Interferon at 50: New Molecules, New Potential, New (and Old) Questions

Science's STKE  25 Sep 2007:
Vol. 2007, Issue 405, pp. pe53
DOI: 10.1126/stke.4052007pe53


Type I interferons (IFNs) are a family of cytokines defined by their antiviral activity but with a broad spectrum of biological activities, including antiproliferative, antitumor, and immunomodulatory effects. Mirroring these activities are diverse therapeutic applications to viral infections, antitumor therapy, and multiple sclerosis. The type I IFNs all signal through a common heterodimeric receptor. The existence of such a large family of cytokines (17 human IFNs) activating a common receptor is unusual. Moreover, the IFNs vary in their relative potency in different assays and are not functionally equivalent. How this functional variation is mediated through a common receptor has not been understood. Reports have now highlighted the interaction of IFNs with the low-affinity receptor subunit IFNAR-1 as a surprising key to their differential activity, particularly regarding antiproliferative and antitumor activities. Two groups have used contrasting approaches to produce variant IFN-α proteins with novel activity profiles. These advances portend enhanced therapeutic possibilities based on the better understanding of IFN-receptor interactions, while raising interesting mechanistic questions.

Two papers report new and improved type I interferons (IFNs) with novel activity profiles that may translate into better therapeutics (1, 2). These papers illustrate the power of both a rational, targeted approach to influencing IFN-activity relationships and a "mutate and screen" approach. They also highlight substantial progress in understanding how closely related type I IFNs can produce dramatically different biological effects, despite signaling through a common heterodimeric type I IFN receptor.

IFN was described 50 years ago by Isaacs and Lindenmann as a substance secreted by virus-treated cells that can elicit viral resistance (3). Mammalian IFNs are now classified into three types (4). Type I IFNs are encoded by a multigene family consisting in humans of 13 closely related IFN-αs and single copies of the more divergent IFN-β, IFN-ω, IFN-κ, and IFN-ε. The evolutionary origin and physiological importance of the large number of type I IFNs are unknown. However, these proteins signal through the heterodimeric type I IFN receptor, IFNAR, triggering the JAK (Janus kinase)-STAT (signal transducer and activator of transcription) pathway, with particular usage of Jak1 and Tyk2 to activate Stat1 and Stat2 (5, 6). Type II IFN is IFN-γ, the antiviral activity of which is secondary to its immunomodulating functions; it signals through a receptor composed of IFNGR-1 and IFNGR-2. Type III IFNs [IFN-λs; also known as interleukin-28 (IL-28) and IL-29] thus far have effects similar to those of type I IFNs, but signal through a unique receptor composed of IFNLR-1 (CRF2-12) and IL-10R2 (CRF2-4) (4).

In addition to their broad-spectrum antiviral activity, type I IFNs produce antiproliferative effects, pro-apoptotic activity (at least in virus-infected cells), and some antitumor effects. Produced in response to viruses or other inducers by most nucleated cells and by plasmacytoid dendritic cells (7), type I IFNs act on numerous cell types and help coordinate the responses of early, innate immunity with the later responses of acquired immunity mediated by B and T cells (8, 9). The broad cellular target range and spectrum of type I IFN activities underlie their therapeutic use as antiviral and antitumor agents, but also elicit side effects that can limit clinical use. Can desired IFN potencies be increased? Can the spectrum of activities be tailored for particular applications to decrease unwanted effects? The papers discussed here suggest a qualified "yes."

The relative potency of different type I IFNs can vary when tested in different cellular assays (10). Thus, when normalized for antiviral activity, IFN-β has a much higher antiproliferative activity than do the IFN-αs (1113). Similarly, two IFN-αs with high antiviral activity and receptor-binding affinity differ greatly in their activation of natural killer cells (14), and there are other examples (15). If all type I IFNs act through the common receptor IFNAR, how can they exert effects of such different relative potencies? Of practical importance, can we design IFNs to have high therapeutic activities and few undesirable effects? The answer should lie in the interaction of IFN with its receptor, but specific models were lacking until recently.

