Induced proximity at the cell surface

Abstract

Molecular proximity is a governing principle of biology that is essential to normal and disease-related biochemical pathways. At the cell surface, protein–protein proximity regulates receptor activation, inhibition and protein recycling and degradation. Induced proximity is a molecular engineering principle in which bifunctional molecules are designed to bring two protein targets into close contact, inducing a desired biological outcome. Researchers use this engineering principle for therapeutic purposes and to interrogate fundamental biological mechanisms. This Review focuses on the use of induced proximity at the cell surface for diverse applications, such as targeted protein degradation, receptor inhibition and activating intracellular signaling cascades. We see a rich future for proximity-based modulation of cell surface protein activity both in basic and translational science.

Induced proximity has emerged as a powerful mode of modulating protein–protein interactions using exogenous bifunctional molecules. This engineering principle has been extensively applied to targeted protein degradation (TPD)^1–3, but its utility is expanding to include modulating posttranslational modifications^4–6 and subcellular localization^7,8. Following the initial reports of lysosome-targeting chimeras (LYTACs)⁹, antibody-based proteolysis-targeting chimeras (PROTACs), known as AbTACs¹⁰ and the related PROTABs (proteolysis-targeting antibodies)¹¹, multiple technologies have been developed around the concept of recruiting lysosome-trafficking receptors and cell surface E3 ligases for targeted degradation of cell surface proteins¹². These TPD platforms join a growing list of therapeutic modalities that induce proximity of cell surface proteins to modulate protein function. Molecular inducers of proximity elicit diverse biological outcomes such as inhibiting or activating receptor signaling, activating cellular cytotoxicity and causing surface protein transfer between cells.

A goal of this Review is to unify these emerging technologies into a broader framework and stimulate discussion around new applications for induced proximity at the cell surface. We organize this Review around four broad mechanisms by which molecular inducers of proximity exert their biological effects at the cell surface. First, we cover modalities for TPD through both lysosome-trafficking receptor and E3 ligase pathways (Fig. 1a). We then focus on bifunctional molecules that inhibit receptor signaling by inducing proximity between two cell surface proteins (Fig. 1b). Next, we highlight the opposite effect: activating receptor signaling through enforced proximity (Fig. 1c). Finally, we detail the nascent field of induced proximity modalities acting on two cells in trans (Fig. 1d). These various modalities were inspired by naturally occurring biological processes. For this reason, engineering efforts are contextualized with discussions of the natural processes on which they are based. Each section is accompanied by a forward-looking perspective focused on outstanding challenges including open mechanistic questions and gaps in current technology.

Fig. 1 |. Bifunctional inducers of proximity can elicit diverse biological effects at the cell surface through multiple mechanisms of action.

a, Degradation through lysosomal trafficking or E3 ligase-mediated ubiquitination. b, Receptor signaling inhibition. c, Activation of receptor signaling. d, By engaging multiple cells.

Targeted degradation via lysosome-trafficking receptors and cell surface E3 ligases

