IL-2 is critical for peripheral tolerance mediated by regulatory T (Treg) cells, which represent an obstacle for effective cancer immunotherapy. Although IL-2 is important for effector (E) T cell function, it has been hypothesized that therapies blocking IL-2 signals weaken Treg cell activity, promoting immune responses. This hypothesis has been partially tested using anti–IL-2 or anti–IL-2R Abs with antitumor effects that cannot be exclusively attributed to lack of IL-2 signaling in vivo. In this work, we pursued an alternative strategy to block IL-2 signaling in vivo, taking advantage of the trimeric structure of the IL-2R. We designed an IL-2 mutant that conserves the capacity to bind to the αβ-chains of the IL-2R but not to the γc-chain, thus having a reduced signaling capacity. We show our IL-2 mutein inhibits IL-2 Treg cell–dependent differentiation and expansion. Moreover, treatment with IL-2 mutein reduces Treg cell numbers and impairs tumor growth in mice. A mathematical model was used to better understand the effect of the mutein on Treg and E T cells, suggesting suitable strategies to improve its design. Our results show that it is enough to transiently inhibit IL-2 signaling to bias E and Treg cell balance in vivo toward immunity.
This article is featured in In This Issue, p.3315
Interleukin-2 is a 15 kDa cytokine produced by activated CD4+ and to a lesser extent by activated CD8+ T lymphocytes. The IL-2R is formed by different combinations of three subunits and exists in two functional configurations. The β (CD122)- and γc (CD132)-chains are responsible for signaling and together form the dimeric intermediate affinity receptor highly expressed on memory (M) CD8+CD44hi and NK cells. The α-chain (CD25) is expressed constitutively on CD4+Foxp3+ regulatory T (Treg) cells and transiently on activated CD4+ and CD8+ T cells, and together with β− and γc-chains, they form the high affinity receptor with 100-fold higher affinity for IL-2 (1).
IL-2 was initially linked with the activation of the immune response because of its capacity to induce T cell growth in vitro (2). However, the demonstration of the crucial role of IL-2 in the generation and homeostasis of Treg cells has underscored the main nonredundant function of IL-2 in vivo (3). Treg cells are responsible for peripheral tolerance and are able to suppress the immune response against self- or non–self-antigens (4). Therefore, it becomes natural to hypothesize that blocking IL-2 signal will weaken Treg cell activity and promote immune responses.
One strategy to experimentally test the latter hypothesis is the use of Abs anti–IL-2 to block IL-2 signaling in vivo (5). However, results obtained with this strategy are nonconclusive because available mAbs have shown different effects as a consequence of blocking different regions of the IL-2 (6, 7). They happen to form immune complexes with the IL-2 in vivo, which preferentially redirect the IL-2 signal to different T cell subsets. A second strategy is the administration of anti–IL-2R Abs (8). However, results obtained are again nonconclusive because they might derive from directly lysing IL-2R+ T cells by the mAbs rather than just blocking IL-2 signaling in vivo.
In this work, we explore a novel approach for blocking IL-2 signal and targeting Treg cells. We designed, obtained, and characterized a new IL-2 mutein with antagonistic properties. This IL-2 mutein is able to bind to the IL-2R on the cell surface but not to the γc-chain. In vitro, the mutein inhibits the IL-2–driven T cell proliferation as well as Treg cell proliferation, differentiation, and expansion. In vivo treatment of mice with the mutein reduces Treg cell numbers and impairs tumor growth. A mathematical model was used to better understand the effect of the mutein on Treg and effector (E) T cells, suggesting suitable strategies to improve their design. Overall, our results show that it is enough to transiently inhibit IL-2 signaling to bias Th and Treg cell balance in vivo toward immunity. Our IL-2 antagonist, or an improved version, could be useful for cancer immunotherapy based on Treg cell inhibition.
Materials and Methods
Design and production of IL-2 muteins
Several IL-2 muteins capable to antagonize the IL-2 signal in vivo through all configurations of the IL-2R were designed. The design was based on the crystal structure of human IL-2/IL-2R (trimeric form) (9, 10). Evolutionary conservation of the residues in the interface with the γc-chain (residues located at <5 Ǻ apart from γc-chain) was evaluated using ClustalW software. Residues highly conserved in more than 20 species were selected. Rosetta and FoldX softwares were then used to evaluate the energetic contributions of conserved positions to the interaction of the IL-2 with the γc-chain. Five positions, Q126, Q13, L18, Q22, I129, and S130, were selected as the most relevant to attempt mutein construction. Residues with shorter side chains or with different chemical natures mutated the selected positions.
The synthetic genes encoding the IL-2 muteins as well as the human wild type (wt) IL-2 were obtained from Geneart, Germany, and were cloned into the commercial vector pET28a, from Novagen. Escherichia coli cells BL21(DE3) strain were transformed with the expression plasmids using the manufacturer’s protocol. Transformed cells were allowed to grow in 200 ml of lysogeny broth medium, and the protein expression was induced with IPTG 4 mM. Isolation of inclusion bodies and further purification were carried out following the procedure described by Moya et al. (11) for the wtIL-2.
