8+CD45RO cells were cultured as in Fig. 1. (A) Expression of IFN-γ, Granzyme-B and TRAIL on CFSE-labeled cells (day 4). Numbers in brackets refer to the percentage of divided cells expressing that molecule. (B and C) Average levels (Mean±SEM ) of IFN-γ (B) or CXCL10 (C) on day-4 culture supernatant (n=14). (D and E) CD3-redirected cytotoxicity against OKT3-loaded p815 cells. (D) Percentage of specific lysis (calculated as detailed in Materials and methods) at varying E:T ratios. (E) Average data (Mean±SEM) of specific lysis depicted as LU (n=6). (F and G) TRAIL-mediated cytotoxicity on Caki-1 target cells at different E:T ratios (F) or at 50:1 ratio in the presence or absence of neutralizing anti-TRAIL and/or anti-FASL mAb (G). (G) The dotted line indicates the percentage of lysis corresponding to unstimulated CD8+ T cells co-cultured with Caki-1. (B, C and E) Numbers in italics indicate p values obtained by Wilcoxon signed-rank (B) or paired T (C and E) tests. One subject representative of six (A, D) or four (F, G) tested. (D, F and G) “B” stands for beads.

The CD3-redirected cytolytic assay using OKT3 mAb-coated p815 target cells is commonly used to evaluate the TCR/CD3-triggered cytotoxicity that entails release of perforin and granzymes, and surface relocation of CD107a. Furthermore, Caki-1 cells, sensitive to TRAIL- but not to FasL-induced cell death, can be used as target cells to assess TRAIL-mediated cytotoxicity 15. Figure 3 strikingly shows that IFN-α enhanced CD3-redirected cytotoxicity (Fig. 3D–E) as well as TRAIL-mediated cytolysis (Fig. 3F–G). Neutralizing anti-TRAIL and anti-FasL mAb revealed the exclusive contribution of TRAIL in the lysis of Caki-1 cells (Fig. 3G).

No significant differences were found between the IFN-α2b and IFN-α5 subtypes in any of these assays (Figs. 1–3 and Supporting Information Figs. 1–4).

Distinct effects of IFN-α on naïve and memory CD8+ T cells

Following CD27- and CD45RA-based phenotypic classifications of CD8+ T cells 16, negatively selected total CD8+ T cells were sorted into naïve (CD45RAhighCD27high), memory (CD45RACD27+) and effector (CD45RA+CD27 and CD45RACD27) cells.

For comparative studies, naïve and memory CD8+ T cells were stimulated as above. Regardless of whether cells were naïve or memory, cell division was not noticeable before 72 h of culture and required CD3/CD28-triggering (Supporting Information Fig. 5A and B). At day 4 of culture, naïve CD8+ T cells from some individuals (3/8) showed a transiently delayed proliferation in the presence of IFN-α (Supporting Information Fig. 5C). However, from day 5, the extent of division was always higher in cells receiving CD3/CD28/IFNAR-derived signals (observed in 8/8 individuals) (Fig. 4A and Supporting Information Fig. 5A and C). By contrast, once division started, CD3/CD28-induced proliferation of memory cells was always delayed by IFN-α (Fig. 4A and Supporting Information Fig. 5B). Interestingly, IFN-α increased the survival of both CD3/CD28-triggered naïve and memory CD8+ T cells (Supporting Information Fig. 5D and E). IFN-α-derived type-3 signals significantly increased the expansion of human naïve CD8+ T cells whereas they reduced the fold expansion of memory CD8+ T cells (Fig. 4B).

Figure 4.

Differential effects of IFN-α on human naïve and memory CD8+ T cells. Purified CD45RAhighCD27high (naïve) and CD45RACD27+ (memory) CD8+ T cells were cultured as in Fig. 1. (A) Proliferation was monitored by CFSE dilution (day 5). Percentages of divided cells are indicated. “B” stands for Beads. (B) Fold cell expansion of purified naïve and memory CD8+ T cells (day 5). Each line represents an individual subject. (C) IFN-γ concentrations (Mean±SEM) determined on 4-day culture supernatants (n=11). (D) CD3-redirected cytotoxicity against OKT3-loaded p815 cells. (E) TRAIL-mediated cytotoxicity (E:T=50:1) in the presence or absence of neutralizing anti-TRAIL and/or anti-FASL mAb. The orange line indicates the percentage of lysis corresponding to unstimulated CD8+ T cells co-cultured with Caki-1 cells in the presence of IgG. (B and C) Numbers in italics indicate p values obtained by the paired T test (B) or the Wilcoxon signed-rank test (C). One subject representative of six (A) or four (D and E) tested.