PROTACs and molecular glues were the original launch point of the TPD field, and several such molecules are undergoing clinical evaluation^13,14. These small-molecule proximity inducers recruit ubiquitination machinery to initiate proteasomal degradation, a mechanistic paradigm that has found broad applicability for intracellular protein targets². A new class of bifunctional molecules comprising LYTACs^9,15, AbTACs¹⁰, PROTABs¹¹ and cytokine receptor-targeting chimeras (KineTACs)¹⁶ has expanded the degradable proteome to include cell surface transmembrane proteins. These cell surface protein degraders can be divided into two major classes: (1) those that recruit a lysosomal trafficking receptor to induce target degradation and (2) bifunctional molecules that recruit plasma membrane-associated E3 ligases with intracellular ubiquitination activity (Fig. 2a). Degraders in the first class operate by hijacking natural lysosomal trafficking processes responsible for membrane protein turnover. For example, LYTACs comprise a target-binding arm, typically a monoclonal antibody, and a second arm that recruits a lysosomal trafficking receptor such as the cation-independent mannose-6-phosphate receptor (CI-M6PR or insulin-like growth factor (IGF) 2 receptor (IGF2R)). Mannose 6-phosphate, the natural ligand for CI-M6PR, is a glycan epitope installed on lysosomal enzymes that marks them for lysosomal localization and recycling¹⁷. Asialoglycoprotein receptor (ASGPR)-targeting LYTACs^15,18,19, KineTACs¹⁶ and related technologies¹² (Fig. 2b) rely on analogous mechanisms by recruiting alternative lysosome-trafficking receptors. By contrast, AbTACs^10,20, PROTABs¹¹ and REULRs (receptor elimination by E3 ubiquitin ligase recruitment)²¹ (Fig. 2c) recruit ubiquitin E3 ligases for target degradation by intracellular ubiquitination and proteasomal clearance. The latter technologies take inspiration from the natural function of RNF43 (ring finger protein 43) and ZNRF3 (zinc and ring finger 3) in degrading Wnt receptors, which in turn regulates Wnt–β-catenin signaling²². Leveraging this Wnt biology, Surrozen Wnt signal enhancers engineered for tissue specificity (SWEETs)²³ are engineered fusion proteins (anti-ASGPR–RSPO2RA) that target the degradation of the E3 ubiquitin ligases RNF43 and ZNRF3 in hepatocytes, leading to enhanced Wnt signaling and liver function restoration; this approach is currently being tested in human clinical trials²³.

Fig. 2 |. Strategies for TPD via lysosome-trafficking receptors and cell surface E3 ligases.

a, Bifunctional cell surface protein degraders act by recruiting lysosome-trafficking receptors for lysosomal clearance (left) or by ubiquitin E3 ligase-mediated proteasomal degradation (right). b, Examples of lysosome-trafficking cell surface TPD modalities. c, Examples of ubiquitin E3 ligase-recruiting degraders. d, Scheme for cell type-selective target degradation enabled by tissue-specific expression of an E3 ligase or lysosome-trafficking receptor.

Among these seemingly distinct modalities, co-opting either lysosomal or proteasomal degradation pathways, there is substantial mechanistic overlap. Programmed cell death ligand 1 (PD-L1) degradation with E3 ligase recruiting AbTACs does not depend on proteasomal function based on chemical inhibitor experiments¹⁰. Furthermore, PROTAB-mediated IGF1 receptor (IGF1R) degradation via ZNRF3 recruitment is only partially dependent on proteasome function¹¹. These findings open the door to mechanistic questions that are important for progress in the field of cell surface protein degradation as noted below. For many applications of TPD, tissue-specific function would offer improved safety profiles, thus expanding the scope of possible targets. For example, a protein driving disease in one tissue but critical to normal physiology in another tissue is not a clinically viable target for degradation unless tissue-selective degradation can be achieved. Recently developed targeted degradation platforms with selectivity for liver^15,18,19, tumor²⁴ and immune cell subtypes²⁵ aim to address this challenge (Fig. 2d). For example, tumor immune cell-targeting chimeras (TICTACs) co-opt the mannose receptor (CD206) to selectively degrade immune checkpoint molecules in tumor-associated macrophages²⁵. In addition to bespoke degraders, traditional bispecific^26,27 and biparatopic²⁸ antibodies can also induce receptor degradation. Although, in this Review, we focus on those induced proximity modalities that operate in the extracellular space, cell surface proteins can also be targeted for degradation through E3 ligase recruitment to intracellular domains²⁹. Our aim here is to highlight broad mechanistic concepts under which induced proximity at the cell surface can be used for TPD, but comprehensive reviews of this rapidly evolving field can be found elsewhere^12,30,31.

Several outstanding challenges persist within the cell surface TPD space that call for further mechanistic investigation and technology development. One notable challenge lies in the maximal degradation achievable with current TPD strategies. The available repertoire of bifunctional degraders relying on lysosomal trafficking or cell surface E3 ligase recruitment rarely achieve total target degradation above 80% (D_max < 80%) due to competing rates of target biosynthesis^9–11,15. For some targets, incomplete knockdown might not produce a desired phenotypic outcome. Recent work demonstrated that introducing cleavable linkers to bispecific degraders is a potential design element for improving target degradation³². Moving forward, defining the minimal D_max required to achieve a desired phenotype will be an important consideration for degrader optimization efforts.