Abs and flow cytometry
All fluorochrome-conjugated mAbs used were from eBioscience unless otherwise stated. For mouse experiments, the following were used: FITC-conjugated anti-CD3 (145-2C11), PECy5.5-conjugated anti-CD4 (L3T4), PE-conjugated anti-Foxp3 (NRRF-30), PE-conjugated anti-CD25 (3C7), PerCP Cy5.5–conjugated anti-CD4, PE Cy7–conjugated anti-CD25, APC-conjugated anti-Foxp3, FITC-conjugated anti–IFN-γ, APC-conjugated anti-CD8, Alexa Fluor 647–conjugated anti-ki67 from BioLegend, and PE-conjugated anti–His-tag from R&D Systems. Intracellular Foxp3 staining sets were purchased from eBioscience. Anti–phospho-STAT5 mAb (D47E7) was obtained from Cell Signaling Technology. Fc receptor binding was blocked by preincubating cells with Fc receptor block reagent from BD Biosciences. For experiments with human lymphocytes, the following were used: FITC- conjugated anti–human-CD3, Alexa Fluor 700–conjugated anti–human-CD4, PE-conjugated anti–human-CD127, PECy7-conjugated anti–human-CD25, PE-CF594–conjugated anti–human-FOXP3. Samples were measured using a FACScan or an LSRFortessa (BD Biosciences) flow cytometer and analyzed using FlowJo software (Tree Star). The mAbs anti-CD3, anti-CD28, and anti–IL-2 (S4B6) used for in vitro stimulation were purchased from BD Biosciences, and the anti-human IL-2(5334) was obtained from R&D Systems.
Cell lines and culture conditions
The murine CTLL-2 and the human Kit225 T cell lines were cultured in RPMI 1640 (Life Technologies) supplemented with 10% heat-inactivated FBS, 50 IU/ml IL-2, 2 mM l-glutamine, 50 U/ml penicillin, and 50 μg/ml streptomycin. Murine MB16F0 melanoma and 4T1 cells were maintained in DMEM F12 (Life Technologies) supplemented with 10% heat-inactivated FBS, 2 mM l-glutamine, 50 U/ml penicillin, and 50 μg/ml streptomycin. All cells were maintained at 37°C under a humidified 5% CO2 atmosphere. For in vivo experiments, tumor cells were harvested using trypsin/EDTA and resuspended in PBS.
Seven- to eight-week-old female C57BL/6 and BALB/c mice were obtained from The National Center for Laboratory Animal Breeding, Havana, Cuba. Foxp3gfp knockin mice (H-2b) were obtained from the University of Washington (Seattle, WA), bred, and maintained in specific pathogen-free conditions at the Instituto Gulbenkian de Ciencia in Oeiras, Portugal. Food and water were administered ad libitum. The experiments were performed according to the guidelines of the International Laboratory Animals Research using standardized procedures in the Centro de Inmunología Molecular, Cuba, or at the Instituto de Medicina Molecular, Portugal.
Cell suspensions and in vitro assays
Before proliferation assays, CTLL-2 cells were harvested by centrifugation, washed twice, and deprived of IL-2 for 5 h. Kit225 cells were deprived of IL-2 for 2 d. After that, 104 cells per well were incubated with serial dilution of either wtIL-2 or IL-2 mutein in RPMI 1640 and allowed to grow for 48 h. Then, 20 μl per well of Alamar blue dye (Invitrogen) was added, and plates were incubated 12 h for CTLL-2 and 6 h for Kit225. Finally, plates were read at 540 and 620 nm, and the proliferation curves were obtained following manufacturer’s recommendation.
For differential effect experiments, the E cells (CD4+Foxp3−) and Treg cells (CD4+Foxp3+) were purified from inguinal, mesenteric lymph node or spleen cell suspensions using the CD4+CD25+ Regulatory T Cell Isolation Kit, from Miltenyi Biotec. For STAT5 phosphorylation and Treg cell proliferation, the CD4+CD25+Foxp3+GFP+ cells were isolated from Foxp3−GFP knockin mice or from Foxp3hCD2 B6 by sorting using a high-speed sorting cytometer (FACSARIA, BD Biosciences). In both cases, purity was tested by flow cytometry. For the proliferation assays, Treg cells were labeled with 2.5 μM CellTrace Violet (CTV).
For the human Treg cell proliferation inhibition assay, PBMCs from healthy donors were cultured in the presence of 2 μg/ml of anti-human CD3 mAb (OKT-3) and either IL-2 mutein or anti-human IL-2 (5334) mAbs at different concentrations. After 120 h in culture, the level of Treg cell proliferation was assessed by flow cytometry based on Ki-67 expression.
For in vitro inhibition assay, total LN population or CD4+Foxp3− cells were labeled with 1 μM CFSE, followed by incubation in 96-well plates previously coated with 3 μg/ml anti-CD3 mAb (2C11) in the presence of different concentrations of IL-2 mutein or anti–IL-2 (S4B6) mAb. For coproliferation assays with purified populations, 5 × 104 labeled E cells were coincubated with different numbers of Treg cells and 105 irradiated splenocytes used as APCs with the addition of 0.5 μg/ml anti-CD3 mAb. After 72 h, the E cell proliferation was assessed by flow cytometry based on CFSE dilution.
