Radiation inactivation of human y-interferon: Cellular ... - Europe PMC

Proc. Nati. Acad. Sci. USA Vol. 91, pp. 5818-5822, June 1994 Biophysics

Radiation inactivation of human y-interferon: Cellular activation requires two dimers (target size/receptor)

JEROME A. LANGER*t, ABBAS RASHIDBAIGI*t, GIANNI GAROTTA§, AND ELLIS KEMPNER¶ *Department of Molecular Genetics and Microbiology, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, 675 Hoes Lane, Piscataway, NJ 08854; §Department of Immunology, Hoffmann-La Roche, Basel, Switzerland; and ILaboratory of Physical Biology, National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health, Bethesda, MD 20892

Communicated by Martin Rodbell, March 22, 1994 (received for review November 17, 1993)

ABSTRACT -Interferon (IFN-y) is a 17-kDa broadspectrum cytokine which exerts its effects on a variety of target cells through its interaction with the IFN- y receptor. Although physicochemical studies of Escherichia coUl-derived IFN-y, as well as its crystal structure, demonstrate that it is a homodimer in solution (Mr 34,000), previous radiation inactivation studies yielded a functional size for IFN-yof 63-73 kDa in an antiviral assay. To understand the relationship between the solution form of IFN-y and the moiety that actually binds to the cellular receptor and activates cells, we examined irradiated nonradioactive and 32P-labeled IFN-y for its migration in SDS/polyacrylamide gels (to determine its physical integrity), its binding to cells, its reactivity in an ELISA, and its antiviral activity. The functional size of IFN-y differed in the assays, being 22 ± 2 kDa for the physical destruction of IFN-y, 56 ± 2 kDa for the cellular binding assay, 45-50 kDa for reactivity in the ELISA, and 72 ± 6 kDa for antiviral activity. The results from the binding assays constitute direct evidence that IFN-y binds to its cellular receptor as a dimer. However, for antiviral activity, the functional mass is equivalent to a tetramer. This is consistent with models involving ligand-induced receptor dimerization, whereby two dimers acting in concert (equivalent to the target size of a tetramer) are required to activate cells in the antiviral

receptor (7, 15-17) and the biochemical pathways leading to cellular activation (e.g., refs. 18-22). Irradiation of macromolecules leads to their destruction and the consequent reduction of measured parameters, such as biological activity (23). Quantitative measurements of the dose-response curve allow the calculation of a "target size," understood as the mass of the unit responsible for the measured parameter. Radiation inactivation techniques have been applied to several IFNs to determine their target size for stimulating cellular responses (24, 25). In a standard cytopathic-effect (antiviral) assay, the functional size for purified recombinant IFN-y from E. coli was 73 ± 6 kDa, about 4 times the monomer mass. It was concluded that a molecular tetramer of IFN-y is required for activation of cells in the antiviral assay; however, it was noted that " . . . it is not clear how these functional sizes related to . . . their receptor interactions that lead to generation of the antiviral state" (25). However, one IFN-y dimer binds to two molecules of a soluble form of the receptor, although this seems to be in equilibrium with a complex of one IFN-y dimer with one molecule of soluble receptor (15). These studies suggesting the binding of an IFN-y dimer appear to be inconsistent with the previous radiation inactivation studies implicating a functional IFN-y tetramer for cellular activation. We have now irradiated highly active 32P-labeled IFN-y (26, 27) to compare directly the functional size of IFN-y for binding to its cell surface receptor and for triggering biological (e.g., antiviral) activity. The finding that IFN-y binds to cells as a dimer but appears to trigger biological responses as a tetramer leads to the conclusion that biological activity requires more than the simple binding of IFN-y to its receptor, leading us to suggest models consistent with these observations.

assay.

