Molecular Engineering of Therapeutic Cytokines

Over the past three decades, a large body of work has been directed at the development of therapeutic cytokines. Despite their central role in immune modulation, only a handful of cytokine therapeutics has achieved regulatory approval. One of the major challenges associated with the therapeutic use of cytokines relates to their short serum half-life and low bioavailability. High doses are required to overcome these problems, which often result in dose-limiting toxicities. Consequently, most cytokines require protein engineering approaches to reduce toxicity and increase half-life. For this purpose, PEGylation, fusion proteins, antibody complexes and mutagenesis have been utilized. Here, we summarize past, recent and emerging strategies in this area.


Introduction
Cytokines are a large family of mainly soluble proteins and glycoproteins that function as key modulators of the immune system.They encompass interleukins (ILs), interferons (IFNs), growth factors, colony stimulating factors (CSFs), the tumour necrosis factor (TNF) family and chemokines (or chemotactic cytokines).These small signalling molecules (<30 kDa) are mostly secreted by leukocytes, but can also be produced by other cell types (e.g., endothelial cells, epithelial cells and fibroblasts).Cytokines can function in an autocrine, paracrine or endocrine manner to stimulate or suppress the activity of target cell populations.Signals conveyed by cytokines are essential for generation, survival and homeostasis of immune cells, as well as for the generation of immune responses upon external stimuli.Due to their natural role as immune modulators, many cytokines have OPEN ACCESS been identified as suitable therapeutic candidates for the treatment of a number of infectious, inflammatory, autoimmune and malignant diseases.
The discovery of cytokines was guided by initial observations that soluble biological factors could mediate inflammation, cellular stimulation and anti-viral activities.For instance, the pro-inflammatory cytokine, interleukin-1 (IL-1), was initially described as a pyrogenic factor isolated from cultured rabbit leukocyte supernatants [1,2].Similarly, reports of a T-cell growth factor present in mixed leukocyte cultures led to the identification of IL-2, a master regulator of T-cell function and development [3][4][5].Furthermore, IFNs were first reported as soluble secreted factors mediating viral interference in chick chorioallantoic membranes [6,7].In fact, pioneering cytokine immunotherapy in humans was performed with IFN-containing (<1%) protein fractions extracted from growth medium of virus-stimulated human leukocytes [7,8].
The emergence of recombinant DNA technology in the 1970s represented a turning point for exploiting the therapeutic potential of cytokines.Similarly important was the development of high performance liquid chromatography, which eased the laborious task of obtaining cytokine preparations pure enough for characterization [7].Using these techniques, the first cytokines to be cloned and synthesized in E. coli were IFN-α and IFN-β in 1980 [7,9,10] and IL-2 in 1983 [11,12].Initial regulatory approval of recombinant IFN-α for the treatment of hairy cell leukaemia was obtained in 1986, while recombinant IL-2 was approved for treating metastatic melanoma in 1992.Additional therapeutic applications for IFN-α, IL-2 and a number of other cytokines have since been approved or are currently under trial, as summarized in Table 1.

Cytokine Toxicity
Cytokine immunotherapy often results in the development of severe dose-limiting side effects.Two properties shared by most cytokines are thought to play a crucial role in the development of treatment-associated adverse effects.Firstly, cytokines are pleiotropic, meaning they are able to influence more than a single cell type.In fact, some cytokines are able to stimulate cell types that mediate opposing biological effects.Furthermore, cytokines have a short serum half-life and, thus, need to be administered at high doses to achieve their therapeutic effects [13].While effectively enhancing therapeutic efficacy, high doses exacerbate pleiotropic activities that manifest as adverse effects in patients.Molecular engineering of cytokines with prolonged half-life, enhanced specificity or localized activity is therefore required to enhance the pharmacological properties of these proteins.Engineered cytokine therapeutics that have been approved for clinical use or are undergoing clinical trials are summarized in Table 2.
A prominent example of cytokine-associated toxicity is the adverse effects observed in patients receiving high-dose (HD) IL-2 cancer immunotherapy.IL-2 stimulates the proliferation of cytotoxic CD8 + T-cells and NK cells, which both act to promote tumour regression [14].Due to its anti-tumorigenic activity, HD IL-2 therapy has been approved for the treatment of metastatic melanoma and renal cancer, with overall objective responses observed in approximately 16-17% of patients [14][15][16].However, HD IL-2 therapy can cause excessive vascular permeability, which leads to an adverse state known as vascular leak syndrome (VLS).Symptoms arising from VLS can be life-threatening, making intensive patient management a requirement for the use of IL-2 immunotherapy [16].