For Treg cell conversion assay, CTV-labeled CD4+Foxp3− cells were cultured in the presence of 5 ng/ml of TGF-β, 3 μg/ml of plate-bound anti-CD3 mAb, 3 μg/ml of soluble anti-CD28 mAb, and different concentrations of or either wtIL-2 or IL-2 mutein.
Antitumor activity in vivo
C57BL/6 or BALB/c mice received 200 μg of IL-2 mutein, 200 μg of anti-CD25 (PC61) mAb, or PBS i.p. during 5 d. On day 6, mice received 2 × 104 of either MB16F10 or 4T1 cells on the right flank, and the tumor growth was followed.
To assess the effect of treatments on tumor-infiltrating lymphocytes, a 4T1 tumor model was used in the same conditions and 2 wk after tumor implantation, total lymphocytes from tumor infiltrates and tumor-draining lymph node (TDLN) were isolated, counted, and stained for analysis by flow cytometry of the accumulation of CD8+ T cells and Treg cells. To assess IFN-γ production by CD8+T cells, total lymphocytes from tumor and TDLN were incubated for 4 h with 50 ng/ml PMA, 0.5 μg/ml ionomycin, brefeldin A, and GolgiStop. In vivo proliferation of Treg cells was assessed by the expression of Ki-67.
For statistical analysis, we used the GraphPad Prism 4.0 software. For comparison of tumor volume curves, a two-way ANOVA was performed. For in vivo Treg cell accumulation and measurement of tumor infiltrates, a parametric ANOVA followed by Bonferroni’s multiple comparison test was applied.
Mathematical model to simulate the effect of IL-2 muteins on T lymphocytes
The mathematical model used in this paper was developed and calibrated to describe the interaction between IL-2 and CD4+CD25+Foxp3+ Treg cells, E CD4+ T cells, M CD8+ T cells, and M NK cells (12). In particular, it takes into account our current knowledge of the differential and dynamic expression of α-, β-, and γc-chains of the IL-2R on these different cell populations. IL-2 muteins (no-γ muteins) are modeled as soluble molecules, which share all the properties of wtIL-2 but whose conjugation to the β-chain of the IL-2R does not mobilize the γc-chain (CD132), leading to no effective signal. We also study in silico, no-γ muteins that also have increased affinity for α-chain of IL-2R (CD25), the affinity for CD25 reduced to zero, increased affinity for the β-chain of IL-2R (CD122), or the affinity for CD122 reduced to zero.
Treatments are always simulated as continuous infusion of the desired molecule in the blood for a defined period. Two parameters always control treatment application: the dosage, which sets up the total amount of IL-2 mutein infused per day and the treatment duration, which sets the time period in which continuous infusion is maintained. The majority of the model parameters were fixed to values directly taken or derived from available independent experimental data (see Appendix D in Ref. 12); just a few parameters remain unknown, and their influence in result was explored inside a range of biologically reasonable values, and none of them change the qualitative results of the simulations.
IL-2 mutein design and production
Different IL-2 muteins were designed to antagonize the IL-2 signal in vivo as tools for Treg cell inhibition. The design was based on the crystal structure of human IL-2/IL-2R (trimeric form) (9, 10). The approach was to mutate the interface between IL-2 and the γc-chain, trying to keep intact the IL-2 interface with the α- and β-chains of the receptor. Such design shall theoretically guarantee that muteins retain the ability to bind IL-2R similar to wtIL-2 but with no signal transduction. Three mutein configurations with three or four mutations each were selected based on their predicted reduction of the energy of interaction with γc-chain. Fig. 1 shows the mutated residues for the mutein M1 located on the three-dimensional IL-2 structure. The mutations corresponding to the three muteins were initially assayed, and the corresponding changes on KD are listed in Table I. As observed, the theoretical calculation of the KDs suggests a high reduction of the interaction with the human and mouse receptors for the three designed muteins and not high differences among the different muteins.
|Label .||Mutations .||KD mut/KD wt Human IL-2R .||KD mut/KD wt Mouse IL-2R .|
|M1||Q22V, Q126A, I129D, S130G||5 × 103||1.5 × 103|
|M2||L18N, Q126Y, S130R||1 × 103||2 × 103|
|M3||Q13Y, Q126Y, I129D, S130R||2 × 103||1 × 103|
|Label .||Mutations .||KD mut/KD wt Human IL-2R .||KD mut/KD wt Mouse IL-2R .|
|M1||Q22V, Q126A, I129D, S130G||5 × 103||1.5 × 103|
|M2||L18N, Q126Y, S130R||1 × 103||2 × 103|
|M3||Q13Y, Q126Y, I129D, S130R||2 × 103||1 × 103|
IL-2 muteins conserve binding capacity to the IL-2R on the cell surface
The recombinant proteins fused to a 6His target motif were expressed as inclusion bodies and were purified to homogeneity following the purification protocol described by Moya et al. (11) for human rIL-2 purification. An in vitro refolding process recovered the native structure and function of the expressed proteins during the purification procedure. To show the feasibility of the refolding, we tested whether the M1 mutein was able to bind to IL-2R on the cell surface. We used CTLL-2 and Kit225 T cell lines as well as freshly isolated murine lymphocytes. In all cases, 2 × 105 cells were incubated for 20 min on ice with a solution of 0.1 mg/ml of wtIL-2 or the M1 mutein diluted in FACS buffer. The cells were washed, and the bound molecules were detected with an anti–6His-tag–PE mAb. We found that the mutein was able to bind to freshly isolated lymphocytes and CTLL-2 and Kit225 cell surfaces in a similar way to wtIL-2, indicating that the mutations did not affect its global conformation (Fig. 2A).
