CD28, ICOS, and 4-1BB each play distinct roles in the CD8 T cell response to influenza virus. CD28−/− mice are severely impaired in primary CD8 T cell expansion and fail to mount a secondary response to influenza. Influenza-specific CD8 T cells expand normally in ICOS−/− mice, with only a small and transient defect late in the primary response and an unimpaired secondary response. Conversely, 4-1BB/4-1BBL interaction is dispensable for the primary CD8 T cell response to influenza, but maintains CD8 T cell survival and controls the size of the secondary response. Previous results showed that a single dose of agonistic anti-4-1BB Ab at priming allowed partial restoration of primary CD8 T cell expansion and full recovery of the secondary CD8 T cell responses to influenza in CD28−/− mice. In this study we show that anti-4-1BB fails to correct the CD8 T cell defect in CD28−/−ICOS−/− mice, suggesting that ICOS partially compensates for CD28 in this model. In support of this hypothesis, we found that anti-4-1BB enhances ICOS expression on both T cell subsets and that anti-4-1BB and anti-ICOS can synergistically activate CD4 and CD8 T cells. Furthermore, ICOS and 4-1BB can cooperate to directly stimulate isolated CD28−/− CD8 T cells. These results reveal a novel interaction between the ICOS and 4-1BB costimulatory pathways as well as unexpected redundancy between CD28 and ICOS in primary CD8 T cell expansion. These findings have implications for costimulation of human T cell responses in diseases such as AIDS or rheumatoid arthritis, in which CD28− T cells accumulate.
Activation of T cells requires signals through the Ag-specific TCR as well as through costimulatory molecules. CD28 is constitutively expressed on naive T cells and plays a key role in the primary expansion and survival of T cells (1). In addition to CD28, it has become increasingly apparent that a series of costimulatory molecules from both the CD28 and TNFR families can enhance T cell activation and survival at different stages of the immune response (2, 3, 4).
ICOS is a member of the CD28/CTLA-4 family (5). It is expressed on activated T cells and binds to ICOS ligand, which is expressed on APC, as well as on multiple nonlymphoid cells (6, 7, 8, 9, 10). ICOS is important for Th responses, germinal center responses, and Ab class switch (11, 12, 13, 14). However, unlike CD28, ICOS induces only minimal amounts of IL-2 (15). Rather, ICOS functions to enhance T cell effector function in a variety of experimental settings (2, 16, 17). Although ICOS has been primarily studied as a costimulator of CD4 T cell responses, there is also evidence for a role in CD8 T cell responses (18, 19, 20, 21).
4-1BB is an inducible TNFR family member found on activated T cells (22). It is also constitutively expressed on dendritic cells (23, 24). Its ligand, 4-1BBL, is found on activated APC (23, 25, 26). In contrast to ICOS, much of the in vivo analysis of 4-1BB/4-1BBL suggests a more prominent role on CD8 T cells (27, 28, 29, 30), although direct costimulatory effects on CD4 T cells have also been demonstrated (31, 32, 33, 34).
Influenza virus infection of mice has been a powerful model to study the role of costimulatory molecules in the immune system and has revealed distinct roles for several costimulatory molecules (35). CD28 is critical for primary CD4 and CD8 T cell responses to influenza virus as well as for Ab class switch (27, 30, 36). Ab class switch and CD4 cytokine production in response to influenza are also impaired in ICOS-deficient mice (37). In contrast, ICOS is expendable for primary CD8 T cell expansion, showing only a transient decrease in CD8 T cell numbers late in the primary response, with full recovery of the CD8 response upon secondary challenge (37). Conversely, 4-1BBL is dispensable for primary CD8 T cell expansion as well as for CD4 or Ab responses to influenza virus. Rather, 4-1BBL-deficient mice show defects in maintenance of CD8 memory T cells late in the primary response and an impaired secondary CD8 T cell response (30).
4-1BB/4-1BBL can provide a costimulatory signal to resting T cells from CD28−/− mice, allowing them to make IL-2 and survive (26, 38). Furthermore, delivery of anti-4-1BB to CD28−/− mice can restore defective primary CD8 T cell expansion in response to influenza virus (39). Even a single dose of anti-4-1BB delivered during priming with influenza virus led to partial recovery of primary CD8 T cell expansion and full recovery of the memory response (40). These data suggested that 4-1BB signals could replace CD28 signals for CD8 T cell priming, and that once T cells have expanded, CD28 was not required again in the secondary response.