Cytokine PEGylation
One of the earliest strategies used for half-life extension of therapeutic cytokines was PEGylation, in which the cytokine is conjugated to polyethylene glycol (PEG) chains (5-40 kDa, Figure 1).PEGylation can be achieved chemically or enzymatically, with biophysical and biochemical properties of PEGylated proteins depending on structure, size, number and location of the PEG chains [17].PEGylation prolongs the circulation half-life of proteins by masking proteolytic cleavage sites and, more importantly, by increasing their hydrodynamic radii, thereby effectively reducing renal clearance [18,19].2).Granulocyte colony stimulating factor (G-CSF) is a 19.6 kDa monomeric cytokine produced by monocytes and fibroblasts that functions to stimulate the growth and differentiation of neutrophils.Recombinant human G-CSF produced in E. coli (filgrastim) is used for the treatment of neutropenia, a pathological reduction in neutrophil numbers often observed in patients receiving myelosuppressive chemotherapy.Daily doses of filgrastim over the course of a chemotherapy cycle are required to prevent neutropenia, mainly because its short serum-half life (t 1/2 = 3.5 h) only allows for transient therapeutic action.By significantly increasing serum half-life (t 1/2 = 42 h), a PEGylated form of G-CSF (pegfilgrastim) has been used to mediate an equivalent therapeutic effect with only a single dose per chemotherapy cycle [20].Similarly to pegfilgrastim, PEGylated forms of IFN-α-2a (Pegasys ® , branched 40 kDa PEG) and IFN-α-2b (Pegintron ® , linear 12 kDa PEG) show approximately 10-fold increases in circulation half-life, correlated with improved therapeutic efficacy for the treatment of chronic hepatitis C [21,22].
Although PEGylation is an effective method for half-life extension, toxicity of PEGylated cytokines remains largely unchanged [20,22,23].While an overall benefit is observed in therapies that use cytokines at relatively low doses (e.g., G-CSF, IFN-α), PEGylation may be detrimental when high doses are required (e.g., HD IL-2), as increased half-life could prevent rapid reversal of toxicity [24,25].

Antibody-IL-2 Immune Complexes
Neutralizing monoclonal antibodies (mAbs) directed against cytokines are commonly used for blocking interactions between cytokines and their receptors.However, several studies in the early 1990s showed dramatic increases in the biological activity of a number of cytokines when mice were administered with a mixture containing cytokine/anti-cytokine mAb at a 2:1 molar ratio [14,[26][27][28][29].These observations suggest that some anti-cytokine mAbs act as cytokine carriers in vivo, identifying a potential use of antibody-cytokine immune complexes in therapeutic applications (Figure 1).Subsequent studies have extensively characterized IL-2/anti-IL-2 mAb complexes, demonstrating that these complexes not only display enhanced biological activity, but also modulate the pleiotropic effects of IL-2 [30].
IL-2 is a major modulator of the immune system capable of stimulating cells bearing either dimeric or trimeric IL-2 receptors (IL-2Rs).The low-affinity dimeric IL-2R (K D = 1 nM), composed of IL-2Rβ (CD122) and γ c subunits, is found in CD122 high populations, such as memory phenotype (MP) CD8 + T-cells and NK cells.In contrast, the high-affinity (K D = 10 pM) trimeric IL-2R consisting of IL-2Rα (CD25), CD122 and γ c subunits is found in CD25 high populations, such as activated T-cells, activated B-cells and regulatory T-cells (Tregs).Notably, signal transduction in both types of IL-2Rs is mediated by CD122 and γ c , while the CD25 chain acts to confer high-affinity binding of IL-2 to trimeric IL-2Rs [32].
Further to modulating IL-2 activity, IL-2/mAb complexes can also prolong the biological effects of IL-2.While reduced renal clearance is suggested to play a major role in the increased half-life observed in IL-2 complexes, additional mechanisms are thought to contribute to this feature [14].These mechanisms include reduced receptor-mediated endocytosis (for IL-2/mAb CD122 complexes) and indirect recycling of IL-2 by the neonatal Fc receptor (for IL-2/mAb CD25 complexes), which rescues IgG antibodies from cellular metabolism [41,42].
Based on the enhanced properties of IL-2/mAb complexes in animal models, the use of mAbs as cytokine carriers has emerged as a potential strategy for improving pharmacodynamics and pharmacokinetics of therapeutic cytokines.However, the therapeutic value of this approach is yet to be tested in clinical trials.

