A point mutation in interleukin-2 that alters ligand internalization.

In previous studies, we have identified an interleukin-2 (IL-2) analog containing a point mutation at position 51 (T51P) that expresses nearly wild-type bioactivity, yet has ~10-fold lower receptor binding affinity. Since ligand-dependent receptor internalization may be the rate-limiting step controlling the duration of IL-2 receptor signaling, a reduction in the receptor internalization rate could contribute to the observed response enhancement for this analog. To evaluate this possibility, we compared the internalization of IL-2 and T51P in three separate assays. While the internalization rate for IL-2 agreed with values determined by others, the internalization of T51P was markedly reduced. The receptor binding rate constants for this analog were only slightly different; thus, altered binding kinetics could not explain the decreased internalization rate. The effects of reduced internalization were also observable in bioassays, where T51P maintained T-cell proliferation for a longer period compared with IL-2. These results indicate that the T51P point mutation reduces the receptor internalization rate compared with IL-2 in a fashion that is independent of the dissociation rate. This analog may represent a new approach to the preparation of cytokine analogs with potentiated agonist and antagonist properties.

In previous studies, we have identified an interleukin-2 (IL-2) analog containing a point mutation at position 51 (T51P) that expresses nearly wild-type bioactivity, yet has ϳ10-fold lower receptor binding affinity. Since ligand-dependent receptor internalization may be the rate-limiting step controlling the duration of IL-2 receptor signaling, a reduction in the receptor internalization rate could contribute to the observed response enhancement for this analog. To evaluate this possibility, we compared the internalization of IL-2 and T51P in three separate assays. While the internalization rate for IL-2 agreed with values determined by others, the internalization of T51P was markedly reduced. The receptor binding rate constants for this analog were only slightly different; thus, altered binding kinetics could not explain the decreased internalization rate. The effects of reduced internalization were also observable in bioassays, where T51P maintained T-cell proliferation for a longer period compared with IL-2.
These results indicate that the T51P point mutation reduces the receptor internalization rate compared with IL-2 in a fashion that is independent of the dissociation rate. This analog may represent a new approach to the preparation of cytokine analogs with potentiated agonist and antagonist properties.
Interleukin-2 (IL-2) 1 is an important immunomodulator that regulates T-cell growth as well as the activities of natural killer cells and B-cells (1). For activated T-cells, IL-2 acts through a receptor system (IL-2R) composed of at least three cell-surface subunits: p55 (␣-subunit) (2)(3)(4), p75 (␤-subunit) (5), and p64 (␥ c -subunit) (6,7). These three chains cooperate to generate the high affinity IL-2R complex (K d Ϸ 10 Ϫ11 M), whereas the ␣-chain alone and the ␤⅐␥ c -subunit complex form low affinity (K d Ϸ 10 Ϫ8 M) and intermediate affinity (K d Ϸ 10 Ϫ9 M) receptor sites, respectively. The ␣-chain displays no ability to transduce intracellular signals. Rather, it forms a heteromeric complex with the ␤-subunit that serves to capture IL-2 on the surface of activated T-cells (8). In contrast, the ␤and ␥ c -subunits associate in a ligand-dependent fashion and form the signaling complex (6,9).
Interaction between IL-2 and the high affinity receptor complex triggers a series of intracellular events, including activation of the JAK-STAT (10 -14) and Ras-Raf-MAPK pathways (15), that lead to T-cell proliferation. Another early event associated with IL-2 receptor occupancy is ligand-dependent receptor internalization (16 -19). Similarly, the intermediate affinity IL-2R (␤⅐␥ c ) can mediate both signaling and internalization (20 -22), while the low affinity IL-2R (␣) induces neither function (23). Some reports suggest that inhibition of IL-2 internalization using anti-receptor antibodies can influence IL-2-dependent cell growth (24,25). However, other studies contradict the requirement of internalization for signaling. Kinetic analyses have shown that some early events induced by IL-2 take place very rapidly after addition of IL-2 to cells and peak within 1 or 2 min (26 -28). Internalization, however, occurs more slowly, with a t1 ⁄2 of ϳ15 min (22,25,29). Furthermore, mutations in the cytoplasmic domain of IL-2R␤ apparently affect IL-2 signaling without altering internalization (30). Although it is unlikely that internalization is required for signaling, it has been well documented that the internalization of the IL-2⅐IL-2R complex results in down-regulation of high affinity IL-2 receptors, presumably attenuating IL-2 responsiveness in T-cells (31). Endocytosis, as well as biosynthesis, controls the time of residency of receptors on the cell surface (32) and therefore modulate cellular responsiveness to IL-2.