Recently, mice deficient in both CD28 and ICOS have been generated by gene targeting (41). Mice lacking both molecules have a more substantial reduction in the Ab response to viruses and exhibit greater defects in T cell proliferation and Th2 cytokine production than observed in the absence of either molecule alone (41). Based on these findings, it was possible that the ability of anti-4-1BB to correct the CD8 T cell response defect in CD28−/− mice was due in part to compensatory effects of ICOS.
In this paper we address the issue of redundancy between CD28 and ICOS functions in the CD4 and CD8 T cell response to influenza virus. We show that anti-4-1BB-induced restoration of the response to influenza in CD28−/− mice depends on the presence of ICOS, suggesting that ICOS partially compensates for CD28 in this model. Further, we show that anti-41BB augments anti-CD3-induced ICOS expression on CD4 and CD8 T cells and that the two molecules cooperate in the direct stimulation of CD8 T cells in the absence of CD28 or CD4 T cell help.
Materials and Methods
CD28−/− (42), ICOS−/− (11), CD28−/−ICOS−/− (DKO)4 mice (41) and littermate controls on the C57BL/6 background were maintained at the Ontario Cancer Institute Facility, University Health Network. For ICOS expression and in vitro costimulation studies, CD28−/− and ICOS−/− mice were also bred and housed at the Division of Comparative Medicine, University of Toronto, and additional wild-type (WT) controls were purchased from Charles River Laboratories. All experiments involving animals were approved by the institute animal care committee in accordance with the guidelines of the Canadian Council for Animal Care.
The anti-CD3-producing hybridoma 145-2C11 was provided by Dr. J. Bluestone (University of Chicago, Chicago, IL). The anti-CD28-secreting hybridoma 37.51 was provided by Dr. J. Allison (University of California, Berkeley, CA). The anti-ICOS Ab C398.4A (43) is available through Dr. U. Dianzani’s laboratory (Novara, Italy). Anti-4-1BB (M6, rat IgG2a) was produced at Immunex. The anti-4-1BB hybridoma 3H3 was provided by Dr. R. Mittler (Emory University, Atlanta, GA). Pilot experiments indicated that 3H3 and M6 behaved identically, so initial experiments used M6 Abs, and some replicates were performed using 3H3 with similar results. Anti-rat/anti-hamster Ab (RG7) and hamster IgG were purchased from Sigma-Aldrich.
Influenza infection of mice
Six- to 8-wk-old mice were infected i.p. with 200 hemagglutinin units (HAU) of influenza A HKx31 (H3N2) and with 100 μg of anti-4-1BB Ab (M6 or 3H3) or control rat IgG Ab (Sigma-Aldrich). The dose of anti-4-1BB was determined in pilot experiments to give the maximal effect in the CD28−/− mice. Influenza virus was produced as previously described (44). Mice were killed at the indicated time points, and their spleens were harvested for single-cell suspensions.
Spleen cell suspensions were prepared in PBS/2% FCS/0.01% sodium azide on ice. Cells were surface-stained with one or more of the following Abs: allophycocyanin-conjugated anti-mouse CD8, allophycocyanin-conjugated anti-mouse CD4, FITC-conjugated anti-mouse CD62L, PE-conjugated anti-mouse ICOS, PE-conjugated anti-mouse CD137, PE-labeled anti-mouse CD69 (eBioscience), and PE-labeled tetramers consisting of murine class I MHC molecule H-2Db, β2-microglobulin, and influenza nuclear protein (NP) peptide, NP366–374 (National Institute of Allergy and Infectious Diseases, MHC Tetramer Core Facility). For each experiment, appropriate isotype control mAbs were used.
Intracellular IFN-γ staining
Spleen cell suspensions were restimulated in culture medium (RPMI 1640/10% FCS with antibiotics and 2-ME) for 6 h at 37°C with 1 μM NP366–374 peptide and GolgiStop (BD Pharmingen). Cells were harvested, resuspended in PBS/2% FCS/azide, and surface-stained with PE-anti-CD8 and FITC-anti-CD62L as described above. After surface staining, cells were fixed in Cytofix/Cytoperm solution (BD Pharmingen) and then stained with allophycocyanin-conjugated anti-mouse IFN-γ diluted in 1× Perm/Wash solution (BD Pharmingen). Samples were analyzed using a FACSCalibur (BD Biosciences) and FlowJo software (TreeStar).