Considerable strides have been made in rendering cell surface TPD technologies tissue specific^{11,15,21,23,24}, but tissue type diversity still far surpasses the available degrader toolkit. Further technology development is needed to address these challenges and others facing the field. As mentioned previously, the mechanistic distinction between lysosome trafficking and E3 ligase-recruiting technologies is unclear, as E3 ligase recruitment by AbTACs and PROTABs appears to mediate lysosomal trafficking of the protein target based on chemical inhibitor studies^10,11. The development of technologies capable of cell surface TPD through a purely proteasomal pathway would offer mechanistic clarity and vet the utility of this approach. Finally, expanding the mechanistic toolbox beyond these two degradation paradigms holds potential for developing new tissue-specific TPD platforms and maximizing degradation efficiency beyond the approximately 80% ceiling.

Receptor inhibition through induced phosphatase recruitment

Protein phosphorylation is a major mode of signal transduction within the cell. Such pathways are particularly important for relaying information from the extracellular space into action at the transcriptional level³³. For example, immune cells use protein phosphorylation cascades to transmit activation and suppression signals in the context of immune responses to infection and disease³⁴. Immunoreceptor tyrosine-based activation motifs (ITAMs) associated with various immune cell receptors (T cell receptor (TCR)³⁵, B cell receptor (BCR)³⁶, Fc receptors³⁷) are phosphorylated upon clustering. Intracellular signaling molecules are then recruited to the phosphorylated ITAM, precipitating a signaling cascade that terminates in transcriptional activation, driving the immune response (Fig. 3a). Inhibitory receptors with phosphatase activity, such as CD45, or that recruit phosphatases through intracellular immunoreceptor tyrosine-based inhibitory motifs, such as programmed cell death receptor 1 (PD-1), oppose activating immunoreceptor signaling by dephosphorylating signaling intermediates. This delicate balance determines the nature and duration of an immune response and crucially relies on proximity. Inhibitory receptors dephosphorylate proximal activating receptors, and antigen recognition induces proximity by clustering activating receptors.

Fig. 3 |. Strategies for receptor inhibition through induced phosphatase recruitment.

a, Overview of activating and inhibitory receptor signaling regulated by proximity at the immune cell surface. APC, antigen-presenting cell; ITIM, immunoreceptor tyrosine-based inhibitory motif. b, Enforced proximity strategy to quell allergic reaction by colocalizing FcεRI and FcγRIIB in mast cells. c, Induced proximity modalities for inhibiting FcεRI signaling. d, Liposomes co-displaying SIGLEC-2 ligand and the peanut allergen Ah2 inhibit BCR signaling in peanut allergen-reactive B cells. L, ligand; Sig, SIGLEC. e, Antibody–CD33 ligand (CD33L) conjugates targeting IgE induce SHP-1 and/or SHP-2 phosphatase recruitment to quell FcεRI signaling. f, Liposomes co-displaying trinitrophenol (TNP) hapten and a ligand for SIGLEC-3, SIGLEC-7 or SIGLEC-9 reveal the roles SIGLECs play in dampening Fc receptor signaling. g, Structures of synthetic SIGLEC ligands. h,i, Receptor inhibition by the RIPR platform for modulating immune receptor signaling through direct dephosphorylation (h) and different available modalities (i).

Recognizing the critical role that proximity plays in regulating receptor signaling, multiple research groups have exploited this principle to synthetically modulate signaling for therapeutic and basic research applications. This approach has been particularly fruitful in suppressing mast cell activation in allergic responses by enforcing proximity between inhibitory receptors and the immunoglobulin E (IgE) receptor FcεRI^38,39. One seminal study used a protein fusion composed of Fcε and Fcγ to induce proximity between FcεRI and the inhibitory Fc receptor FcγRIIB⁴⁰ (Fig. 3b). Administration of this fusion protein to a mouse model of allergic reaction blocked mast cell degranulation and histamine release. This strategy is conceptually rooted in previous work demonstrating that co-aggregation of FcγRIIB and FcεRI dampens IgE-mediated activation and is mechanistically understood to proceed through Src homology region 2 domain-containing phosphatase (SHP)-1 phosphatase recruitment to FcεRI⁴¹ (Fig. 3b). Upon protein fusion administration, decreased phosphorylation of SYK, a substrate of SHP-1 and a critical kinase for transmitting activation signals, was observed^40,42.