When the expression of IFN-γ, Granzyme-B and TRAIL was assessed by flow cytometry analysis, we found that IFN-α enhanced the expression of these three effector molecules both in naïve and memory CD8+ T cells (Supporting Information Fig. 6). However, the fold-change increases in protein induction attributable to IFN-α were markedly higher in naïve cells (Supporting Information Fig. 6). Figure 4C shows that regardless of whether the cells were naïve or memory, the amounts of secreted IFN-γ were higher in cells receiving IFN-α as a signal-3.

The lytic activities driven either by TCR/CD3-triggering (Fig. 4D) or delivered by TRAIL (Fig. 4E) were enhanced by IFN-α-derived type-3 signals both on naïve and memory cells. Lysis of Caki-1 cells was completely mediated by TRAIL in naïve CD8+ T cells, while in memory cells there was a slight contribution of FasL (Fig. 4E).

To further confirm the effects of IFN-α on human naïve CD8+ T cells and to completely exclude Ag-experienced CD8+ T cells, umbilical cord blood mononuclear cells (UCBMC) were used as a source of neonatal CD8+ T cells. Figure 5 shows that IFN-α2b with concomitant CD3/CD28-signaling clearly enhanced proliferation, IFN-γ secretion as well as the cytolytic activity (both CD3-redirected and TRAIL-mediated) of human neonatal CD8+ T cells.

Figure 5.

Direct IFN-α stimulation confers full effector functions to human neonatal naïve CD8+ T cells triggered via CD3/CD28. Human neonatal CD8+ T cells from UCBMC were cultured as in Fig. 1. (A) Extent of proliferation by CFSE dilution (day 5). (B) IFN-γ concentrations (Mean±SEM) on 4-day culture supernatants (n=5). (C) CD3-redirected cytotoxicity depicted as LU. Each box represents the Mean±SEM (n=5). (D) TRAIL-mediated cytotoxicity. One out of five (A) or three (D) subjects tested. (B and C) Numbers in italics indicate p values (paired T test). (A and D) “B” stands for Beads.

Effects of IFN-α on effector CD8+ T cells from human peripheral blood

Circulating CD45RA+/−CD27 CD8+ T cells cells behave as effector CTL since they abundantly express FasL mRNA, contain perforin and Granzyme-B, and are able to kill ex vivo target cells. These cells are characterized by their low proliferative potential 16. As shown in Fig. 6A, CD45RA+CD27 effector cells did not divide after stimulation with Beads even in the presence of IFN-α. However, a weak cell division was observed in CD3/CD28-triggered CD45RACD27 CTL that was delayed by IFN-α (Fig. 6A).

Figure 6.

Direct effects of IFN-α on human CD45RA+CD27 and CD45RACD27 effector CTL. (A) CFSE-labeled purified CD45RA+CD27 and CD45RACD27 CTL were cultured as in Fig. 1. Proliferation was monitored by CFSE dilution (day 5). “B” stands for Beads. (B and C) Freshly purified CD45RA+CD27 and CD45RACD27 CTL were co-cultured (E:T ratio=10:1) with IgG- or OKT3-loaded p815 target cells [(IgG)p815 and (OKT3)p815, respectively] in the presence or absence of IFN-α2b or IFN-α5. (C) Expression of IFN-γ, CD107a and TRAIL after 4 h (IFN-γ, CD107a) or 18 h (TRAIL) of culture. P815 cells were gated out as H-2Kd positive cells. (D) Purified CD45RA+CD27 and CD45RACD27 CTL were cultured as in (A) and IFN-γ levels (Mean±SEM) were determined on day-4 culture supernatants (n=8). Numbers in italics indicate p values (Wilcoxon signed rank test). (E) Purified CD45RA+CD27 and CD45RACD27 CTL were stimulated as in (A) and 18 h later cells were harvested and tested against Caki-1 cells (E:T=10:1) in the presence of neutralizing anti-TRAIL or anti-FASL mAb or mouse IgG. The orange line indicates percentage of lysis corresponding to unstimulated CTL co-cultured with Caki-1 in the presence of IgG. One subject representative of six (A), five (C) or four (E) tested.