,-Interferon (IFN-y) is a cytokine identified as an IFN by virtue of its broad-spectrum antiviral activity. It also has a range of immunomodulatory activities, being a primary activator of macrophages and a physiological inducer of class II major histocompatibility complex (MHC)-encoded molecules (1, 2). Recombinant human IFN-y expressed in Escherichia coli is a physiologically active, unglycosylated molecule with a mass of 17 kDa (3, 4). Purified natural (glycosylated) IFN-y or recombinant human IFN- yfrom E. coli behaves as a dimer in solution, at or near neutral pH, with no evidence for higher molecular weight species or for dissociation to monomers (5-7). IFN-y in crystals is a dimer (8, 9), and a genetically engineered covalent IFN- ydimer, with monomers connected tail-to-head by an IgA hinge peptide, has high biological activity (10). IFN-y stimulates a variety of responses in target cells through interaction with its cellular receptor (1, 2). The IFN-y receptor (IFN-yR) cDNA has been cloned and expressed (11). A receptor-associated accessory factor required for IFN-a activation has been identified and cloned (12, 13), and the existence of additional accessory factors has been deduced (14). Biochemical and molecular biological studies have begun to elucidate the interaction of IFN-y with its

EXPERIMENTAL PROCEDURES IFNs and Antiviral Assays. Recombinant human IFN- ywas isolated from E. coli and was radiophosphorylated with [y-32P]ATP by using cAMP-dependent protein kinase (26, 27). The ability of irradiated or unirradiated IFN- y to protect cells from virus killing was assayed on human WISH cells with vesicular stomatitis virus (VSV) or with encephalomyocarditis virus (EMCV) as described (28). In brief, serially diluted irradiated or unirradiated IFN-y was incubated with cells at 370C for 6-16 hr. Virus was then added and the IFN-y titer was determined 24-48 hr later as the concentration of IFN-y that provided 50% protection of cells, as judged by Abbreviations: IFN, interferon; IFN-yR, IFN-y receptor; VSV, vesicular stomatitis virus; EMCV, encephalomyocarditis virus; BSA, bovine serum albumin. tTo whom reprint requests should be addressed. tPresent address: Interferon Sciences, 783 Jersey Avenue, New Brunswick, NJ 08901.

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

5818

Biophysics: Langer et al. staining live cells with crystal violet. Laboratory standards calibrated against international reference standards were included. Irradiation and Processing of Samples. Samples of nonradioactive or 32P-labeled IFN-y were prepared for radiation inactivation as follows. For IFN-y, samples were at 10o units/ml (10 ug/ml) in either Dulbecco's modified Eagle's medium (DMEM) or Eagle's minimal essential medium, each containing 10%o fetal bovine serum. For 32P-labeled IFN--y, -3 pg of IFN-y was phosphorylated in the presence of 1 mCi (37 MBq) of [y.32P]ATP to about 0.6 mol Of 32p per mol of IFN- y (170 t&Ci/,pg) (26, 27). This was diluted to 10-11 ml in phosphate-buffered saline (PBS) containing 0.5% bovine serum albumin (BSA). Samples of 0.5 ml were placed in 2-ml siliconized glass ampules and frozen on dry ice. Vials were rapidly sealed with an oxygen/gas flame, with no thawing observed. Storage of samples was at -70'C or lower prior to and following irradiation. Shipment of samples was on dry ice. Samples were not thawed until ready for assay. In several experiments, samples seemed to be at least twice as sensitive to radiation in all assays (i.e., to have higher calculated functional sizes). The addition of ascorbate to the PBS/BSA formulation or the use of an entirely different medium reduced or eliminated spurious and variable radiation sensitization. Thus, improved buffer conditions are (i) PBS with BSA, as above, but supplemented with 10 mM ascorbate; or (ii) 10 mM Tris-HCl, pH 7.5, with apotransferrin at 4 mg/ml (Sigma catalogue no. T2252). It is likely that the divergent results arose from a radiosensitizing effect of phosphate, as observed in several other systems (29). The data reported here were from those extensive data sets utilizing PBS/BSA consistent with the final, less extensive data collected with samples in PBS/BSA/ascorbate or Tris/ apotransferrin. The details of irradiation have been described (30). In the present experiments, frozen samples were irradiated at -1350C. After irradiation, IFN-y samples were thawed and transferred to polypropylene tubes, and radiolabeled samples were checked by liquid scintillation counting to ensure uniform sample recovery from the irradiated vials. Generally, initial analysis by SDS/PAGE, cell binding, and antiviral assay were performed immediately on the thawed samples. Remaining material was quickly frozen in liquid nitrogen and stored in the vapor phase of liquid nitrogen. Samples were thawed as necessary for subsequent experiments, but care was taken to quickly thaw and freeze samples. No systematic or significant decrease in activity was found in the several (fewer than three) cycles of rapid thawing/quick freezing under this regimen. SDS/PAGE. Irradiated and unirradiated samples of 32plabeled IFN-y were analyzed by SDS/15% PAGE (31). A constant volume (and amount of radioactivity) of each irradiated sample was added to sample buffer and boiled with 2-mercaptoethanol. After electrophoresis, gels were dried and subjected to autoradiography. Pieces corresponding to the IFN band were cut from the dried gel by using the autoradiograph as the cutting template. Pieces were added to 2 ml of Bray's scintillation fluid and the radioactivity was measured in a liquid scintillation counter. Pieces of gel of the same size from proximal lanes having no sample were used to determine background radioactivity. For each experiment, at least two gels were run with duplicate samples, and the results were averaged. Binding of IFN-'yto Cells. Binding of irradiated and control IFN-y to HL-60 human promyelocytic leukemia cells was assayed approximately as described (27, 32). Nonspecific binding, measured in the presence of >100-fold excess nonradioactive IFN-'y, was approximately constant for irradiated and nonirradiated samples, constituting 2-5% (100-200 cpm)

