Engineered antibody fusion proteins for targeted disease therapy

Payload immunocytokines are therapeutic fusion proteins that comprise a cytokine fused to an antibody or antibody fragment that is specific to a disease-associated antigen or other cellular marker. These proteins allow safe, systemic, and targeted delivery of cytokine payloads in a manner that potentiates their therapeutic efficacy. Payload immunocytokines have been preclinically and clinically investigated for a range of disease applications [ 5 , 8 , 9 ]. This section discusses the most recent applications of these novel biopharmaceuticals in cancer, chronic inflammation, and autoimmunity ( Figure 1A - J and Table 1 ). Payload immunocytokines have been most widely investigated as cancer therapeutics. In recent formulations, antibodies have been fused to cytokine payloads including interleukin (IL)-4, IL-15, TNFα, IL-2, IL-12, IL-21, interferon (IFN)-α, and IFN-β. These cytokines may have immunostimulatory, immunosuppressive, or pleiotropic roles in homeostasis. This section focuses on payload immunocytokines that exploit the immunostimulatory activities of the fused cytokine for cancer immunotherapy ( Table 1 ). The antitumor mechanisms of immunocytokines discussed herein are detailed in Table 1 . To deliver cytokine payloads to the tumor microenvironment (TME), recently developed immunocytokines have incorporated antibodies that target extracellular matrix (ECM) components, specifically disease-associated isoforms of fibronectin. One such antibody, denoted F8, recognizes the alternatively spliced extra domain (ED)A of fibronectin, which is highly expressed in neovasculature at sites of inflammation but is undetectable in most healthy tissues [ 10 ]. Several cytokines have been fused to recombinant fragments of the F8 monoclonal antibody and have achieved preclinical and clinical success in cancer. One scFv-based fusion molecule containing IL-4, denoted F8-IL4-F8 ( Figure 1A ), was found to localize to the tumor and inhibit the growth of murine teratocarcinoma [ 11 ]. A diabody-based fusion protein that linked F8 to IL-15 (F8F8-IL15; Figure 1B ) showed preferential localization to solid tumors and significantly reduced tumor progression, both as a monotherapy and in combination with soluble TNFα [ 12 ]. Finally, combination treatment with a diabody-based F8/IL-2 fusion protein (F8-IL2; Figure 1C ) and a scFv-based F8/TNFα fusion protein (F8-TNF; Figure 1D ) cured a proportion of teratocarcinoma, fibrosarcoma, and lung carcinoma tumors, although no protective immunity against rechallenge was acquired [ 13 ]. Interestingly, an F8 scFv-based dual cytokine fusion protein incorporating both TNFα and IL-2 (IL2-F8-TNF; Figure 1E ) localized to neoplastic lesions in vivo , eradicated F8-IL2- and F8-TNFα-resistant sarcomas, and produced durable and protective responses against colon, myeloid leukemia, and teratocarcinoma tumors [ 14 ]. Another ECM-targeting antibody that has been incorporated in recently developed immunocytokines, denoted L19, engages the alternatively spliced EDB of fibronectin, enabling selective localization to tumor-associated blood vessels [ 15 ]. Recently, fusion proteins comprising the L19 scFv tethered to TNFα [ 16 - 18 ], IL-12 [ 17 , 19 ], or IL-2 [ 20 , 21 ] have demonstrated antitumor efficacy. L19-TNFα ( Figure 1D ) was found to improve murine tumor remission in sarcoma lesions when combined with either chemotherapy or immune checkpoint inhibition [ 16 ]. Furthermore, both L19-TNFα and L19-IL12 ( Figure 1B ) were found to accumulate in, and cure, a percentage of murine orthotopic gliomas [ 17 ]. L19-IL12 has also demonstrated potent antitumor activity against murine sarcoma and colon carcinoma, with durable tumor clearance achieved when the immunocytokine was combined with immune checkpoint blockade [ 19 ]. Another study demonstrated the elimination of murine myeloma tumors by L19-TNFα and L19-IL2 ( Figure 1B ) both independently and combinatorially [ 18 ]. Another L19 scFv/IL-2 fusion protein (L19L19-IL-2; Figure 1B ) was found