IFN-α Subtypes Differentially Affect Human T Cell Motility¹

The type I IFNs are a family of closely related proteins and glycoproteins that are produced in response to viral infections (1, 2). The type I IFN family consists of 12 different IFN-α subtypes (3) in addition to an IFN-β (3) and an IFN-ω subtype (4). All type I IFNs have similar amino acid sequences and bind to a common receptor (5). They are all antiviral cytokines, but their antiviral potencies differ (6), and the reason for the persistence of so many similar cytokines has been the subject of debate for many years (7).

It has long been recognized that the type I IFNs play a role in the development of the immune response during viral infections (8), and over the last few years it has become clear that type I IFNs also play a role in the initiation of an immune response in the absence of viral infections (9). Thus, the interaction between an activated T cell and a dendritic cell leads to the release of IFN-α (10), and, in humans, IFN-α is critical for the further maturation of Th1 T cells as it up-regulates cell surface expression of the IL-12R via activation of STAT4 (11). It is noteworthy that the largest differences between type I IFNs are seen in immunological assay systems, perhaps suggesting that all of the IFNs have similar antiviral properties, but that different subtypes may have different immunomodulatory roles (reviewed in Ref.12).

The key role of chemokinetic factors in the development of immune responses is well established: the development of immune-mediated inflammation requires the recruitment and activation of lymphocytes at antigenic sites (13), and polarization of T lymphocytes is needed for the T cells to focus on the APCs and establish adhesive interactions that allow TCR triggering (14). Both of these events are mediated by chemotactic cytokines, the best characterized of which are the chemokines. Other mediators produced at the site of inflammation may also play a role in chemokinesis, although this has not yet been established. In this context, the influence of type I IFNs on T cell motility remains to be clarified. Previous studies focusing on the effects of IFN-β have generated conflicting results. High doses of IFN-β (1 μg/ml) were shown to reduce T cell production of the chemokinetic cytokines RANTES and MIP-1α and to reduce lymphocyte chemokinesis (15), but Cremer et al. (16, 17) found that IFN-β induced the production of RANTES in a variety of cell lines. However, the expression of the chemokine receptor CCR5 was down-regulated by IFN-β, and the net effects of IFN-β on T cell migration were not examined. To date, direct effects of type 1 IFNs on T cell motility have not been reported.

To investigate the role of type I IFNs on recruitment of lymphocytes, we have compared the effects on T cell chemokinesis and transendothelial migration of two IFN-α subtypes (IFN-α2 and IFN-α8). These two subtypes are potent antiviral subtypes, although IFN-α8 is a little more potent than IFN-α2 with an ED₅₀ of 2 pg/ml compared with 9.7 pg/ml for IFN-α2 in fibroblast-derived cell lines (18). We find that IFN-α2 induces chemokinesis of both CD4⁺ and CD8⁺ T cells at various stages of differentiation, and IFN-α2 induces functional changes that result in enhanced T cell motility, including up-regulation of integrins. By contrast, IFN-α8, despite having similar antiviral properties, does not affect T cell motility. Although quantitative differences between the different type I IFNs have been reported previously, this is the first report of a qualitative difference between two closely related and equally potent antiviral IFN-α subtypes.

A number of different IFN-α preparations were used, and all had similar effects. Recombinant, pharmaceutical grade IFN-α2a (Roferon; Hoffmann-LaRoche, Basel, Switzerland; sp. act., 2 × 10⁸ IU/mg) was purchased; recombinant IFN-α2 (sp. act., 2.7 × 10⁸ IU/mg) and IFN-α8 (sp. act., 4.95 10⁸ IU/mg) were purchased from PBL Biosciences (Piscataway, NJ); bacterial recombinant IFN-α2 and IFN-α8 were prepared from Escherichia coli; and mammalian cell-derived IFN-α2 and IFN-α8 were prepared as described (19) and were a kind gift from P. Patten (Maxygen Pharmaceuticals, Santa Clara, CA). The human chemokines CXCL-12 (stromal-derived factor-1α) and CCL5 (RANTES) (PeproTech, Peterborough, U.K.) were used at a concentration of 50 and 100 ng/ml, respectively, in chemokinesis assays (see below).