IL-2 muteins have impaired capacity to induce IL-2–dependent T cell proliferation
We expected that the inserted point mutations were sufficient to abrogate the interaction with the γc-subunit and, as a consequence, the signaling capacity through the IL-2R. Therefore, the designed muteins were tested in a classical proliferation assay with the murine cell line CTLL-2 and the human line Kit225. Both cell lines are dependent on IL-2 for growth and constitutively express the trimeric high affinity form of the IL-2R. In both cases, the behavior was similar: none of the muteins was able to induce cell proliferation in concentrations near the wtIL-2 active range (Fig. 2B, 2C). However, at higher concentrations, the muteins were able to induce cell proliferation, albeit in concentrations 1000–3000-fold higher than wtIL-2 EC50. Therefore, in all T cell lines tested, the three IL-2 muteins have dramatically reduced IL-2R signaling, behaving as very weak IL-2 agonists. Our data show mutein M1 clearly has the lowest capacity to induce the CTLL-2 and Kit225 cell line proliferation; therefore, it was selected to scale up its production and to perform subsequent studies.
When considering the position of the mutated residues, the poor signaling capacity shown by mutein M1, and the conserved binding capacity to the cell surface, we can anticipate that the mutein has similar capacity to wtIL-2 to bind to the α- and β-chains but has decreased binding to the γc-chain.
IL-2 mutein M1 does not induce proliferation or STAT5 phosphorylation of Treg cells in vitro
To test the M1 mutein effect on Treg cells, CD4+Foxp3+GFP+ cells were purified from reporter mice. To assess the effect of M1 on Treg cell proliferation, the sorted cells were labeled with CTV and cultured for 96 h with 2 μg/ml plate-bound anti-CD3 mAb, 2 μg/ml soluble anti-CD28 mAb in the presence of different concentrations of the M1 mutein or wtIL-2. As expected, wtIL-2 induced the proliferation of Treg cells at concentrations near 1 ng/ml (Fig. 2D). On the contrary, the mutein M1 did not induce Treg cell proliferation even at 5 μg/ml.
After IL-2R engagement, three major intracellular signaling pathways are initiated: MAPK, PI3K, and STAT5, but these are differentially regulated in the Treg cell compartment. On Treg cells, IL-2 dependent STAT5 phosphorylation is the most important pathway inducing upregulation of Foxp3 and CD25 (13). We determined whether the mutein is able to induce STAT5 phosphorylation on isolated Treg cells. In line with the proliferation data, the presence of M1 mutein did not lead to STAT5 phosphorylation in any of the assayed concentrations (Fig. 2E).
IL-2 mutein M1 inhibits IL-2–driven T cell proliferation
The mutein M1 and the anti–IL-2 mAb S4B6 were compared regarding their capacity for inhibiting IL-2–driven T cell proliferation. Total lymph node cells were labeled with CFSE and stimulated for 72 h with soluble anti-CD3 in the presence of S4B6 mAb or the IL-2 mutein M1 (Fig. 3A). A high proliferation level was measured for stimulated CD4+ lymphocytes reaching 85%; this proliferation is induced by IL-2 that is produced after T cell activation. As expected, both the neutralizing anti–IL-2 S4B6 mAb and the M1 mutein reduced T cell proliferation. In addition, both the IL-2 mutein and the anti–IL-2 mAb induced cell death, as expected for activated T lymphocytes after IL-2 deprivation (Fig. 3B). The level of IL-2 produced in the supernatant of CD4+ T lymphocytes was 100 pg/ml, but in the presence of the mutein, the IL-2 concentration was reduced to 20 pg/ml. Therefore, the mutein behaves as an IL-2 antagonist inhibiting IL-2 production, T cell proliferation, and survival.
As the mutein was constructed based on the human wtIL-2 molecule, we were also able to test the effect on human lymphocytes. We stimulated total human PBMC isolated from a buffy coat, with anti-CD3 mAb (OKT3) and determined the mutein effect on T cell proliferation. Similar to the mouse experiment, the mutein inhibited human T CD4+Foxp3− proliferation measured as number of cells ki67+ (Fig. 3F).
The mutein inhibits Treg cell differentiation and expansion
We tested in vitro the ability of the M1 mutein to induce de novo Treg cell differentiation and expansion out of naive CD4+ T cells. FACS-purified CD4+CD25−Foxp3− cells were cultured in the presence of anti-CD3 and anti-CD28 mAbs, 5 ng/ml TGF-β, and different concentrations of the wt or the mutated IL-2. The wtIL-2, but not the mutein, was able to increase the frequency of CD4+Foxp3+ T cells recovered after 3 d of culture up to 84% (Fig. 3C). Moreover, Treg cells that recovered after treatment with wtIL-2 showed stronger signs of proliferation than those recovered after mutein treatment, as assessed by CTV dilution. For the higher mutein concentrations, the absolute number of Treg cells recovered was also decreased with respect to the control condition (Fig. 3D). These data suggest that the mutein is not only unable to substitute wtIL-2 to promote de novo differentiation, but it antagonizes the effect of endogenous IL-2, produced upon anti-CD3 stimulation of naive CD4+ T cells, hampering the differentiation and expansion of new Treg cells.