The type I IFN receptor, IFNAR, is a heterodimer composed of IFNAR-1 and IFNAR-2, both members of the class 2 cytokine receptor family (16) (Fig. 1). IFNAR-2 binds to type I IFNs with different affinities, mostly in the nanomolar range (17, 18), and its low-affinity partner, IFNAR-1, which binds to IFNs with micromolar affinities, also differentially recognizes the type I IFNs (18). IFNs bind initially to IFNAR-2, and this complex then recruits IFNAR-1 (19, 20). Because the biological and biochemical activities of IFNs do not correlate neatly with their affinities for the cellular receptor, the studies discussed here have dissected the interactions of native and mutant IFNs with the individual receptor subunits. These studies combine measurement of the binding rates and affinities of IFNs for the soluble ligand-binding domains of IFNAR-1 and IFNAR-2; comparison of diverse native type I IFNs; extensive mutagenesis of the commonly used IFN-α2; measurement of biological and biochemical activities of the IFNs; and hypothesis-testing by generation of novel IFN mutants that are subjected to the same battery of measurements.

Fig. 1.

Schematic of the formation of the type I IFN-receptor complex and its activation of different pathways. The stepwise formation of the ternary complex is depicted, followed by the relatively rapid activation of responses that require only low-affinity or short-lived association of IFNAR-1. IFNs that have high affinity for IFNAR-1 form longer-lived ternary complexes that are associated with additional biochemical and cellular responses. These slower responses often require sustained exposure to IFNs in vitro. The time line requires experimental validation. [Schematic inspired by figures in (32)]

The extracellular ligand-binding domain of IFNAR-2, by virtue of its smaller size and higher affinity for IFNs than IFNAR-1, is a more tractable target for mutagenesis and structural studies. Schreiber and colleagues assembled a detailed functional map of the amino acid residues of both IFNAR-2 and IFN-α2 at their binding interface (21, 22) [see also (23)]. This was complemented by a structural map derived from nuclear magnetic resonance studies (2426). The affinity of IFNAR-2 for native type I IFNs and IFN mutants correlated well with cellular antiviral activity, which is a sensitive and rapidly initiated response, and with activation of the transcription factor interferon-α–stimulated gene factor 3 (ISGF-3), suggesting that these activities are dominated by the interaction between IFNAR-2 and IFN. However, measurements of the binding of IFN to IFNAR-2 did not correlate well with antiproliferative activity (18, 27).

The critical insight was that the strength of IFN binding to the low-affinity subunit IFNAR-1 determines, to a first approximation, the relative activity of type I IFNs in antiproliferative and other important activities (Fig. 1). This observation arose through mapping the binding site in IFN-α2 for IFNAR-1: Roisman et al. (27), in addition to finding alanine substitutions in IFN-α2 that decreased its affinity for IFNAR-1, unexpectedly found three neighboring conserved residues which, when individually substituted with alanine, increased the affinity of IFN-α2 for IFNAR-1. These substitutions increased antiproliferative activity more than they did antiviral activity, which suggested that the affinity of IFN-α2 for IFNAR-1 could be improved. Moreover, when these three alanine substitutions were combined into a triple mutant (residues H57, E58, and Q61; HEQ), the resulting variant IFN-α [IFN-α2(HEQ)] showed greatly increased affinity for IFNAR-1 compared to IFN-α2 (28), making it closer to IFN-β in its affinity for IFNAR-1. This increased affinity for IFNAR-1 resulted in a greatly enhanced antiproliferative activity compared to IFN-α2, and an overall pattern of IFN-stimulated gene expression more similar to that of IFN-β than to that of IFN-α. In contrast, the antiviral activity of IFN-α2(HEQ) and its ability to stimulate a reporter gene controlled by a Stat-1–dependent promoter resembled that of the unmodified IFN-α2. These increased activities of IFN-α2(HEQ) and IFN-β also correlated with an increased half-life of the ternary complex of IFNAR-2, IFN, and IFNAR-1 relative to that mediated by IFN-α2. Similarly, Piehler’s laboratory observed that the affinity of native type I IFNs for soluble IFNAR-2 correlates with antiviral activity and activation of ISGF-3, but not with antiproliferative activity (18). However, they observed deviations from the correlation between affinity for IFNAR-1 and antiproliferative activity, and suggested that these deviations result from an additional effect of the relative affinity for IFNAR-2 (18). Thus, although the trends are now clear, a quantitative treatment of the dependence of different activities of IFNs on their affinities for the two receptor subunits remains. It is also evident that the abundance of the receptor, particularly IFNAR-1, on the cell surface markedly affects the sensitivity of cells, especially with regard to their antiproliferative response and their global transcriptional response.