The following mAbs were used to purify peripheral blood T cell subsets: anti-CD56 (Leu-19; BD Biosciences, San Jose, CA), mouse anti-human Ig (Fab specific; Sigma-Aldrich, Poole, Dorset, U.K.), anti-HLA-DRα (L243; American Type Culture Collection (ATCC), Manassas, VA), anti-CD8 (OKT8; ATCC), anti-CD4 (OKT4; ATCC), anti-CD45RO (UCHL1; ATCC) (20), and anti-CD45RA (SN 130; gift of G. Janossy, Royal Free Hospital, London, U.K.) (21). For phenotypic analysis, the following Abs were used: anti-CD11a (LFA-1, clone 38; Serotec, Kidlington, U.K.) and anti-CD49d (VLA-4, clone HP2.1; Serotec). mAbs directed against the two components of the IFN receptor (IFNAR1 and IFNAR2) have been previously described (22) and were used at a concentration of 5 μg/ml.

Endothelial cells (HUVEC) were isolated from human umbilical cord veins and cultured, as previously described (23). The HSB-2 CD4/CD8 double-negative human thymocyte line was a kind gift of D. Palmer (Royal Veterinary College, London, U.K.).

PBMC were obtained by Ficoll-Hypaque (Pharmacia Biotech, Uppsala, Sweden) centrifugation of heparinized blood, washed twice, and resuspended in RPMI 1640 medium supplemented with 10% FCS, 2 mM glutamine, 50 IU/ml penicillin, and 50 μg/ml streptomycin. The cell preparation was then depleted of adherent cells by two 45-min rounds of adherence to plastic on tissue culture dishes at 37°C. The nonadherent cells were subsequently collected and incubated with a mixture of purified mAbs at saturating concentrations for 30 min at 4°C. The cells were then washed twice to remove excess Ab and further enriched by magnetic immunodepletion. Briefly, mAb-treated cells were incubated with magnetic microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) coated with sheep anti-mouse Ig for 15 min at 4°C, and bead/mAb-coated cells were removed by passage through a magnetic column (miniMAC system; Miltenyi Biotec).

An alloreactive CD4⁺ T cell line was generated to represent activated T cells by stimulation of purified peripheral blood CD4⁺ T cells with HLA-DR-mismatched irradiated (60 Gy) PBMC. The line was maintained in culture by weekly stimulation with allogeneic PBMC, rIL-2 (10 U/ml; Roche, Mannheim, Germany) in RPMI 1640 medium supplemented with 10% human serum, 2 mM glutamine, 50 IU/ml penicillin, and 50 μg/ml streptomycin. For use in experiments, the T cells were purified by isolation on a Ficoll-Paque (Pharmacia Biotech) gradient 7 days after restimulation.

The transmigration experiments were conducted using HUVEC monolayers grown on Costar transwell tissue culture well inserts (diameter 12 mm) that contained polycarbonate membranes with a 3 μm pore size (Costar, High Wycombe, U.K.). A total of 5 × 10⁴ endothelial cells (EC)⁴ was seeded onto fibronectin-coated (50 μg/ml; Sigma-Aldrich, Poole, U.K.) polycarbonate membranes overnight to form a monolayer. In experiments assessing migration induced by integrin ligands, transwell inserts were incubated with either human fibronectin (10 μg/ml; Sigma-Aldrich) or human rICAM-1 (2 μg/ml, human ICAM-1 Fc chimera; R&D Systems, Abingdon, U.K.) in PBS for 2 h at 37°C, and then washed twice with PBS. T cells (5–7 × 10⁵/well) were added into each insert and left to migrate. The number of migrated T cells was determined by counting the cells present in the well medium at different time points over the next 24 h. Results are expressed as percentage of transmigrated cells in three independent counts.

For chemokinesis assays with CXCL-12, T cells were seeded (5–10 × 10⁵/well) in the upper chamber of a 3-μm-pore polycarbonate transwell (Costar). A 0.5 ml vol of the chemokinesis medium (RPMI 1640 2% FCS) containing CXCL-12 (50 ng/ml), or CCL5 (100 ng/ml), or control medium was added to the bottom chamber of the transwell, while 0.2 ml of cell suspension was added to the top chamber. Transwells were incubated for 6–24 h at 37°C with 5% CO₂. The number of migrated cells was evaluated, as described above.

A total of 10⁵ T cells was incubated with the indicated mAb at 4°C for 30 min. As a control, T cells were incubated with an isotype-matched irrelevant Ab. After washing twice in PBS with 2.5% FCS, the cells were incubated for an additional 30 min at 4°C with 100 μl of 1/50 dilution of fluoresceinated sheep anti-mouse Ig (Amersham, Amersham, U.K.). After two additional washes, stained cells were analyzed using a FACSCalibur (BD Biosciences) flow cytometer.

