Molecular engineering of interleukin-2 for enhanced therapeutic activity in autoimmune diseases

Abstract

The interleukin-2 (IL-2) cytokine plays a crucial role in regulating immune responses and maintaining immune homeostasis. Its immunosuppressive effects have been harnessed therapeutically via administration of low cytokine doses. Low-dose IL-2 has shown promise in the treatment of various autoimmune and inflammatory diseases; however, the clinical use of IL-2 is complicated by its toxicity, its pleiotropic effects on both immunostimulatory and immunosuppressive cell subsets, and its short serum half-life, which collectively limit the therapeutic window. As a result, there remains a considerable need for IL-2-based autoimmune disease therapies that can selectively target regulatory T cells (Tregs) with minimal off-target binding to immune effector cells in order to prevent cytokine-mediated toxicities and optimize therapeutic efficacy. In this review, we discuss exciting advances in IL-2 engineering that are empowering the development of novel therapies to treat autoimmune conditions. We describe the structural mechanisms of IL-2 signaling, explore current applications of IL-2-based compounds as immunoregulatory interventions, and detail progress and challenges associated with clinical adoption of IL-2 therapies. In particular, we focus on protein engineering approaches that have been employed to optimize the Treg bias of IL-2, including structure-guided or computational design of cytokine mutants, conjugation to polyethylene glycol (PEG), and development of IL-2 fusion proteins. We also consider future research directions for enhancing the translational potential of engineered IL-2-based therapies. Overall, this review highlights the immense potential to leverage the immunoregulatory properties of IL-2 for targeted treatment of autoimmune and inflammatory diseases.

1. Introduction

IL-2 is a four-helix bundle cytokine that orchestrates the differentiation, proliferation, activation, and survival of various immune cell subsets, particularly lymphocytes [1]. Its pleiotropic nature enables IL-2 to activate both immunostimulatory immune effector cells (i.e., CD4⁺/CD8⁺ effector T cells and natural killer (NK) cells) and immunosuppressive Tregs. The role of IL-2 in regulating the balance between promotion and repression of immune responses has made this cytokine an attractive target for immunotherapy. However, its pleiotropic activities lead to unwanted toxicities and limit potency, and its short serum half-life further hinders clinical performance. There is thus immense interest in engineering IL-2 to decouple its immunostimulatory versus immunoregulatory effects, and thereby harness its potential to prevent and treat a variety of immune conditions, ranging from cancer to autoimmune disorders. This review will detail the well-established role of IL-2 in regulating the immune response by maintaining Treg homeostasis and also describe how IL-2 may be used for prevention and treatment of autoimmune and inflammatory diseases. The review will cover early therapeutic efforts with the natural cytokine and progress through emerging engineering approaches that overcome the limitations of native IL-2. These innovative approaches preferentially stimulate Treg activity and/or extend cytokine half-life, helping to realize the therapeutic potential for IL-2 in treating autoimmune and inflammatory conditions.

2. IL-2 as a Therapeutic

IL-2 was the first cytokine to be molecularly cloned [2,3] and it was initially described as a T cell growth factor [4,5]. IL-2 is primarily secreted by CD4⁺ T cells, although it can also be produced by CD8 ⁺ T cells, NK cells, NKT cells, activated dendritic cells (DCs), and mast cells [6–12]. Due to its potency in activating effector T cells as well as NK cells, early IL-2 therapeutic studies focused on its use as an immune stimulant, culminating in the approval of IL-2 by the FDA to treat metastatic renal cell carcinoma in 1992 [13] and metastatic melanoma in 1998 [14]. However, just as IL-2 was gaining traction as an anti-cancer drug, it was discovered that mouse models deficient in either IL-2 or components of its receptor developed severe autoimmune conditions and displayed lymphoproliferation, revealing IL-2’s pivotal role in tolerance and protection from autoimmune diseases [15–17]. Subsequent work showed that adoptive transfer of normal Tregs into an IL-2Rβ knockout mouse model prevented lethal autoimmunity, providing the first evidence that Tregs critically depend on IL-2 for activation, proliferation, differentiation, and survival [18].

2.1. IL-2 Signaling

The duality of IL-2 is based on the differential receptor expression patterns between immune effector cells and Tregs. IL-2 signals through formation of either an intermediate-affinity (K_D≈1 nM) heterodimeric receptor complex, comprising the IL-2 receptor beta (IL-2Rβ, CD122) and common gamma (γ_c, CD132) chains, or a high-affinity (K_D≈10 pM) heterotrimeric receptor complex, consisting of IL-2 receptor alpha (IL-2Rα, CD25), as well as the IL-2Rβ and γ_c chains (Fig. 1) [3,12,19]. As revealed by the crystal structure, IL-2 interacts with the extracellular domains of all 3 receptor subunits, and receptor/receptor contact between the IL-2Rβ and γ_c chains stabilizes complex formation [19]. Whereas IL-2 independently binds IL-2Rα (K_D≈10 nM) and IL-2Rβ (K_D≈200 nM), the cytokine only binds γ_c in the presence of IL-2Rβ [20–22]. Due to the 100-fold higher affinity of IL-2 for the heterotrimeric versus the heterodimeric receptor, expression of the nonsignaling IL-2Rα subunit dictates the sensitivity of cells to circulating IL-2. IL-2Rα is constitutively expressed at high levels on Tregs but is virtually absent from naïve immune effector cells, which results in significantly greater potency of the cytokine on Tregs versus immune effector cells (Fig. 1) [23–25]. Additionally, the transcriptional positive feedback loop for IL-2Rα, as well as other amplifying effects contribute to the high sensitivity of Tregs to the IL-2 cytokine [26,27]. It should be noted that IL-2Rα is also expressed on activated effector T cells and innate lymphoid cells (ILC) such as ILC-2, albeit at lower levels compared to Tregs, making IL-2 a complex target for immunotherapy [28–31].

Fig. 1.

Biology of IL-2 signaling. IL-2 signals through formation of either an intermediate-affinity heterodimeric complex (K_D≈1 nM), consisting of the IL-2Rβ and γ_c subunits, or a high-affinity heterotrimeric receptor complex (K_D≈10 pM), comprised of the IL-2Rα, IL-2Rβ, and γ_c subunit. IL-2Rα is highly expressed on Tregs but not immune effector cells, thus rendering the IL-2 more potent on Tregs. IL-2 cytokine/receptor crystal structure obtained from PDB ID 2B5I.

Formation of IL-2 cytokine/receptor complexes activates intracellularly associated Janus kinases (JAKs) on the IL-2Rβ and γ_c subunits, which then phosphorylate key intracellular residues on these receptor chains [32,33]. These residues in turn serve as docking sites for signal transducer and activator of transcription (STAT) 5, which becomes phosphorylated, dimerizes, and translocates to the nucleus to regulate gene expression [33,34]. Although this pathway (termed the JAK-STAT pathway) is the principal signaling outlet for IL-2, the cytokine also secondarily activates the phosphatidylinositol 3-kinase (PI3K) and mitogen-activated protein kinase (MAPK) pathways [2,27,29]. Of note, in Tregs, but not CD8⁺ effector T cells, the PI3K pathway is inhibited by phosphatase and tensin homolog (PTEN) due to high expression of PTEN, enabling distinct signaling profiles for IL-2 on Tregs versus CD8⁺ T cells [35].

2.2. Role of IL-2 in Treg Activity and Function

IL-2 promotes differentiation of CD4⁺ T cells [36], as well as proliferation and cytolytic activity of CD4⁺ T cells, CD8⁺ T cells, and NK cells [37], supporting inflammatory immune responses. Following antigen-specific activation, effector T cells upregulate IL-2Rα to perpetuate signaling and compete with IL-2Rα^High Tregs for engagement of circulating IL-2 [28–30]. However, the IL-2Rα subunit is expressed on activated effector T cells at lower levels compared to Tregs [29,38]; thus, Tregs maintain their competitive advantage for IL-2 consumption, allowing for continued inhibition of immune effector cell activity. Extensive work has further elucidated the role of IL-2 in maintaining the activity and function of Tregs to effectively suppress immune responses [11,4,39,40,33,12].