To test the IL-2 antagonistic capacity of the M1 mutein over Treg cells, mouse CD4+CD25+Foxp3+GFP+ Treg cells were isolated by flow cytometry and labeled with CTV. Then, sorted Treg cells were stimulated for 72 h with anti-CD3, anti-CD28, and 0.5 ng/ml of wtIL-2 in the presence of different concentrations of the mutein M1 or anti-CD25 mAb (PC61). The concentration of IL-2 present in the culture was selected from previous experiments, and it was able to induce Treg cell proliferation to the maximum level obtained in the experiment. We found that both the mutein and the anti-CD25 mAb inhibited Treg cell proliferation in a dose-dependent mode (Fig. 3E). In the case of the mutein, this inhibition means that the mutein was able to compete with the IL-2 available in the culture.
The ability of the mutein to inhibit human Treg cell proliferation in vitro was also assayed by flow cytometry; the mutein was able to totally inhibit the TCD4+Foxp3+ (Fig. 3G). In this case, the mutein was able to compete with the IL-2 produced on the culture after T cell stimulation.
The mutein is preferentially trapped by Treg cells releasing E cells from suppression
The IL-2 mutein was able to bind to the surface of cells expressing the αβγ form of the IL-2R as well as wtIL-2. Therefore, we expected that in a T cell mixture, Treg cells harboring a higher affinity trimeric IL-2R would trap the mutein in their surface more efficiently, concentrating the antagonistic effect over them. To evaluate the preferential antagonism of IL-2 mutein over Treg cells, we used a T cell proliferation assay in which sorted CD4+Foxp3+ Treg and CD4+Foxp3− E T cells were mixed back in different proportions. We kept the number of E T cells constant and varied the number of Treg cells. We always kept the ratio of Treg:T E cells suboptimal (<1:2) to avoid the high levels of suppression reported on classical in vitro suppression assays. T cells were activated with anti-CD3 and were cultured in the presence of two different concentrations of IL-2 mutein (5 and 10 ng/ml). The CD4+Foxp3− cells were labeled CFSE, and their proliferation was followed by flow cytometry.
In the absence of Treg cells, the mutein inhibits E T cell proliferation to baseline level (Fig. 4). The addition of increasing amounts of Treg cells restored E T cell proliferation back to maximal levels in the presence of the mutein. It is important to note that the observed release of the suppression exerted by the mutein over E T cells occurred when the number of Treg cells added to the culture was barely half of the number of E T cells present. Thus, Treg cells seem to be preferentially trapping the IL-2 mutein, preventing the mutein from inhibiting the IL-2 signal on E T cells. We can therefore conclude that, in presence of a mixture of Treg and E T cells, the mutein binds preferentially to the Treg cells, releasing E cells from the suppression.
Treatment with IL-2 mutein in vivo reduces Treg cell accumulation
To determine if the mutein was able to inhibit Treg cell proliferation or survival in vivo, we evaluated Treg cell accumulation in mutein-treated healthy mice. Healthy mice were injected with either PBS, wtIL-2, or the IL-2 mutein during a week. The wtIL-2 and mutein were injected twice a day to compensate for the short t1/2 of these small molecules; at the end, the accumulation of Treg cells was determined on spleens. As it has been previously described (14), the group treated with the wtIL-2 showed an increase in Treg cell frequency (p < 0.05) and number (p < 0.01), being statistically different from the groups treated with either PBS or mutein. In contrast, the mice treated with the mutein displayed a significant reduction of Treg cell numbers when compared with PBS-treated controls (p < 0.05) (Fig. 5A, 5B).
In vivo treatment with the IL-2 mutein prevents tumor growth and reduces Treg cell accumulation in tumor infiltrates
We decided to compare the antitumor effect of the IL-2 mutein with the anti-CD25 (PC61) mAb, which has been previously shown to reduce Treg cell number and tumor growth in several models when it is administered prior to tumor inoculation (8). We chose two different experimental murine tumor models previously reported to be sensitive to Treg cell suppression: MB16F10 melanoma (15) and 4T1 mammary tumor (16). Mice were treated with 200 μg of IL-2 mutein or 200 μg anti-CD25 mAb 5 d before tumor cell injection. Growth of the s.c. implanted tumors was then followed. In both models, mice treated with the mutein showed a significant reduction in tumor growth when compared with the control group (p < 0, 01); the mutein effect was similar to that observed on the anti-CD25 mAb–treated mice (Fig. 5C, 5D). Following the tumor growth for 30 or 35 d was enough to demonstrate that transiently blocking the IL-2 signal was sufficient to impact Treg cell homeostasis and to induce a significant delay on tumor growth. None of the treated mice were cured; the tumors continued to grow but grew very slowly.