Synthetically inducing proximity between an inhibitory receptor and an activating immune receptor to dampen immune activation has since been extended to different biological contexts using various bifunctional formats. For example, it was shown that bispecific antibodies targeting FcγRIIB and IgE could also suppress IgE-mediated signaling in mast cells and basophils⁴³. In this case, targeting IgE, rather than FcεRI directly, circumvents the challenge of Fcε competing with endogenous IgE for FcεRI binding. Further studies have demonstrated antigen-specific mast cell suppression⁴⁴ and alternative formats for inducing proximity between FcγRIIB and FcεRI based on Fc fusions⁴⁵ and bispecific designed ankyrin repeat proteins (DARPins)⁴⁶ (Fig. 3c). FcγRIIB recruitment has also been applied to inhibit BCR signaling by targeting subunits of the BCR complex CD79b⁴⁷ and CD19 (refs. 48,49). This approach offers a non-depletive strategy for quelling B cell activity and is currently being clinically investigated for multiple autoimmune indications with obexelimab (XmAb5871)⁵⁰.

Sialic acid binding lectins (SIGLECs) are glycan-binding proteins primarily expressed on the surface of immune cells⁵¹. While glycan specificity varies across the SIGLEC family, many SIGLECs share immunoreceptor tyrosine-based inhibitory motifs in their cytosolic domains, giving rise to immune inhibitory signaling. Immune activating receptors including FcγRIA^52,53, the TCR⁵⁴ and the BCR⁵⁵ are inhibited by colocalization with such inhibitory SIGLEC receptors. SHP-1 and SHP-2 phosphatase recruitment is a unifying mechanistic feature of immune inhibition connecting the SIGLEC family members with previously discussed inhibitory receptors that suppress IgE signaling. Synthetic SIGLEC ligands have been developed⁵⁶, and these have been incorporated into various proximity-inducing constructs. For example, using SIGLEC-engaging tolerance-inducing antigenic liposomes (STALs) co-displaying synthetic SIGLEC-2 ligands and B cell antigen, Paulson and Nemazee achieved selective apoptosis of antigen-specific B cells in model systems^57,58 and later extended this strategy to peanut allergen-specific B cell depletion⁵⁹ (Fig. 3d). BCR signaling is initiated by antigen engagement and receptor clustering, after which phosphorylated ITAMs within CD79a and CD79b recruit phosphorylated SYK, the central kinase for downstream B cell activation⁶⁰. SIGLEC-2 (also known as CD22) opposes BCR activation by recruiting SHP-1 to the signaling complex, which then dephosphorylates SYK, shutting down signaling. STALs co-engaging CD22 and the BCR synthetically induce this natural process to block B cell signaling (Fig. 3d). Subsequent work co-engaging IgE with either SIGLEC-3 (also known as CD33) or SIGLEC-8 using antibody–ligand conjugates⁶¹ (Fig. 3e) or STALs^62,63 showed that SIGLEC recruitment to FcεRI suppresses mast cell degranulation. In addition to proof-of-concept therapeutic strategies, SIGLEC ligand-displaying liposomes have helped elucidate the details of SIGLEC-mediated FcγRIA signaling inhibition. FcγRIA (CD64) is the high-affinity IgG-binding Fc receptor expressed on multiple immune cell subtypes and signals through ITAM phosphorylation and SYK recruitment³⁷. Liposomes co-displaying trinitrophenol hapten with either SIGLEC-3, SIGLEC-7 or SIGLEC-9 ligands (Fig. 3g) were used to compare the inhibitory effects of SIGLEC proximity on FcγRIA signaling (Fig. 3f)⁵³.

