Can Targeting Peptides Improve the Safety of Immunotherapies?

By exploiting disease-restricted receptors for delivery of cytotoxic or immuno-stimulatory payloads, targeting peptides narrow the therapeutic index: they enhance local concentration of the drug but limit activation of by-stander lymphoid organs, thus minimizing the risk of systemic cytokine storms, neurotoxicity or autoimmune flare that are commonly associated with conventional checkpoint blockade or CAR-T therapies. As peptides can be cyclized, D-amino-acid stabilized and cleavage-activated in matter of days, they also offer a rapidly iterative platform to "tune" affinity, off-rate and proteolytic release to ensure that engagement is only triggered in the presence of antigen, and thus convert blunt immune activation into a postcode-restricted signal.

Why Safety Remains a Major Limiting Factor in Immunotherapies

Safety is still one of the most important bottlenecks in immunotherapies. In part, this is because these therapies often lack selectivity for specific immune cell subsets, and rather induce a systemic immune response. As a consequence, dose-limiting toxicities can emerge from this systemic immune activation. The broad and non-specific nature of these therapies often leads to off-target effects and poor therapeutic selectivity, causing unwanted inflammation and damage to healthy tissues.

Systemic Immune Activation and Dose-Limiting Toxicities

Pan-immune targeting agents, such as checkpoint inhibitors or CAR-T cells, are myeloid, innate lymphoid and B as well as T-cell products. Because they lack selectivity, once the dose reaches a level above the activation threshold of the most refractory cell subset within the circulation or tissue, CRS, neurotoxicity, colitis or fulminant hepatitis will result. The therapeutic window is therefore small and, as the effect of the therapy is largely irreversible, the investigator is forced to abort dosing at the first sign of systemic flare in the hope of salvaging response. Thus, many patients are under-treated. Also, as the circulating drug concentrations must be greater than the activation threshold of the most refractory cell population, it follows that peak plasma levels will be higher than those required to induce endothelial activation and complement recruitment or neuronal inflammation. Attempts to expand the window via alternate high-low schedules are moot, as non-target organs will still be exposed to the peak concentration, a consideration that is why spatial control is far superior to temporal control.

Fig. 1 Innate immune response.^1,5

Why Potent Immunotherapies Often Lack Precision

CD3, CD28 or 4-1BB agonists activate ～100% of T cells, including regulatory subsets that inhibit tumor immunity; similarly, STING agonists induce IFN-β from every myeloid cell, causing a cytokine storm. As these receptors are shared by protective and pathogenic populations, in vitro potency begets pan-immune activation in vivo, reducing tumor specificity while exacerbating off-tissue damage. So even "tumor-specific" antigens like HER2 or GD2 are still expressed on normal epithelia/neurons, just at a low level, such that CAR-T cells/bispecific antibodies can still target heart or brain tissue, once there's high enough systemic exposure. Since our current platforms cannot discriminate between high-antigen tumor cells and low-antigen healthy cells, the same potency that can push a tumor into remission can also push you into dose-limiting cardiotoxicity/neurotoxicity, and clinicians are forced to choose between efficacy and safety.

Table 1 Conceptual comparison: conventional vs peptide-guided immunotherapy safety

How Targeting Peptides Could Improve Immunotherapy Safety

Targeted peptide drug delivery flips the pharmacokinetic script. Instead of flooding the system with high potency but non-selective effectors, a small amino-acid carrier specifically delivers drugs to and within leukocytes relevant to the disease or tumor microenvironments. Metabolically unstable and<3 kDa in size, unbound peptide or conjugates that miss their target are quickly cleared by the kidneys, reducing overall exposure but not intra-lesional activity. This change in site of action can convert an all-on switch of the immune system to a zip code specific signal, with decreased potential for off-tissue cytokine release syndrome, neurotoxicity, or autoimmunity.

Restricting Drug Exposure to Relevant Immune Cell Populations

Small cyclic peptides mimicking endogenous ligand loops (e.g., RGD for integrins or mannose for CD206) lodge to activation-associated receptors outside of the native pocket and thus do not compete with serum ligands, yet still recruit clathrin. Since the same peptide can be lipidated or thiolated with no impact on the binding face, a single sequence can be ported across liposomes, polyplexes or virus-like particles which simplifies scale-up while preserving modular adjustability of the delivery system for siRNA, STING agonists or CRISPR ribonucleoproteins. The conjugate is ferried across the cytosol inside recycling endosomes and inclusion of three to four histidines lowers the pKa of the vector, inducing vesicle leakage before lysosomal fusion. Since any unconjugated peptide is filtered renally within 15 min, plasma concentrations of free immunomodulator remain below the threshold for endothelial activation, aligning potent local modulation with negligible systemic exposure.

Improving Localized Immune Modulation

Chemokine fields produce concentric rings around the area of inflammation; presenting chemokine-mimetic loops to this environment allows peptides to hitch-hike on the migrating leukocytes, and in effect to use the cells themselves as Trojan horses, reaching the deeper tissue layers that are unavailable to passive diffusion. Because the peptide is further cleaved by cathepsin S once inside the leukocyte, release of the payload is further delayed until after the carrier has transmigrated to the perivascular niche, thereby synchronizing delivery depth with chemotactic flux. Peptide–drug conjugates utilize a self-immolative spacer that fragments within acidified endosomes, and therefore ensures that immunostimulatory payloads such as STING agonists or IL-12 are released prior to vesicle maturation into a degradative lysosome. Because the peptide itself is further metabolized into amino acids, no residual polymer accumulates in liver or spleen, and transient immune modulation is balanced with repeat-dosing safety by avoiding off-target activation of by-stander organs.