to induce superior tumor growth inhibition compared with L19-IL2 both alone and in combination with checkpoint blockade [ 20 ]. In a recent Phase II clinical trial, L19-IL2 was found to be well tolerated and was synergistic with the chemotherapeutic agent dacarbazine, increasing the proportion of patients whose tumors were reduced or eradicated [ 21 ]. L19-TNFα, L19-IL12, and L19-IL2 are currently being explored in various stages of human clinical trials as monotherapies and/or in combination with other immunocytokines and established cancer therapies ( Table 1 ). An alternative tumor targeting strategy capitalizes on the free DNA fragments found in necrotic tumor regions. Specifically, the human lgG1 antibody NHS76 [ 22 ], which binds histones, has been conjugated to the IL-12 heterodimer (IL-12p70) to enhance tumor localization (NHS-IL12; Figure 1F ). Fusion to the full-length antibody increased the plasma half-life and antitumor efficacy of IL-12 [ 23 ]. NHS-IL12 demonstrated synergistic effects with radiation, chemotherapy, a cancer vaccine, or immune checkpoint inhibition, once again underscoring the therapeutic potential of immunocytokines in combination therapy approaches [ 23 , 24 ]. In a Phase I clinical trial, NHS-IL12 was found to be well tolerated and immunostimulatory [ 25 ]. Later-stage human clinical trials are currently focused on combination therapies that integrate NHS-IL12 with non-cytokine-based antibody fusion proteins and other cancer treatments ( Table 1 ). Other immunocytokine approaches have also been employed for the targeted delivery of proinflammatory cytokines to the TME with preclinical success. For many of these approaches, an affinity-impaired cytokine has been used to ensure that the antibody targeting domain, rather than the cytokine payload, directs the localization of the fusion protein. A full IgG antibody specific to programmed cell death protein 1 (PD-1) was fused to an affinity-impaired IL-21 variant ( Figure 1F ) to deliver IL-21 to PD-1-expressing T cells while also increasing the serum half-life of IL-21 and minimizing off-target effects on local antigen-presenting cells [ 26 ]. The resulting anti-PD-1-IL-21 immunocytokine provided protection in an anti-PD-1 -refractory humanized mouse model of melanoma [ 26 ]. In another example, the full human epidermal growth factor receptor 2 (HER2)-targeted antibody trastuzumab (Herceptin®) was fused to a glycoengineered IFN-β variant ( Figure 1F ). The resulting anti-HER2-IFN-β immunocytokine localized to HER2 + tumors in vivo and potentiated the effects of either IFN-β or trastuzumab alone to promote tumor regression in xenograft and humanized HER2 + tumor models [ 27 ]. Activity on target cytokines (AcTakines), IFNs (AcTaferons), and TNFs (AcTafactors) represent a novel class of immunocytokines wherein a targeted sdAb is fused to an affinity-impaired cytokine. A fusion protein comprising a CD20-targeted sdAb and a reduced-affinity IFN-α variant ( Figure 1G ) significantly reduced growth of CD20 + lymphoma and melanoma tumors without inducing systemic toxicity [ 28 ]. An anti-C-type lectin domain containing A (Clec9A)-IFN-α AcTakine ( Figure 1G ) targeted reduced-affinity IFN-α to Clec9A + dendritic cells (DCs), significantly attenuating tumor growth in several syngeneic and humanized mouse tumor models in the absence of toxicity. Combining the AcTakine with immune checkpoint blockade, chemotherapy, or low-dose soluble TNFα led to complete tumor regressions [ 29 ]. In another study, the anti-CD13-TNFα Ac Tafactor ( Figure 1G ) was found to localize to the tumor neovasculature, resulting in robust control or complete eradication of tumors when combined with either engineered T cell therapy or a CD8-targeted IFN-α AcTakine [ 30 ], further accentuating the potential for payload immunocytokines in oncology, particularly when used in combination therapies. Due to their inherent localization to sites of inflammation, ECM-targeting antibodies have also been used to direct anti-inflammatory cytokine activity to sites of inflammation in chronic