In 2020, the phosphatase recruitment concept was advanced a step further with the development of the receptor inhibition by phosphatase recruitment (RIPR) platform⁶⁴. Although conceptually analogous to previous work recruiting phosphatases for receptor inhibition, the RIPR platform induces proximity between a target receptor and the phosphatase CD45 directly, rather than through indirect recruitment of a cytosolic phosphatase such as SHP-1 or SHIP-2 (Fig. 3h). Bifunctional RIPR molecules are composed of two arms: a single-chain variable fragment (scFv)-based CD45-binding arm and a fused target-binding arm that aims the dephosphorylation activity. The PD-1-targeting RIPR-PD-1 induces contact between CD45 and PD-1 to dampen PD-1-based inhibitory signaling by directly dephosphorylating PD-1, thus activating TCR signaling in T cells (Fig. 3h). PD-1 is an immunoinhibitory receptor responsible for modulating TCR signaling⁶⁵. Upon T cell engagement with a target cell (tumor cell or antigen-presenting cell), PD-1 is recruited to the immune synapse by interaction with its ligands, PD-L1 and/or PD-L2. Phosphorylated PD-1 recruits the phosphatase SHP-2 to the synapse, which in turn negatively regulates the TCR⁶⁵. Preventing SHP-2 recruitment by dephosphorylating PD-1 releases the brakes on TCR signaling (Fig. 3h). This study also demonstrated receptor dephosphorylation, with RIPR molecules targeting signal regulatory protein α (SIRPα), cytotoxic T lymphocyte-associated protein 4 (CTLA-4) and CD28, the last of which demonstrates the potential for also inhibiting activating receptors through this approach⁶⁴. A follow-up study showed that RIPR molecules targeting natural killer group 2 member A (NKG2A), Ly49C and Ly49I potentiate natural killer cell killing function (Fig. 3i)⁶⁶. Direct phosphatase recruitment also appears to be a viable path to block B cell activation based on a patent detailing the use of CD45 × CD79-bispecific antibodies⁶⁷. Recently reported DNA aptamer-based RIPR molecules targeting protein tyrosine phosphatase receptor type F (PTPRF) can induce dephosphorylation of c-Met and epidermal growth factor receptor (EGFR)⁶⁸.

Although direct or indirect phosphatase recruitment is gaining traction as a strategy for modulating receptor signaling, additional mechanisms are needed to extend this approach beyond receptors that signal through tyrosine phosphorylation. For example, steric exclusion of signaling molecules or ligands from immune synapses is a possible avenue to modulate receptor function. This potential strategy directly relates to a mechanistic question still outstanding regarding the RIPR platform, in which CD45 recruitment may enforce synapse exclusion of the target receptor. Owing to its large size, CD45 is excluded from productive immune synapses⁶⁹; so it is possible that RIPR molecules recruit target receptors away from the immune synapse. Ligand-blocking antibodies and lysosome-trafficking degraders are alternative methods to inhibit receptor signaling. At this stage, it is still unclear which approach is superior or whether the answer to this question is context dependent. Comparative studies between these different platforms would advance our understanding of optimal approaches to inhibit receptor signaling.

Receptor activation through induced proximity

Receptor subunit proximity regulates signaling in many cell surface receptor–ligand interactions. Interleukin (IL)-2 signaling is a hallmark example of this phenomenon, in which IL-2 signals by recruiting two to three subunits of the IL-2 receptor (IL-2R) complex. IL-2 induces proximity between IL-2Rβ and IL-2Rγ subunits to initiate signaling through the Janus kinase (JAK)1–JAK3–signal transducer and activator of transcription (STAT)5 pathway (Fig. 4a). The third subunit, IL-2Rα, is expressed on a subset of immune cells and serves to increase the receptor complex affinity for IL-2 (ref. 70). In this way, IL-2 and other cytokines⁷¹ are nature’s induced proximity modalities used to tailor responses to infection through receptor subunit expression patterns and cytokine concentrations. The importance of proximity for switching on signaling extends beyond cytokine receptors, as can be seen with Wnt⁷² and fibroblast growth factor (FGF)⁷³ signal transduction. In this section, we highlight engineering efforts that exploit the principle of receptor activation through induced proximity.