Consequently, inhibition of ligand-dependent receptor internalization should be reflected by changes in biological response. In this study, we report the first example of ligandinduced inhibition of receptor internalization that is not solely due to altered dissociation kinetics.

MATERIALS AND METHODS
Preparation and Characterization of IL-2 Analogs-Proteins were generated via cassette mutagenesis on a synthetic IL-2 gene (33) and were expressed in Escherichia coli, refolded, and purified as described previously (34). All proteins were determined to be homogeneous by reverse-phase HPLC (4.5 ϫ 25-mm duPont NEN Protein Plus C 18 column) and monomeric upon HPLC size-exclusion chromatography (4.5 ϫ 25-mm TSK-Gel G3000W XL column; buffer containing 25 mM NaPO 4 , pH 7.0, and 100 mM KCl). The concentration of purified proteins was determined by UV absorption in 6 M guanidine HCl (⑀ 280 nm ϭ 9.53 ϫ 10 3 ) (35).
Bioassays-Bioassays were carried out on normal human PBLs, isolated as described above and stimulated with OKT3 monoclonal antibody (Ortho Pharmaceuticals) for 3 days followed by 2 days of incubation in the absence of OKT3 (36). Quantitative bioassays were performed on these day 5 human PBLs, with a 48-h incubation in the presence of a full range of ligands as described previously (34). Ex-tended bioassays were performed on the same population of cells with a single protein concentration (EC 50 ). Cells were harvested every 12 h after the regular 48-h incubation period. The biological response was measured by [ 3 H]thymidine incorporation.
Radioiodination-Purified monomeric recombinant IL-2 and T51P were labeled with Na 125 I using lactoperoxidase/glucose oxidase as described (37). Briefly, 30 l of concentrated IL-2 (66.7 M in 5 mM NH 4 OAc, pH 5) or T51P analog (40 M in 25 mM NaH 2 PO4, pH 4.5) were incubated with 1 mCi of Na 125 I (100 mCi/ml; Amersham Corp.), 15 l of 5% glucose, 30 l of 0.2 M NaPO 4 , pH 7.0, and Enzymobeads (Bio-Rad) for 5 min following the manufacturer's instructions. The reaction was quenched by adding 7.5 l of 10% NaN 3 and 30 l of 0.1 M unlabeled NaI to terminate the enzymatic activity. An aliquot of 30 l of 10% fetal calf serum was added to the reaction mixture prior to separation on a Sephadex G-10 column (1-ml syringe) equilibrated with phosphatebuffered saline, pH 6.8, containing 10% fetal calf serum. The column was eluted with the same buffer. The fractions were collected, and those containing peak radioactivity were pooled. The specific activity was determined by bioassay and expressed as molecules/cpm (38).
Radioreceptor Binding Assays-Equilibrium receptor binding of 125 I-IL-2 and 125 I-T51P was performed on activated PBLs after 3 days of OKT3 stimulation as described (8). Thus, the activated PBLs were washed with and preincubated in the metabolic inhibitors 2-deoxy-Dglucose (50 mM) and NaN 3 (15 mM) in phosphate-buffered saline, pH 7.0, containing BSA to reduce ligand internalization (39). Serial 2-fold dilutions of labeled IL-2 or T51P (beginning at concentrations 10-fold greater than the approximate K d value) were added to the cell suspensions (2 ϫ 10 6 cells) and incubated for up to 90 min in the same buffer. Nonspecific binding, assessed by addition of a 400-fold molar excess of unlabeled ligand, was subtracted from all data points. Equilibrium dissociation constants were determined by Scatchard analysis (40). The results presented are the average of four replicate determinations repeated in two to four independent experiments.