Cytokine Fusion Proteins
Genetic fusion to other proteins is a useful strategy for improving or modifying the biophysical properties of cytokine therapeutics (Figure 1).A large body of work has been directed towards the development of cytokine fusion proteins that display extended half-life (e.g., fusion to Fc, albumin or transferrin), increased activity (e.g., fusion to cytokine agonists), enhanced cytotoxicity (e.g., fusion to bacterial toxins or lytic Fc) and localized delivery (e.g., fusion to antibodies or antibody fragments), with several of them currently undergoing clinical trials (Table 2).

Fusion of Cytokines to the Fc Domain
Several advantages are introduced by fusing the Fc domain of IgG antibodies to therapeutic proteins (Figure 1).These include increased avidity of the fused partner, cost-effective purification via protein G/A affinity chromatography and, more importantly, prolonged circulation half-life [43][44][45][46][47]. Notably, half-life extension of Fc fusion proteins results from a combination of reduced renal clearance (due to increased molecular size) and FcRn-mediated recycling, where Fc binds FcRn at acidic pH in endosomes and is then released at physiological pH to the plasma [41].Moreover, expression of FcRn in lung epithelium has been shown to mediate pulmonary uptake of Fc fusion proteins, thereby providing an alternative route of administration particularly relevant to the treatment of respiratory conditions [45,46].
In addition, effector functions mediated by Fc can significantly alter the biological activity of Fc fusion proteins.These functions include antibody dependent cell-mediated cytotoxicity (ADCC), which occurs through binding of the Fc domain to Fc receptors (FcR) on immune cells, as well as complement-mediated cytotoxicity (CDC), induced by binding to C1q [48].Notably, Fc effector functions can be modulated through mutagenesis, modification of glycosylation status and choice of Fc subclass [44,48,49].For instance, etanercept (Enbrel ® ), a TNFRII-Fc fusion protein approved for the treatment of rheumatoid arthritis and other inflammatory conditions, functions through both blocking the action of the pro-inflammatory cytokine TNF-α and inducing ADCC of TNF-α expressing cells [49].By contrast, the Fc domain of abatacept (Orencia ® ), an inhibitor of T-cell co-stimulation, contains mutations that prevent activation of ADCC and CDC [50].In addition, an engineered Fc domain with reduced effector functions has been utilized for the generation of an IL-21R-Fc fusion protein, which antagonizes the action of the pro-inflammatory cytokine IL-21 and shows therapeutic potential for the prevention of allograft rejection and the treatment of inflammatory and autoimmune conditions [51][52][53].
While Fc fusion technology has been successfully implemented for the blockade of cytokine action via cytokine receptor-Fc fusions, direct fusion of Fc to cytokines is a useful strategy for enhancing cytokine activity.For instance, fusion of Fc to interleukin-10 (IL-10) has been shown to significantly prolong the biological activity of IL-10 in mice [54][55][56][57].IL-10 is an anti-inflammatory and immunosuppressive homodimeric cytokine mainly produced by monocytes, but also by B-cells, NK cells and a number of T-cell subsets.Treatment with murine IL-10-Fc containing a modified, non-lytic Fc domain (i.e., abrogated ADCC and CDC) resulted in extended protection against endotoxin-induced septic shock [57] and in prevention of spontaneous autoimmune diabetes in non-obese diabetic (NOD) mice [56].While possessing the same biological functions of unfused IL-10, the prolonged serum half-life of IL-10-Fc resulted in superior therapeutic activity [56].Based on these animal models, similar approaches could be attempted to enhance human IL-10 activity, since this cytokine displayed insufficient therapeutic efficacy in clinical trials of rheumatoid arthritis [58].Other developed cytokine-Fc fusions displaying enhanced biological activity in mice include a cytolytic IL-2-Fc, which targets destruction of activated T-cells in autoimmune conditions [59] and a nonlytic IL-7-Fc with potential as a vaccine adjuvant for induction of T-cell mediated immune responses [60].However, whether analogous fusion proteins have the potential of promoting comparable biological effects in humans remains unclear.