inflammatory and autoimmune disorders ( Table 1 ). Despite being less clinically advanced than immunocytokines for oncological applications, immunocytokines designed to suppress inflammation by the incorporation of F8, L19, and other antibodies have shown preclinical promise in arthritis, chronic skin inflammation, endometriosis, chronic cardiac allograft rejection, atherosclerosis, and inflammatory bowel disease [ 5 , 8 ]. In recent advances, novel immunocytokines that incorporate IL-4, IL-9, IFN-α, and IL-10 have been explored preclinically for treatment of chronic inflammation and autoimmune diseases. In a mouse model of collagen-induced arthritis, F8-IL4 ( Figure 1C ), in combination with dexamethasone, induced a complete cure of the disease [ 31 ]. In a mouse model of experimental pulmonary hypertension, F8-IL9-F8 ( Figure 1A ) decreased right ventricular systolic pressure and improved echocardiographic parameters [ 32 ]. As an alternate approach to ECM-based targeting, a Clec9A-targeted AcTaferon successfully localized IFN-α activity to Clec9A + DCs, resulting in protection from disease in a mouse model of multiple sclerosis without the toxicity associated with soluble IFN-α [ 33 ]. In another study, a fusion protein comprising an anti-CD86 scFv fused to an Fc domain and an impaired-affinity monomeric IL-10 ( Figure 1H ) induced the differentiation of CD86 + tolerogenic DCs, which could be therapeutically exploited for graft-versus-host disease (GVHD) [ 34 ]. Taken together, the results of these recent preclinical and clinical investigations highlight the therapeutic potential of immunocytokines in cancer, chronic inflammation, and autoimmunity. Immunocytokines overcome challenges associated with cytokine-based therapies, including short serum half-life, pleiotropy, and systemic toxicity, to permit safe, targeted, and potent cytokine activity at the site of disease. Importantly, these fusion proteins have demonstrated the most preclinical and clinical promise when used in combination with other immunocytokines or other immunotherapy approaches, underscoring their potential for synergy in emerging therapeutic strategies.

Developments in structure-guided molecular engineering have facilitated the design of biased payload immunocytokines that modulate the activities of the incorporated cytokine. This section discusses signal modulating immunocytokines, focusing largely on those that incorporate IL-2, which is currently the most actively studied and advanced cytokine with respect to immunocytokine development ( Table 1 ). Due to its vital role in orchestrating the immune response, IL-2 has been employed for the treatment of both cancer and autoimmune disorders. However, the clinical performance of IL-2 is hampered by its harmful systemic toxicities and short serum half-life (<5 min) as well as its activation of both immunostimulatory cells [i.e., CD8 + T cells and natural killer (NK) cells] and immunosuppressive cells [regulatory T cells (T Reg S)] [ 35 - 37 ]. IL-2 signals through either an intermediate-affinity heterodimeric receptor comprising the IL-2 receptor (IL-2R)β and common gamma (γ C ) chains, or a high-affinity heterotrimeric receptor, which also contains the nonsignaling IL-2Rα (CD25) chain ( Figure 2A ) [ 38 ]. In contrast to naïve effector cells, T Reg s robustly express IL-2Rα and therefore have dramatically higher sensitivity to IL-2. The design of IL-2 immunocytokines that improve safety by biasing cytokine activity towards particular immune cell subsets, while also increasing IL-2 half-life, introduces an enticing opportunity for targeted cytokine therapy. One promising approach, first proposed by Boyman et al. , is the development of anti-IL-2 anti-bodies that selectively direct IL-2 activity towards either effector cells or T Reg s [ 39 ]. Specifically, the S4B6 antibody biased IL-2 activity towards the expansion of effector cells, and IL-2/S4B6 complexes effected potent antitumor activity in mice without inducing systemic toxicity [ 40 - 42 ]. The JES6-1 antibody induced preferential T Reg