Proc. Natl. Acad. Sci. USA 91 (1994)

5819

of total binding for unirradiated samples (2000-7000 cpm in various experiments). Cells and IFN-y were incubated at either 40C for 2.5 hr or 230C for 1-1.5 hr with gentle rocking. The slopes ofthe radiation decay curves were the same at 40C and 230C. In some experiments with unirradiated IFN-'y, preliminary saturation binding curves were measured. A concentration of 32P-labeled IFN-y corresponding to s60% of saturation was then used for the subsequent binding experiments with irradiated ligand, to ensure a linear doseresponse curve. ELISA. The ELISA for IFN-y (33) is a sandwich assay, employing the same monoclonal antibody (y69; ref. 34) for binding to the solid support and for detection. The ELISA detects the biologically active dimer, but not denatured

IFN-'y (33, 34). Data Analysis. In the simplest cases, the survival of activity as a function of radiation dose is expected to be a simple exponential decay (23, 35). From both fundamental considerations and the details of the experimental protocol, it has been demonstrated that the target size (mass, M, in kilodaltons) can be calculated from the exponential activity decay curve from the equation: M = (1.79 x

106)/D37

where D37 is the radiation dose (in megarads; 1 rad = 0.01 Gy) at which the activity is reduced to 37% of that in the unirradiated control. Values reported here were derived from a linear leastsquares fit, with the calculated survival curve constrained to pass through the measured value at zero dose (i.e., unirradiated). Additional analyses were performed to examine the sensitivity of the results to calculational manipulations, such as removing the constraints, eliminating outlying data points, using variance weighting, and/or truncating the data sets. All calculated target sizes from these manipulations were within 10% of one another. Values used here are from the constrained analysis. In some instances where statistical analysis showed only fair correlation with a straight line, systematic deviations from linearity were checked by calculating lines for various parts of the curve and comparing their slopes. Data significantly deviating from linearity were not included in the final analysis (see results regarding binding curves). Further statistical treatment was performed by pooling data from three experiments, calculating error envelopes of each curve at the 95% level, and performing a heterogeneityof-slopes analysis.

RESULTS The target size of 32P-labeled recombinant human IFN-yin an antiviral assay was 72 ± 5 kDa (n = 6) (Figs. 1 and 2; Table 1). In most data sets, the linearity was very good, as was the reproducibility. One data set was omitted because of striking nonlinearity in the assay. The measured value is in good agreement with the previously reported value of 73 + 6 kDa for nonradioactive recombinant human IFN-y (24). Values obtained in two laboratories using two different viruses, VSV and EMCV, differ only slightly [EMCV, 68 + 4 kDa (n = 3); VSV, 76 ± 3 kDa (n = 3); see footnote *, Table 1]. The effect of irradiation on 32P-labeled IFN-y was monitored by the radioactivity in the IFN-y monomer band in SDS/polyacrylamide gels. Irradiated samples showed a progressive loss of radioactivity (fewer intact IFN-y molecules) as the radiation dose increased. This corresponds to a target size of 22 ± 2 kDa (n = 3) (Figs. 1 and 2; Table 1). This value is in reasonable agreement with the calculated size of the monomer, 16.9 kDa. The implication is that each monomer within the strongly associated molecular dimer is independently destroyed. Following a hit to one monomer of the