The most critical transcription factor for Treg development and function is Forkhead box P3 (FOXP3). FOXP3-deficient patients display the fatal immune dysregulation polyendocrinopathy, enteropathy, X-linked syndrome (IPEX) disorder, highlighting the essential role of FOXP3 in protection against autoimmune diseases [41]. Signaling through the IL-2 complex promotes FOXP3 expression [23] by activating STAT5, which binds to the FOXP3 promoter and increases production of the transcription factor [42]. Accordingly, IL-2-knockout mice show significantly reduced FOXP3⁺ Treg percentages in peripheral lymphoid organs [43]. In addition to its role in activating Treg-supporting FOXP3 transcription, IL-2 signaling also directly expands Tregs, and indeed, transfer of IL-2-producing cells into IL-2-deficient mice has been shown to increase Treg numbers [44]. As an additional immunoregulatory mechanism, IL-2 promotes expression of immune-suppressing checkpoint proteins on Tregs, such as cytotoxic-T-lymphocyte-associated antigen 4 (CTLA-4) [43] and programmed cell death 1 (PD-1) [45].

2.3. Dysfunction of Immune Activity in Inflammatory and Autoimmune Diseases

Breakdown and dysregulation of immune activity, and specifically dysfunction of Tregs, leads to the development of autoimmune and inflammatory diseases. Treg deficiency enables hyperactivation of immune effector cells [46], which leads to inflammatory disorders, as seen with FOXP3 mutations that lead to Treg depletion in mice [47] and IPEX syndrome in humans [41]. Impaired Treg suppressive function is linked to autoimmune diseases such as active systemic lupus erythematosus (SLE) [48] and rheumatoid arthritis (RA) [49]. Depletion of Tregs in otherwise healthy patients triggers onset autoimmunity in the form of diabetes, lymphoproliferation, and tissue lesions [46,50]. As previously mentioned, IL-2-deficient mice develop autoimmune conditions due to Treg depletion [15,16], illustrating the vital role for IL-2 in supporting Treg survival and function in order to maintain homeostatic balance with immune effector cells.

2.4. Low-Dose IL-2

Although the first FDA approval for IL-2 was granted in the context of cancer treatment, the low IL-2 sensitivity of immune effector cells necessitated the use of high doses (~120 million IU per day dose [39,51]) of the cytokine [23–25]. Unfortunately, high-dose IL-2 treatment is accompanied by severe toxicities, including hypotension, tachycardia, diarrhea, hypoxia, and thrombocytopenia [13,14,52]. One of the most prominent toxicities of IL-2 is vascular leak syndrome, which can lead to potentially fatal pulmonary edema and organ failure [11,53–55]. Since Tregs are significantly more sensitive to IL-2 compared to immune effector cells due to their elevated levels of IL-2Rα expression [23–25], low-dose IL-2 administration (<3 million IU per day dose [39,56–58]) preferentially stimulates Tregs to allow for improved treatment of autoimmune and inflammatory diseases while also circumventing the harmful side effects associated with high-dose IL-2 administration. Low-dose IL-2 treatment has also been shown to inhibit the differentiation of T helper 17 cells (Th17) and T follicular helper cells (Tfh), thereby enhancing its anti-inflammatory activities [59,60]. Multiple preclinical and clinical studies have demonstrated promising results for low-dose IL-2 treatment in a range of autoimmune conditions [39,61–69,56,66,70–72].

One of the first clinical trials that used low-dose IL-2, led by Saadoun et al., involved daily subcutaneous dosing at 1.5–3 million IU on a specified treatment schedule in patients with hepatitis C virus (HCV)-induced vasculitis [61]. Treatment resulted in significant expansion of Tregs in HCV-infected patients with no evidence of the adverse effects commonly observed for high-dose IL-2 treatment. Clinical improvement was observed in 8 of 10 patients [61]. Another pioneering clinical trial was conducted by Koreth et al., in which they administered low-dose IL-2 (daily subcutaneous dosing at 0.3–3 million IU per square meter of body surface area for 8 weeks) in patients with chronic graft-versus-host disease (GVHD), and their results showed a maximum tolerated dose of 1 million IU per square meter, increases in Treg abundance and Treg to CD4⁺ effector T cell ratios, and significant amelioration of symptoms in 12 of 23 evaluated patients [62]. Humrich and colleagues delivered low-dose IL-2 (subcutaneous dosing of 1 million IU on a specified treatment schedule) to patients with SLE, which was found to induce preferential expansion of Tregs [63]. The investigators went on to conduct a Phase II trial with a higher dosage of 1.5 million IU IL-2 combined with standard treatment in refractory SLE patients, and a significant improvement (31.6%) was observed in the response rate relative to standard treatment alone [64].

In a landmark low-dose IL-2 study encompassing 11 autoimmune diseases, including Bechet’s disease, ulcerative colitis (UC), and psoriasis, Rosenzwajg and colleagues assessed the effects of subcutaneous dosing of 1 million IU IL-2 daily for 5 days and fortnightly for 6 months. They observed robust Treg activation in the absence of immune effector cell activation in all treated patients, demonstrating promise for further clinical evaluation of low-dose IL-2 treatment in a broad range of autoimmune diseases [65]. Allegretti et al. administered low-dose IL-2 (subcutaneous dosing of 0.3–1.5 million IU per square meter over 8 weeks) in patients with moderate to severe UC [66]. They found that a dose of 1 million IU per square meter was the maximum tolerated dose and treatment led to peripheral Treg proliferation and remission or response in 9 of 17 patients [66]. Snapper and colleagues have extended these findings by initiating a clinical trial using low-dose IL-2 in patients with moderate to severe Crohn’s disease (NCT04263831). In another instance, Zhang et al. delivered low-dose IL-2 (subcutaneous dosage of 1.0 million IU on a specified schedule involving twice daily delivery) combined with methotrexate (MTX) in patients with RA and observed significant improvement in RA responses [67]. A recent pilot study by Miao et al. demonstrated that administering low-dose IL-2 (subcutaneous dosing of 1 million IU every 2 days in combination with standard treatment for 12 weeks) in patients with idiopathic inflammatory myopathies (IIMs) in combination with standard therapy resulted in substantial Treg expansion, and 14 of 18 patients exhibited improved clinical scores [68]. Additionally, low-dose IL-2 treatment (0.5 million IU dosing for 5 days in one study and 1 million IU dosing on specified treatment cycles in a subsequent study) in patients with primary Sjögren Syndrome led to Treg expansion and showed promise for improving clinical responses [56,69]. Allegretti et al. also showed that administration of low-dose IL-2 (at the maximum efficacious dose of 1 million IU per square meter for 8 weeks) resulted in selective Treg expansion and improved Mayo scores in a majority of patients with UC [66].

Despite these successes, other trials have demonstrated limitations for low-dose IL-2 therapy. A study led by Bluestone and colleagues infused autologous polyclonal Tregs and administered low-dose IL-2 (subcutaneous dosing of 0.33–1.0 million IU on a specified schedule) into patients with type 1 diabetes (T1D), which led to expansion of Tregs. However, undesired expansion of NK and CD8⁺ T cells was also observed, hindering efficacy [70]. Additionally, recent studies investigating the effects of low-dose IL-2 in children with T1D (0.125–0.500 million IU per square meter on a specified treatment schedule) and patients with amyotrophic lateral sclerosis (ALS) (daily dosing of 1–2 million IU per square meter for 5 days followed by fortnightly dose for 1 year) observed that cytokine therapy led to Treg proliferation but did not lead to clinical improvement [71,72]. Moreover, although low-dose IL-2 is considered safe for use in patients, side effects such as injection site reactions and fever-like symptoms occur [73], and off-target activation of undesirable or autoreactive effector cells, coupled with the cytokine’s short serum half-life (13–85 min) [40,74], result in a restrictively narrow therapeutic window. Optimization of dosing is complicated by the delicate balance between administering sufficient levels for induction of Treg expansion and avoiding undesirable activation of immune effector cells and cytokine-mediated toxicities [75]. To overcome these barriers, there are growing engineering efforts to selectively bias IL-2 effects towards Tregs [4]. Excitingly, many of these approaches are now coming to bear clinically [76]. In this review, we discuss progress in harnessing protein design strategies to empower the therapeutic translation of IL-2 to treat autoimmune and inflammatory diseases (Table 1).