Next we examined the effects of IFN-α on the effector functions of CD45RA+/−CD27 CTL. As these cells are endowed with immediate effector functions, freshly purified CD45RA+CD27 and CD45RACD27 CTL were co-cultured with control IgG- or OKT3-loaded p815 target cells in the presence or absence of IFN-α without any previous step of in vitro restimulation (Fig. 6B). As depicted in Fig. 6C, IFN-α markedly enhanced the expression of IFN-γ upon encounter of OKT3-loaded target cells. Similarly, IFN-α also increased the levels of secreted IFN-γ upon stimulation of CD45RA+CD27 and CD45RACD27 CTL with Beads (Fig. 6D). By contrast, IFN-α did not alter the surface expression of CD107a as attained by the co-culture with OKT3-loaded target cells (Fig. 6C).

Freshly purified CD45RA+CD27 or CD45RACD27 CTL did not express TRAIL on their surface (data not shown). However, expression of TRAIL became apparent after 18 h of culture with OKT3-loaded p815 cells combined with IFN-α (Fig. 6C). This expression correlated with enhanced TRAIL-mediated killing of Caki-1 cells (Fig. 6E).

Human CMV-specific CD8+ T cells display enhanced effector functions upon expansion with IFN-α

CD8+ T cells specific for the CMVpp65495–503 epitope were sorted from HLA-A2+ subjects that showed a detectable positive staining in PBL with the HLA-A2/CMVpp65495–503-pentamer (CMVpent). The patterns of CD45RA/CD27 expression within the CMVpent+ cell population varied among individuals (Supporting Information Fig. 7A and B). Freshly purified CMVpent+ cells resembled the surface phenotype ascribed to effector or recently activated CTL, rather than to resting memory lymphocytes. CMVpent+ cells paralleled effector CTL since they expressed Granzyme-B (Supporting Information Fig. 7C) and were able to kill (Supporting Information Fig. 7D) and to produce high amounts of IFN-γ (Fig. 7A) upon ex vivo recognition of target cells.

Figure 7.

IFN-α enhances effector functions of CMV-specific CD8+ T cells. Sorted CMVPent+ cells were co-cultured (4 h) with control or CMVpp65495–503 peptide-loaded T2 cells in the presence or absence of IFN-α2b or IFN-α5 (A), or were expanded (4–5 days) in the presence of IL-2 with anti-CD3/CD28-Beads alone [B(IL2)] or together with IFN-α2b [B(IL2)+IFN-α2b] or IFN-α5 [B(IL2)+IFN-α5] (B–G). After expansion, cells were deprived of IL-2 (18 h) and then co-cultured with control or CMV peptide-loaded T2 cells (E:T=5:1) (E–G). (A) Expression of IFN-γ and CD107a by CMV-specific CD8+ T cells. (B) Proliferation of CMV-specific CD8+ T cells monitored by CFSE dilution. CMVPent+ cells cultured in the absence of (w/o) IL-2 are shown to indicate the starting point for CFSE dilution. (C) Fold cell expansion (days 4–5). (D) IFN-γ concentrations determined on day-4 culture supernatants. (E) Expression of IFN-γ and CD107a by expanded CMV-specific CD8+ T cells. For the last 4 h prior to staining, expanded cells were co-cultured with control or CMV peptide-loaded T2 cells. (A and E) T2 cells were gated out as H-2Kb positive cells. (F and G) Expanded CMV-specific CD8+ T cells were mixed at different E:T ratios (F) or at 5:1 ratio (G) with T2 cells pulsed with control or CMV peptide. (C, D and G) Each line represents an anonym donor re-coded as CMV-digit. Numbers in italics indicate p value (paired T test). One subject representative of six (B), five (A) or four (E) tested.

As CMV-specific cells were endowed with features of effector CTL, freshly purified CMVPent+ CTL were directly co-cultured with HLA-A2-expressing T2 cells loaded with control or CMVpp65495–503 peptide (CMV peptide), in the presence or absence of IFN-α. IFN-α enhanced the production of IFN-γ, but did not affect the surface expression of CD107a (Fig. 7A). Accordingly, IFN-α did not alter the immediate lytic activity of CMV-specific CTL (Supporting Information Fig. 7D).