Fusion to Serum Albumin
Similarly to IgG antibodies, serum albumin (SA) displays an unusually long circulation half-life (~3 weeks in humans).Interestingly, half-life prolongation of these functionally and structurally unrelated proteins results primarily from interaction with FcRn.Although SA binds FcRn at a different site than IgG, both interactions are pH-dependent and result in FcRn-mediated rescue from cellular catabolism [61].Due to its high potential for half-life prolongation, the use of SA as a carrier protein has been implemented in a number of therapeutics [13].Some of the strategies utilized for this purpose include genetic fusion to SA, conjugation to SA-binding moieties and fusion to SA-binding antibodies or antibody fragments [13].Notably, cytokines that have been genetically fused to SA include IL-2 [62], G-CSF [63], IFN-α [64][65][66] and IFN-β [67].
Albinterferon-α2b (albIFN), a genetic fusion of IFN-α2b and SA, is a prominent example of cytokine pharmacokinetic enhancement through combination with albumin (Figure 1 and Table 2).Phase III clinical trials demonstrated that the therapeutic efficacy of albIFN was superior to that of IFN-α2b and not inferior to PEGylated IFN-α2b (PEGintron ® ) in mediating sustained virologic responses in chronic hepatitis C patients [64,66].Furthermore, fusion to SA extended the elimination half-life of IFN-α2b to ~200 h, thus allowing administration of albIFN in two-or four-week dose intervals, as opposed to weekly (PEGintron, t 1/2 ~40 h) or tri-weekly dosing regimes (IFN-α2b, t 1/2 ~4 h) [21,65].
Recently, cytokines have been fused to domain antibodies (dAbs) that bind SA (Figure 1).Often referred to as nanobodies, dAbs are single heavy (V H ) or light (V L ) chain variable domains, which are only 11-13 kDa in size and capable of binding antigen [68].Albumin-binding dAbs (AlbudAbs TM ) have been used to generate genetic fusions to IFN-α2b and IL-1 receptor antagonist (IL-1Ra), a naturally occurring anti-inflammatory cytokine that blocks IL-1 signalling (Anakinra ® , approved for treatment of RA) (Table 1).The resulting fusion proteins displayed a marked increase in half-life, which correlated with superior efficacy [69,70].IL-1Ra-AlbudAb mediated enhanced reduction of arthritis scores in a mouse model of collagen-induced arthritis, when compared to IL1-Ra [69], while IFN-α2b-AlbudAb showed superior activity than that of an IFN-α2b-SA fusion protein, as measured by in vitro anti-viral activity and mediation of tumour regression in a mouse model of melanoma [70].