Table 1.

Preclinical and clinical engineered IL-2-based compounds discussed in this review

Compound Type	Compound Name	Description	Development Status	Clinical Indications	References
IL-2 muteins	Proleukin (Iovance Biotherapeutics)	Recombinant human IL-2 with C125S mutation	Approved	Metastatic renal cell carcinoma, metastatic melanoma	[74]
	M1	Treg-biased IL-2 mutein containing mutations at V69A and Q74P to enhance affinity to IL-2Rα	Preclinical	-	[79, 80]
	M6	Treg-biased IL-2 mutein containing mutations at V69A, Q74P, and I28T to enhance affinity to IL-2Rα	Preclinical	-	[79, 80]
	2–4 IL-2	Treg-biased IL-2 mutein containing mutations at N29S, Y31H, K35R, T37A, K48E, V69A, N71R, Q74P, N88D, I89V to enhance affinity to IL-2Rα	Preclinical	-	[83]
	BAY 50–4798 AIC284 (Aicuris)	Treg-biased IL-2 mutein containing N88R mutation to reduce affinity to IL-2Rβ	Phase I (terminated)	Melanoma, renal cell carcinoma	[85, 86]
	IgG-(IL-2N88D)2 RO7049665 (Roche)	IgG1-conjugated, Treg-biased IL-2 mutein containing N88D mutation to reduce affinity to IL-2Rβ	Phase II (terminated)	Ulcerative colitis, autoimmune hepatitis	[87, 150, 151]
	Fc.Mut24 AMG 592 (Amgen)	IgG2a Fc domain-conjugated, Treg-biased IL-2 mutein containing N88R and V91D mutations to reduce affinity to IL-2Rβ	Phase II	GVHD, SLE, rheumatoid arthritis, ulcerative colitis	[93, 153–157]
	H9/Super-2	IL-2 mutein containing mutations at L80F, R81D, L85V, I86V, I92F to enhance affinity to IL-2Rβ	Preclinical	-	[22]
	H9-RETR	IL-2 mutein containing mutations at L80F, R81D, L85V, I86V, I92F to enhance IL-2Rβ affinity, L18R, Q22E, Q126T, S130R mutations to reduce γc affinity	Preclinical	-	[94]
IL-2 muteins	IL-2-REH	IL-2 mutein containing mutations at L18R, Q22E, and Q126H to reduce γc affinity	Preclinical	-	[95]
	orthoIL-2 + orthoIL-2Rβ STK-009+SYNCAR (Synthekine)	Orthogonal IL-2 containing E29D, Q30N, M33V, D34L, Q36K, and E37A mutations to ablate binding wild type IL-2Rβ and preserve binding to orthogonal IL-2Rβ counterpart	Phase I	CD19+ hematologic malignancies	[96–98]
	V91K, D20A, M104V	IL-2 mutein containing mutations at V91K and D20A to reduce IL-2Rβ affinity, M104V mutation to reduce IL-2Rα affinity	Preclinical	-	[100]
	Erb-sumIL2	Asymmetric Fc-conjugated heterodimer composed of a CD8+ T cell-biased IL-2 mutein (mutations: F42A, L80F, R81D, L85V, I86V, and I92F) and an anti-human EGFR Fab	Preclinical	-	[177]
	CC-92252 (Bristol-Myers Squibb, Celgene)	Treg-biased IL-2 mutein-Fc fusion protein	Phase I (terminated)	Psoriasis	[152]
	PT101/MK-6194 (Merck)	Treg-biased IL-2-Fc fusion protein with mutations that reduce IL-2Rβ binding affinity and enhance affinity to IL-2Rα	Phase I/II	Ulcerative colitis, atopic dermatitis	[158–162]
	XmAb27564 (Xencor)	Treg-biased IL-2 mutein-Fc-fusion protein with enhanced binding affinity to IL-2Rα	Phase I	-	[163, 164]
	mRNA 6231 (Moderna)	Lipid nanoparticle-encapsulated mRNA encoding for HSA-IL-2m, a human serum albumin (HSA)-conjugated IL-2 mutein with N88D mutation to reduce affinity to IL-2Rβ, V69A and Q74P mutations to enhance affinity to IL-Rα	Phase I	-	[101, 165]
PEGylated IL-2	NKTR-214 Bempegaldesleukin (Nektar Therapeutics)	IL-2 variant containing an average of 6 releasable PEG chains that limit binding to the heterotrimeric IL-2 receptor	Phase III (terminated)	Prostate cancer, melanoma, head and neck squamous cell cancer, metastatic renal cell carcinoma, solid tumors, sarcoma	[115, 117, 118]
	NKTR-358 Rezpegaldesleukin (Nektar Therapeutics)	IL-2 variant conjugated with PEG chains that attenuate binding to IL-2Rα and IL-2Rβ	Phase II	SLE, atopic dermatitis, psoriasis, ulcerative colitis	[109, 119–121, 166–169]
	Dual-31/51–20K	Treg-biased IL-2 variant containing 20 kDa PEG chains conjugated at residues Y31 and T51	Preclinical	-	[110]
	KKC80	Treg-biased IL-2 with 80 kDa PEG conjugated to an azide-containing lysine derivative at residue 129	Preclinical	-	[111]
	THOR-809 (Synthorx)	Treg-biased IL-2 variant containing PEG conjugated to a non-proteinogenic amino acid	Phase I	-	[112]
IL-2 fusion proteins	IL-2/CD25	IL-2/IL-2Rα fusion protein via a non-cleavable amino acid linker	Preclinical	-	[122, 124, 125]
	EHD2-scTNFR2	Mouse IL-2 fused to an engineered tumor necrosis factor (TNF) domain	Preclinical	-	[128]
IL-2/anti-IL-2 antibody complexes and fusion proteins	IL-2/JES6–1	Treg-biased complex of mouse IL-2 and an anti-mouse IL-2 antibody	Preclinical	-	[135–140, 143, 148]
	IL-2/F5111.2	Treg-biased complex of human IL-2 and an anti-human IL-2 antibody	Preclinical	-	[141]
	IL-2/UFKA-20	Treg-biased complex of human IL-2 and an anti-human IL-2 antibody	Preclinical	-	[142]
	F5111 IC	Treg-biased “immunocytokine” comprised of a single-chain human IL-2/anti-human IL-2 antibody fusion	Preclinical	-	[144]
De novo IL-2	Neo-2/15 NL-201 (Neoleukin)	CD8+ T cell-biased de novo IL-2 with an erased IL-2Rα binding interface and enhanced interactions with IL-2Rβγc	Phase I (terminated)	Refractory cancer	[104, 105]
	S1–S17	CD8+ T cell-biased IL-2 muteins with enhanced IL-2Rβγc interactions	Preclinical	-	[106]
Tolerogenic particles	ImmTOR+F5111 IC (Selecta Biosciences)	Biodegradable nanoparticles encapsulating rapamycin administered with a Treg-selective single-chain IL-2/anti-IL-2 antibody fusion protein	Preclinical	-	[175]
	Tol-MPs+F5111 IC	Biodegradable microparticles loaded with rapamycin and functionalized with a biased IL-2 fusion protein and MHC class II/myelin peptide complexes	Preclinical	-	[176]
IL-2 prodrugs	ProIL2	CD8+ T cell-biased IL-2 mutein-Fc fusion protein conjugated to IL-2Rβ by a matrix metalloproteinase-cleavable linker	Preclinical	-	[178]
	XTX202 (Xilio Therapeutics)	Masked IL-2 prodrug with bias towards IL-2Rβγc	Phase I/II	Solid tumors	[179, 180]
	WTX-124 (Werewolf Therapeutics)	Conditionally active IL-2 prodrug containing an inactivating Fab domain and a half-life extension domain	Phase I	Solid tumors	[181, 182]
	IL-2 β/γ (Ascendis Pharma)	IL-2 prodrug with a 40 kDa mPEG carrier and a methoxy polyethylene glycol moiety in the IL-2Rα binding interface	Phase I/II	Solid tumors	[183, 184]