Enumeration of Cell-surface Receptors after Internalization-To examine the effect of internalization on high affinity IL-2R sites, day 3 activated PBLs were treated with unlabeled IL-2 or T51P (at their K d concentrations) in regular supplemented RPMI 1640 medium at 37°C for 45 min. As a control for no internalization and for incomplete acid-induced dissociation, an equivalent number of cells were incubated with unlabeled T51P at its K d concentration at 4°C for 45 min. Cells were then cooled to 4°C and washed twice with ice-cold supplemented RPMI 1640 medium and pelleted. To dissociate surface receptor-bound ligand, the cell pellets were then resuspended in low pH buffer (10 mM citrate, 0.14 M NaCl, and 50 g/ml BSA, pH 4) and incubated for 3 min at 4°C (22). The acid wash was repeated, and the cells were washed twice with cold supplemented RPMI 1640 medium. The number of cell-surface receptors remaining were then determined via radioreceptor binding at 4°C using 125 I-IL-2 followed by Scatchard analysis as described above.
Direct Internalization of 125 I-Labeled IL-2 and T51P-The binding and internalization of 125 I-labeled IL-2 and T51P were performed on PBLs expressing high affinity receptor sites as described (22). The cells were first incubated for 5 min at 37°C in supplemented RPMI 1640 medium containing 100 M chloroquine, a lysosomotropic agent that prevents degradation of internalized proteins (16,41). 125 I-IL-2 or 125 I-T51P was then added at a final concentration equaling ϳ10 times the K d value (160 pM and 1.3 nM for IL-2 and T51P, respectively) and incubated for 30 min at 4°C. The cells were then washed twice with ice-cold supplemented RPMI 1640 medium to remove unbound ligand. Internalization was initiated by resuspending cells (10 7 cells/ml) in prewarmed (37°C) supplemented RPMI 1640 medium containing 100 M chloroquine. At selected time intervals, 100-l aliquots of the cell suspension were removed and diluted in 1.2 ml of ice-cold supplemented RPMI 1640 medium. The cells were pelleted, and the radioactivity in the supernatant was measured to determine the amount of 125 I-protein that had dissociated from the receptors. The cell pellet was then resuspended in 200 l of 10 mM citrate, pH 4, containing 0.14 M NaCl and 50 g/ml BSA. After incubation at 4°C for 3 min, the cells were centrifuged through a 400-l layer of silicone/paraffin oil. The radioactivity in the cell pellet and that in the supernatant above the oil layer were measured to determine the level of pH 4-resistant (internalized) and pH 4-sensitive (cell surface-bound) protein, respectively.
Ligand Recycling Assay-To determine the extent of recycling of internalized ligand, a modification of the assay reported by French et al. (42) was employed. Day 3 activated PBLs were washed three times with supplemented RPMI 1640 medium and resuspended at 1 ϫ 10 7 cells/ml in the same medium containing 125 I-IL-2 (48 pM; specific activity ϭ 1,215,000 molecules/cpm) or 125 I-T51P (360 pM; specific activity ϭ 201,000 molecules/cpm). A second series of samples were prepared and kept at 4°C throughout the entire assay (non-internalizing conditions) to control for release of ligand bound nonspecifically to the cell surface. Since a significantly greater amount of total radioactivity was employed in the case of T51P, an equivalent sample was prepared containing a large excess of unlabeled ligand (4.8 nM) to determine nonspecific binding. All of the cell samples, except the two internalization-negative controls, were incubated at 37°C for 1 h to allow internalization. All of the samples were cooled to 4°C; washed twice with ice-cold RPMI 1640 medium; and then washed twice with ice-cold 10 mM citrate, pH 4, containing 0.14 M NaCl and 50 g/ml BSA (after a 3-min incubation) to remove any surface receptor-bound ligand. After two additional washes with ice-cold supplemented RPMI 1640 medium, the cells were resuspended in the same medium at 37°C containing an excess of unlabeled ligand (either 1.6 nM IL-2 or 12 nM T51P) to prevent rebinding of released labeled ligand. At selected time points, 500-l aliquots (5 ϫ 10 6 cells) were removed to Eppendorf tubes. and the cells were pelleted. The supernatant was collected, and the cell pellet was resuspended in 500 l of ice-cold RPMI 1640 medium and pelleted again. The supernatants were combined and counted by solid scintillation to determine the amount of radioactivity released into the medium. The cell pellet was counted to determine the remaining cell-associated radioactivity.