Fusion to Human Transferrin
Genetic fusion to serum transferrin (Tf) represents another strategy for cytokine optimization primarily resulting in enhanced pharmacokinetics.Serum transferrin is a 80 kDa glycoprotein that mediates iron transport from systemic circulation into cells and tissues [71].When bound to ferric ions, transferrin displays high affinity for the transferrin receptors (TfRs) displayed on the surface of most cell types [71].Upon interaction, the Tf/TfR complex is internalized via receptor-mediated endocytosis into endosomes, where iron is released and Tf/TfR is then recycled to the cell surface [71,72].Fusion of protein therapeutics to Tf or TfR-binding antibodies has been utilized for a number of applications, including half-life extension, targeting of malignant cells overexpressing TfRs and targeting of the brain capillary endothelium for transport of therapeutics across the brain blood barrier [47,[72][73][74].Moreover, expression of TfRs in the gastrointestinal epithelium provides potential for oral delivery of Tf-associated drugs [75].
In a study from Bai et al. in 2005 [76], subcutaneous administration of a G-CSF-Tf fusion protein to BDF1 mice resulted in comparable biological activity to unfused G-CSF (filgrastim), as measured by increase in absolute neutrophil counts (ANC).Given that the G-CSF-Tf fusion only displayed one tenth of the activity of filgrastim in vitro, half-life prolongation was suggested to elevate in vivo biological activity to match that of filgrastim.Notably, oral administration of G-CSF-Tf resulted in significant increases in ANC, while ANC after oral administration of unfused G-CSF remained at baseline levels.More recently, introduction of a helical linker between G-CSF and Tf resulted in further improvements in biological activity [77].

Fusion to Agonistic Receptors
While most soluble cytokine receptors act as natural cytokine antagonists, some of them can potentiate cytokine activity [78].One such receptor is the IL-15 receptor α sub-unit, which has been shown to augment the biological activity of IL-15 [79].IL-15 is a 15 kDa monomeric common γ c cytokine that displays many overlapping functions with IL-2, as they share γ c and IL-2Rβ for signalling.Notably, IL-15 does not share IL-2Rα with IL-2, but instead has its own high-affinity (K D = 100 pM) receptor α chain, IL-15Rα.The dominant mechanism of IL-15 action in vivo appears to occur via IL-15Rα-mediated trans-presentation of IL-15 from antigen presenting cells to neighbouring cells bearing the dimeric IL-2Rβγ [80].Based on its immunostimulatory properties, the use of IL-15 is being explored in clinical trials of melanoma and renal cancer (Table 1).
Administration of soluble IL-15 in mouse models stimulates the expansion of MP CD8 + T-cells and NK cells, which express high levels of IL-2Rβγ [81].Interestingly, this effect was significantly enhanced when IL-15 was administered in combination with a soluble IL-15Rα-Fc [79].In agreement with this observation, a fusion protein composed of IL-15 and a functional domain (Sushi domain) of IL-15Rα showed significant increases in IL-15 activity [82].The conversion of IL-15 into a superagonist following IL-15α binding appears to result from conformational changes that increase the interaction with IL-2Rβγ [79].
Other cytokines that have been genetically fused to agonistic soluble cytokine receptors include the hematopoietic cytokines, IL-6 and IL-11, which have shown therapeutic potential for hepatic regeneration and prevention of thrombocytopenia, respectively [83][84][85].