Note: Trade names of clinical stage IL-2-based compounds are denoted by italics and their manufacturers are indicated in parentheses

3. Cytokine Mutation Approaches (Muteins)

The development of cytokine mutants, also known as muteins, was one of the first strategies implemented for engineering IL-2 to selectively expand Tregs. Various rational design and directed evolution approaches, along with computational methods, have been employed to manipulate the specific interactions between IL-2 and its receptor subunits. These efforts have resulted in biased IL-2 variants that selectively potentiate Tregs and lead to favorable outcomes in the treatment of autoimmune conditions. One approach to cytokine engineering involves rational design, based on structural prediction and/or mutational scanning (Fig. 2a) [77]. In this strategy, selected amino acids, typically in the cytokine/receptor interface, are mutated to alter cytokine function. As an alternative approach, cytokines may be site-specifically or randomly mutagenized and the resulting cytokine variant libraries may be screened using directed evolution platforms, such as phage display or yeast display, to identify clones with desirable binding properties (Fig. 2b) [78]. Directed evolution enables screening of larger mutagenic libraries and can also provide an unbiased approach to cytokine engineering. In practice, a combination of structure-based approaches and directed evolution is often employed to identify mutations that functionally bias cytokine behavior in a therapeutically advantageous manner. Innovative computational tools, such as de novo engineering, are also becoming integral parts of the cytokine engineering workflow (Fig. 2c) Although muteins have been designed for both mouse and human IL-2, this review focuses on human IL-2 muteins.

Fig. 2.

Cytokine mutation approaches for Treg-biased IL-2 mutein development. IL-2 muteins or mimetics can be engineered through a rational design and site-directed mutation; b random or site-specific mutagenesis and directed evolution using a platform such as yeast surface display; or c de novo computational design. d Treg-biased IL-2 muteins can be developed through modulation of binding to the IL-2Rα, IL-2Rβ, and/or γ_c subunits. e IL-2 muteins with increased IL-2Rα dependency show enhanced specificity for Tregs, which more prominently express the heterotrimeric receptor (right), compared to immune effector cells, which primarily express the heterodimeric receptor (left).

3.1. IL-2 Muteins with Enhanced Binding to IL-2Rα

Early approaches to Treg-biased IL-2 mutein development aimed to improve IL-2 binding to the IL-2Rα subunit (Fig. 2d), favoring the formation of the heterotrimeric receptor complex that is highly expressed on Tregs (Fig. 2e). To this end, directed evolution of an error-prone mutagenic library of IL-2 variants against the IL-2Rα subunit was performed using the yeast display platform, isolating two muteins, denoted M1 (V69A, Q74P) and M6 (V69A, Q74P, I128T), that exhibited a 15–30-fold increase in affinity for IL-2Rα compared to natural IL-2 [79,80]. Of note, residues V69 and Q74 are located at the IL-2/IL-2Rα binding interface, whereas I128 is found at the IL-2/γ_c interface [81]. Notably, although M1 and M6 had higher affinity for IL-2Rα relative to IL-2, they also showed 2–5-fold weaker binding affinity for IL-2Rβ. In terms of functional outcomes, the engineered variants exhibited greater potency in inducing T cell proliferation compared to IL-2, which the investigators attributed to a “cell surface ligand reservoir effect,” whereby the high affinity mutants persisted longer on the cell surface and contributed to an increased integrated growth signal [82]. Building on this progress, Rao et al. employed a library shuffling workflow to derive IL-2 mutants with even higher affinity for IL-2Rα [83]. The resulting muteins induced T cell proliferation with potency comparable to IL-15, a cytokine that also signals through the IL-2Rβ and γ_c subunits but has a 1,000-fold higher affinity for its private alpha chain, IL-15Rα. In addition to increased binding affinity, covalent binding of IL-2 to IL-2Rα can selectively promote Treg activation. Using an artificial fluorosulfate-L-tyrosine (FSY) amino acid, Zhang et al. found that FSY-bearing IL-2 variants covalently bound to IL-2Rα via sulfur-fluoride exchange, prolonging the engagement of IL-2 on Tregs and improving the cytokine’s therapeutic efficacy in murine models of SLE and GVHD [84].

3.2. IL-2 Muteins with Diminished Binding to IL-2Rβγ_c

In addition to enhancing IL-2 binding to IL-2Rα, modulating the binding interactions between IL-2 and the IL-2Rβ or γ_c subunits can also augment the affinity of IL-2 for Tregs (Fig. 2d). Shanafelt et al. hypothesized that IL-2 toxicity was primarily driven through the cytokine’s activation of NK cells, and thus sought to skew the activity of IL-2 towards T over NK cells to mitigate these effects [85]. The investigators postulated that by compromising the interaction of IL-2 with IL-2Rβ and/or γ_c, the resulting mutein would favor binding to cells (i.e., Tregs) that preferentially express the heterotrimeric receptor over the heterodimeric receptor (Fig. 2e). In one study, a human IL-2 mutein, denoted BAY 50–4798, was designed using site-directed mutagenesis to contain a single mutation, N88R, in the IL-2/IL-2Rβ interface [85]. This mutation ablated cytokine binding to IL-2Rβ and weakened interactions to the IL-2Rβγ_c heterodimer by five orders of magnitude, while maintaining the same affinity for IL-2Rα, resulting in >3,000-fold greater selectivity in stimulating proliferation of T cells over NK cells compared to wildtype IL-2. In a follow-up study, AIC284 (renamed from BAY 50–4798) preferentially expanded Treg cells over CD4⁺ effector T cells in vivo while ameliorating clinical symptoms in experimental autoimmune encephalomyelitis (EAE), a rodent model of multiple sclerosis (MS) [86]. Taking a similar approach, Peterson et al. developed an IL-2 mutein (IL-2N88D) that also reduced affinity for IL-2Rβγ_c. The mutein was then fused to a full human immunoglobulin G (IgG) antibody of irrelevant specificity, creating the bivalent fusion protein, IgG-(IL-2N88D)₂ [87]. Conjugation to a full-length antibody or fragment crystallizable (Fc) region extends the half-life of the cytokine through Fc domain interactions with the neonatal Fc receptor (FcRn) that enable FcRn-mediated recycling [88–92]. IgG-(IL-2N88D)₂ demonstrated increased selectivity in stimulating Tregs over immune effector cells compared to Proleukin^® (native human IL-2 with a stabilizing C125S mutation) and a control IL-2/IgG fusion protein in human peripheral blood mononuclear cells (PBMCs). A single low dose of IgG-(IL-2N88D)₂ elicited more robust expansion of functional Tregs than multiple high doses of Proleukin^® in cynomolgus monkeys. In another study, Khoryati et al. created Fc.Mut24, an IL-2 mutein with the IL-2/IL-2Rβ interaction-disrupting N88R and V91D mutations, fused to an IgG2a Fc domain for half-life extension [93]. The fusion protein was more selective than a wildtype IL-2/Fc fusion protein at expanding the Treg cells in vivo. Furthermore, Fc.Mut24 was found to arrest ongoing T1D development in nonobese diabetic (NOD) mice.

As an alternative strategy to biasing Treg expansion for autoimmune disease protection, Mitra et al. engineered IL-2 variants that functioned as “receptor signaling clamps,” with distinct T cell activation thresholds [94]. Using a previously developed IL-2 “superkine” (denoted H9, or Super-2) that was evolved to exhibit 200-fold enhanced affinity for IL-2Rβ [22] as a template, muteins with impaired γ_c binding were developed to attenuate IL-2Rβ/γ_c heterodimerization, thus acting as IL-2 partial agonists or antagonists. One of these IL-2 analogs, H9-RETR, introduced 4 disruptive mutations in the IL-2/γ_c interface (L18R, Q22E, Q126T, and S130R) to H9. H9-RETR potently antagonized IL-2 function in vitro, and a H9-RETR/IgG4 Fc fusion protein prolonged survival in a murine model of GVHD following allogeneic bone-marrow transplantation.