expansion and IL-2/JES6-1 complexes prevented diabetes [ 43 , 44 ] and other autoimmune diseases [ 45 , 46 ] in mice. Fusion of IL-2 to anti-IL-2 antibodies has shown tremendous promise in overcoming translational challenges for cytokine/antibody complexes, such as dissociation and dosing optimization. Both S4B6 and JES6-1 have been genetically fused to IL-2 ( Figure 2B ) to produce intramolecularly assembled immunocytokines that are more potent than the respective complexes [ 47 , 48 ]. However, both S4B6 and JES6-1 specifically recognize mouse IL-2 and do not cross-react with the human cytokine. Recently, human IL-2 was grafted into a complementarity-determining region (CDR) of NARA1 [ 49 ], an antihuman IL-2 antibody that recapitulates the effector cell biasing activities of S4B6. The resulting immunocytokine, NARA1 leukin ( Figure 2A ), demonstrated superior pharmacokinetic properties to, and therapeutically outperformed, IL-2/NARA1 complexes in a mouse model of melanoma. These promising findings illustrate the therapeutic advantage of cytokine-antibody fusion proteins [ 50 ]. As an alternative to anticytokine antibodies, mutagenically biased cytokines, termed muteins, can be fused to antibodies targeting disease antigens to create biased payload immunocytokines. For example, fusion of an IL-2 mutein with abolished IL-2Rα binding (IL-2v) to an anticarcinoembryonic antigen (CEA) antibody (CEA-IL2v; Figure 2B ) inhibited tumor growth and enhanced survival in CEA + tumor-bearing mice [ 51 ]. CEA-IL2v also synergized with other antibody immunotherapies to reduce the tumor burden in mouse models of CEA-expressing breast and colon cancers [ 52 ]. Furthermore, CEA-IL2v was found to synergize with CEA-targeting chimeric antigen receptor (CAR) T cells to induce complete tumor eradication and protective immunity from rechallenge in a mouse model of CEA + colon carcinoma [ 53 ]. Another fusion protein replaced one Fab of an anti-epithelial growth factor receptor (EGFR) antibody with super mutant IL-2 (sumIL-2), which has reduced IL-2Rα binding activity ( Figure 2B ) [ 54 ]. The resulting molecule inhibited tumor growth in a mouse model of melanoma. Similar to IL-2, IL-15 has also been incorporated in biased payload immunocytokines. IL-15 shares the IL-2Rβ and γ C chains of IL-2R, while also engaging the private high-affinity IL-15Rα chain (CD215). To enhance the NK and memory T cell-potentiating activities of IL-15, receptor-linker-IL-15 (RLI) was developed, wherein IL-15 is linked to the minimal binding domain of the IL-15Rα chain to enhance the interaction with the IL-2Rβ and γ C chains [ 55 ]. A fusion protein that linked RLI to an anti-CD20 antibody extended survival in a mouse model of lymphoma while also substantially increasing the circulation half-life [ 56 ], illustrating the multifactorial advantages of biased payload immunocytokines. Conditional steric blocking, or ‘masking’, of cytokines can also improve biasing. For example, both an effector cell-biased IL-2 variant and the IL-2Rβ extracellular domain (ECD) were separately fused to a Fc domain ( Figure 2C ) [ 57 ] such that the IL-2Rβ chain bound and ‘masked’ the fused IL-2, which was attached to its respective Fc domain via a cleavable linker. The masked IL-2 payload was thereby selectively released in the presence of tumor-associated matrix metalloproteinases, allowing the molecule to exhibit superior efficacy compared with IL-2 as well as powerful synergy with checkpoint blockade therapy. Two recent patents described similar IL-2 fusion proteins using antibody fragments as masks and reported their successful use in mouse models of colon cancer [ 58 , 59 ]. Masking strategies have been extended beyond the IL-2 system. A recent patent described the generation of masked IL-12 in various formats, including IL-12 fused to a cleavably linked mask as well as anti-CD20 scFvs ( Figure 2C ) [ 60 ]. One of these masked IL-12 fusion proteins significantly reduced tumor growth in a mouse model of colon cancer, highlighting the specificity and potency of masking technologies.