Fig. 4 |. Strategies for receptor activation through induced proximity.

a, Cytokine receptor subunit complexation induced by natural IL-2 and mimetics and key intracellular signaling molecules involved. p, phosphorylated. b, Bispecific antibody formats used as cytokine mimetics. iAb, i-shaped antibody. c, FGFR1c homodimerization induced by the natural ligand FGF21 or a bispecific antibody FGF surrogate targeting FGFR1c and KLB. d, β-catenin signaling with and without natural or synthetic Wnt ligands that heterodimerize FZD and LRP to induce β-catenin stabilization. GSK3, glycogen synthase kinase 3. e, Selection of bispecific antibody and protein fusion formats used as Wnt ligand surrogates. f, Canonical PTH signaling involving Gα_s–cAMP signaling, Gα_q activation and Ca²⁺ mobilization, and β-arrestin recruitment (left) and highly biased agonism involving Gα_s–cAMP signaling elicited by PTHR1 homodimerization using a biparatopic PTH–nanobody conjugate (right). VHH indicates a nanobody.

Antibody-based cytokine⁷⁴ and cytokine receptor⁷⁵ blockers have shown noteworthy clinical success, with multiple drugs on the market. By contrast, cytokines that serve as potential immune agonists for anti-viral and anti-cancer applications have been much less successful⁷¹. Two major hurdles in developing cytokines as therapeutic agents are their poor pharmacokinetic properties and pleiotropic signaling effects in different tissue and immune cell types. Over the past two decades, engineered forms of IL-2 (ref. 76), IL-4 (ref. 77) and other cytokines⁷¹ have been developed with biased signaling capacities, but addressing the poor pharmacokinetic profiles and other developability issues associated with the cytokine format remains a challenge. Recognizing that IL-2 can be abstracted as an IL-2Rβ–IL-2Rγ heterodimerizer and reconstructed using a bispecific antibody that recruits IL-2Rβ and IL-2Rγ represented a key advance in engineering cytokine mimetics (Fig. 4a). First demonstrated by the laboratories of Trinklein⁷⁸ and Garcia⁷⁹, the use of bispecific antibodies (UniAbs⁷⁸ or scFv fusions⁷⁹) in place of endogenous IL-2 simultaneously addresses the developability challenges associated with cytokine pharmacokinetic properties and transforms the cytokine-engineering problem into a more tractable antibody engineering problem. These antibody-based cytokine mimetics (ACMs) are selective for the intermediate-affinity IL-2Rβγ receptor, sparing the high-affinity IL-2Rαβγ receptor responsible for undesired side effects⁷⁸. Subsequent work from the groups of Sockolosky and Lazar demonstrated that i-shaped antibodies and Contorsbodies (Fig. 4b) recognizing IL-2Rβ and IL-2Rγ can be used as IL-2 surrogates, in which their unique antigen-binding geometries are critical to mimic natural cytokine signaling⁸⁰. This concept was extended to induce IL-2Rβ–IL-10Rβ heterodimerization, creating a completely unnatural receptor interaction with the potential to modulate tissue selectivity⁷⁹.

The importance of FGF21 in regulating energy balance and glucose and lipid homeostasis has rendered mimetics of this stress-inducible hormone an attractive drug target for use in metabolic and fibrotic diseases^81,82. FGF21 activity relies on the proximity of its receptor, FGF receptor (FGFR), and its obligate co-receptor β-klotho (KLB). This reliance limits FGF21 effects to the pancreas, liver and adipose tissues where KLB is primarily expressed⁸³. FGF21 binding to FGFR–KLB causes FGFR dimerization and phosphorylation of its intracellular kinase domains, initiating signaling through STAT1, phospholipase Cγ (PLCγ) and FGFR substrate 2 (FRS2) (Fig. 4c)⁸⁴. Based on the poor plasma half-life of endogenous FGF21 (ref. 85), bispecific avimers⁸⁶ and antibodies⁸⁷ targeting FGFR1c and KLB were generated to mimic the action of FGF21 in formats suitable for therapeutic development. The FGF21 mimetic bFKB1 reverses insulin resistance and obesity in mice by inducing thermogenic activity in brown adipose tissue in a tissue-selective manner. bFKB1 enhances FGFR1–KLB heterodimerization in biophysical experiments⁸⁷, but previous work demonstrating that FGFR1–KLB heterodimerization is FGF21 independent⁸⁸ calls into question whether bFKB1 induces heterodimerization in addition to FGFR1 homodimerization. Because bFKB1 activity requires both KLB and FGFR1, it may serve first as a proximity sensor and then as a proximity inducer in FGFR1 homodimerization. The FGFR1–KLB-bispecific agonist BFKB8488A (fazpilodemab) has undergone phase 1b trials in non-alcoholic fatty liver disease and type 2 diabetes mellitus⁸⁹ and is currently being investigated for use in non-alcoholic steatohepatitis.