In a follow-up study, Glassman et al. exploited cell type-intrinsic differences in IL-2 signaling to develop mechanism-based IL-2 partial agonists with reduced pleiotropy and improved Treg selectivity [95]. Employing a structure-based design approach, the authors identified an IL-2 mutein (IL-2-REH) with reduced γ_c affinity through mutations L18R, Q22E, and Q126H. This mutein showed increased dependence on IL-2Rα expression compared to wildtype IL-2. In vivo, IL-2-REH induced proliferation of Tregs with reduced activity on NK cells and CD8⁺ T cells, leading to improved recovery in a dextran sulfate sodium (DSS)-induced mouse model of colitis.

Adoptive transfer of Tregs has proven effective for both autoimmune disease treatment and promotion of organ transplantation tolerance; however, ex vivo Treg expansion is clinically challenging. To expand Tregs in vivo following adoptive transfer more selectively, Sockolosky et al. devised an innovative two-step approach to engineer orthogonal IL-2/IL-2Rβ pairs that bind to one another and transmit native IL-2 signals, but do not interact with the wildtype cytokine or receptor [96]. The authors first created an IL-2Rβ double mutant (denoted orthoIL-2Rβ) with no detectable binding to wildtype IL-2, and subsequently leveraged yeast surface display-based evolution to identify an IL-2 mutein (denoted orthoIL-2) with five mutations in the IL-2/IL-2Rβ interface that bound to orthoIL-2Rβ but not wildtype IL-2Rβ. In later work, Hirai et al. demonstrated that orthoIL-2 treatment selectively expanded orthoIL-2Rβ-transduced Tregs in vivo, resulting in improved tolerance of heart allografts following adoptive transfer in a mixed hematopoietic chimerism mouse model [97]. The orthoIL-2/orthoIL-2Rβ system was also found to enhance the efficacy of Treg therapy following allogeneic hematopoietic stem cell transplantation, leading to GVHD protection while maintaining graft-versus-tumor (GVT) responses in mouse models [98].

3.3. IL-2 Muteins that Simultaneously Modulate Binding to IL-2Rα and IL-2Rβγ_c

IL-2 muteins with synergistic mutations in both the IL-2Rα and the IL-2Rβγ_c binding interfaces of the cytokine have also been developed to induce targeted Treg expansion (Fig. 2d). Using the aforementioned high-affinity IL-2Rα mutein (2–4 IL-2) developed by Rao et al. as a template [83], Liu et al. developed selective Treg antagonists by incorporating either V91R or Q126T residue substitutions at the IL-2Rβ and γ_c subunit binding interfaces, respectively [99]. The resulting mutants competitively antagonized wildtype IL-2 signaling. While Treg antagonism is not desirable for autoimmune disease therapy, this study presented a promising approach to target Tregs with engineered human IL-2 analogs. In another example, Ghelani et al. generated a panel of engineered IL-2 muteins with attenuated affinities for both IL-2Rα and IL-2Rβ to assess the threshold IL-2 signal required for induction of biased Treg cell responses [100]. One mutein, which contained a single IL-2Rα interface mutation (M104V) and 2 IL-2Rβ interface mutations (V91K, D20A), exhibited a >25-fold reduction in signaling potency on CD25^HighFOXP3⁺ gated Tregs within human PBMCs, and led to significant decreases in expression of FOXP3, IL-2Rα, and CTLA4. However, this mutein was still found to promote Treg-mediated suppression of effector T cells.

Picciotto et al. also targeted the IL-2Rα and IL-2Rβ interfaces simultaneously, developing an IL-2 mutein with the affinity-enhancing V69A and Q74P IL-2Rα interface mutations and the disruptive N88D IL-2Rβ interface mutation [101]. This mutein was developed as a fusion to human serum albumin (which also extends half-life through FcRn-mediated recycling) creating HSA-IL2m, which induced STAT5 phosphorylation exclusively on Tregs within a mixed population of human PBMCs [101,102]. Subcutaneous delivery of lipid nanoparticles with mRNA encoding for HSA-IL2m led to selective expansion of Tregs in both mice and non-human primates, corresponding with reduced disease severity in mouse models of acute GVHD and EAE.

4. Computational Approaches for IL-2 Engineering

In addition to rational design and directed evolution approaches, there has been growing interest in developing computational algorithms to engineer biased cytokines based on underlying biophysical concepts. One such computational approach is that of de novo design, in which principles of protein folding are used to sample the full potential amino acid sequence space and engineer proteins with desired structural and functional properties whose sequences are unrelated to those observed in nature (Fig. 2c) [103]. Physical constraints are captured in an energy function, which is used to derive the lowest energy state for each possible sequence, and the sequences whose lowest energy state matches most closely to the target structure are selected for experimental investigation. Using computational de novo design approaches in conjunction with subsequent rounds of experimental evolution employing the yeast surface display platform [79], Silva et al engineered a variant of IL-2 that eliminates the natural bias of IL-2 towards IL-2Rα^High cells by erasing the interface with IL-2Rα and enhancing the cytokine’s interaction with the IL-2Rβγ_c receptor heterodimer [104]. They demonstrated that their engineered molecule, denoted Neo-2/15 preferentially activates immune effector cells relative to Treg cells. This molecule advanced to a Phase 1 clinical trial for cancer (NCT04659629) sponsored by the company Neoleukin, but the study was halted prematurely [105]. Although immune effector cell bias represents the opposite of what is desirable for autoimmune disorders, the same fundamental principles could be used to design IL-2 variants that would accentuate bias towards IL-2Rα^High cells, based on appropriate receptor binding or structural conformation criteria.

In an alternative approach, Ren and colleagues built on the natural topology of IL-2 and focused on stabilizing the core protein structure rather than protein-protein interfaces to favor interaction with the IL-2Rβγ_c heterodimer [106]. Using this approach, they derived variants of IL-2 (denoted S1–S17) that were predicted to have enhanced interaction with IL-2Rβγ_c and recapitulate the immune effector cell bias of the aforementioned H9 mutein [22], which exhibits 200-fold higher affinity for the IL-2Rβ subunit compared to wildtype IL-2. Importantly, the computationally designed IL-2 variants functioned as intended directly from computational predictions, and did not require in vitro engineering to elicit immune bias [106]. Again, although the bias was designed to favor immune effector cell engagement, this same approach could theoretically be used to stabilize conformations of the cytokine that favor interaction with IL-2Rα^High Tregs.

5. PEGylation

Another emerging strategy in altering and optimizing cytokine activity is polymer conjugation, which involves attaching PEG chains to cytokines (known as PEGylation). This enhances the cytokine’s pharmacokinetic profile by increasing the cytokine’s hydrodynamic volume, reducing clearance of the PEGylated cytokine to extend its serum half-life [107,108]. PEGylation also has the capability to modulate cytokine/receptor interactions. In the specific case of IL-2, bias towards particular receptor subunits has been pursued to achieve selective activation of either immune effector cells or Tregs [109–112].

5.1. Nonspecific PEGylation

Whereas early work in IL-2 PEGylation focused on extending half-life of natural IL-2 in the context of cancer treatment [113,114], more recent efforts have emphasized modulation of IL-2 signaling in addition to half-life extension. Nektar and Bristol Myers Squibb (BMS) co-developed a PEGylated IL-2 variant called NKTR-214 (BEMPEG) [115], which was the first PEGylated IL-2 intended to bias function in addition to extending serum half-life. NKTR-214, which contained an average of 6 releasable PEG chains, was shown to favor activation of immune effector cells over Tregs by preferentially blocking the epitope on IL-2 that interacts with IL-2Rα through conjugating bulky PEG groups to the cytokine via amine reactivity. These PEG groups were designed to release over time, allowing for interaction of NKTR-214 with activated effector cells (which upregulate IL-2Rα), creating a positive feedback loop for immune effector cell activation. In addition to its IL-2 biasing activity, NKTR-214 also significantly improved the pharmacokinetic profile of the cytokine, leading to promising results in multiple mouse tumor models [115]. Unfortunately, human trials showed no clinical benefit for NKTR-214 as a monotherapy or in combination with immune checkpoint inhibitor antibodies (NCT03729245; NCT03635983), and toxicities were also observed; thus, clinical investigation of this molecule has been discontinued [116–118].