Antibody-based fusion proteins have demonstrated extraordinary clinical value, and the explosion of innovative molecular modifications, fusion formats, and protein payloads in preclinical development forecasts further therapeutic advancement. Antibody-based targeting of payloads to specific cells or tissues can dramatically increase the potency of the payload, allowing the administration of lower doses and preventing undesirable side-effects. The protein payloads fused to antibodies also exhibit reduced renal clearance and increased circulating half-life, diminishing the need for repeated dosing. Genetic fusion of the payload to an antibody ensures a fixed stoichiometric ratio of antibody to payload, eliminating formulation optimization, dosing, and consistency concerns during clinical translation. In this review, we summarized engineered antibody fusion proteins that incorporate cytokine, enzyme, toxin, neuroprotective agent, and soluble factor trap payloads. These fusion proteins have the versatility to address a vast array of disease indications. Immunocytokines, RITs, and ADEPTS have all advanced to clinical trials, with several achieving full FDA approval. While the most clinically advanced molecules have been focused on oncological applications, immunocytokine- and RIT-based treatment of infectious, chronic inflammatory, and autoimmune diseases are certain to soon follow. Use of antibody fusion proteins to treat neurological conditions and hereditary degenerative diseases is also close on the horizon [ 107 - 115 ]. To date, numerous Fc-fused therapeutic proteins have been FDA approved, but clinical progress has been comparatively limited for the analogous targeted antibody fusion proteins. Minimizing off-target activity of the antibody and payload is an engineering and design challenge for all classes of fusion proteins reviewed (see Outstanding questions ). The development of novel fusion strategies and the incorporation of peptide and protein motifs that provide context-dependent activity will further realize the potential for this emerging category of drugs. One of the most prominent challenges for payload immunocytokines that incorporate natural cytokines is that the multifarious activities of cytokines stimulate both immunostimulatory and immunosuppressive programs. The incorporation of modified cytokines that have attenuated or biased signaling shows expansive potential for new, safer forms of immunotherapy. Additionally, modifications to the antibodies themselves to conditionalize activity on stimuli such as pH or ionic content may enhance payload efficacy [ 119 ]. An ongoing challenge for RITs and ADEPTS is that their therapeutic activity often wanes rapidly with repeated administration due to immunogenicity. Elimination of immunogenic epitopes and the use of endogenous toxins and enzymes can overcome this limitation. More generally, antibody fusion protein development is sure to benefit from, and advance, combination treatment approaches that incorporate other immunotherapeutic modalities. Overall, the future looks bright for antibody fusion proteins, as ongoing efforts promise to empower the clinical coming of age for this innovative class of immunotherapeutics.

Various antibody fusion proteins involving non-cytokine payloads have been developed for therapeutic applications. Fc fusions are generally the most clinically advanced among these approaches, with more than a dozen having received FDA approval, and many more in various stages of therapeutic development [ 7 , 61 , 62 ]. These fusion proteins are designed to capitalize on the extended half-life of Fc domains or to serve as a scaffold to bring two proteins (e.g., receptor subunits) into close proximity, but they do not leverage the targeting capacity of antibodies. Nonetheless, Fc fusions laid the groundwork for the development of newer antibody fusion proteins that employ the targeting activity of antibodies. This section focuses on translationally promising antibody fusion proteins that take advantage of the targeting activity of antibodies, including immunotoxins, enzyme-antibody fusion proteins, neuroprotective protein-antibody fusions, and targeted soluble factor traps ( Table 2 and Figure 3 ). RITs are fusions of toxic moieties to antibody fragments ( Figure 3A - C ). RITs have primarily been developed as cancer therapies [ 63 - 65 ], although in some research RITs have been designed for infectious disease applications [ 66 - 68 ]. In both cases, the antibody portion of the RIT functions to target the toxic payload to diseased cells. The toxic moieties can be categorized by their mechanisms