To determine the extent of degradation of the labeled ligand released after internalization, 500-l aliquots of supernatants at selected time points were applied to Sephadex G-10 columns (0.8 ϫ 20 cm) preequilibrated in 0.2 M NaPO 4 , pH 7.0, containing 50 g/ml BSA. The columns were eluted with the same buffer, and 500-l fractions were collected and counted. The columns were precalibrated with 125 I-IL-2 to determine the elution volume of undegraded ligand.
Receptor Dissociation Kinetics-The receptor dissociation kinetics of 125 I-labeled IL-2 and T51P were examined on day 3 activated PBLs expressing high affinity receptor sites as described (8). Briefly, the cells were preincubated with metabolic internalization inhibitors as described above. All dilutions and subsequent manipulations were made with the same inhibitor buffer used in pretreating the cells. Cells were then incubated in the presence of 125 I-IL-2 or 125 I-T51P at concentrations ϳ10 times greater than their K d values (160 pM and 1.2 nM for IL-2 and T51P, respectively) for 30 min at 37°C. Dissociation was initiated by addition of a 1000-fold molar excess of unlabeled ligand. Aliquots (100 l; 2 ϫ 10 6 cells) were removed at selected intervals and centrifuged through an oil layer to determine the cell-associated and cell-free radioactivities. Nonspecific binding, determined separately by addition of excess unlabeled ligand during the initial incubation, was subtracted from each determination. Results are the average of two independent experiments, each carried out in triplicate. Dissociation rate constants (k off ) were determined from the slopes of lines generated by plotting the natural logarithm of (specific bound molecules/cell)/(maximum specific bound molecules/cell) versus time. The dissociation half-life was determined from the relationship t1 ⁄2 ϭ 0.693/k off , where t1 ⁄2 is the dissociation half-life and k off is the dissociation rate constant (43).
Surface Plasmon Resonance Analysis of Protein Binding to IL-2 Receptor Complexes-Surface plasmon resonance (SPR) analysis of IL-2 and T51P binding to an immobilized heteromeric IL-2 receptor ␣⅐␤ pseudo high affinity complex (IL-2R␣␤⅐cc) was performed as described (44). SPR instrumentation (BIAcore TM ), CM5 sensor chips, and aminecoupling reagents containing N-hydroxysuccinimide, N-ethyl-NЈ-(3-diethylaminopropyl)carbodiimide, and ethanolamine HCl were obtained from Pharmacia Biotech Inc. Purified IL-2R␣␤⅐cc complex was diluted to a concentration of Ϸ30 nM in NaOAc buffer (10 mM), pH 4.5, and coupled to the dextran-modified gold surface of a CM5 sensor chip using the manufacturer's amine coupling chemistry as described in the BIAcore TM systems manual. Surfaces densities of 600-2000 resonance units of IL-2R␣␤⅐cc complex were employed, and regeneration was achieved by injection of 10 mM HCl (4 l). Prior to SPR analysis, stock solutions of IL-2 and T51P were dialyzed against phosphate buffer (10 mM sodium phosphate, pH 7.4, and 150 mM NaCl), and the protein concentrations were determined from A 280 nm values (34). The samples were then diluted to the desired concentrations in analysis buffer (10 mM sodium phosphate, pH 7.4, 150 mM NaCl, and 0.005% surfactant P-20) containing 100 g/ml BSA. Five to eight serial dilutions of each protein were injected over the IL-2R␣␤⅐cc surfaces at a flow rate of 8 l/min. Sensorgrams were recorded and normalized to a base line of 0 resonance unit. Equivalent volumes of each protein dilution were also injected over an activated and ethanolamine-quenched surface for subtraction of bulk refractive index background. For the determination of dissociation rate constants, analysis buffer containing 1 M IL-2R␣␤⅐cc complex was injected during the dissociation phase to minimize rebinding to the biosensor surface. Sensorgrams were analyzed by nonlinear least-squares curve fitting using BIAevaluation 2.1 software (Pharmacia Biotech Inc.) to yield the apparent kinetic rate constants as described (44).