To evaluate the effect of the mutein on Treg cell accumulation in tumor-bearing mice, we selected the mammary carcinoma 4T,1 which has been reported as a high-infiltrated tumor model inducing high accumulation of Treg cells. We repeated the treatments and tumor inoculations identical to the previous experiment, but we sacrificed the mice 15 d after tumor inoculation and evaluated the Treg cell and T CD8+ lymphocyte accumulation in the TDLN and in the tumor infiltrates. As expected, the 4T1 tumor induced an increment on the percentage of Treg cells in the TDLN, reaching more than 25% of Foxp3+ cells among CD4+T lymphocytes in comparison with the normal level of Treg cells in naive mice. In the mice treated with the mutein or with the anti-CD25 mAb, we observed a reduction of Treg cell percentages until 20%; this effect was statistically significant (p < 0.05). The reduction of Treg cell percentages on tumor infiltrates was even higher; particularly, the mice treated with the mutein showed the lowest accumulation of Treg cells (7%). When the proliferation level on Treg cells was assessed by ki67 staining, a reduction was detected both in the TDLN and in the tumors (Fig. 5I). Both in the lymph nodes and in the infiltrates, the ratio of CD8+/Treg cells was duplicated. It means that the mutein was able to prevent the development of a suppressive microenvironment inside the tumor and in the secondary lymphoid organs where the antitumor response was started. Moreover, the level of IFN-γ–producing CD8+ cells inside the tumors was higher in the group treated with the mutein (34%) in comparison with the control group (24%) (Fig. 5J). In general, the mutein induced similar effects to the anti-CD25 mAb.
Modeling the therapeutic impact of no-γ IL-2 muteins
To better understand the effect of blocking IL-2 signal with different IL-2 muteins, we used a mathematical model previously developed and calibrated in our group (12). This model focuses on the complex interplay of IL-2 with the dynamics of CD4+CD25+Foxp3+ Treg cells, E CD4+ T cells, M CD8+ T cells, and M NK cells. To address the immune-stimulating potential of different therapies, the model is set to steady-state equilibrium, where Treg cells effectively dominate the expansion of auto-reactive E T lymphocytes and M cells. Such steady-state is interpreted in the model as natural tolerance. Then, the system dynamics are perturbed by simulating the injection of a desired agent, further evaluating whether it is driven away from the natural tolerance steady-state.
Initially, we explored in the model the effect of the injection of the ideal no-γ IL-2 mutein during 5 d over T lymphocyte dynamics. The injection of the mutein induces a fast initial reduction in the number of Treg cells, which allows the expansion of both E and M T cells. A rebound of Treg cell numbers is then observed following the massive production of endogenous wtIL-2 induced by E T cell expansion (Fig. 6A), but for high doses, the system runs away into a new steady-state characterized by a large number of E and M cells and a small basal amount of receptor cells. This new steady-state has been typically interpreted, in our model, as an autoimmune steady-state. Qualitatively, the most striking feature of the dynamics in Fig. 6A is that the IL-2 mutein seems to negatively affect the Treg cell dynamics and not that of E and M cells that also might use IL-2, resulting in a strong immune stimulation. These in silico results are in accordance with the reduction of Treg cells and the increment of the CD8/Treg cell ratio observed in vivo in the mice inoculated with the 4T1 tumor cells and treated with the mutein.
The model reproduces many practical observations of IL-2 manipulations quite well and particularly reproduces the properties of the antagonist mutein. Therefore, we decided to explore how changes in the different characteristics of the mutein could contribute to the potency of its resulting immune stimulatory activity; the in silico results could be used for future improvement of the molecule to be used as a therapeutic tool. We tested the importance of the molecule t1/2, simulating the effect of two types of no-γ muteins with lifespans in blood for 9 min (just like rIL-2 produced in E. coli) or 7 h (like the fusion protein IL-2-Fc), respectively. Fig. 6B shows the proportion E + M cells versus receptor cells in day 6, just 1 d after finishing treatment, depicted as a function of the dose of injected muteins. In both cases, increasing the injected dose increased the system imbalance to favor the E + M cell population, and increasing the t1/2 decreased the dose required for obtaining the described effect. Interestingly, the observed imbalance is maximal at some intermediate dose and declines slightly afterward. Such decline derives from the fact that at larger concentrations, the injected muteins also prevent M cell expansion by directly inhibiting IL-2 signals through the intermediate affinity IL-2R.
We explored the relevance of interactions with both α- and β-chains of the IL-2R for the potency of the mutein immune stimulatory activity. Fig. 6C shows the result of simulating treatments with no-γ muteins that can only bind to either the β- or the α-chain of the IL-2R, referred to in this article, respectively, as no-γ_no-α and no-γ_no-β IL-2 muteins. Interestingly, all variants of no-γ muteins appear to have immune stimulatory activity, increasing, in a dose dependent way, the ratio of E + M/receptor in the simulations. However, the no-γ mutein, which binds to both α- and β-chains, is more active at lower doses, suggesting that it is quite important to efficiently block all forms of the IL-2R.
Finally, Fig. 6D shows the activity of no-γ muteins that have increased (100 times) the affinity for CD25 (no-γ_α+) or for CD122 (no-γ_β+), respectively. The results show that increasing the affinity for CD122 does not increase the stimulatory activity of the mutein. It actually reduces its activity at higher doses, following its increased capacity to block IL-2 signal equivalently through all forms of IL-2R. However, increasing the affinity by CD25 does significantly increase the no-γ mutein activity, particularly at low doses. This is a direct consequence of the increased affinity of this mutein but only for the high affinity form of the IL-2R and with it, its preferential inhibition over the Treg cells that overexpress it.