5820

=U

fi

Biophysics: Langer et A

0.1

0

I00.01

Proc. Natl. Acad. Sci. USA 91 (1994) Table 1. Summary of calculated functional molecular sizes for IFN-y from target-size analysis Functional size, kDa Assay SDS/PAGE (32P-IFN-y band) 22 + 2 (n = 3) 32P-IFN-y binding to cells 56 + 2 (n = 3) Antiviral activity 32P-labeled IFN-y (VSV) 76 + 3 (n = 3) Nonradioactive IFN-y (EMCV) 78 + 20 (n = 4)* ELISA 45-50 (n = 2)t *This value includes one measurement that produced a calculated value of 108 kDa. If this unusual value is omitted, the average of the three remaining determinations for nonradioactive IFN-y is 68 ± 4 kDa (n = 3). tData not shown.

The target size for the nonradioactive IFN-y moiety recognized in the ELISA was 45-50 kDa, consistent with a dimer. This is expected from the design of the ELISA, in which the same antibody is used for both binding the IFN to the plate and for detection (33), thus requiring that the Dose (M"d) recognized IFN-y epitope be present in more than one copy. When data from three binding experiments were pooled FIG. 1. Fractional activity of IFN-y in several assatys as a function of radiation dose from one set of irradiated samples . Values and compared with the pooled data from the antiviral assays, for analysis by SDS/PAGE (o), binding to HL-60 cells (o), and the differences in functional size calculated from the receptor antiviral assays (A) represent the average from duplicate gel lanes, at binding and from the antiviral assays were highly significant least triplicate binding replicates, and duplicate antiviral assays. as demonstrated by the heterogeneity-of-slopes test (Fig. 2). Lines were fit by a linear regression with the line constr Dained to a 1.0 at zero dose. intercept fractional activity of 120

DISCUSSION

rpmaine1 intatrt i1tA,IDErA hu. 1U1 111u ads JUUg5 UY mobility in an SDIS/polyacrylamide gel. For the binding of 32P-labeled IFN-yto cells, the target size was 56 ± 2 kDa (ni = 3) when binding assays were performed at subsaturating c Concentrations of IFN-y. At these concentrations, the meas;ured decay curves were clearly defined by a single exponent ial function.

Aimar than athar Illullulilcl mnnwnnar

QC

Almost all radiation inactivation studies of receptor-ligand ligand to interactions have involved the binding of native

irradiated receptors. The present study considers the reciprocal protocol: irradiated ligand (IFN-y) was examined for the survival of its structural integrity, for its remaining ability to bind to receptors, and for the fraction of biological (antiviral) activity still expressed. The target sizes for human IFN- yare 22 ± 2 kDafor SDS/PAGE analysis, 45-50 kDafor ELISA, 56 ± 2 kDa for cell surface binding, and 72 ± 5 kDa for a biological antiviral assay (Figs. 1 and 2; Table 1). These a1 1 differences emphasize that the "functional molecular size" is Eldependent on the function being measured. While our results provide direct evidence that IFN-y binds to its cell surface D x0 receptor as a dimer, the accompanying observation of a higher functional mass for biological activity forces us to consider new models for the mechanism of cellular activation by IFN-y. 'The target size for the physical destruction of IFN-y is 22 c kDa, equivalent to a monomer. If remaining undamaged monomers had retained either significant affinity for the receptor or biological activity, the functional target size for 0.01 cellular binding and biological activity would have been that of the monomer. When one subunit of an oligomer is destroyed by radiation, the overall structure is generally maintained: the fragmented monomer is held together with the surviving subunits (36). This precludes the assembly ofactive J , l ^ 1 oligomers from surviving monomers (37), a process which L 0.001 would have shifted the functional target size to that of the 20 40 60 80 100 120 monomer (38). Thus, surviving monomers generated during Dose (Mrad) irradiation of IFN-ydimers neither have intrinsic activity nor associate to form active oligomers. FIG. 2. Fractiorial activity of IFN-e in several assays as a The 56-kDa functional size in the receptor binding assay is from three independent irradiation experfunction of radiation dose 2.5 times the experimentally observed value for the destruci iments. Values for ; tion of the physical moiety (22 kDa) and 3 times the true cells (e), and antivii ml assays(b) represent the avedge from duplimonomer mass (17 kDa). The cell binding data are, however, cate gel lanes, at lelast triplicate binding replicates, and duplicate only slightly higher (statistically insignificant) than the results antiviral assays. Limmes were fit by a linear regression with the line from the ELISA (45-50 kDa), which is known to measure a constrained to interrcept a fractional activity of 1.0 at zero dose. dimer (33, 34). Moreover, since IFN-y in solution and in a Shaded areas repressent envelopes for 95% confidence levels about the mean for the po oled data. crystal is a dimer (see below), and since there are no instances 0.1