T cells (10⁶) were treated with 0.5 and 5 ng/ml of the appropriate IFN-α subtype, and total RNA was harvested after 6 h using TRIzol (Invitrogen Life Technologies, Paisley, U.K.). Extracted RNA was DNase I treated using RQ1 RNase-free DNase and reverse transcribed using Moloney murine leukemia virus reverse transcriptase (both from Promega, Madison, WI). The PCR was performed and analyzed on the Rotorgene (Corbett Research, Mortlake, New South Wales, Australia) using QuantiTect SYBR green (Qiagen, Crawley, U.K.). Primers were as follows: MxA forward, 5′-AACAACCTGTGCAGCCAGTA, and MxA reverse, 5′-AAGGGCAACTCCTGAGAGTG; IFN-responsive factor (IRF)7 forward, 5′-GAGCCCTTACCTCCCCTGTTAT, and IRF7 reverse, 5′-CCACTGCAGCCCCTCATAG. All reactions were normalized against the housekeeping gene RPL13A (RPL13a forward, 5′-CCTGGAGGAGAAGAGGAAAGAGARPL13A, and RPL13a reverse, 5′-TTGAGGACCTCTGTGTATTTGTCAA; Sigma-Aldrich). The 2′-5′-oligoadenylate synthetase (2′5′OAS) was amplified with 30 cycles: 90°C for 15 s, 58°C for 30 s, and 72°C for 30 s using Taq polymerase (Promega) with the following primers: 2′5′OAS forward, 5′-AACTGCTTCCGACAATCAAC, and reverse, 5′-CCTCCTTCTCCCTCCAAA.

The activity of IFN-α2 and IFN-α8 was further examined by treating the fibroblast-derived cell line HL116 (a kind gift from G. Uze, Institute Pasteur, Paris, France) with differing concentrations of type I IFN. This cell line is derived from human HT1080 cells and contains a luciferase cDNA under the control of the IFN-inducible promoter derived from the human 6-16 gene. Treatment of these cells with IFN readily induces production of luciferase. Cells were grown to confluence in 96-well plates before treatment with the appropriate concentration of IFN-α for 12 h, followed by analysis of luciferase activity, as described (24).

The ability of IFN-α2 and IFN-α8 to induce T cell migration was investigated by a transwell assay (see Materials and Methods). The effects of different concentrations of IFN-α2 and IFN-α8 on migration of a thymocyte line and recently activated alloreactive CD4⁺ T cells are shown in Fig. 1. The chemokinetic effect was measured at 24 h, as T cell migration was not measurable at earlier time points for the thymocyte line. However, migration was observed as early as 2 h following the exposure to IFN-α2 in the highly mobile T cell line (see below). A dose-dependent increase in migration of the cells was observed following addition of IFN-α2 (Fig. 1, A and B), similar to that induced by 50 ng/ml CXCL-12 (optimal concentration determined from preliminary experiments; data not shown). The dose-response curve was bell shaped, with maximum migration induced by concentrations between 1 and 5 ng/ml IFN-α2 (∼1000 IU/ml). By contrast, IFN-α8 did not enhance T cell motility (Fig. 1, C and D, shows the effects of IFN-α8 from three separate experiments). To exclude the possibility that these differences were due to a different potency of IFN-α2 and IFN-α8, the ability of IFN-α8 to induce chemokinesis at a dose range between 0.01 and 1 ng/ml was assessed, but no significant effect on T cell migration was observed (Fig. 1, C and D). Note that the background (unstimulated) migration of the HSB-2 cells was highly variable (ranging from <5% to greater than 10%) and probably dependent upon the passage number and previous culture conditions of the cells. All experiments with a particular stimulant were therefore performed with the same batch of cells, and parallel experiments with IFN-α2 and IFN-α8 using the same cells and identical doses of IFN (0.5–50 ng/ml) showed chemokinetic activity with IFN-α2, but not IFN-α8 (data not shown).