Current adoptive therapies developed to treat CMV infection after allogenic bone marrow transplantation involve isolation of circulating CMV-specific CD8+ T cells from healthy donors, in vitro expansion and infusion into the patients 17. To explore how IFN-α could affect the process of in vitro expansion, sorted CMVPent+ cells were cultured for 4–5 days with IL-2-conditioned medium alone or together with IFN-α2b, Beads or Beads in combination with IFN-α2b or IFN-α5. IL-2 was absolutely necessary for proliferation and survival of isolated CMV-specific cells (Supporting Information Fig. 7E). As shown by the CFSE dilution profiles of CMVPent+ cells from five individuals, cells underwent division in a synchronized manner regardless of the starting differentiation stage of sorted cells (Fig. 7B and Supporting Information Fig. 7F–G). CMV-specific cells in the presence of IL-2 divided without CD3/CD28-stimulation (Supporting Information Fig. 7F), indicating that the CMVpent used for the sorting sufficiently signaled through TCR/CD3. Overstimulation with Beads retarded proliferation of CMVpent-triggered cells (Supporting Information Fig. 7F). IFN-α slightly delayed the division driven by CMVpent-mediated TCR engagement either alone (Supporting Information Fig. 7G) or together with CD3/CD28-triggering (Fig. 7B). The cell expansion upon stimulation with CMVpent and Beads was clearly lowered by IFN-α (Fig. 7C).

In the presence of IL-2, CMVpent-triggered cells secreted IFN-γ (Supporting Information Fig. 7H), and the levels of secreted IFN-γ increased if the cells were further stimulated with Beads. Addition of IFN-α enhanced the amounts of IFN-γ secreted (Fig. 7D and Supporting Information Fig. 7H).

Next, we examined the IFN-α effects on the effector functions of the expanded CMV-specific cells. Hence, CMV-specific CTL cultured for 4–5 days with Beads+IL-2 in the presence or absence of IFN-α were deprived overnight of IL-2 and subsequently co-cultured with T2 target cells loaded with control or CMV peptide. Figure 7E shows that cells expanded in the presence of IFN-α produced higher amounts of IFN-γ and mobilized more efficiently CD107a to the surface than cells expanded without IFN-α. Similarly, there was a minor but significant enhancement of the cytolytic activity against peptide-loaded targets (Fig. 7F and G). Both IFN-α subtypes tested showed similar behavior (Fig. 7).

Discussion

We show that IFN-α supports human naïve CD8+ T-cell expansion while favoring the differentiation of naïve cells into effector CTL. By contrast, on Ag-experienced CD8+ T cells we found that whereas IFN-α enhances the effector functions, it decreases fold cell expansion. No differences were found between IFN-α2b and IFN-α5 subtypes, suggesting redundancy in the system.

The magnitude of the stimuli and the inputs from different stimulatory/inhibitory receptors are critical parameters for the outcome of the T-cell response. Thus, the need of choosing a fixed dose of stimuli, a single costimulatory signal and few time points for the array analyses provides a limited and static picture of the transcriptional changes induced on human T cells. Despite this limitation, our array data provided a baseline definition of the IFN-α transcriptional effects on human CD8+ T cells and will form the basis for further and more detailed studies.

The results of the transcriptional analysis of human CD8+ T cells stimulated with IFN-α alone agree with previous studies of IFN-α stimulation of unfractionated PBL 18, 19. The overall similarity suggests that IFN-α imprints a common transcriptional signature on the peripheral blood immune cell populations. Despite induction of relevant genes for effector functions, human CD8+ T cells treated only with IFN-α experienced no sign of activation. However IFN-α-derived signals synergize with signals elicited by CD3/CD28-triggering and promote the acquisition of effector functions on human CD8+ T cells. The biological meaning of the regulation of all these genes relevant for CD8+ T-cell functions by IFN-α itself is still unknown. One possibility is that pre-exposure to IFN-α induces mRNA that facilitate T-cell activation upon an eventual Ag encounter.