Treg cells have been described as potent in vivo immunosuppressors responsible for the maintenance of peripheral tolerance (3). Their normal function becomes a problem in cancer patients because Treg cells can dampen the desired antitumor response. It has been documented that increased numbers of Treg cells correlate with advanced stages of several malignancies (17), and, although the topic is controversial, several works suggest a correlation between Treg cell accumulation and bad prognosis (18, 19). Thus, the idea of eliminating or inactivating Treg cells while preserving E T cells has been considered as a promising strategy for cancer therapy or as adjuvant strategy for cancer vaccines (20).
IL-2 plays a central role in the generation and maintenance of an effective T cell response in vivo. IL-2 signaling controls both the primary and secondary expansion of the CD8+ T cell population in vivo (21); it could rescue exhausted CD8+ T cells, promoting its proliferation (22), and it regulates CD4 T cells differentiation in Th1, Th2, and Th17 based on its modulation of the expression of IL-4R, IL-12R, or IL-6R (23). However, the crucial role of IL-2 in the generation, homeostasis (24), and function (25) of the CD4+CD25+ Treg cells is considered its main nonredundant function in vivo. Therefore, blocking IL-2 signaling should weaken Treg cell activity and promote immune responses.
An initial attempt to prove the latter hypothesis was carried out by Setoguchi et al. (5). They injected anti–IL-2 mAbs in naive mice, reducing the number of CD4+CD25+ Foxp3+ Treg cells and inducing autoimmunity. However, in these studies, they used the S4B6 anti–IL-2 mAb, which was first hypothesized (6) and later demonstrated (26) to block only the interaction of IL-2 with the IL-2Rα. Indeed, the mAb forms immune complexes with IL-2 that signal through the intermediate affinity IL-2Rβγ. A different anti–IL-2 mAb, JES-61A2, which blocks the interaction of IL-2 with IL-2Rβ- and IL-2Rγ-chains, has been studied more recently. This Ab forms immune complexes with IL-2 and induces a large expansion of Treg cells in vivo (6). As a consequence, it has been shown effective to prevent transplant rejection and to treat autoimmune disorders in mouse models (27). A different strategy to address the latter hypothesis can be drawn from the use of therapies based on anti–IL-2R mAbs. Indeed, the administration of the anti-CD25 mAb (PC61) before tumor inoculation reduces the number of Treg cells and reduces tumor growth in mice (8). In humans, the administration of anti-CD25 mAb (Daclizumab) has been combined with cancer vaccines (28), and it is an approved immunosuppressive treatment for some autoimmune disorders like multiple sclerosis (29). However, the effect of these anti–IL-2R mAbs derives mainly from directly lysing CD25+ T cells rather than blocking IL-2 signal in vivo. Thus, overall, the results of treatments based on anti–IL-2 or anti–IL-2R mAbs are so far ambiguous and inconclusive, most likely because the observed effects cannot be exclusively attributed to the blocking of IL-2 signaling in vivo.
In this work, we tested a different approach for blocking IL-2 signaling in vivo and to target Treg cells. We describe the use of IL-2 muteins, designed to antagonize IL-2 signal in T cells through all forms of the IL-2R. These recombinant proteins include mutations in the region contacting the IL-2R γc-chain without modifying the interfaces with α– and β–IL-2R chains. It was previously demonstrated that deletion of the region F124-Q126 (30) and more specifically deletion of Q126 (31) leads to IL-2 muteins with severely reduced biological activity. Such works constituted great contributions for the understanding of IL-2/IL-2R interactions, but it was beyond their scope to demonstrate the possible therapeutic use of those molecules. We have used bioinformatics tools and the three-dimensional structure of the complex IL-2/IL-2R (9, 10) to select mutations on different residues inside the γc-chain interface. These mutations include the Q126 but also other close residues with relevant contributions to IL-2 γc-chain interaction, namely Q22, S130, and I129. Three mutein variants were constructed, and all behaved as very weak IL-2 agonists on the CTLL2 cell line.
Out of the three muteins, the weakest agonist, M1 (Q22V, Q126A, I129D, S130G), was able to inhibit IL-2–driven proliferation of E and Treg cells in vitro, inducing cell death. M1 was also able to inhibit de novo Treg cell differentiation and abrogate in vivo tumor growth in a similar extent to anti-CD25 mAb. Moreover, the mutein was able to decrease Treg cell numbers in vivo; the reduction was particularly high in the tumor infiltrates. This impact of M1 over Treg cells is likely the main mechanism underlying the antitumor effect observed both in melanoma and in 4T1 murine tumor models. To our knowledge, this is the first time in which an IL-2 mutein with antagonistic behavior has been tested in vivo for antitumor effect. Liu et al. (32) developed human IL-2 antagonists constructed by adding Q126 or V91 mutations onto the backbone of an IL-2 analog with incremented affinity by IL-2Rα. Those muteins behave as IL-2 antagonists in vitro for both E and human Treg lymphocytes, but their in vivo effects remain to be addressed. More recently, Mitra et al. (33) also developed a human IL-2 antagonist with several mutations in the interface with the IL-2R γc-chain but with incremented affinity by IL-2R β-chain. This antagonist inhibits IL-2 signaling as effectively as anti-CD25 or anti-CD122 Abs on different in vitro settings. Moreover, it was shown to cause significant immunosuppression in vivo, prolonging survival in a mouse model of graft-versus-host disease.