RESULTS AND DISCUSSION
We have previously reported point mutations in interleukin-2 analogs that result in apparent signaling abnormalities (34). One of these analogs contains a point mutation at position 51 (Pro for Thr) that produces an enhancement of the biological response. Although this protein has ϳ10-fold lower receptor binding affinity, it expresses nearly wild-type activity and is thus able to generate a significantly greater biological response than the protein at equivalent receptor occupancy. The objective of this study was to investigate the mechanism of the apparent facilitation in response observed for the analog T51P.
Protein Preparation-The cDNA encoding the IL-2 analogs was prepared via cassette mutagenesis on a synthetic IL-2 gene as described (33). Expression, refolding, and purification of the proteins were carried out using techniques that had been optimized for IL-2 (45). Purity and aggregation state of all of the analogs were monitored by reverse-phase and size-exclusion HPLC. In all cases, the products were determined to be homogeneous and monomeric at the concentrations employed for the bioassays and receptor binding studies (34). Periodic examination of the aggregation state of stock solutions was performed to check for self-association. For all of the proteins examined, no detectable aggregation was noted upon storage.
Receptor Binding and Internalization-The apparent anomaly in biological response of T51P when compared with IL-2 was based on a direct comparison of bioactivity (EC 50 ) and receptor affinity (K d ) values of the analog and IL-2 (34). The K d values were determined from competitive binding assays under steady-state conditions established for IL-2. The validity of the observation is critically dependent upon the accuracy of the values determined for EC 50 and K d . Since this study focused on the relative rates of ligand internalization, it was necessary to radiolabel T51P for these measurements. Employing this labeled ligand in direct equilibrium saturation binding assays should provide independent confirmation of previously determined K d values. Therefore, we radiolabeled the T51P analog, determined its bioactivity and specific radioactivity, and carried out equilibrium binding analysis on normal human T-cells. The Scatchard analysis of IL-2 compared with the analog T51P is shown in Fig. 1. The K d values for IL-2 and the analog T51P determined from these data were 13.4 Ϯ 1.4 and 112.3 Ϯ 1.8 pM, respectively. These values are consistent with K d values previously determined from competitive binding assays (16 and 130 pM) (34) and further support a facilitated response for T51P.
Upon confirmation of K d values, we investigated the relative internalization rates of the two ligands to help elucidate the possible mechanism of the observed enhancement of the biological response for T51P. Three parameters determine T-cell responsiveness in this system: the concentration of IL-2, the IL-2R density on cells, and the duration of the IL-2R interaction (32). The density of surface receptors is determined by their rate of synthesis, internalization, and recycling. Internalization, along with ligand dissociation, also determines duration of occupancy of the IL-2R. Although the unliganded IL-2R also undergoes internalization, the rate is rather slow (t1 ⁄2 Ϸ 1-2 h) (19,20,31). In the presence of IL-2, the IL-2⅐IL-2R complex is rapidly internalized (t1 ⁄2 ϭ 15 min), resulting in a down-regulation of high affinity IL-2 receptors that dampens IL-2 responsiveness in T-cells (31). A response that is enhanced compared with the wild-type ligand could result from longer receptor occupancy due to a reduced rate of internalization of the ligand-bound receptor complex. Therefore, we determined the internalization rate of the T51P analog and compared it with IL-2. The internalization rates of both 125 I-IL-2 and 125 I-T51P were examined on day 3 PBLs prepared as described under "Materials and Methods." As shown in Fig. 2A, 125 I-IL-2 initially bound to the cell surface was rapidly internalized at 37°C. The t1 ⁄2 determined from these data was ϳ17 min, a value that compares favorably with the t1 ⁄2 of ϳ15 min reported by others (22,25,29). In contrast, little internalization could be detected for T51P over the duration of the assay (Fig. 2B). In comparison with IL-2, where little dissociation from surface receptors could be detected, the majority of 125 I-T51P initially receptor-bound was found in the dissociated fraction. Therefore, for T51P, unlike IL-2, the rate of dissociation from the high affinity receptor greatly exceeds the rate of internalization.