Transcriptional analyses performed in human CD8+CD45RO cells stimulated with Beads and either IFN-α2b and/or IFN-α5 show that, as a signal-3 cytokine, IFN-α regulates outstanding genes involved in the overall activation of T cells. Among these genes we found IL2. IL-2 is an important cytokine for survival, clonal expansion and differentiation of T cells 20. The fact that IFN-α also promotes the surface expression of CD25 strengthens the idea that IFN-α may promote the CD8+ T-cell response, at least in part, by inducing additional cytokines that could further stimulate CD8+ T cells in an autocrine manner.

Importantly, the chief transcriptional signature of IFN-α, as a third signal, encompasses the up-regulation of transcripts involved in effector functions (IFNG, GZB and TRAIL) as well as production of chemokines (CXCL10 and CXCL11). A similar transcriptional signature has been found in OT1 cells stimulated in vitro with artificial DC and IFN-α 14, suggesting that IFN-α may promote the conversion of CD8+ T cells not only into highly effector cells but also into efficient chemotactic attractants of additional effector cells. This transcriptional effect was substantiated at the protein level and verified by functional assays. Our results contradict data recently published by Ramos et al. 21, who found that IL-12 but not IFN-α enhances human CD8+ T-cell effector functions as promoted by CD3/CD28-triggering. These authors added recombinant IL-2 to the cultures and neutralizing mAb against IL-4, IL-12 and IFN-γ that may have masked the immunostimulatory properties of IFN-α.

The induction of genes coding for effectors proteins suggests that IFN-α may also control the expression of transcription factors involved in CD8+ T-cell differentiation 22. Recently Mescher's group has shown that both IL-12 and IFN-α enhance the expression of T-bet, Eomes and blimp-1 coding genes in OT1 cells 14. However, we observed that whereas CD3/CD28-triggering regulates the expression of T-bet, bcl-6, Id2 and blimp-1, IFN-α does not exert any marked effect on the expression of these genes. We do not have any transcriptional data about Eomes, since Eomes is not represented on the HG-U133A 2.0 array. The heterogeneity in human population and different expression kinetics are likely involved in this discrepancy between human and mouse studies. The group of Mescher has also reported that several genes coding for TNF receptors, such as CD27, OX40, 4-1BB and GITR, are regulated by IFN-α-derived type-3 signals in OT1 cells 14, 23. However, we did not observe any transcriptional effect of IFN-α on the expression of these molecules by human CD8+ T cells, suggesting species-related differences.

We also show that IFN-α-derived signal-3 enhances IFN-γ production as well as Granzyme-B- and TRAIL-mediated cytotoxicity both in naïve and memory CD8+ T cells, although naïve CD8+ T cells are more dependent on IFN-α. The relative IFN-I independence of memory CD8+ T cells could be related to the ready state of their TCR signaling machinery 24. With regard to the effector subset, IFN-α enhances IFN-γ production and TRAIL-mediated cytotoxicity of CD45RA+/−CD27 effector CTL. This is an interesting point because it is thought that effector-type cells have reached a terminal differentiation stage 16. Our findings suggest that these cells may also be targets for IFN-α-based therapy.

The IFN-α effects on CD3/CD28-triggered fold expansion vary depending on the CD8+ T-cell subpopulation. Whereas IFN-α enhances the expansion of naïve CD8+ T cells, it delays the proliferation of Ag-experienced cells. This is reminiscent of reports showing that IFN-α exerts opposing functions on the proliferation depending on the cell type, the context of its action and/or the presence of other stimuli 5, 13, 25, 26. STAT1 seems to be mediating the direct anti-proliferative effects of IFN-I in mice 5, 27. It has been reported that Ag-triggered murine naïve CD8+ T cells down-regulate the levels of STAT1 to counter the anti-proliferative effect of IFN-I during viral infection 28. Ongoing experimentation will elucidate the role of STAT1 and other signaling molecules in the control of human CD8+ T-cell proliferation by IFN-α and its dichotomy of effects on naïve and memory cells.

IFN-α may also provide pro-apoptotic or survival signals depending on the type and the stage of the cell 13, 29–33. We show that IFN-α prevents CD3/CD28-triggered cell death in human naïve and memory CD8+ T cells. This is in agreement with previous experiments both in humans 30, 32, 33 and in mice 13. The reported increased survival seems to be associated with elevated levels of Bcl-xL 32, 34, and with the prevention of PKC-δ translocation to the nucleus 33.