Spleen cells (5 × 106 cells) were incubated with 250 HAU/ml heat-killed (56°C, 30 min) influenza A HKx31 for 3–4 days to stimulate CD4 T cell responses (45). Supernatants were removed, and the levels of IFN-γ were measured by ELISA as described previously (30). We also used intracellular cytokine staining after 72-h stimulation (Golgi stop added in the last 6 h) to confirm that heat-killed virus results in IFN-γ production by CD4 T cells with little or no IFN-γ production by CD8 T cells (data not shown).
For T cell isolation, adherent cells were depleted by incubation at 37°C in polystyrene tissue culture plates. Total T cells or CD8 T cells from naive C57BL/6 and CD28−/− mice were isolated by depletion of unwanted subsets using magnetic beads (EasySep; StemCell Technologies).
In vitro stimulation of T cells to induce ICOS expression
Wells in 24-well plates were coated with combinations of the following Abs (as indicated in the figure legends): anti-rat/anti-hamster IgG (RG7; 10 μg/ml) at 37°C for 2 h, followed by washing and incubation with anti-CD3 (1 μg/ml), anti-4-1BB (10 μg/ml), anti-CD28 (10 μg/ml), and rat IgG control (10 μg/ml). After a 72-h incubation, the cells were analyzed by flow cytometry. They were stained with PE-labeled anti-mouse ICOS, FITC-labeled anti-mouse CD69, Cy5-PE-labeled anti-mouse CD8, and allophycocyanin-labeled anti-mouse CD4.
CFSE-based assay to measure ICOS and 4-1BB-mediated costimulation of cell division
T cells were stained with CFSE (Molecular Probes) as previously described (46). In brief, cells were resuspended in PBS at 5 × 107/ml. CFSE was added to the cell suspension at a final concentration of 2.5 μM and incubated for 10 min at 37°C. Cells were washed twice in PBS/10% FCS and recounted. A total of 1 × 106 T cells or CD4 and CD8 T cells (as indicated in the figure legends) were cultured in 24-well, flat-bottom plates (Falcon; BD Biosciences) that were previously coated with Abs as follows. The plates were first coated with anti-hamster/anti-rat IgG (RG7; 10 μg/ml) at 37°C for 2 h. The plates were then washed with PBS and bound with a mixture of following Abs (as indicated in the figure legends): anti-ICOS mAb (5 μg/ml), anti-4-1BB (5 μg/ml), anti-CD28 (5 μg/ml), hamster IgG (5 μg/ml), rat IgG (5 or 10 μg/ml to ensure same total Ab in each well), and anti-CD3 mAb (0.25 μg/ml) and incubated at 37°C for 2 h. Cells were incubated for 3 days and were analyzed by flow cytometry. Pilot studies were used to establish optimal CD3 and anti-4-1BB or anti-ICOS concentrations to observe minimal effects with anti-CD3 alone and enhancement with costimulation.
Anti-4-1BB Ab enhances Db/NP366–374-specific CD8 T cell responses to influenza virus in CD28−/− mice, but not in CD28−/−ICOS−/− mice
Previous results have shown that provision of an extra costimulatory signal via systemic administration of agonistic anti-4-1BB Ab allows a CD28-independent CD8 T cell response to influenza virus in B6 mice (39, 40). To test whether ICOS contributes to this CD28-independent response, we infected WT littermate controls, CD28−/−, ICOS−/−, and CD28−/−ICOS−/− DKO mice with influenza virus in the presence of anti-4-1BB Ab or rat Ig control. We examined the CD8 T cell response to the immunodominant (NP366–374) epitope at the peak of the primary response (day 7). As expected, WT mice responded to influenza virus in the presence or the absence of anti-4-1BB. CD28−/− mice showed a partial restoration of the primary response to influenza when treated with anti-4-1BB, confirming previous results (40). In contrast, there was no augmentation of the CD8 T cell response to influenza by anti-4-1BB treatment of DKO mice (Fig. 1, A and C). When percentages of CD8 T cells were converted to Ag-specific T cell numbers, the same trend was observed (Fig. 1 B). Similar results were obtained in three experiments with three mice per group. Analysis of the response on day 21 after priming as well as after secondary challenge confirmed the absence of response in DKO mice (data not shown), arguing that it is not simply a kinetic effect that leads to the apparent lack of response in DKO mice.