Despite the failure of the immune effector cell-biased PEGylated IL-2, Nektar went on to develop a PEGylated IL-2 variant denoted NKTR-358 (rezpegaldesleukin or LY3471851), which was engineered to instead direct cytokine effects towards Tregs over effector cells [109]. The investigators designed a panel of PEGylated IL-2 variants via amine reactivity and selected NKTR-358 based on its molecular binding characteristics; this IL-2 variant showed attenuated binding affinities for both IL-2Rα and IL-2Rβ compared to native IL-2. The limited interaction with IL-2Rβ led to significantly impaired activity on IL-2Rα^Low immune effector cells, whereas IL-2Rα^High Tregs retained sufficient affinity for NKTR-358 to preserve activation, thus leading to Treg bias [109]. The investigators showed that NKTR-358 induced sustained and selective Treg activity with minimal activation of immune effector cells in vitro [109], and a single cumulative dose of NKTR-358 in non-human primates also showed improved pharmacokinetic properties compared to continuous dosing of low-dose IL-2 [109]. Furthermore, NKTR-358 treatment ameliorated disease progression in mouse models of SLE [109]. This molecule is currently undergoing investigation in clinical studies (see Section 7.1) [119–121].

5.2. Site-specific PEGylation

Zhang et al. addressed an important challenge associated with nonspecific PEGylation of cytokines, which was the undesired production of heterogeneous and random conjugates, making it difficult to precisely modulate receptor binding profiles [110]. The investigators sought to eliminate variability while also enhancing Treg bias by performing PEGylation of IL-2 in a site-specific and well-characterized manner [110]. They specifically coupled PEG chains of variable molecular weights to key cytokine residues in the interface with IL-2Rβ by incorporating site-specific azide-bearing amino acids and performing copper-free click reactions [110]. Their most promising candidate was the variant with dual 20 kilodalton (kDa) PEG chains positioned at the Tyr31 and Thr51 sites on IL-2 (denoted dual-31/51–20K). Dual-31/51–20K demonstrated a mildly attenuated affinity for IL-2Rα and an ~80-fold decrease in affinity for IL-2Rβ, resulting in selective binding to and preferential activation of Tregs over immune effector cells, similar to NKTR-358 [110]. In vivo, this molecule showed enhanced pharmacokinetic properties and improved therapeutic efficacy compared to native IL-2 in murine models of SLE, GVHD, and collagen-induced arthritis [110].

Ikeda and colleagues also engineered a novel site-specific Treg-biased PEGylated IL-2 variant, denoted as KKC80, using similar methods [110,111]. Like dual-31/51–20K, KKC80 exhibited a moderately reduced affinity for IL-2Rα and a significant decrease in binding affinity to IL-2Rβ, biasing IL-2 towards Treg activation and improving upregulation of Treg markers, while also extending serum half-life. These enhanced properties led to amelioration of GVHD in mouse models and suppression of inflammation in monkey models of delayed-type hypersensitivity [111].

Ptacin et al. from Synthorx, which was acquired by Sanofi, also developed site-specific PEGylated IL-2 variants that are biased towards Treg activation. Their approach was unique in that they incorporated unnatural amino acids, which were synthesized using their Expanded Genetic Alphabet platform, into specific sites on IL-2, and then covalently attached PEG chains to these residues [112]. They identified the mono-PEGylated THOR-809 (SAR444336) as the optimal candidate, and showed that this variant induced preferential expansion and activation of Tregs with minimal stimulation of immune effector cells, while also improving cytokine half-life [112]. Collectively, these various approaches to PEG-mediated cytokine bias offer an exciting opportunity for controlled regulation of IL-2 activity through polymer conjugation.

6. Treg-Biased IL-2 Fusion Proteins

Another strategy for biasing IL-2 is through fusion to a receptor subunit, allowing for intramolecular engagement and/or cytokine redirection. One such approach is fusion of IL-2 to the IL-2Rα receptor subunit via a non-cleavable amino acid linker [122–125]. In addition to biasing cytokine behavior, increasing the overall size of the resulting molecule also extends the cytokine’s in vivo half-life [122]. Ward et al. pursued this approach to pre-engage IL-2 with IL-2Rα, bypassing the cellular requirement for trimeric IL-2 receptor expression and theoretically favoring engagement of cells that express the dimeric IL-2 receptor (i.e., immune effector cells) [122]. However, they unexpectedly found that in fact their IL-2/IL-2Rα fusion protein (denoted IL-2/CD25) primarily formed inactive head-to-tail dimers wherein the IL-2 and IL-2Rα bind on adjacent molecules. Slow dissociation of the dimer results in a low concentration of active monomer that simulates low-dose IL-2 therapy and leads to selective activation of Tregs over immune effector cell populations (Fig. 3a) [122,124,125]. Work by DeOca et al. on a separate IL-2/IL-2Rα fusion protein demonstrated that their molecule also formed higher order multimers that were biased towards Treg activation [123]. The molecule designed by Ward et al. demonstrated preclinical efficacy in mouse models of T1D, as well as SLE [124,125]. However, at high doses, this fusion protein in fact activated and expanded immune effector cells; thus careful dosing optimization will be required for translation of this approach [126,127]. Additionally, the fusion proteins described herein used the mouse cytokine and receptor proteins; thus, clinical translation will require characterization of human IL-2/IL-2Rα fusion protein function.

Fig. 3.

Treg-biased IL-2 fusion proteins and IL-2/anti-IL-2 antibody complexes. a IL-2/IL-2Rα fusion proteins form inactive head-to-tail dimers which slowly dissociate, resulting in a low concentration of long-lived active monomer that behaves similarly to low-dose IL-2 and selectively activates Tregs. b A dual-acting cytokine fusion protein comprised of IL-2 and a tumor necrosis factor receptor 2 (TNFR2)-selective single-chain TNF mutein (scTNF_R2). The EH-domain containing 2 (EHD2) protein was used to form a dimeric molecule. The fusion protein stimulates both the IL-2 and TNFR2 signaling pathways on Tregs. c IL-2 can be complexed with an anti-IL-2 antibody that sterically occludes IL-2 from engaging with either IL-2Rβ or γ_c, thereby preventing activation of immune effector cells (left). In the presence of IL-2Rα (right), the antibody dissociates from IL-2, allowing the cytokine to engage with cells expressing the trimeric IL-2R, thus favoring Tregs. d Treg-biased cytokine/antibody fusion proteins (immunocytokines) function in the same manner as IL-2/anti-IL-2 antibody complexes but covalently tether IL-2 to the antibody via a flexible linker. Immunocytokines overcome limitations of IL-2/anti-IL-2 antibody complexes, such as dosing optimization, increased regulatory hurdles, and concerns of complex dissociation leading to off-target effects and toxicity.

In another strategy, mouse IL-2 was fused to a tumor necrosis factor receptor 2 (TNFR2)-selective single-chain TNF mutein (denoted scTNF_R2), using an EH-domain containing 2 (EHD2) protein to form a dimeric molecule. The resulting fusion protein simultaneously activated the IL-2 and TNFR2 signaling pathways, acting through multiple mechanisms to induce biased expansion and strong potentiation of Tregs (Fig. 3b) [128]. First, TNFR2 promotes Treg expansion, and a maximally suppressive subset of Tregs is defined by TNFR2 expression [129–132]. Second, IL-2 activity on Tregs enhances TNFR2-dependent expansion [133,134]. This work provides a promising dual-acting synergistic mechanism that can be harnessed to bias IL-2 towards Tregs.