0

LL

0

Biophysics: Langer et A known to us of a soluble protein dimer rearranging or aggregating to an active trimer, it is extremely unlikely that the binding form is a trimer. Therefore, the value of 56 kDa most likely represents a dimer with some systematic error. Our results provide a direct demonstration that the dimer is the moiety bound by the IFN-y receptor on cells. This evidence underscores the importance of the dimer. Previously it was shown that (i) IFN-y, whether natural or recombinant, glycosylated or unglycosylated, appears to be a stable dimer in solution (5, 6, 39-47),11 with no evidence at neutral pH for association to a tetramer, or for dissociation to a monomer (5); (ii) the molecule crystallizes as a dimer, where strong interactions arise both from the extensive surface contacts between the monomers and from the intermingling ofhelices between the two monomers (8, 9); and (iii) a dimer appears to be the moiety that binds to the extracellular binding domain of the IFN-y receptor (15). Nevertheless, the results of Pestka et al. (24), confirmed here, clearly demonstrate that IFN- yin an antiviral assay has a tetramer target size. It is significant that the functional mass of IFN-y for binding and for antiviral activity was measured on the same batches of irradiated 32P-labeled IFN-'y. Taken together, the radiation inactivation studies suggest that IFN-y binds to its receptor as a dimer, but the simple binding of the dimer is not sufficient to elicit an antiviral state. The "functional tetramer" required for antiviral activity is formally equivalent to the concerted action of two dimers.** Consider, for example, classes of models in which two IFN-y dimers act in concert to produce cellular activation. (i) A dimer binds to a receptor. Binding of the first dimer results in cooperative binding of a second IFN-y dimer, leading to cellular activation. (ii) Each of two dimers independently binds to each of two receptor molecules, and subsequent association of two IFN--y/receptor complexes is required for cellular activation. Both classes of models would be consistent with a tetramer as the functional size of IFN-,y in the antiviral assay. Specific models involving two dimers for cellular activation might be distinguished by evidence for positive cooperativity in the dependence of cellular activation on IFN-y concentration. IFN-yR is a member of the class 2 family within the cytokine receptor superfamily (48). Ligand-induced receptor dimerization of receptors is thought to be a common mechanism for initiating cellular activation within this family (e.g., refs. 49 and 50) and may be a general feature of polypeptide growth factor/cytokine receptor signaling (51). There are at least two mechanisms for this dimerization: (i) the ligand (L) can form a "bridge" between two receptors (R), producing a complex (R/L/R), or (ii) complexation of the ligand with one "The few discrepancies with these conclusions come from crude preparations of IFN--y. There are some indications of a Mr 65,00070,000 form in some crude preparations of IFN--y (40, 43). In one case, the predominant form of IFN-y activity in crude serumcontaining conditioned medium had a Mr - 65,000-70,000, whereas after partial purification the IFN--y migrated at Mr 40,000 (43). It is not known whether the Mr 65,000-70,000 form is a higher aggregate of IFN-y itself or whether it represents a complex with another serum polypeptide. Nevertheless, in purified preparations, the solution form is the dimer. **If the probability of an IFN-y dimer binding to a receptor is proportional to the concentration of active IFN dimer, and the biological activity is the result of the independent (or related) binding of two dimers, then the activity is proportional to the product of the binding probabilities and hence to the square of the concentration of the dimers. On a semilogarithmic plot, the slope of the inactivation curve for antiviral activity would be twice that for binding; the calculated functional size would be the tetramer. Qualitatively, the inactivation of antiviral activity, being the result of two independent probabilities of binding, is more sensitive to a decrease in the concentration of active IFN- y caused by radiation.