Receptor proximity plays a central role in the endogenous Wnt–β-catenin signaling pathway⁹⁰. Soluble Wnt proteins bind to the cell surface, inducing complex formation between Frizzled (FZD) and low-density lipoprotein receptor-related protein (LRP)-5 and LRP-6 receptors, thus initiating a signaling cascade that results in nuclear localization of the transcriptional coactivator β-catenin⁹⁰. In the nucleus, β-catenin regulates cell growth and differentiation. Modulating this pathway holds promise for applications in tissue repair and regeneration^90,91. Poorly understood tissue-specific effects and posttranslational lipidation make endogenous Wnt proteins challenging to develop as drugs⁹². In a quest to construct synthetic Wnt surrogates, bifunctional fusion proteins, each composed of an FZD-binding subunit (anti-FZD scFv or the bacterial FZD-binding protein B12) and an LRP-5- and LRP-6-binding subunit based on the C-terminal domain of the human Wnt antagonist DKK1, were generated⁹³. These Wnt surrogates mimic canonical Wnt signaling and promote organoid growth that typically depends on natural WNT3A⁹³. Mechanistically, Wnt proteins and their surrogates initiate β-catenin signaling by sequestering the multicomponent β-catenin degradation complex, increasing steady state β-catenin concentration and facilitating its nuclear localization⁹⁰ (Fig. 4d). Additional synthetic Wnt agonists based on the FZD–LRP-5–LRP-6-induced proximity principle have been reported using diabodies⁹⁴, DARPins⁹⁵ and other antibody-based formats^92,96 with enhanced affinities and specificities for FZD family members (Fig. 4e).

G protein-coupled receptors (GPCRs) make up the largest class of drug targets approved by the Food and Drug Administration⁹⁷, but drugs that target them for certain indications, such as chronic pain, have been hindered by pleiotropic effects of GPCR signaling. This simultaneous opportunity and challenge fuels the quest for biased GPCR agonists⁹⁸. It was recently demonstrated that a biparatopic type 1 parathyroid hormone (PTH) receptor (PTHR1) agonist composed of a peptide–nanobody conjugate homodimerizes PTHR1, inducing highly biased signaling through the Gα_s–cyclic AMP (cAMP) pathway without β-arrestin 2 recruitment or Gα_q activation and Ca²⁺ mobilization^99,100 (Fig. 4f). Although more mechanistic details are needed to fully understand this biased agonism, the use of induced proximity at the cell surface to elicit biased GPCR signaling is a promising new direction for therapeutic development.

ACMs offer a promising opportunity to modulate signaling while maintaining the benefits of antibody pharmacokinetics. However, it is critical to compare the signaling strength, signal transduction pathway and duration of ACMs to those of cytokines, as minor differences in binding dimensions can cause significant alterations in pharmacodynamics. Additionally, ACMs tends to be larger than cytokines in size, which could alter biodistribution and function of ACMs. Similar questions of biodistribution apply to FGF21 and Wnt mimetics, given the low molecular weight of their endogenous counterparts. A key question arising from using antibodies as cytokine mimetics is whether receptor heterodimerization geometry resembles the natural pose induced by native IL-2. Structural data including cryo-electron microscopy can be used to design ACMs, but extensive empirical testing is required to identify molecules with optimal properties given the complex nature of signaling pathways in cells.

Induced proximity involving multiple cells

TCR binding to major histocompatibility complex (MHC) molecules is a natural example of induced proximity, where a cognate peptide serves as the molecular inducer of proximity that bridges the TCR and MHC expressed on opposing cell surfaces (Fig. 5a). Upon TCR binding to cognate peptide–MHC, spatial reorganization on T cells is initiated where TCRs cluster at the immunological synapse, and key signaling molecules such as LAT and SLP76 are recruited to these microclusters, facilitating the formation of a signalosome that orchestrates downstream signaling pathways essential for T cell activation. Additional examples of trans-acting molecular inducers of proximity, such as antibody opsonization and effector cell recruitment, abound in the immune system. These natural processes serve as the key inspiration for redirecting the immune system against cancer cells using T cell engagers (TCEs). TCEs are bifunctional proteins that bind cancer-specific antigens with one arm and recruit cytotoxic T cells with a second arm, typically engaging the CD3ε subunit of the TCR complex (Fig. 5b). TCEs have found clinical success in mainly hematological malignancies and are increasingly being tested in solid tumors such as prostate cancer. Cell engagers are now expanding to include trifunctional molecules and are being used to recruit alternative immune effector cells¹⁰¹. Comprehensive reviews of this class of therapeutics can be found elsewhere^101–103.