7. IL-2/anti-IL-2 Antibody Complexes and Fusion Proteins

It has been found that complexing IL-2 with certain anti-IL-2 antibodies can enhance therapeutic efficacy and reduce toxicity of the cytokine by biasing the cytokine towards specific immune cell subsets while also increasing its in vivo half-life. A pioneering study by Boyman and colleagues discovered that the JES6–1 antibody, which binds to mouse IL-2, biases the activity of the cytokine towards Tregs over immune effector cells [135]. Subsequent structural studies demonstrated the molecular rationale for this biased signaling. Specifically, JES6–1 sterically obstructs IL-2 binding to the IL-2Rβ and γ_c subunits, but also allosterically exchanges IL-2 with the IL-2Rα subunit, leading to strong bias towards IL-2Rα^High cells (i.e., Tregs) (Fig. 3c) [135,136]. IL-2/JES6–1 complexes showed promise in mouse models of autoimmune disease [135–140], but they are limited in translatability since they utilize mouse IL-2. Recently, 2 antibodies against human IL-2 were reported that bias IL-2 towards Treg activation: the F5111.2 antibody [141]; and the UFKA-20 antibody [142]. F5111.2, when complexed with human IL-2, sterically blocks IL-2 from binding to IL-2Rβ and also induces disruptive perturbations that favor binding to the IL-2Rα interaction site on the cytokine. Similar to the JES6–1 antibody, dissociation of the F5111.2 antibody is required for selective activation of Tregs through a release/exchange mechanism (Fig. 3c). IL-2/F5111.2 complexes demonstrated therapeutic efficacy in mouse models of T1D, xenogeneic GVHD, and MS [141]. The UFKA-20 antibody preferentially stimulates Tregs through a similar release/exchange mechanism; however, this antibody sterically obstructs IL-2Rα binding in addition to obstructing IL-2Rβ binding (as does F5111.2). However, the interference with IL-2/IL-2Rβ binding is greater than that with IL-2/IL-2Rα binding, which functionally biases IL-2/UFKA-20 complex activity towards biased expansion of Tregs in both mice and rhesus macaques [142].

To overcome the translational limitations of IL-2/anti-IL-2 antibody complexes, such as dosing optimization, increased regulatory hurdles, and concerns of complex dissociation leading to off-target effects and toxicity of the free cytokine, single-agent fusion proteins (immunocytokines) that intramolecularly fuse IL-2 to anti-IL-2 antibodies have been developed (Fig. 3d) [143,144]. In one instance, mouse IL-2 was fused to the N-terminus of the JES6–1 antibody light chain via a flexible linker, which led to enhanced Treg bias in vitro and in vivo, as well as improved efficacy in a mouse model of colitis compared to the IL-2/JES6–1 complex [143]. In another instance, the F5111 antibody (the parent antibody of F5111.2) was fused in a similar manner to human IL-2. The resulting immunocytokine (F5111 IC) showed improved Treg bias as compared to the human IL-2/F5111.2 antibody complex in both cellular and animal models. Promisingly, this fusion protein conferred robust protection in mouse models of colitis and immune checkpoint inhibitor-induced diabetes mellitus [144].

8. Outlook and Perspectives

Low-dose IL-2 has shown considerable promise in the treatment of various autoimmune and inflammatory conditions, as well as in transplantation medicine [39,73,145,146]. In a single major histocompatibility complex (MHC)-mismatched skin transplant model in humanized mice, low-dose IL-2 treatment preferentially expanded donor‐specific Tregs, thereby improving allograft survival [147]. In another study, pre-treatment with the Treg-biasing IL-2/JES6–1 antibody complex facilitated long-term acceptance and function of fully MHC-mismatched lung allografts in mice [148]. However, low-dose IL-2 therapy has been limited by harmful toxicities, a narrow therapeutic window due to off-target activation of immune effector cells, and short serum half-life. To overcome these barriers, there has been immense activity in engineering IL-2 to improve its pharmacological properties through such approaches as direct manipulation of IL-2 to create muteins, conjugation to PEG, and development of fusion proteins or cytokine/antibody complexes. Each of these approaches carries unique advantages and disadvantages, but all have shown promise in preclinical studies of autoimmune diseases, and several are under active investigation in clinical trials.

8.1. Clinical Progress of Engineered IL-2-Based Compounds

Early clinical data evaluating the ability of engineered IL-2 therapies to selectively stimulate effector cells and induce anti-cancer immunity have been disappointing [149], indicating that biased IL-2 variant activity may differ between preclinical models and clinical experience. These results underscore the importance of rigorously characterizing the clinical performance of Treg-biased IL-2-based therapies in autoimmune diseases and transplantation medicine. As of 2023, engineered IL-2 compounds have not received regulatory approval in any country for the treatment of autoimmune or inflammatory diseases [76]. A total of 9 next-generation IL-2-based compounds have entered Phase I or Phase II clinical trials for a range of indications including SLE, atopic dermatitis, psoriasis, ulcerative colitis, autoimmune hepatitis, RA, GVHD, and T1D. However, most of these trials to date have yielded disappointing or mixed results. Of the 17 trials completed thus far, 5 were terminated due to a lack of efficacy, and there are currently 6 IL-2-based therapies undergoing clinical trials, all of which are IL-2 muteins with increased IL-2Rα subunit dependency.

RO7049665 (Roche), which was developed from IgG-(IL-2N88D)₂, is an IgG1-conjugated, IL-2Rα-biased IL-2 mutein containing the N88D mutation [87]. The compound was studied in a Phase I trial for treatment of ulcerative colitis (NCT03943550) and a Phase II trial for treatment of autoimmune hepatitis (NCT04790916), yet both studies were terminated due to a lack of efficacy [150,151]. Similarly, an IL-2 mutein-Fc fusion protein, CC-92252 (Bristol-Myers Squibb and Celgene), was tested in a Phase I trial in healthy individuals and patients with psoriasis (NCT03971825), but progression criteria were not met, thus the study was terminated early [152].

AMG 592 (Amgen) is an IL-2 mutein Fc fusion protein with increased IL-2Rα dependence that has been assessed in 6 Phase I or Phase II clinical trials [153]. In a Phase Ib multiple ascending dose study in patients with SLE (NCT03451422), substantial increases in Treg cell numbers were observed following treatment with AMG 592 [154]. A Phase II study of AMG 592 in patients with ulcerative colitis (NCT04987307) is ongoing, despite recent termination of a Phase Ib/IIa study in patients with active RA (NCT03410056) and a Phase II study in patients with SLE (NCT04680637) due to lack of efficacy [155–157]. A similar mutant IL-2-Fc fusion protein, with mutations that increase affinity to IL-2Rα and decrease IL-2Rβ binding affinity (MK-6194, Merck), was tested in a Phase Ia single ascending dose clinical trial [158,159]. The protein was well-tolerated and selectively expanded total Tregs (particularly IL-2Rα^High Tregs) [159,160], and MK-6194 is now being evaluated in 2 Phase Ib/IIa clinical trials in patients with ulcerative colitis (NCT04924114) and atopic dermatitis (NCT05450198) [161,162]. XmAb27564 (Xencor), an IL-2 mutein-Fc fusion protein with enhanced binding affinity for IL-2Rα, was well-tolerated in a Phase I study in healthy volunteers (NCT04857866) and induced Tregs (particularly IL-2Rα^High Tregs) [163,164], motivating ongoing Phase Ib multiple-ascending dose studies in patients with atopic dermatitis and psoriasis. In an alternative approach to protein-based IL-2 therapies, Moderna developed mRNA6231, a lipid nanoparticle (LNP)-encapsulated mRNA encoding for a Treg-selective IL-2 mutein fused to human serum albumin [101]. A Phase I dose escalation study of mRNA6231 is ongoing in healthy adult participants (NCT04916431) [165].

NKTR-358 (Nektar) (see Section 4.1), a PEGylated IL-2 with attenuated binding to both IL-2Rα and IL-2Rβ, has been assessed in 9 trials to date, the most of any engineered IL-2 therapeutic [109]. Overall, Phase I clinical trials showed that NKTR-358 treatment led to increased Treg numbers in patients with SLE [166], psoriasis, and atopic dermatitis [120], and led to improved disease scores in the cases of psoriasis and atopic dermatitis [119,167]. In a Phase II follow-up study in adults with SLE (NCT04433585), a mid-level dose of NKTR-358 improved SLE Disease Activity Index (SLEDAI), but did not achieve the primary endpoint (4-point SLEDAI score reduction) [168,169]. Therefore, use of NKTR-358 for SLE is no longer being explored; however, based on the promising Phase I data, Nektar intends to conduct a Phase IIb study in patients with atopic dermatitis [120,121].