To assess the potential of IFN-α to condition specific Ag-experienced CD8+ T cells, we have examined the effects of IFN-α on CMV-specific CD8+ T cells isolated from healthy CMV carriers. Our data show that the TCR- and/or CD3/CD28-triggered proliferation of CMV-specific cells is diminished by IFN-α. By contrast, exposure to IFN-α during the in vitro expansion enhances IFN-γ production and, to a lesser extent, the cytolytic capabilities of CMV-specific cells. For the in vitro conditioning of Ag-experienced CD8+ T cells to be used in adoptive immunotherapy this could be advantageous, but the IFN-α-induced reduction of expansion might be a handicap.

As a whole, our studies show that IFN-α directly communicates with human CD8+ T cells and that the biological effects derived from this stimulation vary depending on the CD8+ T-cell population. Our data provide important information to understand and improve IFN-α-based therapies for viral and malignant diseases.

Materials and methods

Human IFN-α

Recombinant human IFN-α2b (Realdiron) and IFN-α5 were from Sicor Biotech UAB (Vilnius, Lithuania). Both IFN were produced following GMP requirements and contained ≤5.8 IU of endotoxins/mg of protein (Gel Clot method), ≤1.2 ng of host-cell-derived proteins/mg of total protein (ELISA) and ≤25 pg of host-cell- and vector-derived DNA/mg of protein (real-time PCR). The antiviral activity of IFN-α2b and IFN-α5 was 1.66 108 and 1.01 108 IU/mg of protein, respectively.

Cell isolation

PBL were eluted from leukocyte filters provided by the blood Bank of Navarra (Spain). UCBMC were isolated by repeated centrifugation of cordon blood cells and treatment with Ammonium-chloride lysing buffer until almost complete lysis of erythrocytes. All blood and UCBMC donors gave written informed consent (Ethics Committee from the University Clinic of Navarra 007/2007 and 013/2009). For purification of CD8+CD45RO cells, PBL were labeled with the human CD8+ T-cell Isolation kit-II (Miltenyi) and sorted in an autoMACS Separator (DEPLETEs). Purified total CD8+T cells (≥75% of purity) were labeled again as before and then with anti-CD45RO microbeads (Miltenyi). Cells were sorted once more (DEPLETEs) (purity of CD3+CD8+CD45RO cells ≥95%). For purification of different CD8+ T-cell subsets, purified total CD8+ T cells were stained with the biotin mAb cocktail for CD8+ T-cell isolation (Miltenyi) and then with anti-CD27-FITC (M-T271), anti-CD45RA-PECy5 (HI100) mAb and Streptavidine-PE (to gate out contaminating non-CD8+ T cells). Cells were sorted into CD45RAhighCD27high, CD45RACD27+, CD45RA+CD27 and CD45RACD27 CD8+ T cells in a FACSAria (BD Biosciences) (95% of purity regardless the CD8+ T-cell subset). To purify CMVpp65495–503-specific CD8+ T cells, purified total CD8+ T cells from CMVpent+ subjects were stained with the biotin mAb cocktail for CD8+ T-cell isolation and subsequently with Streptavidin-PE and CMVpent-APC (Proimmune). CMVpent+ cells were sorted in a FACSAria to 95% purity. Human neonatal CD8+ T cells from UCBMC were labeled with anti-CD8 microbeads (Miltenyi) and purified using POSELD2 program (purity of CD3+CD8+≥90%).

Cell culture

Purified CD8+ T cells were cultured (5×105 cells/mL) with medium alone (RPMI-glutamax medium (Invitrogen) supplemented with 10% FCS (Sigma) and 1% penicillin/streptomycin (Invitrogen)) or medium containing IFN-α2b, IFN-α5, anti-CD3/CD28-Beads (Beads coated with anti-human CD3 and CD28 mAb) (Invitrogen) alone or together with IFN-α (IFN-α2b or IFN-α5). The IFN-α dose was 500 IU/mL. Beads were used at a 1:10 Beads:cell ratio. Purified CMVpp65495–503-specific CD8+ T cells were left unstimulated or stimulated with anti-CD3/CD28-Beads alone or together with IFN-α, in IL-2-conditioned medium (50 IU/mL) (Peprotech). In some cases, previously to stimulation, CD8+ T cells were labeled with 1.25 μM of CFSE (Sigma-Aldrich).