Fusion to Toxins
A number of cytokines have been fused to bacterial toxins in order to promote targeted cytotoxicity of specific cell subsets (Figure 1) [86,87].Ontak ® (denileukin diftitox) is a fusion protein composed of IL-2 and the enzymatically active portion of diphtheria toxin (DT) approved for the treatment of cutaneous T-cell lymphoma (CTCL) (Table 2) [88].Malignant cells in CTCL lesions overexpress IL-2Rβγ and IL-2Rαβγ, with greater denileukin diftitox sensitivity observed in cells displaying the IL-2R heterotrimer [89].The IL-2 portion of denileukin diftitox directs DT to cells bearing IL-2Rs, ultimately resulting in cell death through inhibition of protein synthesis [90].
Partial and complete responses are observed in approximately 20% and 10% of CTCL patients after denileukin diftitox treatment, respectively [91].Nevertheless, treatment-associated toxicities are common and include the development of VLS (in ~26% of patients [91]) and, in rare cases, visual loss [92].Although the mechanisms mediating the latter effect are unknown, a direct effect on Tregswhich express high levels of IL-2Rαβγ-has been proposed to promote autoimmune retinopathy and subsequent visual impairment [93,94].
Other cytokines that have been fused to toxins for targeted cell cytotoxicity include the functionally related cytokines, IL-4 and IL-13.Fusion proteins composed of either IL-4 or IL-13 and a truncated form of Pseudomonas exotoxin (PE) were developed to target the elimination of malignant cells within glioblastoma multiforme, a type of solid brain tumour that overexpresses IL-4 and IL-13 receptors (Table 2) [87].Notably, IL-13-PE (cintredekin besudotox) reached phase III clinical trials, where it mediated increased progression-free survival in patients receiving drug infusions within resected tumour regions, when compared to patients receiving Gliadel ® (carmustine) wafers after tumour resection [95].However, no overall survival advantage was achieved with cintredekin besudotox treatment, which may have reflected sub-optimal drug delivery [95,96].

Fusion to Antibodies and Antibody Fragments (Immunocytokines)
The generation of cytokine-antibody fusion proteins is an elegant approach for engineering targeted cytokine activity (Figure 1).The rationale for the development of such fusion proteins, also known as immunocytokines, derives from the superior efficacy and reduced toxicity of localized over systemic administration of therapeutic cytokines [97][98][99].Since physical delivery of cytokines is not feasible in many instances, immunocytokines provide an alternative avenue for achieving localized cytokine action.As such, immunocytokines are designed to target disease antigens through their antibody moieties in order to induce (or potentiate) effector functions through their cytokine components.A number of immunocytokines are currently undergoing clinical trials (Table 2), with many more being evaluated in pre-clinical studies.The vast number of engineered immunocytokines and their therapeutic potential has been recently reviewed in previous issues [98,100] and elsewhere [47,101,102].A brief summary of the major factors determining the therapeutic activity of specific immunocytokines, namely, commonly targeted antigens, antibody formats and employed cytokines, is presented below.

Antibody Format
A large number of genetically engineered antibodies and antibody fragments have been utilized for the generation of immunocytokines (see [47,98,101,102]).Different antibody formats display variations in molecular size, affinity and avidity, which mainly affect pharmacokinetic aspects, such as serum half-life and biodistribution [118].Antibody formats that have been assessed in clinical trials of immunocytokines include full-length IgG, single-chain Fv (scFv) and diabodies (Figure 1) [98,101].Fusion to full-length IgG (~150 kDa) provides bivalent antigen binding and prolonged serum half-life, due to reduced renal clearance and FcRn-mediated recycling [13,114].Furthermore, effector functions mediated through the Fc domain of IgG (ADCC and CDC) can be maintained in order to potentiate targeted cell cytotoxicity [119].However, IgG-based immunocytokines display slow and non-homogeneous tissue penetration, particularly when targeting solid tumours [120].Tumour penetration can be enhanced by utilization of the smaller scFv (~30 kDa) fragments, which are composed of single V L and V H domains joined by a ~15 residue flexible polypeptide linker [121].Nevertheless, the monovalent nature of scFv fragments is correlated with poor tumour retention, while their reduced molecular size results in rapid clearance from blood circulation [122].Generation of scFv fragments with shorter linkers (~5 residues) prevents V H and V L from the same polypeptide chain from interacting and promotes the formation of bivalent scFv homodimers, also known as diabodies (~55 kDa).Similarly to scFv, diabodies display rapid tissue penetration, while displaying enhanced tumour retention and circulation half-lives [123].
Engineering of novel antibody formats is likely to continue expanding the repertoire of immunocytokines with optimized disease-targeting properties.A recent advance illustrating this trend was the generation of an IFN-α-based immunocytokine displaying multivalent and dual antigen binding, with therapeutic potential for the treatment of non-Hodgkin lymphoma [124,125].