Analysis of splenocytes from infected mice after restimulation for 6 h with NP366–374 peptide to detect IFN-γ-producing effector T cells revealed that coinjection with anti-4-1BB could induce an influenza-specific CD8 effector population in CD28−/− mice, but failed to induce this population in the DKO mice (Fig. 2). Thus, ICOS is required for anti-4-1BB therapy to correct the primary expansion defect observed in the CD8 T cell response to influenza in CD28−/− mice.
Role of anti-4-1BB in CD4 responses in the presence or absence of CD28 and ICOS
To examine the effect of anti-4-1BB on the CD4 T cell response to influenza virus, we restimulated splenocytes from infected mice with heat-killed influenza virus for 96 h and analyzed IFN-γ production. Stimulation with heat-killed virus results in presentation in the MHC class II, but not the MHC class I, pathway (45) (see Materials and Methods). Splenocytes from control Ig-treated WT and ICOS−/− mice produced IFN-γ (Fig. 3), whereas those from CD28−/− and DKO mice showed no detectable IFN-γ production. Anti-4-1BB Ab treatment during priming resulted in a gain of IFN-γ production by CD4 T cells from CD28−/− mice after restimulation, but had no effect on DKO mice. Thus, anti-4-1BB can enhance CD4 T cell activation, leading to a cytokine response in CD28-deficient mice, and ICOS is required for this effect.
Anti-4-1BB up-regulates ICOS expression on CD4 and CD8 T cells
Previous results have shown that CD28 signaling can enhance the levels of ICOS induced by TCR signaling (47, 48, 49). The finding that ICOS was required for anti-4-1BB to correct the anti-influenza T cell response in CD28−/− mice led us to hypothesize that in the absence of CD28, 4-1BB may act in part by up-regulating ICOS, which then may compensate for the lack of CD28. To test this hypothesis, we stimulated CD28−/− T cells with anti-CD3 and either control Ab (rat IgG) or anti-4-1BB in vitro (Fig. 4). At 24 h, anti-CD3 alone induced ICOS expression on the CD4 T cell subset, but had minimal effects on CD8 T cells. At this time point, there was no enhancement with anti-4-1BB. By 48 h, ICOS expression was enhanced by anti-4-1BB treatment on CD8 T cells and also had a small, but reproducible, effect on CD4 T cells. This expression was limited to the CD69+ activated T cells and persisted at 72 h, declining by 96 h. Thus, anti-4-1BB enhances ICOS expression on CD8 and CD4 T cells in the absence of CD28.
Synergistic effects of anti-4-1BB and anti-ICOS in enhancing anti-CD3-induced division of CD4 and CD8 T cells
Anti-4-1BB and anti-ICOS have been shown individually to enhance division of T cells in the presence of a suboptimal CD3 signal (5, 29). To test for possible additive or synergistic effects of the two costimulatory pathways on T cell responses, total T cells were labeled with CFSE, and the division of CD4 and CD8 T cells was analyzed by gating on the specific subset. In the presence of a suboptimal concentration of anti-CD3 (0.25 μg/ml), anti-ICOS had little or no effect on CD4 or CD8 T cell proliferation, whereas anti-4-1BB stimulated a weak response in the WT CD4 and CD8 T cell subsets (Fig. 5,A). In contrast, when the two costimulatory signals were combined, it was clear that there was more extensive division of CD4 and CD8 T cells in the presence or the absence of CD28 (Fig. 5, A and B). We consistently found that CD8 CD28−/− T cells showed a greater response to anti-ICOS plus anti-4-1BB than CD4 CD28−/− T cells, whereas with WT T cells, the responses of CD4 and CD8 T cells to the two costimulatory molecules were similar and weaker than those obtained with anti-CD3 plus anti-CD28. These results show that in the presence or the absence of CD28, under suboptimal conditions of TCR stimulation, costimulation through both ICOS and 4-1BB can synergistically contribute signals leading to division of CD4 and CD8 T cells.