This apparent deficiency was further documented by enumeration of cell-surface receptors by Scatchard analysis after internalization induced by unlabeled ligand (Fig. 3). When cells were treated at 37°C with unlabeled IL-2 at its K d concentration (16 pM) for 45 min, the high affinity receptor sites were reduced by ϳ50% (from 1717 to 834 sites/cell), a result that is comparable to previous studies employing the same assay (31). However, cells similarly treated with T51P at its K d concentration (130 pM) displayed no significant reduction in high affinity receptors (1580 sites/cell, or 92% IL-2R retained) when compared with cells treated with 130 pM T51P at 4°C (to control for acid dissociation and no internalization). Of note, the K d values determined in this equilibrium binding assay were greater than expected for the high affinity receptor (K d ϭ 217, 154, and 185 pM for control cells, IL-2-treated cells, and T51P-treated cell, respectively). These values are typical of the dissociation constant reported for the "pseudo high affinity" IL-2R ␣⅐␤ site (46 -49). In these assays, acid washes were employed to remove non-internalized receptor-bound ligand. In independent expression experiments, we have consistently noted that the soluble IL-2R ␥ c -subunit ectodomain is highly unstable at low pH and is irreversibly denatured. 2  ing the IL-2R ␣⅐␤ pseudo high affinity site as a measure of non-internalized high affinity receptors.
Kinetics of Ligand Dissociation-The observed deficiency of internalization could result from the failure to initiate this process due to conformational effects. Divergent conformational requirements for internalization and signaling have been proposed for other ligand-receptor systems (50). Consistent with this model is the finding that T51P possesses altered tertiary structure as revealed in the near-UV CD spectrum (34). Alternatively, reduced internalization could simply be due to altered dissociation kinetics. A faster off-rate would cause a greater fraction of receptor-bound ligand to dissociate rather than internalize. This phenomenon has been reported for murine IL-2 analogs (29). To examine this possibility, kinetic dissociation rate assays were performed on day 3 activated human PBLs. The results of these experiments are shown in Fig. 4. The dissociation rate constants (k off ) calculated from these data were 2.4 ϫ 10 Ϫ4 s Ϫ1 for IL-2 and 3.6 ϫ 10 Ϫ4 s Ϫ1 for T51P. The dissociation half-lives calculated from the dissociation rate constants were 48.1 and 32.7 min, respectively. The values obtained for IL-2 are similar to those previously reported (8,37), and although the dissociation rate of the T51P analog is slightly faster than that of IL-2, the difference cannot account for the observed internalization deficiency. Furthermore, an increased dissociation rate would reduce the duration of receptor occupancy, resulting in a decrease in biological response (32) and not the enhancement observed for T51P.
SPR Analysis of Ligand Binding to IL-2R␣␤⅐cc Complexes-The affinity of T51P for the high affinity receptor was ϳ10-fold weaker than that of IL-2 as determined in both competitive and equilibrium binding assays, yet the dissociation rate was only slightly faster compared with IL-2 (30% increase). Therefore, a commensurate decrease in association rate would be expected for T51P in order to explain the differences observed in the equilibrium dissociation constants. To determine the influence of the T51P mutation on the association rate, we employed SPR (BIAcore TM ) analysis to examine the on-rate kinetics of this analog to an immobilized pseudo high affinity heteromeric receptor complex (IL-2R␣␤⅐cc). We have previously characterized the binding of IL-2 to this biosensor surface and have demonstrated that the kinetic and equilibrium binding constants are similar to those reported for the cell-surface pseudo high affinity receptor, the preformed complex that serves to capture ligand (44). Representative sensorgrams for the binding of IL-2 and T51P to this IL-2R␣␤⅐cc biosensor surface are shown in Fig. 5 (A and B, respectively). The association rate constants (k on ) determined from the slopes of the plots of k s versus C (Fig. 5C) were 3.93 Ϯ 0.05 ϫ 10 6 M Ϫ1 s Ϫ1 for IL-2 and 1.23 Ϯ 0.03 ϫ 10 6 M Ϫ1 s Ϫ1 for T51P. These values are extremely fast and are certainly limited by mass transport to the surface of the biosensor. Thus, the actual on-rate constants are, in all likelihood, more rapid. Nevertheless, these results indicate that the on-rate constant for T51P is at least 3-fold slower than that for IL-2 to this receptor complex. In contrast, the off-rate constants (k off ) from the ␣⅐␤ site in this complex were similar  4. Receptor dissociation kinetics. Shown is the dissociation of 125 I-labeled IL-2 (q) and the T51P analog (E) from high affinity receptor sites on activated human T-lymphocytes. Each point represents the mean of triplicate determinations performed as described under "Materials and Methods." The specific binding at t ϭ 0 (HR 0 ) was determined just prior to the initiation of dissociation, whereas subsequent determinations of bound receptor (HR) were performed at the intervals indicated. The respective dissociation rate constants (k off ) for IL-2 and T51P, representing the mean of two independent assays, were 2.4 ϫ 10 Ϫ4 and 3.6 ϫ 10 Ϫ4 s Ϫ1 . The corresponding t1 ⁄2 values were 48.1 and 32.7 min, respectively.