of cytotoxicity: pore-forming toxins (PFTs) ( Figure 4A ); ribosome-inactivating proteins (RIPs) ( Figure 4B ); and microtubule-disrupting proteins (MDPs), which prevent mitosis and vesicular transport and thereby prompt apoptosis [ 63 , 68 , 69 ]. RIPs are the most diverse and developed class of toxins used in RITs, whereas MDPs are the least. Both categories require the antibody portion of the RIT to facilitate endocytosis to effect cell killing. By contrast, most PFTs directly integrate with the cell membrane and therefore do not require antibody-mediated endocytosis to exert cytolytic functions ( Figure 4A ) [ 70 , 71 ]. The integration of PFTs in the cell membrane is often sensitive to the overabundance of lipids that are synthesized by certain tumor cells, which restricts toxicity to those target cells. RIT activity also usually requires the dense antigen expression that is preferentially observed on tumor cells, as is the case for the HER2-targeted gigantoxin-4-4D5scFv ( Figure 3A ) [ 70 ]. ‘Caged’ RITs were developed to further restrict PFT activity to the context of target cells. For instance, DHFR-(furin)-FraC-Nb and DHFR-(furin)-ClyA-Nb ( Figure 3A ) fuse a highly water-soluble, small, nontoxic dihydrofolate reductase (DHFR) protein to the PFT RIT via a furin cleavage site-containing linker, ensuring that the PFT is active only in the context of the furin-rich TME ( Figure 4A ) [ 71 ]. RITs that are currently in clinical trials or are FDA approved utilize one of three bacterially derived toxins, all of which are RIPs: Pseudomonas aeruginosa exotoxin A (PE) [ 72 - 82 ], diphtheria toxin (DT) [ 83 - 85 ], and Shiga toxin 1 (Stx1) [ 86 , 87 ] ( Figure 3B ). PE has been modified extensively for use in RITs to increase cytotoxicity, reduce immunogenicity, and increase resistance to protease degradation [ 65 , 68 , 88 ]. The two most clinically advanced RITs, moxetumomab pasudotox (LUMOXITI®) [ 73 ] and oportuzumab monatox [ 75 ], both incorporate optimized PEs. DT has also been optimized for use in RITs and serves as the payload for several molecules in clinical trials [ 64 , 83 , 85 ]. Only one Stx1 payload-bearing RIT is in clinical trials [ 86 ]. A major outstanding challenge for RITs is that many patients develop antidrug antibodies to the toxic moiety, which can accelerate clearance and/or neutralize the toxic moiety, dramatically hindering therapeutic performance. Moreover, the size of many immunotoxins has limited their clinical application in oncology to the treatment of hematological cancers, as most RITs are unable to achieve adequate penetration of solid tumors. Additionally, inefficient cytosolic delivery of RITs necessitates extraordinary potency of the cytotoxic payload. To circumvent immunogenicity challenges, researchers have sought to humanize the antibody portion of the RIT, eliminate immunogenic epitopes from exogenous toxins, and identify endogenous human toxin alternatives [ 65 , 88 ]. To address tumor penetration issues, new streamlined RIT formulations employing alternative antibody fragments such as nanobodies have been developed [ 89 , 90 ]. To enhance RIT potency, protein features such as endoplasmic reticulum localization sequences (KDEL) and/or protease cleavage sites have been incorporated to facilitate the release of the toxin into the cytosol [ 91 ]. Although the vast majority of the development and optimization to date has focused on exogenous toxins, endogenous RIP toxins such as granzyme B and granulysin have recently been preclinically developed and are tremendously appealing since they have no immunogenicity [ 65 , 92 - 94 ]. Another RIT approach targets a modified superantigen (SAg) to the tumor [ 95 - 98 ]. Unlike the other toxins described, which directly disrupt cellular function, SAgs such as the modified staphylococcal enterotoxin A used in naptumomab estafenatox recruit and activate T cells to kill target cells ( Figures 3C and 4C ). Overall, RITs show tremendous promise for the targeted elimination of cancerous and/or malignant cells. The toxins in many of the RIP-based RITs discussed above are enzymes, but a number of other enzymes that are not directly cytotoxic have also been employed in antibody fusion proteins. Antibody-directed enzyme prodrug therapy (ADEPT) is a strategy in which a cancertargeting antibody-enzyme fusion protein and an enzyme-activatable prodrug are administered successively ( Figure 4D ) [ 99 ]. In contrast to most RITs, the antibody in ADEPT fusion proteins should not trigger receptor-mediated endocytosis, as the enzyme must remain outside the cell to access the prodrug. A fusion protein comprising