In some cases, freshly purified CD8+ T cells were directly co-cultured (4 h) (i) with control IgG- or anti-CD3 OKT3 mAb-loaded p815 target cells (E:T ratio=10:1) or with (ii) HLA-A2+ T2 cells (E:T=5:1) loaded with HLA-A2-restricted control peptide (Leukocyte Proteinase-3169–177) or CMV peptide (Proimmune), in the presence or absence of IFN-α.

To facilitate IFN-γ detection by intracellular staining, cells were cultured in the presence of Brefeldin A (10 μg/mL) (Sigma-Aldrich) for the last 6 h of culture or along the culture (in the case of 4 h short-term assay). For the detection of CD107a, cells were cultured in the presence of anti-human CD107a-PE mAb (H4A3) or mouse IgG1-PE (10 μg/mL) (BD Biosciences) and Monensine (1 μg/mL) (Sigma-Aldrich).

Real-time RT-PCR

Total RNA was extracted using the nucleic Acid Purification lysis solution and the semiautomated ABI Prism 6100 Nucleic Acid PrepStation system (Applied Biosystems). Total RNA was treated with DNase prior to RT with M-MLV reverse transcriptase in the presence of RNaseOUT (all from Invitrogen). Real-time RT-PCR was performed using the CFX96 Real-time system, the IQ SYBR Green Mix (BioRad) and specific primers for each gene (Supporting Information Table 3). Results were normalized to β-actin. The amount of each transcript was expressed by the formula: 2Δct [Δct=ct(β-actin)-ct(gene], with ct as the point at which fluorescence rises appreciably above background fluorescence.

Flow cytometry analysis

Cells stained with fluorochrome-labeled mAb and/or CFSE were acquired on a FACSCalibur (BD Biosciences) and analyzed using FlowJo (Tree Star). Fold expansion was calculated as the output/input ratio of the absolute numbers of cells determined using Trucount beads (BD Biosciences). Before intracellular staining, cells were fixed and permeabilized with Cytofix/CytoPerm solution (BD Biosciences). To assess the percentage of cell death, cells were first stained for surface markers and then with TOPRO-3 (Invitrogen) (10 nM).

Cytotoxicity assays

Following culture (day 4) CD8+ T cells were mixed with 51Chromium-labeled p815 cells in the presence or absence of anti-CD3 OKT3 mAb (5 μg/mL) or with Caki-1 cells. In some experiments, CD8+ T cells and Caki-1 cells were co-cultured in the presence of neutralizing anti-human TRAIL (RIK-2) and/or FasL (NOK-2) mAb (10 μg/mL). Cytotoxity activity of CMVpp65495–503-specific CD8+ T cells was assayed against control or CMV peptide-pulsed 51chromium-labeled HLA-A2+ T2 cells. 51Chromium release was counted in a Topcount (Packard). Lysis percentage was calculated as [(experimental release-spontaneous release)/(maximum release-spontaneous release)]×100. Lysis by CD3-redirected cytotoxicity was also depicted as Lytic units (LU) (number of effector cells needed to lyse 3000 targets cells) calculated by the formula LU=[1/(E:T50%)]×3000, where E:T50% is the E:T ratio at which 50% of lysis occurred. E:T50% was inferred from the killing curve (Lysis versus E:T ratio). The percentage of specific lysis was calculated after deduction of the non-specific lysis (in the presence of control peptide or IgG) from the total lysis in the presence of specific peptide or OKT3 mAb.

Statistical analysis

Data were analyzed first by the Shapiro Wilk Normality test and then by Paired T or Wilcoxon signed-rank test, depending on whether the data were or were not from a normally distributed sample, respectively. All tests were two-tailed and conducted at 95% of confidence.

Acknowledgements

Financial support was from Ministerio de Ciencia e Innovaciœn (MCI) (SAF2008-03294 y TRA2009_0030), Departamento de Salud (Gobierno de Navarra), Redes Temñticas de Investigación Cooperativa (RD06/0020/0065), Fondo de Investigación Sanitaria (PI060932), SUDOE (IMMUNONET) and UTE Project CIMA. S.H.-S. was supported by AECC and by MCI (RYC-2007-00928). The authors thank Blood Transfusion Center of Navarra (Spain) and Paul Miller for editing.

Conflict of interest: Grant support and reagents from DIGNA-Biotech (Madrid, Spain). I.G., U.M. and J.R. are full time employees of DIGNA-Biotech.

Ancillary