Cytokine Partners
A number of cytokines, including IL-2, IL-7, IL-12, IL-15, GM-CSF, IFN-α, IFN-γ and members of the TNF-superfamily, have been utilized for the development of immunocytokines with potential for cancer immunotherapy [47,101].While targeted delivery of most of these cytokines into cancerous lesions aims for the recruitment and activation of cytotoxic immune cells, such as T-cells and NK cells; some others also promote anti-proliferative activities (IFN-α) [111] or induce cell death upon binding to their cognate receptors (TNF-superfamily members) [126].By contrast, the anti-inflammatory cytokine IL-10, which suppresses the expression of inflammatory mediators by immune cells, has been utilized for the development of immunocytokines targeting the neo-vasculature of chronic inflammation sites [58,127].

Cytokine Mutagenesis
Mutagenesis can be employed to engineer cytokines with enhanced stability, half-life, specificity and activity (Figure 1).In early studies, scanning and deletion mutagenesis allowed the development of cytokines with modified activities [128][129][130][131][132].These approaches have gradually been replaced by protein engineering techniques, such as rational design, computational modelling and directed evolution, which allow for more efficient cytokine optimization.

Stability
Most protein therapeutics suffer from sub-optimal stability and cytokines are no exception [133].
Poor stability can result in high levels of protein unfolding, degradation and aggregation, which can lead to many issues affecting formulation, shelf life, immunogenicity and therapeutic efficacy [134][135][136].A simple strategy for improving cytokine stability is the mutation of free cysteines to serines, thereby preventing undesired disulphide bonds that could lead to protein aggregation.This approach has been applied to a number of cytokines, including the commercially available form of recombinant human IL-2, Proleukin ® (Table 2) [134,[137][138][139]. Cytokine stability can also be enhanced by the introduction of stabilizing mutations within α-helices.This strategy is particularly relevant to the 4-helix bundle cytokines and requires the mutation of residues with low α-helical propensity (e.g., proline, glycine) to residues with high α-helical propensity (e.g., alanine).Using this approach, the introduction of double and triple glycine-to-alanine substitutions resulted in the generation of a number of G-CSF variants displaying significantly enhanced resistance to chemical denaturation [140].Furthermore, a computational method developed to identify optimal sequences for a defined three-dimensional structure was used to engineer a G-CSF variant with 10 core amino acid substitutions that displayed decreased aggregation and increased thermal stability [141].

Half-Life Extension
Degradation via receptor-mediated endocytosis plays a major role in the clearance of therapeutic cytokines [13,142].Evidence suggests that after internalization, cytokines that remain bound to their receptors in endosomal compartments (pH ~ 5-6) are destined to lysosomal degradation, while those that are released from their receptors are recycled back to the cell surface [143][144][145].Based on this premise, rationally designed G-CSF variants with reduced affinity for G-CSF receptor at endosomal pH, but unaffected binding to the receptor at physiological pH, showed increased cytokine recycling correlated with prolonged half-life and enhanced potency in vitro [146].This pH-specific reduction in affinity was achieved by mutating a number of residues involved in receptor binding to histidine, a process known as "histidine switching".Since histidine (pK a = 6.5) has a net positive charge at endosomal pH, but is neutral at physiological pH, these substitutions exclusively targeted disruption of receptor binding in endosomes, thereby promoting G-CSF recycling.