ICOS and 4-1BB cooperate in the direct costimulation of CD28−/− and CD28+/+ CD8 T cells
To date, the majority of the evidence suggests that ICOS is a costimulator for Th responses of CD4 T cells (1). In contrast, 4-1BB shows preferential effects on CD8 T cells in some (29, 30, 50, 51), but not all (34), models. Thus, the effects of ICOS on CD8 T cells observed in the T cell cultures (Fig. 5) could be due to indirect effects via CD4 T cells. Thus, it was of interest to determine whether 4-1BB and ICOS could cooperate directly on isolated CD8 T cells in the presence or the absence of CD28.
We first tested whether 4-1BB and ICOS could be coexpressed on isolated CD8 T cells. Upon anti-CD3 treatment, both ICOS and 4-1BB were induced on CD28−/− CD8 T cells in response to anti-CD3 stimulation, and a significant proportion of CD8 T cells coexpressed 4-1BB and ICOS (Fig. 6 A).
To test the direct effect of anti-4-1BB and anti-ICOS on CD8 T cells, highly purified CD8 T cells from WT and CD28−/− mice were labeled with CFSE and tested for responses to costimulation in the presence of suboptimal anti-CD3. Costimulation with anti-CD3 and anti-4-1BB alone induced the division of both WT and CD28−/− cells (Fig. 6,B, second row), whereas anti-CD3 and anti-ICOS had only a marginal effect on cell division (Fig. 6,B, third row). In contrast, dual costimulation with anti-4-1BB and anti-ICOS greatly enhanced the division of WT and CD28−/− CD8 T cells (Fig. 6 B, fourth row). Again, as seen in total T cell cultures, we consistently observed a greater effect of dual costimulation on the CD28−/− CD8 population. Thus, costimulation with both anti-4-1BB and anti-ICOS can cooperate to enhance CD8 T cell proliferation in a CD4 T cell- and CD28-independent manner.
The studies we report reveal a redundancy between CD28 and ICOS in the immune response to influenza virus and a previously unrecognized relationship between 4-1BB and ICOS in activation of CD8 T cells. Although mice deficient in CD28 were able to mount an influenza-specific CD8 T cell response in the presence of agonistic anti-4-1BB Ab, we show that mice lacking both CD28 and ICOS failed to produce Ag-specific immune responses under the same conditions. We also show that anti-4-1BB can induce higher levels of ICOS expression on CD4 and CD8 T cells and that the two molecules can act synergistically to induce cell division. Moreover, experiments on isolated CD8 T cells indicated that ICOS and 4-1BB can be coexpressed on anti-CD3-activated CD28−/− CD8 T cells and can cooperate to induce CD8 T cell division independently of CD28 signaling or CD4 T cell help.
Initial analysis of ICOS-deficient mice suggested that ICOS played a minor role in CD8 T cell responses. For example, in the CD8 T cell response to influenza virus, there was no defect in initial CD8 T cell expansion, and only a minor defect in CD8 T cell numbers late in the primary response, which fully recovered upon rechallenge with virus (29, 37). In contrast, experiments with CD28−/− mice showed a profound defect in the CD8 T cell response to influenza (27, 30), which could be fully compensated by treatment with agonistic anti-41BB Abs (39, 40). The present study shows that at least part of this effect was due to compensation by ICOS. Similarly, in analyzing antiviral humoral responses, Suh et al. (41) showed a greater effect of removal of both CD28 and ICOS than observed with mice lacking one or the other costimulatory pathway. Furthermore, blockade of the ICOS costimulatory pathway increases susceptibility to Toxoplasma gondii in CD28−/− mice, a response previously thought to be CD28 independent (52). Taken together, these results show that ICOS can play an important role in both CD4 and CD8 T cell responses, particularly when signals through CD28 are limiting.
Previous studies have shown that anti-CD28 mAb can induce higher levels of ICOS than stimulation with anti-CD3 alone (47, 48, 49). In this report we show that anti-4-1BB can partially replace this effect of CD28 by increasing the level of ICOS on anti-CD3-activated T cells. Previous reports also demonstrated that ICOS is expressed on activated CD4 and CD8 T cells, with higher expression on the CD4 T cells (5). In the present study there was greater expression of ICOS on anti-CD3-treated CD4 T cells compared with CD8 T cells, but anti-4-1BB appeared to compensate for this difference by enhancing ICOS expression to a greater extent on CD8 T cells (Fig. 4). This result is in keeping with previous reports that anti-4-1BB has greater effects on CD8 than on CD4 T cells (29, 50).