Proc. Natl. Acad. Sci. USA 91 (1994)

5821

receptor (R/L) can be followed by "side-to-side" association of two complexes, with a final stoichiometry (R/L)2. Examples of both types of complexes exist. Growth hormone monomer forms a bridge between two molecules of its receptor binding protein (52, 53). Interleukin 4 (IL-4) and members of its subfamily [IL-5, granulocyte/macrophagecolony-stimulating factor (GM-CSF), IL-3) also may form active complexes by bridging two receptors, regardless of whether the receptor is a monomer (e.g., IL-4 receptor), or an a/P heterodimer (IL-5, GM-CSF, and IL-3 receptors) (54). Other members of the cytokine receptor superfamily (e.g., the erythropoietin receptor) also appear to undergo ligandinduced dimerization. Side-to-side dimerization may be exemplified by the active complex of epidermal growth factor and its receptor and possibly other receptor tyrosine kinases (e.g., refs. 51, 55-57). Our results suggest that the activation of cells requires the mass of four IFN-y monomers. It is well established that native IFN-y is a dimer (see above), and we have demonstrated here that it binds to its receptor as a dimer. If ligand-induced receptor dimerization is also involved in cellular activation, either a side-to-side model or a bridging ligand model could be consistent with the radiation inactivation data. For a side-to-side complex, each IFN-y dimer could bind to a single IFN-y receptor ((IFN-y)2/IFN--yR], without acting as a bridging ligand; this complex could then associate with a similar complex, to form the complex [(IFN-'y)2/(IFN-yR)h2. In the alternative model, each IFN-y dimer could bridge two IFN-yR molecules, forming the complex IFN-yR/(IFN-y)2/IFN-yR; evidence for this complex comes from both hydrodynamic and crosslinking studies (15, 17). In addition, to account for the four monomers of IFN-y implicated by the radiation inactivation results, two of these complexes would then associate as [IFN-yR/(IFN-y)2/ IFN-'yR]2 to initiate cellular signaling. We provisionally suggest that this latter model, which utilizes the symmetry of the IFN-y dimer as a bridging molecule, is preferred. These general models are independent of the precise subunit structure of the receptor. It now appears that the active IFN-'yR consists of the high-affinity IFN-yR polypeptide (11), as well as one or more accessory factors required for intracellular signaling (12-14). This multisubunit structure is similar to that of other cytokine receptors where signaling (and sometimes high-affinity binding) requires at least two subunits as an a/,B heterodimer (49, 50, 54). Tests and constraints on these models might be obtained from in vitro studies of the interaction of IFN-'y with the IFN-yR/ accessory factor complexes and from radiation inactivation studies where the functional size of the receptor, rather than the ligand, is studied. The irradiation of 32P-labeled IFN-y permits the coordinated analysis of physical integrity, cell-surface or soluble receptor binding, and cellular activation. From these results, it is clear that the antiviral state does not result from the binding of a single IFN-y dimer to its receptor. Rather, our results are consistent with models invoking the action of two IFN-y dimers for biological activity. Preliminary radiation inactivation data suggest a similar requirement for two IFN-y dimers in the IFN-y-induced activation of human macrophages, as measured by superoxide release, hydrogen peroxide release, or bacteriocidal activity (unpublished data). It would be of interest to look at early cellular events, such as the activation of specific transcription factors. Thus, our measurement of several parameters has clarified previous radiation inactivation studies of IFN-y and suggested mechanistic possibilities for cellular activation by IFN-y. We appreciate the assistance of Ms. Patrice Gregory (Academic Computing Services, Robert Wood Johnson Medical School) for statistical analysis and interpretation. We thank Dr. Sidney Pestka

5822

Proc. Natl. Acad. Sci. USA 91 (1994)

Biophysics: Langer et al.