To determine whether the differential effect of IFN-α2 and IFN-α8 on T cell motility was due to a global modification in T cell responses to IFN-α subtypes, we examined the induction of a number of different IFN-α-inducible genes, MxA, IRF7, and 2′5′OAS, in an alloreactive CD4⁺ T cell line. MxA was induced by both IFN-α2 and IFN-α8 (Fig. 2, A), but IRF7 and 2′5′OAS were induced by IFN-α2, but not by IFN-α8 (Fig. 2, B and C, respectively). Hence, lack of a chemokinetic effect by IFN-α8 is not due to a global down-regulation of T cell responsiveness to IFN-α8, as MxA is induced by this subtype; rather, in human T cells, there are differences in the response to two closely related IFN-α subtypes.

The chemokinetic properties of IFN-α subtypes were analyzed on distinct T cell subpopulation at various stages of differentiation, including CD45R0- and CD45RA-expressing CD4 and CD8 T cells, recently activated alloreactive T cells, and a double-negative thymocyte line, HSB-2. We had previously established by flow cytometry that all of these T cell populations expressed low levels of IFN-α receptor (data not shown). Migration was compared with optimal induction by a chemokine (CXCL12 for CD45RA-expressing T cells, recently activated T cells and thymocytes, and CCL5 for CD45RO-positive T cells) (25). As shown in Fig. 3, all T cell subsets tested responded to the presence of IFN-α2 in the lower chamber with enhanced T cell migration, whereas IFN-α8 had no effect. A dose response was determined for all of the T cell subsets, and a similar range of optimal doses for all of the populations analyzed was observed (data not shown).

To establish whether IFN-α2 induces directed T cell migration (chemotaxis) or random T cell motility (chemokinesis), T cells were exposed to a CXCL12 chemotactic gradient in the presence of either CXCL12 or IFN-α2 in the upper chamber in a short-term chemotaxis assay. As shown in Fig. 4,A, the presence of CXCL12 in the upper chamber temporarily (possibly due to a loss of gradient with time), but significantly, prevented T cell migration. In contrast, the addition of IFN-α2 enhanced (albeit not significantly) CXCL12-induced chemotaxis (Fig. 4 B). These data suggest that the IFN-α2-induced T cell migration is due to a chemokinetic, rather than chemotactic activity.

To confirm that the chemokinetic activity induced by IFN-α2 was due to binding to the classical IFN receptor, experiments were performed in which an alloreactive CD4⁺ T cell line, responsive to both IFN-α2 and the chemokine CXCL12, was treated with neutralizing mAbs directed against the two components of the heterodimeric IFN-α receptor (22). T cells were incubated with 5 μg/ml of either Ab EA12 (directed against IFNAR1) or Ab A9 (directed against IFNAR2) for 30 min, and then exposed to an optimal chemokinetic concentration of IFN-α2 (5 ng/ml). Interestingly, only the A9 Ab completely inhibited the chemokinetic effects of IFN-α2, while Ab EA12 had a minimal effect on IFN-induced chemokinesis (Fig. 5, A–D). Neither Ab affected CXCL12-driven migration (data not shown). These results show that the interaction between IFN-α2 and IFNAR2 is essential for IFN-induced chemokinesis, but because a mAb against IFNAR1 only partially inhibited chemokinesis, either this receptor component is not involved in the chemokinetic response to IFN-α2 or the domain on IFNAR1 that mediates chemokinesis is not completely blocked by binding of Ab EA12. Because previous studies have suggested that different regions within IFNAR1 bind to different IFN-α subtypes (26), we investigated the possibility that IFN-α8 might act as a competitor of IFN-α2-mediated chemokinesis. As shown in Fig. 5 E, IFN-α2-mediated T cell migration was inhibited by the addition, in the upper chamber of the transwell, of nonchemokinetic IFN-α8, suggesting that IFN-α2 and IFN-α8 bind to adjacent or overlapping sites on the IFNAR1. Further studies will be required to determine the role played by IFNAR1 in mediating IFN-α2-induced chemokinesis.

Previous studies from our laboratory using these mAbs have shown that in an antiviral assay, the Ab directed against IFNAR1 reduces the antiviral activity of IFN-α2 by 25-fold, while the Ab against IFNAR2 reduces activity >500-fold (18). To confirm that both the anti-IFNAR Abs remained functional in other assays, we used the IFN assay system developed by Uze et al. (24), in which luciferase production is regulated by an IFN-α-inducible promoter. As shown in Fig. 6, when cells containing IFN-α-inducible luciferase were treated with IFN-α2 or IFN-α8, expression of luciferase increased in response to both subtypes, although IFN-α8 was slightly more potent (Fig. 6,A). Addition of 5 μg/ml of each Ab to HL116 cells treated with 0.5 ng/ml IFN-α led to inhibition of the IFN-inducible luciferase (Fig. 6 B), although, as previously reported by us (18) and others (27), inhibition was less marked with the monoclonal directed against IFNAR1.