The kinetic constants for IL-2 determined in this study were very similar to those previously reported (44), demonstrating the reproducibility of SPR analysis for this system. Although the k on values were not determined for cell-surface receptors and are underestimates of the actual on-rate constants, the slower association rate constant observed for T51P is consistent with its lower K d for the high affinity receptor. Since the IL-2R ␣⅐␤ pseudo high affinity complex captures ligand on the cell surface (8) prior to the recruitment of the ␥-subunit to form the high affinity receptor complex, the on-rate to this complex should reflect the on-rate to cells expressing the high affinity receptor. Thus, the slight increase in the dissociation rate constant from the high affinity site observed for T51P combined with at least a 3-fold decrease in association rate constant measured to the pseudo high affinity complex explain the loss of affinity for the high affinity receptor for this analog.
Biological Activity-The hypothesis that an internalization deficiency contributes to the observed facilitated response was also supported by T-lymphocyte proliferation bioassays. Internalization ultimately results in degradation or recycling of the ligand (32). An analog that is only slowly internalized should, therefore, suffer less degradation and remain functional for a greater period of time. To test this hypothesis, we carried out biological assays on the same population of normal human T-lymphocytes employed in receptor binding assays. As shown in Fig. 6, 2 days after addition of an approximate EC 50 concentration of each ligand, IL-2 and T51P produced ϳ50% of the maximum biological response as determined in regular full titration bioassays carried out in parallel. However, between days 2.5 and 10.5, T51P displayed a greater response than IL-2. In addition, the response generated by IL-2 reached background sooner (7.5 days). A similar result was observed when either day 3 EC 30 or day 3 EC 80 concentrations of each ligand were chosen (data not shown). It should be noted that the decrease in proliferative response is not solely a function of the concentration of undegraded ligand. The loss of high affinity receptors, even in the presence of saturating concentrations of IL-2, ultimately modulates the biological response of IL-2 on activated T-cells. In fact, others have found that the rate of disappearance of the high affinity receptor in the presence of IL-2 results in a 75% loss of these sites by day 8 after activation (51). The rate of loss of receptor expression reported in that study is similar to the rate of decay of the IL-2 response ob-  (44). Apparent association rate constants as determined from the slopes of the plots (C) of k s versus C (where k s ϭ Ϫk on C Ϫ k off ) were as follows: IL-2, k on ϭ 3.93 Ϯ 0.05 ϫ 10 6 M Ϫ1 s Ϫ1 ; and T51P, k on ϭ 1.23 Ϯ 0.03 ϫ 10 6 M Ϫ1 s Ϫ1 . All data were collected as described under "Materials and Methods" on BIAcore TM instrumentation, and data analysis was performed using BIAevaluation 2.1 software.
FIG. 6. Human PBL bioassays. Biological responses generated using approximate EC 50 concentrations of IL-2 (empty bars) and the T51P analog (filled bars) were measured on activated human T-lymphocytes. Activities were measured by [ 3 H]thymidine incorporation at the times indicated. Each bar represents the mean of five determinations. The maximum responses (8889 and 8890 cpm for IL-2 and T51P, respectively) were determined from full titration bioassays carried out on day 2. served in Fig. 6.