for useful discussions. This work was supported by grants to J.A.L. from the American Cancer Society (BC-698), the Foundation of the University of Medicine and Dentistry of New Jersey (4-89), and the Pharmaceutical Manufacturers Association Foundation. 1. Pestka, S., Langer, J. A., Zoon, K. C. & Samuel, C. E. (1987) Annu. Rev. Biochem. 56, 727-777. 2. Farrar, M. A. & Schreiber, R. D. (1993) Annu. Rev. Immunol. 11, 571-611. 3. Gray, P. W., Leung, D. W., Pennica, D., Yelverton, E., Najarian, R., Simonsen, C. C., Derynck, R., Sherwood, P. J., Wallace, D. M., Berger, S. L., Levinson, A. D. & Goeddel, D. V. (1982) Nature (London) 295, 503-508. 4. Devos, R., Cheroutre, H., Taya, Y., Degrave, W., Van Heuverswyn, H. & Fiers, W. (1982) Nucleic Acids Res. 10, 24872501. 5. Yphantis, D. A. & Arakawa, T. (1987) Biochemistry 26, 54225427. 6. Rinderknecht, E., O'Connor, B. H. & Rodriguez, H. (1984) J. Biol. Chem. 259, 6790-6797. 7. Fountoulakis, M., Juranville, J.-F., Maris, A., Ozmen, L. & Garotta, G. (1990) J. Biol. Chem. 265, 19758-19767. 8. Ealick, S. E., Cook, W. J., Vijay-Kumar, S., Carson, M., Nagabhushan, T. L., Trotta, P. P. & Bugg, C. E. (1991) Science 252, 698-702. 9. Samudzi, C. T., Burton, L. E. & Rubin, J. R. (1991) J. Biol. Chem. 266, 21791-21797. 10. Lunn, C. A., Davies, L., Dalgarno, D., Narula, S. K., Zavodny, P. J. & Lundell, D. (1992) J. Biol. Chem. 267, 1792017924. 11. Aguet, M., Dembic, Z. & Merlin, G. (1988) Cell 55, 273-280. 12. Soh, J., Donnelly, R. J., Kotenko, S., Mariano, T. M., Cook, J. R., Wang, N., Emanuel, S., Schwartz, B., Miki, T. & Pestka, S. (1994) Cell 76, 793-802. 13. Hemmi, S., Bohni, R., Stark, G., Di Marco, F. & Aguet, M. (1994) Cell 76, 803-810. 14. Soh, J., Donnelly, R. J., Mariano, T. M., Cook, J. R., Schwartz, B. & Pestka, S. (1993) Proc. NatI. Acad. Sci. USA 90, 8737-8741. 15. Fountoulakis, M., Zulauf, M., Lustig, A. & Garotta, G. (1992) Eur. J. Biochem. 208, 781-787. 16. Gentz, R., Hayes, A., Grau, N., Fountoulakis, M., IaIhm, H.-W., Ozmen, L. & Garotta, G. (1992) Eur. J. Biochem. 210, 545-554. 17. Greenlund, A. C., Schreiber, R. D., Goeddel, D. V. & Pennica, D. (1993) J. Biol. Chem. 26, 18103-18110. 18. Pellegrini, S. & Schindler, C. (1988) Trends Biochem. Sci. 18, 338-342. 19. Larner, A. C., David, M., Feldman, F. M., Igarashi, K.-i., Hackett, R. H., Webb, D. S. A., Sweitzer, S. M., Petricoin, E. F., III, & Finbloom, D. S. (1993) Science 261, 1730-1733. 20. Ruff-Jamison, S., Chen, K. & Cohen, S. (1993) Science 261, 1733-1736. 21. Silvennoinen, D., Schindler, C., Schlessinger, J. & Levy, D. E. (1993) Science 261, 1736-1739. 22. Sadowski, H. B., Shuai, K., Darnell, J. E., Jr., & Gilman, M. Z. (1993) Science 261, 1739-1743. 23. Kempner, E. S. (1993) Q. Rev. Biophys. 26, 27-48. 24. Pestka, S., Kelder, B., Familletti, P. C., Moschera, J. A., Crowl, R. & Kempner, E. S. (1983) J. Biol. Chem. 258, 9706-9709.