Modulation of Specificity
Modulating the specificity and activity of pleiotropic cytokines is necessary for increasing therapeutic efficacy and preventing the development of adverse effects.Accordingly, a number of cytokines have been engineered through rational design and directed evolution in order to mediate selective stimulation of specific cell subsets.
Generated by yeast display, variants of IL-2 with greater than 100-fold increases in affinity to IL-2Rα were shown to mediate enhanced expansion of T-cells in vitro [147,148].Further modification of these variants to disrupt binding to IL-2Rβ and IL2Rγ resulted in the generation of signalling-deficient analogues that act as competitive antagonists of IL-2 in cells displaying IL-2Rαβγ.The therapeutic application of such IL-2 mutants is mainly directed towards the inhibition of immunosuppressive Tregs in malignant conditions [149].
Recently, an IL-2 variant displaying ~250-fold increased affinity to IL-2Rβ was developed through in vitro evolution using yeast display [150].Interestingly, most of the five mutations resided in the core of IL-2 and resulted in a conformational change similar to the one induced after IL-2Rα binding.This mutant, also known as super-2, displayed increased ability to stimulate cells displaying IL-2Rβγ, when compared to unmodified IL-2.Accordingly, super-2 mediated more potent expansion of MP CD8 + T-cells and NK cells, which correlated with increased anti-tumour activity in mice.By shifting the balance of IL-2 stimulation towards cells displaying IL-2Rβγ, super-2 is predicted to mediate tumour regression with a reduced incidence of treatment-associated adverse effects [150].
Modulation of the activity of another common γ c cytokine, IL-4, was achieved by using a combination of rational design and directed evolution [151].IL-4 is a pleiotropic cytokine that can signal through either type I or type II (also responsive to IL-13) IL-4Rs.Importantly, regulatory functions of IL-4 are mostly mediated by cells displaying type I IL-4Rs (e.g., B-cells and T-cells) and drive the development of humoral immune responses (T H 2 responses).In contrast, effector functions of IL-4 are largely mediated by non-hematopoietic cells expressing the type II IL-4Rs and promote the development of allergic inflammation [152].
Interestingly, both types of IL-4Rs display the high affinity (K D = 100 pM) IL-4Rα subunit, differing only in the signalling subunit they incorporate, namely γ c in type I and IL-13Rα in type II IL-4Rs.Increasing the affinity of IL-4 for γ c or IL-13Rα by mutagenesis produced a number of IL-4 variants that mediated selective in vitro expansion of cells bearing either type I or type II IL-4Rs, respectively [151].In particular, IL-4 variants with enhanced selectivity for cell subsets displaying type I receptors show therapeutic potential for immune stimulation with reduced effector functions.

Development of Cytokine Antagonists
Disruption of cytokine binding to signalling receptor sub-units is a useful approach for the development of antagonists with therapeutic potential.An IL-4 variant (pitrakinra) that was rationally designed to have reduced binding to γ c and IL-13Rα, but unimpaired binding to IL-4Rα, is currently undergoing phase II clinical trials for the treatment of allergic asthma and atopic eczema (Table 2) [153][154][155].Pitrakinra effectively targets cells bearing type I and II IL-4Rs without inducing signalling, thereby acting as a dual competitive inhibitor of IL-4 and IL-13 [156].Similarly, by mutating residues in IL-6 to disrupt binding to gp130, the signalling sub-unit of IL-6R, but not affecting the binding to the IL-6Rα chain, resulted in the development of an IL-6R antagonist with therapeutic potential for the treatment of multiple myeloma and lung fibrosis [157][158][159][160].

Concluding Remarks
The immunomodulatory properties of cytokines provide exceptional potential for the treatment of several conditions.However, sub-optimal pharmacokinetics and toxicity have limited their therapeutic potential, as reflected by the reduced number of cytokines currently approved for clinical use.The therapeutic shortcomings displayed by most cytokines have motivated the optimization of these proteins through a number of molecular engineering strategies, including chemical conjugation (e.g., PEGylation), immunocomplexing, development of cytokine fusion proteins and mutagenesis.These strategies have been mainly directed towards the generation of cytokines with prolonged half-life, enhanced specificity/activity and localized action, but also to take advantage of the targeting properties of some cytokines (e.g., cytokine-toxin fusion proteins) and for the generation of cytokine antagonists.
The large number of engineered cytokines currently undergoing pre-clinical and clinical evaluation are likely to constitute the next generation of cytokine therapeutics displaying increased efficacy and reduced incidence of adverse effects.As research continues in this field, we anticipate that optimally designed cytokines will be developed utilizing a combination of molecular engineering strategies in order to simultaneously enhance multiple pharmacokinetic and pharmacodynamic parameters.

Table 2 .
Engineered cytokines approved for clinical use or undergoing clinical evaluation.