Since cells with greater receptor density proliferate sooner (52), reduced internalization for T51P and the associated longer receptor occupancy and duration of signaling could potentially result in a greater absolute maximum response compared with IL-2. Activated T-cells express a normal distribution of high affinity receptors (52); thus, a small fraction of these cells may express a density of receptors sufficient to cause proliferation in response to T51P, but not IL-2, at the time of harvesting. In fact, in eight independent bioassays, we observed a significantly greater maximum response (p Ͻ 0.05) ranging from 103 to 160% for T51P. IL-2 never exhibited a greater response than T51P. Therefore, the fact that this analog can generate a biological response of longer duration and, in some cases, greater magnitude is consistent with the reduced internalization model.
Ligand Recycling-An alternative explanation for the extended bioactivity observed for T51P could be an altered rate of degradation after internalization such that a greater fraction of ligand is recycled back into the medium when compared with IL-2 and thus would be available for repeated receptor interactions. A point mutation (Y31G) in epidermal growth factor was reported to result in increased potency after a single administration despite reduced receptor affinity, an observation not unlike what we have determined for T51P. The origin of this effect was shown to be increased ligand recycling (rather than degradation) when compared with wild-type epidermal growth factor (53). To examine this possibility, we carried out a similar assay designed to determine the extent of ligand recycling. Labeled ligands were incubated with day 3 activated T-cells for 1 h to allow sufficient time for receptor-mediated internalization. The cells were then cooled to 4°C and washed with cold media, and then surface receptor-bound ligand was removed by treatment with low pH buffer as in the previous internalization assay. The cells were then incubated at 37°C in media containing excess unlabeled ligand to prevent rebinding of any released label. Fig. 7A depicts the results of this assay. Although labeled IL-2 was released back into the medium in a time-dependent fashion, little 125 I-T51P was detected in the medium over the same time period. The amount of radioactivity released by the 125 I-T51P-treated cells was similar to that released by equivalently treated cells kept at 4°C during the initial incubation to prevent internalization and by cells treated with labeled T51P plus a 100-fold molar excess of unlabeled ligand to determine nonspecific binding. The diminished release of labeled T51P would be anticipated if less ligand were internalized; therefore, these results are consistent with a reduced internalization mechanism.
In addition, gel filtration analysis of the medium containing released label from the 125 I-IL-2-treated cells (Fig. 7B) indicates that the majority of radioactivity is represented by recycled intact ligand. Previous studies have suggested that IL-2 is degraded after receptor-mediated internalization (54, 55). These studies, however, did not attempt to identify recycled ligand, and little degradation was noted during the first hour after IL-2 treatment.
The results of three independent assays employed here indicate that the IL-2 analog T51P is deficient in inducing internalization of high affinity receptors. Consequently, although it possesses ϳ10-fold lower binding affinity, T51P exhibits close to wild-type biological activity. The dissociation rate constant for this analog was only slightly faster than that for IL-2. This study supports a model in which receptor signaling and internalization are separable events. The dissociation of IL-2 receptor signaling from internalization has been reported previously (30). Deletions in the cytoplasmic domain of the IL-2R ␤-sub-unit resulted in the loss of signaling in response to IL-2 without influencing internalization. Conversely, in another system, mutations in the epidermal growth factor receptor abrogated internalization, but not signaling ability, leading to uncontrolled proliferation in the presence of ligand (56). To our knowledge, the IL-2 analog T51P reported in this study is the first example of a ligand-induced alteration of receptor internalization that is not simply a result of an increased off-rate. This observation suggests that the subtle differences in tertiary structure that exist between T51P and IL-2 are transmitted to the receptor during ligand-induced cross-linking, causing a less efficient internalization signal.
The influence of the T51P mutation was not only reflected in reduced internalization, but was also observable in bioassays. This analog maintained T-cell proliferation for a longer period than IL-2 and often produced a greater maximum response. This effect is a combination of greater cell-surface receptor density and longer duration of receptor occupancy, both resulting from reduced internalization. In addition, prolonged survival of active ligand may also contribute. As such, this analog could provide an approach to the preparation of cytokine analogs with potentiated agonist and antagonist properties that is separate from, but complementary to, receptor affinity maturation.