a CEA-specific antibody fragment and bacterial carboxypeptidase G2 (CPG2), denoted MFECP1 ( Figure 3D ), showed promising safety and localized to patient tumors in early clinical trials, but was limited by immunogenicity [ 99 - 102 ]. CPG2 must form a homodimer for enzymatic activity and it is possible that instability of the dimer in vivo limited the therapeutic efficacy. Preclinical development of ADEPTS continues, but as with MFECP1, clinical translation will require modification to overcome immunogenicity challenges [ 103 , 104 ]. Antibody-enzyme fusion proteins for targeted enzyme-replacement therapy represent another class of molecules that has advanced to clinical trials. The antibodies in these fusion proteins facilitate the transport of the enzyme payload to lysosomes, where they replace the function of the mutated enzymes ( Figure 4E ) to treat genetic conditions such as Pompe disease and mucopolysaccharidosis [ 105 , 106 ]. One fusion protein, VAL-1221, also facilitates the delivery of the enzyme to the cytosol, where it degrades glycogens that have leaked from lysosomes [ 107 , 108 ]. Other Trojan horse’ antibody-enzyme fusion proteins (e.g., HIRMAb-enzyme fusion proteins; Figure 3E ) promote transport across the blood–brain barrier in addition to facilitating endocytosis via the human insulin receptor (HIR). These molecules have shown potential in preclinical studies as well as Phase I and II clinical trials [ 109 - 112 ]. HIRMAb fusion proteins have also been expanded to deliver nonenzymatic agents [ 113 ]. For instance, HIRMAbs fused to neuroprotective payloads such as glial cell line-derived neurotrophic factor (GDNF) and erythropoietin (EPO) ( Figure 3F ) have shown preclinical promise as potential treatments for acute ischemic stroke and traumatic brain injury [ 114 , 115 ]. In another example, fusion of HIRMAb to the ECD of TNFR2 serves as a ‘soluble factor trap’ ( Figure 3G ), depleting cytotoxic TNFα in the brain following traumatic injury [ 116 ]. In a cancer-directed soluble factor trap, bintrafusp alfa sequesters immunosuppressive transforming growth factor beta (TGF-β), while also targeting tumor cells via programmed death-ligand 1 (PD-L1) ( Figures 3G and 4F ) [ 117 , 118 ]. Bintrafusp alpha is being extensively investigated, with more than three dozen active Phase I–III clinical trials.

Antibody (see Glossary) therapeutics have become a cornerstone of the pharmaceutical market owing to their favorable drug-like properties [ 1 - 4 ]. More recent efforts have sought to exploit the power of these targeted therapies to create novel antibody fusion proteins that offer several advantages over other drug formats. Antibodies have a long circulation half-life due to the presence of fragment crystallizable (Fc) domains, which can be leveraged to enhance the duration of fused short-lived therapies such as cytokines [ 5 ]. Furthermore, the highly specific binding activities of antibodies allows targeting of fused therapies that would otherwise be dangerous to administer systemically; for instance, due to their inherent toxicity or their propensity for pleiotropic behavior [ 3 ]. In addition, the unimolecular format of antibody fusion proteins obviates the need for chemical conjugation and circumvents downstream dosing, manufacturing, and optimization challenges that could impede regulatory approval. Several antibody fusion proteins are already approved as therapeutics. As early as 1998, etanercept [a fusion of tumor necrosis factor (TNF) receptor 2 (TNFR2) to an Fc domain] was approved for the treatment of rheumatoid arthritis [ 6 ]. Since this approval, antibody fusion proteins have been diversified by the linking of various payloads to different antibody components, including full-length immunoglobulins, Fc domains, single-chain variable fragments (scFvs), single-domain antibodies (sdAbs), and antigen-binding fragments (Fabs). This review focuses on recent preclinical and clinical progress in the development of antibody fusion proteins for targeted disease therapy. Specifically, we address cytokine-antibody fusion proteins, including those that target cytokine payloads to particular cell types of interest as well as those that directly or indirectly bias cytokine signaling. We also discuss recent developments in the fusion of antibodies to non-cytokine payloads, such as toxins, enzymes, neuroprotective agents, and soluble factor traps. Given the recent surge in both clinical approvals and the market share of antibody-derived drugs [ 7 ], the advances described herein promise to be of broad interest to researchers across the field of drug development.