Several studies show that ICOS plays a significant role in CD4 T cell responses (2, 5). Its role in CD8 T cells remains controversial, although recent studies in infectious and tumor models show that ICOS is necessary for optimal CD8 T cell responses (19, 20, 21, 37). The present report shows that 4-1BB and ICOS can be coexpressed on isolated CD8 T cells and that ICOS can directly stimulate CD8 T cells, particularly in combination with an anti-4-1BB-induced signal.
In this study it appears that anti-4-1BB functions in part by increasing the expression of ICOS on the T cells. However, it seems unlikely that this is the only effect of anti-4-1BB in this model, because anti-4-1BB alone clearly induced some division by the T cells in the absence of a signal through ICOS (Figs. 5 and 6). Thus, it is likely that anti-4-1BB-mediated therapy of the influenza response in CD28−/− mice has two components: a direct effect of 4-1BB on CD8 T cells as well as effects on induction of ICOS, which allows ICOS to compensate for CD28 in both CD4 and CD8 T cell responses and perhaps provide help for the response.
In addition to the well-established effect of anti-4-1BB as a costimulator of CD8 T cell responses (29, 40, 53), agonistic anti-4-1BB Abs have been shown to block humoral immunity and to ameliorate autoimmune disease by inhibiting CD4 T cell responses in vivo (54, 55, 56, 57, 58, 59, 60). Recent evidence suggests that the apparently inhibitory effects of anti-4-1BB could be due to strong stimulation through 4-1BB leading to excessive generation of IFN-γ, which, in turn, leads to immunosuppressive mechanisms such as IDO (60) or TGF-β (61) production. In the present study anti-4-1BB neither inhibited nor augmented the CD4 T cell response in WT mice, consistent with our previous findings that CD4 T cell responses to influenza are unchanged in 4-1BBL-deficient mice (30). However, anti-4-1BB clearly enhanced CD4 T cell responses in CD28−/− mice (Fig. 4). Thus, under conditions of more limited stimulation in the CD28−/− mice, the costimulatory effects of 4-1BB on CD4 T cells predominate over any inhibitory effects.
We consistently observed that the effects of anti-4-1BB plus anti-ICOS on CD8 T cells were greater on the CD28−/− cells than their WT counterparts, even in the absence of CD4 T cells. The reason for this enhanced response to dual costimulation by the CD28−/− T cells is not clear. It is possible that it is due to a different cytokine milieu in CD28−/− vs WT mice or to competition for signaling intermediates when CD28 is present. Nevertheless, the finding that signaling through 4-1BB and ICOS can synergistically activate CD28−/− T cells is of particular interest with respect to the human immune system. In humans, CD28−CD8 memory phenotype T cells accumulate with age and with chronic diseases such as HIV and multiple myeloma (62, 63, 64, 65). In fact, with chronic HIV infection, the majority of HIV-specific memory CD8 T cells are CD28 negative (66, 67, 68). Conversely, in rheumatoid arthritis, CD4+CD28− T cells accumulate (69). Previous studies have shown that 4-1BBL can activate human CD28−CD4 and CD8 T cells (70, 71). The findings reported in this study, showing that 4-1BB and ICOS can synergistically activate CD28−/− T cells, suggest that either stimulating or blocking ICOS and 4-1BB in combination may be important in diseases where CD28− T cells are prevalent.
In summary, the data presented in this report reveal unexpected redundancy between CD28 and ICOS in the primary CD8 T cell response to influenza virus and reveal novel interactions between the 4-1BB and ICOS costimulatory pathways, leading to synergistic activation of CD28+/+ or CD28−/− T cells.
We thank Amgen for provision of the M6 Ab, and Robert Mittler for provision of the 3H3 hybridoma.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by a grant from the Canadian Institutes of Health Research (to T.H.W.), the Canadian Vaccine and Immunotherapeutic Network (to T.W.M), and Associazione Italiana per la Ricerca sul Cancro, Milan (to U.D.). W-K. S. was supported by a fellowship from the Cancer Research Institute (New York).
Abbreviations used in this paper: DKO, double knockout; HAU, hemagglutinin unit; NP, nuclear protein; WT, wild type.