Overall, our results show that it is sufficient to transiently inhibit IL-2 signaling, with an IL-2 antagonist biased to the high affinity IL-2R, to shift the in vivo E/Treg cell balance toward immunity. This manipulation could be of practical use in the therapy of cancer, particularly for those tumors where suppression mediated by Treg cells is a problem. However, there is opportunity for further optimization of this type of treatment because the observed abrogation of tumor growth was similar to the effect achieved following anti-CD25 treatment, and the reduction on Treg cell frequency was only moderate.
To complement our understanding of the effect of the muteins in vivo and to predict whether it is possible to improve their effect by changing some of their basic properties, we used a mathematical model (12). This model has been calibrated with experimental data, and it has been used to study the impact of different IL-2 modulating therapies, like anti–IL-2 mAbs (34), immune complexes of IL-2+ anti–IL-2 mAbs, and different types of IL-2 muteins (35). Our in silico simulations showed that treatment with an ideal no-γ mutein induced a transient and modest reduction of Treg cell numbers and expansion of E CD4, CD8 M, and NK cells. This immune stimulatory effect was maximal at some intermediate dose of the mutein, because at larger doses, the mutein inhibits the expansion of the M T cells as well. Qualitatively, the simulation results support the idea that the mutein shares with wtIL-2 the capacity to preferentially bind to Treg cells (at low concentrations) because these cells express the high affinity IL-2Rαβγ at the cell surface. Only at high concentration, the mutein significantly affects IL-2 signaling on E and M cells expressing the intermediate affinity IL-2Rβγc. Second, the initial reduction of Treg cells releases the suppression exerted by them over E cells. As a result, E cells expand and produce more wtIL-2, which further stimulates the expansion of M CD8 and NK cells. It must be stressed that the preferential binding of the mutein to Treg cells, which is essentially the in silico predicted behavior, was demonstrated in the in vitro experiments showed in this work.
It is known that the small m.w. of rIL-2 represents a disadvantage for therapy because of the fast clearance from the blood. This problem is even more relevant for a molecule expected to be used as an in vivo antagonist. Therefore, we tested in silico the influence of the t1/2 in the efficacy of the mutein. We found that no-γ muteins with longer t1/2 require significantly lower doses to achieve the same biological activity. Actually, increasing the t1/2 of the no-γ mutein to around 7 h is predicted to reduce the therapeutic doses required in two orders of magnitude. This result strongly suggests the convenience of producing an IL-2 mutein fused to the Ig C region in mammalian cells.
The no-γ muteins developed in this work were able to block the IL-2 signal through both the high-affinity IL-2Rαβγ and the intermediate-affinity IL-2Rβγ. We have explored in our in silico simulations the relevance of the interactions with α- and β-chains of the IL-2R for the potency of the immune stimulatory activity of the mutein. The results obtained show that it is very important to bind both α- and β-chains, although increasing the affinity by α-chain, but not by the β-chain, increases the stimulatory properties of the no-γ mutein at low doses. Qualitatively, the latter result is very interesting. It confirms that the preferential binding of the mutein to the high-affinity IL-2Rαβγ, and therefore to the Treg cells that overexpress it, is very significant for the observed dynamics. Overall, our results predict that by introducing a few more mutations into the current no-γ mutein to increase the affinity to the IL-2Rα, for instance those previously described by Rao et al. (36), its potency as an anticancer drug could be significantly increased. On the contrary, a no-γ mutein variant including more mutations to increase the affinity for the IL-2Rβ, for instance, like the one previously described by Mitra et al. (34), will be less effective as an anticancer drug. But it might be more effective to induce CD8 and NK cell immunosuppression at large doses, an effect useful in the treatment of the graft-versus-host disease or in some autoimmune disorders.
IL-2 has a key role in regulating tolerance and autoimmunity. The comprehension of its in vivo function must lead to the design of better therapeutic strategies. After solving the three-dimensional structure of the complex IL-2/IL-2R, several works have explored the possibility of modifying the interaction of the cytokine with the receptor chains aimed to change its efficacy as therapeutic tool. The architecture of the quaternary complex IL-2/IL-2R gives rise to the possibility of independently manipulating the contact regions with the different receptor chains. In the present work, we explored the effect of decreasing the interaction of IL-2 with the γ-chain with an IL-2 antagonist with in vivo antitumor effect. Moreover, in a previous work, we characterized another IL-2 mutein that does not bind to the α-chain of the receptor and shows higher antitumor effect than the wtIL-2 (37). There are other examples of IL-2 muteins with improved properties: for example, the work from Levin et al. (38) in which a mutein with increased affinity by the β-chain behaves as a superagonist. From all this current work of protein engineering with this or other cytokines, more promising alternatives for cancer therapy might arise.
We thank Dr. Alicia Santos, from the Centro de Ingeniería Genética y Biotecnología, La Habana, for kindly donating the IL-2–dependent cell lines.
The authors have no financial conflicts of interest.