Exposure to IFN-α2 enhances T cell mobility, but in vivo recruitment of T cells to an area of inflammation requires migration of the T cells across an endothelial barrier. To determine whether T cell migration through an EC monolayer was modified by exposure to IFN-α, we studied the migration of activated alloreactive CD4⁺ T cells and thymocytes incubated for 24 h with increasing doses of IFN-α2 or CXCL12 (50 ng/ml) across an EC monolayer. As it is shown in Fig. 7, following incubation with IFN-α2 (5 ng/ml), T cell migration through an EC monolayer was enhanced in a dose-dependent manner, and the effect was similar to that elicited by culture with the chemokine CXCL12. In contrast, T cells cultured with antiviral doses of IFN-α8 (0.5–5 ng/ml) did not display enhanced transendothelial migration (data not shown).

Chemokines can indirectly enhance T cell migration by increasing the expression of adhesion molecules (28). To determine whether IFN-α2-induced enhancement of T cell transendothelial migration correlated with up-regulation of adhesion molecules, we studied the expression of the β₂ and β₁ integrins LFA-1 and VLA-4 on T cells exposed to IFN-α2 and IFN-α8. Expression of these molecules on alloreactive CD4⁺ T cell line (Fig. 8) following 24-h incubation in the presence of either IFN-α2 (5 ng/ml), IFN-α8 (5 ng/ml), or CXCL12 (50 ng/ml) was assessed by flow cytometry. As a control, T cells were cultured in medium alone. As shown in Fig. 8, both IFN-α 2 (C and D) and CXCL12 (G and H) induced up-regulation of both LFA-1 and VLA-4 integrins. By contrast, expression of these molecules was not modified by exposure to IFN-α8 (E and F) or culture in medium alone (A and B). The functional significance of IFN-α2-induced up-regulation of these molecules was assessed by comparing T cell migration through fibronectin- or rICAM-1-coated transwells. As shown in Fig. 8, I and J, exposure to IFN-α2 (and to CXCL-12), but not to IFN-α8, significantly increased T cell migration mediated by both these integrin ligands.

The type I IFNs play a central role in the resolution of viral infections. Thus, mice that lack the type I IFN receptor die rapidly when exposed to minute viral doses (29). The antiviral effects of the type I IFNs are mediated to a large degree by their induction of an antiviral state in responsive cells: binding of IFN to its cell surface receptor leads to up-regulation of a large number of genes (IFN-stimulable genes) that have antiviral properties (reviewed in Ref.2) and that inhibit viral replication. In addition to the induction of an antiviral state, the type I IFNs facilitate the generation of an antiviral immune response by enhancing T and B cell activation (9). It is now becoming clear that even in the absence of a viral infection, type I IFNs are produced during the development of an immunological response and play a role in regulating it. Thus, dendritic cells that have been exposed to mycobacteria produce type I IFNs (30), and the interaction between dendritic cells and activated T cells leads to the production of IFN-α (10).

A key event in the development of an immune response is the recruitment and activation of T cells at the site of priming and inflammation. The initial stages of T cell activation are mediated by Ag-independent interactions, which establish areas of focal contact between T cells and APC (31). Such interactions are initiated by chemoattractant-induced cell polarization and subsequent redistribution of adhesion molecules on the T cell surface. These, in turn, allow TCR engagement by MHC:peptide complexes displayed on the APC and lead to the formation of highly organized structures, immunological synapses, which efficiently transduce Ag-initiated signals (14). As type I IFNs are early mediators generated during infective and noninfective immune inflammation, we assessed whether two members of the IFN-α family, α2 and α8, which have similar antiviral potency as well as therapeutic applications, had any influence on T cell motility and migration. Our results show that IFN-α2 can induce T cell chemokinesis and favor T cell migration by enhancing T cell responsiveness to chemokines (see Fig. 4) and adhesion molecule expression and function (see Fig. 7). Interestingly, all of these effects were mirrored by those elicited by the CXCL12 chemokine. Unlike chemokines, however, IFN-α2 could induce T cell chemokinesis irrespective of the subset or differentiation stage, but did not drive directional migration (chemotaxis).