8.2. On the Horizon: Novel Engineering Strategies for IL-2 Therapies

Although engineered IL-2 compounds (Table 1) have demonstrated considerable advantages compared to low-dose IL-2 therapy, their clinical development has faced a number of limitations. When administering Treg-biased IL-2 muteins, higher doses have been required due to their reduced signaling potency through the IL-2Rβ subunit [170]. Consequently, there is a delicate balance between the concentration of engineered IL-2 muteins required to potentiate Tregs and concentrations that will elicit undesirable activation of immune effector cells. The pharmacokinetic properties of IL-2 muteins have been improved through PEGylation, but PEG chains with a molecular weight of greater than 5 kDa can potentially induce anti-PEG antibodies, leading to unwanted immunogenicity [171]. Furthermore, the large hydrodynamic volume of PEGylated IL-2 variants may also limit tissue penetration, hindering their ability to reach the intended target cells. Pharmacokinetic limitations of IL-2 muteins have also been overcome by fusion to IgG Fc domains or through design of IL-2/anti-IL-2 antibody fusion proteins, which increase the half-life, avidity, and specificity of the cytokine [172]. Indeed, an exploration of additional cell surface targets on Tregs also revealed that IL-2Rα itself is a maximally unique surface target for Tregs, making avidity the most promising route to bias IL-2 towards Tregs [172]. However, an inherent limitation for IL-2/anti-IL-2 antibody fusion protein production is the potential formation of intermolecular oligomers [144].

Despite efforts to bias IL-2 towards Tregs, it is important to note that IL-2 therapies evoke non-selective expansion of Tregs, whereas it has been shown that antigen-specific Treg cells are significantly more effective in suppressing effector T cell activity compared to polyclonal Treg cells [173,174]. Antigen specificity can be improved through integration with other immunomodulatory platforms and delivery approaches. For instance, targeting lipid-encapsulated mRNA encoding IL-2 muteins, similar to the approach taken by Moderna with its candidate mRNA-6231, could allow for development of localized, antigen-specific immune tolerance [101]. Recently, Kishimoto et al. developed biodegradable nanoparticles encapsulating rapamycin (ImmTOR) to induce selective immune tolerance to co-administered antigens [175]. When ImmTOR was administered with Treg-selective IL-2 muteins or a single-chain IL-2/anti-IL-2 antibody fusion (F5111 IC, see Section 6.2), they observed an increase in the number of total Tregs, as well as profound synergistic increase in antigen-specific Treg numbers when combined with a target antigen. These results suggest that immunomodulatory compounds could increase the therapeutic window of engineered IL-2 therapies while facilitating durable, antigen-specific immune tolerance. In a similar strategy to expand antigen-specific Tregs for MS treatment, Rhodes et al. designed biodegradable microparticles (Tol-MPs) loaded with rapamycin and functionalized with (F5111 IC) and MHC class II/myelin peptide complexes [176]. By acting as surrogate antigen-presenting cells, Tol-MPs promoted disease prevention and reversal in mouse models of MS.

Fusion with antibodies targeting disease antigens is an alternative method to achieve tissue-specificity with biased IL-2 therapies. Sun et al. created an asymmetric Fc-conjugated heterodimer comprising an immune effector cell-biased IL-2 mutein and an anti-human EGFR Fab (Erb-sumIL2) [177]. The targeted IL-2 fusion protein not only improved specificity for EGFR⁺ tumor tissue but also efficiently activated tumor-infiltrating CD8⁺ T cells. While this strategy was applied in the context of cancer immunotherapy, a similar design could be implemented to treat autoimmune diseases by conjugating a Treg-selective IL-2 variant to an antibody targeting tissue- or disease-associated antigens.

The activity of engineered IL-2 compounds can also be restricted to certain tissues or cell types through a receptor “masking” strategy [102]. By tethering a receptor subunit to an IL-2 variant via a protease- or pH-cleavable linker, the cytokine is rendered inactive until exposed to specific proteases or pH conditions at the disease site. This strategy was implemented to develop a prodrug (ProIL2) comprised of an immune effector cell-biased IL-2 mutein Fc fusion protein tethered to IL-2Rβ via a matrix metalloproteinase-sensitive linker [178]. ProIL2 was activated upon cleavage by tumor-associated enzymes (which allowed for receptor dissociation) and was found to preferentially expand antigen-specific CD8⁺ T cells in the tumor microenvironment. Another masked IL-2 prodrug, XTX202 (Xilio Therapeutics), with bias towards IL-2Rβγ_c is currently being evaluated clinically to treat advanced solid tumors (NCT05052268) [179,180]. WTX-124 (Werewolf Therapeutics), a conditionally active IL-2 prodrug containing an inactivating Fab domain and a half-life extension domain, is also being evaluated in a Phase I trial in patients with advanced solid tumors (NCT05479812) [181,182]. Another prodrug, known as IL-2 β/γ (Ascendis Pharma), was created by permanently attaching a methoxy polyethylene glycol (mPEG) moiety in the IL-2Rα binding site and transiently conjugating a 40 kDa mPEG carrier for sustained release [183,184]. IL-2 β/γ is currently in a Phase I cancer trial in combination with an immune checkpoint inhibitor and/or chemotherapy (NCT05081609). Adaptation of these approaches to engineer Treg-biased IL-2 molecules for autoimmune disease treatment would require design of prodrugs conditionally activated by disease-specific features, such as proteases, pH conditions, or other environmental factors.

Preferential binding of IL-2 to Tregs might also be achieved through engineered scaffolds that stabilize the interactions between IL-2 and the heterotrimeric receptor complex. Such stabilization has been achieved for the heterodimeric IL-2 complex, wherein Spangler et al. used the yeast surface display directed evolution platform to develop “stapler” single-chain variable fragments (scFvs) that stabilized the IL-2/IL-2Rβ interface, leading to a 15-fold enhancement in interaction affinity [185]. Moving forward, this engineering approach could be leveraged to stabilize the interactions between IL-2 and IL-2Rα, and stapler scFvs could potentially be incorporated into bispecific antibodies to modulate IL-2 function in specific tissues for targeted autoimmune disease therapy.

8.3. Conclusions

In summary, this review highlights the immense potential of IL-2 as an immunoregulatory therapeutic for the treatment of autoimmune and inflammatory conditions, as well as in transplantation medicine. Despite its demonstrated potential in autoimmune disease treatment, the clinical use of IL-2 has been complicated by its pleiotropic effects on both immunostimulatory and immunosuppressive cell types. To maximize the specificity of IL-2 for Tregs in the context of autoimmune diseases, while also improving its pharmacokinetic profile, various protein design approaches have been investigated. While many engineered IL-2-based molecules have shown promising results in preclinical and early clinical studies, there is still a need to induce disease-localized, antigen-specific Treg expansion while minimizing cytokine-mediated toxicities. Overall, novel cytokine engineering technologies have led to significant advancements in the field of IL-2 immunotherapy, and have opened the door for new possibilities in the targeted treatment of autoimmune and inflammatory diseases.

Key Points.

1. IL-2 is a pleiotropic cytokine that coordinates immune cell activities and has been used as a therapeutic agent for various diseases including cancer and autoimmune disorders. The clinical use of IL-2 for autoimmune diseases is limited by toxicity, lack of specificity for Tregs, and short serum half-life.

2. Molecular engineering of IL-2 has been explored to bias its effects towards Tregs, employing strategies such as unbiased, structure-guided, or computational design of IL-2 mutants, chemical modification with polyethylene glycol, and fusion with other proteins or antibodies.

3. Engineered IL-2 variants have demonstrated reduced toxicity, improved pharmacokinetics, and enhanced efficacy for treating autoimmune diseases in preclinical and early-stage clinical studies; however, further research and optimization is required for medical translation of these IL-2-based compounds.