Fig. 5 |. Induced proximity modalities acting between two cells.

a, Natural cognate MHC–peptide–TCR recognition. b, T cell engagement for cancer cell elimination. c, Targeted protein transfer using TrogoTACs.

Bispecific cell engagers that carry out functions beyond cell killing are an emerging application of trans-acting induced proximity. Trogocytosis is a process in which intercellular exchange of membrane fragments occurs, first described in immune cells but also more recently in cancer, neuronal remodeling and host–pathogen interactions^104,105. We leveraged this natural process to build bifunctional antibody-based trogocytosis-targeting chimeras (TrogoTACs) that redirect the natural process of trogocytosis^104,106 to induce the transfer of cell surface proteins from a target cell to acceptor B cells¹⁰⁷ (Fig. 5c). These molecules can directly remodel the cell surface proteome in an additive rather than degradative manner without genetic intervention. TrogoTACs can remodel cancer cell surfaces to resensitize them toward the TCE tebentafusp, thus inducing potent cell lysis at levels comparable to those of cells naturally sensitive to this therapeutic¹⁰⁷. This technology complements the expanding toolkit for protein degradation and may have applications in disorders characterized by pathogenic absence of a cell surface protein.

Most trans-acting bifunctional molecules developed to date are used for target cell elimination through immune cell recruitment. Trans-acting induced proximity approaches that engage two cells without inducing cytolysis are a nascent area. Redirecting trogocytosis to directly manipulate the cell surface proteome is an exciting new direction, but mechanistic questions surrounding the natural process of trogocytosis are still outstanding. Beyond protein transfer, leveraging induced proximity at the cell–cell interface is a rich area for developing new ways to directly manipulate cell biology in a context-dependent manner.

A forward-looking perspective for induced proximity at the cell surface

Cell surface-induced proximity modalities have expanded our ability to both manipulate and understand the biology of the cell, but further technology development and mechanistic elucidation are necessary to achieve the full potential of these approaches. For example, new modalities that induce targeted degradation of cell surface proteins by recruiting E3 ligases or lysosome-trafficking receptors are rapidly increasing in number, but we still lack mechanistic understanding of these platforms (see above). From an engineering perspective, it is unclear how parameters such as binding affinity, ternary complex geometry and recycling capacity should be optimized to produce a clinically useful degrader. A global understanding of tonic signaling in antibody-blocked receptors would help define the optimal use cases for a degradation strategy over a conventional antibody blockade approach. Finally, clinical data on this class of degraders will provide crucial evidence of their potential for clearing proteins in a safe and efficacious manner. In the receptor signaling modulation space, cytokine mimetics have great potential to exert desirable immunomodulatory effects with tissue and immune cell specificity. ACMs are an especially exciting format, given the more straightforward development path for clinical use. We are eager to see how these approaches are leveraged for autoimmune and oncology indications in the coming years. Given the importance of protein proximity to signaling cascades originating from the extracellular space, there exists great opportunity to modulate intracellular signaling by acting on receptors in the extracellular space. The first proximity inducers that inhibit signaling have largely leveraged phosphatase recruitment, but potential to expand this toolkit lies in exploring alternative enzyme recruitment strategies. For induced proximity modalities such as TCEs and TrogoTACs that recruit proteins on opposing cell membranes, the potential for new modalities is vast. These bispecific molecules substantially alter the cell surface protein interactome at the new cell–cell interface, but, so far, their use has mainly been limited to cell killing. Next-generation induced proximity modalities targeting cell surface proteins have the potential to radically alter our ability to manipulate the cell surface proteome and enhance our understanding of native biological processes.