Building on these results, Schreiber and colleagues used phage display to produce an IFN-α2 variant optimized for increased affinity to IFNAR-1 (1). Binding of a triple mutant of IFN-α2 [H57Y/E58N/Q61S (YNS)] to IFNAR-1 was much stronger than that of IFN-α2, which also resulted in increased binding of YNS to the native cellular receptor compared to that of IFN-α2. The variant had a greatly increased antiproliferative activity, with only a slight increase in antiviral activity compared to IFN-α2. After demonstrating greatly enhanced in vitro pro-apoptotic activity, the authors showed that the YNS mutant exhibited dramatic antitumor activity against human tumor cells transplanted into mice, as did IFN-β. Thus, pro-apoptotic and antitumor activities, as with antiproliferative activity, are highly dependent on the affinity of IFN for IFNAR-1, and are therefore amenable to manipulation.

IFN-αs with novel and potentially useful activity profiles were also derived by Brideau-Andersen et al. (2). Eschewing design, they combined gene shuffling of sections of different human IFN-α genes to generate novel IFN derivatives, using high-throughput screening to select novel IFN-αs that had high antiviral activity. The authors hypothesized that improved treatment of viral infections might be facilitated by IFNs that have high antiviral and T helper 1 (TH1)–inducing activity, but low antiproliferative activity, which might cause dose-limiting effects such as neutropenia or thrombopenia. Their procedure produced IFN-α hybrids that exhibited greatly increased antiviral and TH1-inducing activities and higher binding affinities for IFNAR, but decreased antiproliferative activities, compared to IFN-α2. These IFN-α hybrids had properties that were almost the opposite of those of the HEQ and YNS mutants. Brideau-Andersen et al. correlated a novel four–amino acid motif in the IFNAR-1–binding site with the high antiviral:antiproliferative ratios that were observed. They suggested that the ternary complex formed with the IFN-α hybrid may signal differently from that formed by IFN-α2. However, it is hoped that this will be better understood if the affinities of these IFNs for soluble IFNAR-1 and IFNAR-2 are carefully measured, whereupon these mutants can also contribute to testing and refining the emerging model of the roles of IFNAR-1 and IFNAR-2 in differential biological responses.

Our knowledge of the interaction between IFN and IFNAR-1 still has large gaps. The structures of the extracellular ligand-binding domain of IFNAR-1, both alone and in complex with different IFNs, are not yet known. Determining these structures and the structure of the ternary complex formed by IFN, IFNAR-1, and IFNAR-2 may be facilitated by the use of type I IFN variants that have a higher affinity than IFN-α2 for IFNAR-1.

How does differential affinity translate into differential cellular activation? The affinity of IFNs for IFNAR-1 generally correlates with the stability (half-life) of the ternary complex composed of IFN, IFNAR-1, and IFNAR-2 (18). Thus, the affinity differences may be surrogates for differences in the stability of active complexes. If so, how do short-lived and long-lived ligand-receptor complexes trigger different pathways? Is it through stronger or differential recruitment of intracellular signaling molecules, production of higher concentrations or different fluxes of intracellular molecules, or some combination of these? What role do differential ligand-induced receptor internalization, receptor down-regulation, and endocytic trafficking of these complexes play (29, 30)? Is there differential mobilization of IFN signal–damping molecules, such as the suppressors of cytokine signaling (SOCS), protein inhibitors of activated Stats (PIAS), and Src-homology 2 domain (SH2)–containing phosphatases (SHPs) (31)? More work is also needed to examine other in vitro and in vivo activities for their response to these novel IFNs and for their adherence to these patterns of differential activation. Despite such questions, these novel IFNs show promise for the development of improved IFN therapeutics.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
View Abstract

Navigate This Article