25. Kempner, E. S. & Pestka, S. (1986) Methods Enzymol. 119, 255-260. 26. Kung, H.-F. & Bekesi, E. (1986) Methods Enzymol. 119, 296-301. 27. Rashidbaigi, A., Kung, H.-f. & Pestka, S. (1985) J. Biol. Chem. 260, 8514-8519. 28. Familletti, P. C., Rubinstein, S. & Pestka, S. (1981) Methods Enzymol. 78, 387-394. 29. Kempner, E. S. & Miller, J. H. (1994) Anal. Biochem. 216, 451-455. 30. Harmon, J. T., Nielsen, T. B. & Kempner, E. S. (1985) Methods Enzymol. 117, 65-94. 31. Laemmli, U. K. (1970) Nature (London) 227, 680-685. 32. Langer, J. A. & Pestka, S. (1986) Methods Enzymol. 119,

305-311.

33. Gallati, H., Pracht, I., Schmidt, J., Haring, P. & Garotta, G. (1987) J. Biol. Reg. Homeostatic Agents 1, 109-118. 34. Kelder, B., Rashidbaigi, A. & Pestka, S. (1986) Methods Enzymol. 119, 582-587. 35. Kempner, E. S. & Haigler, H. T. (1985) in Growth and Maturation Factors, ed. Guroff, G. (Wiley, New York), Vol. 3, pp. 149-173. 36. Potier, M., Thauvette, L., Michaud, L., Giroux, S. & Beauregard, G. (1991) Biochemistry 30, 8151-8157. 37. Boyer, T. & Kempner, E. S. (1992) Anal. Biochem. 207, 51-57. 38. Verkman, A. S., Skorecki, K. & Ausiello, D. A. (1984) Proc. Natl. Acad. Sci. USA 81, 150-154. 39. Falcoff, R. (1972) J. Gen. Virol. 16, 251-253. 40. Langford, M. P., Georgiades, J. A., Stanton, G. J., Dianzani, F. & Johnson, H. M. (1979) Infect. Immun. 26, 36-41. 41. De Ley, M., Van Damme, J., Claeys, H., Weening, H., Heine, J. W., Billiau, A., Vermylen, C. & De Somer, P. (1980) Eur. J. Immunol. 10, 877-883. 42. Yip, Y. K., Pang, R. H. L., Urban, C. & Vilcek, J. (1981) Proc. Natl. Acad. Sci. USA 78, 1601-1605. 43. Nathan, I., Groopman, J. E., Quan, S. G., Bersch, N. & Golde, D. (1981) Nature (London) 292, 842-844. 44. Devos, R., Opsomer, C., Scahill, S. J., Van der Heyden, J. & Fiers, W. (1984) J. Interferon Res. 4, 461-468. 45. Braude, I. (1984) Biochemistry 23, 5603-5609. 46. Tsujimoto, M., Tarutani, T., Ogawa, K. & Higashi, N. (1984) J. Interferon Res. 4, 111-114. 47. Miyata, K., Yamamoto, Y., Ueda, M., Kawade, Y., Matsumoto, K. & Kubota, I. (1986) J. Biochem. (Tokyo) 99, 1681-

1688.

48. Bazan, J. F. (1990) Proc. Nati. Acad. Sci. USA 87, 6934-6938. 49. Kishimoto, T., Taga, T. & Akira, S. (1994) Cell 76, 253-262. 50. Sprang, S. R. & Bazan, J. F. (1993) Curr. Opin. Struct. Biol. 3, 815-827. 51. Ullrich, A. & Schlessinger, J. (1990) Cell 61, 203-212. 52. Cunningham, B. C., Ultsch, M., De Vos, A. M., Mulkerrin, M. G., Clauser, K. R. & Wells, J. A. (1991) Science 254, 821-825. 53. De Vos, A. M., Ultsch, M. & Kossiakoff, A. A. (1992) Science 255, 306-312. 54. Boulay, J.-L. & Paul, W. E. (1992) J. Biol. Chem. 267,2052520528. 55. Bishayee, S., Majumdar, S., Khire, J. & Das, M. (1989) J. Biol. Chem. 264, 11699-11705. 56. Heldin, C. H., Ernlund, A., Rorsman, C. & Ronnstrand, L. (1989) J. Biol. Chem. 264, 8905-8912. 57. Li, W. & Stanley, E. R. (1991) EMBO J. 10, 277-288.