To exclude the possibility that the chemokinetic effect of IFN-α2 was due to the presence of a contaminant, a number of different preparations of IFN-α subtypes, including in-house recombinant subtypes as well as pharmaceutical grade IFN-α2, commercially available laboratory preparations, and subtypes prepared from transfected mammalian (CHO) cells, were used with similar results. In addition, to rule out the possibility that small variations in potency were responsible for the differences seen in our chemokinesis experiments, we examined the different subtypes over a wide concentration range (0.01–50 ng/ml). In general, different IFN-α subtypes have similar effects on the induction of classical IFN-stimulated genes (32), although subtle differences between IFN-α and IFN-β have been seen in human endothelium and fibroblasts (32, 33), and differences between IFN-α1 and IFN-α2 have been reported in human T cells (34). In our human T cell lines, we found a clear distinction between the response to IFN-α2 and IFN-α8: some genes did not appear to be induced by IFN-α8, but were induced by IFN-α2. In line with previous studies (32, 33), our results suggest that two signal transduction pathways are operational in these T cells: one responds to both IFN-α2 and IFN-α8 and mediates induction of MxA and, possibly, other genes, while a second pathway is IFN-α2 specific and mediates chemotaxis as well as the induction of 2′5′OAS and IRF7. Further work will be required to define the characteristics of these signaling pathways in more detail, and studies are in hand to delineate the full array of genes that are differentially regulated by these two IFN-α subtypes.

We were surprised to find that the induction of MxA in human T cells in response to IFN-α followed a bell-shaped dose-response curve. Although such responses are seen with many cytokines, this phenomenon has not been previously observed with type I IFNs: previous studies on the induction of IFN-inducible genes in fibroblast-derived cell lines (32) have shown that the response to differing concentrations of IFN-α is linear. It is not clear why T cells respond to type I IFNs in this way; it may be related to the known antiproliferative effects of IFN-α on human T cells, and it is possible that during IFN-α-induced growth arrest, MxA expression is reduced relative to RPL13A, leading to the observed reduction in the MxA:RPL13A ratio. Alternatively, this may reflect an unusual IFN-α dose-response relationship that is unique to human T cells.

All of the type I IFNs bind to a common receptor that consists of two components: IFNAR1 and IFNAR2. The full details of the interaction between different IFN-α subtypes and the different receptor components have not yet been established, but it is clear that different type I IFNs interact in different ways and engage with different sites on the dimeric receptor. Thus, mutated IFNAR2 proteins that bind to IFN-β, but not IFN-α, have been identified (35), and mAbs directed against IFNAR1 have different effects on the antiviral activity of different IFN-α subtypes (27). It thus seems likely that different IFN-α subtypes interact in a different way with the common IFN receptor, and the observation that mAbs against IFNAR1 have only a minimal effect on the chemokinetic effects of IFN-α2 suggests that this IFN-α subtype engages with an IFNAR1 motif that is not bound by our mAb. This is supported by the observation that both Abs can inhibit reporter luciferase activity induced by both IFN-α2 and IFN-α8.

The molecular mechanism by which IFN-α enhances T cell motility is not yet clear, and it may involve both direct and indirect mechanisms. For example, type I IFNs can induce cell membrane and cytoskeletal changes (36), suggesting a direct effect, but other studies have reported changes in the expression of chemokines (17), suggesting that IFN-α2 may have an indirect effect. Our experiments suggest both a direct chemokinetic effect as well as enhanced T cell motility by up-regulation of integrins. Interestingly, all of these effects are mirrored by those induced by the chemokine CXCL12, suggesting that IFN-α2 and some chemokines share some signaling events. Further work will be required to address this possibility.

Early studies on the differential effects of different IFN-α subtypes have found that the greatest differences between the subtypes are seen when immunological functions are assessed; for example, there are marked differences in the effects of different IFN-α subtypes on NK cell function (37). However, these differential effects were due to quantitative differences; at high concentrations, all subtypes have similar effects. The data presented in this work are the first example of two closely related IFN-α subtypes with differential effects on both transcriptional events and function in human T lymphocytes. The physiological significance of this difference is not yet clear, but because different infections can induce different IFN-α subtypes (27, 38), it is possible that differential production of different IFN-α subtypes leads to a qualitatively different inflammatory infiltrate, thus influencing the overall phenotype of the developing immune response. In vivo models will be required to address this possibility.

We are grateful to Dr. D. Tough (Jenner Institute, Compton, U.K.) for critical review of the manuscript.