Design Requirements for Safety-Oriented Immune Targeting Peptides

Safety-first peptide design is based on the premise that the immune system is a bustling city: the ligand needs to go to the correct postcode (cell type), knock only once (one receptor) and not linger (off-target interactions = cytokine release), so it needs to have moderate affinity, rapid off-rate, built-in cleavage, and an amino-acid alphabet without PRR motifs. Since the peptide itself can be metabolized within minutes, any off-target conjugate will be cleared renally before it can secondarily activate complement or FcγR-bearing cells: which means that any peptide conjugate translates potent local modulation with negligible systemic exposure.

Target Specificity vs Immune System Redundancy

CD11b coats both inflammatory monocytes and neuro-protective microglia; PD-L1 is highly expressed on tolerogenic dendritic cells as well as exhausted T cells. Targeting just one epitope therefore amplifies a mixed population which in part drives disease and in part restrains it. Designers instead use Boolean-logic strategies, fusing together two moderate-affinity peptides separated by a protease-cleavable bridge, such that only cells expressing both epitopes can internalize the full payload, sharpening therapeutic index without expanding the peptide outside renal filtration dimensions. Ultra-high-affinity ligands (sub-nanomolar Kd) bind for hours, resulting in receptor down-regulation and a loss of repeat-dosing sensitivity. The affinity can instead be lowered to 50–200 nM and pH-sensitive histidines added to speed up off-rate when the endosome acidifies, allowing the peptide to turn on and off of the cell multiple times, maximizing cumulative payload without pathway saturation. As the same peptide can be engineered to release at pH 6.0 yet re-bind at pH 7.4, one sequence can therefore deliver multiple drug cargoes per disease flare yet still self-neutralize inflammation once it resolves.

Affinity and Exposure Control

Extended periods of receptor occupancy are capable of inducing the downstream signalling cascades (e.g., NF-κB, IRF3) even with a payload in its inert state. To avoid this, they incorporated reducible disulfide bonds or ester linkers that are susceptible to hydrolysis by enzymes present in the cytosol such that the peptide ligand itself is truncated into non-immunogenic fragments within 30 min of uptake. This "self-neutralizing" architecture thereby limits the window of activation to the short period required for drug release, while tightly coupling receptor engagement with metabolic disappearance. Because peptide is cleared renally within minutes, plasma exposure of free ligand is negligible; the dose–response curves are therefore driven by receptor density rather than systemic concentration. The predictability that results from this allows clinicians to titrate the dose based on biomarker occupancy (e.g., PET tracer) rather than body weight, thereby reducing inter-patient variability and the risk of overshoot that precipitates cytokine storms. Because the same peptide can be programmed to release at pH 6.0 and yet re-bind at pH 7.4, one sequence can cycle on and off of the receptor multiple times, maximizing cumulative payload while maintaining a linear dose–occupancy relationship.

Minimizing Peptide-Induced Immune Recognition

Arginine or lysine-rich linear peptides can resemble bacterial patterns, which can initiate TLR2/TLR4-mediated signalling, transforming the delivery vehicle into a de facto adjuvant. Mutation of every third cationic residue to the neutral norleucine, or addition of D-alanine at positions 2 and 13, prevents docking of pattern-recognition receptors, but retains ability to escape the endosome. Removal of free N- and C-termini via head-to-tail cyclisation also prevents recognition of PAMPs and decreases cytokine secretion, without affecting binding to pattern-recognition receptors. Longer peptides (length > 30 residues) or multiple repeats of proline-rich segments may be presented on MHC-II, which results in production of anti-peptide antibodies and faster clearance on subsequent administration. Limiting the peptide length to 12–24 amino acids and insertion of non-natural β-amino acids in non-contact areas reduce immunogenicity but retain protease resistance. Insertion of a self-hydrolysable ester bond that cleaves the peptide down to individual amino acids within 4–6 h of delivery to the cytosol also ensures that there is no intact ligand left to serve as a subsequent antigenic signal, and transient delivery coincides with long-term tolerability.

Delivery Architectures That Leverage Targeting Peptides for Safer Immunotherapies

Peptide-based architectures combine the addressability of short amino-acid based shuttles with the payload capacity of synthetic or biological scaffolds to create hybrid systems that can be surface-decorated, covalently linked or self-assembled into immune-modulating vehicles. As peptides can be encoded at the DNA level or chemically synthesized in days, the same construct can be produced recombinantly on viral capsids or conjugated to lipids and polymers, and so one molecular blueprint can be used for multiple formulation platforms without changing the core chemistry.

Peptide–Drug Conjugates for Controlled Immune Modulation

Peptide–drug conjugates (PDCs) make use of a short targeting ligand covalently tethered to the therapeutic payload through a self-immolative spacer that cleaves within acidified endosomes. As the peptide itself is<3 kDa, the entire construct falls under the renal filtration cut-off and can achieve fast plasma clearance and high lesion-to-blood ratios in vivo when administered for the targeting of activated T cells or tumor-associated macrophages. The spacer can be tuned to release the parent drug within 30–60 min of uptake to ensure that immunostimulatory payloads such as STING agonists or IL-12 are released prior to the vesicle maturing into a degradative lysosome. PDCs have an in-built dosing ceiling: once surface receptors are saturated, any excess conjugate circulates in its unbound state and is rapidly excreted, preventing the protracted systemic exposure that can occur with liposomal or polymeric carriers. As the peptide is cleaved into amino acids, no residual polymer accumulates in the liver or spleen to offset the safety of repeat dosing with transient immune modulation. Additional ester bonds that hydrolyze at pH 6.0 can be incorporated by designers to confine drug activation to inflamed tissue where lactic acid production from tumors and activated immune cells lowers interstitial pH, thereby tightening spatial selectivity without needing to incorporate any external targeting moieties.

Peptide-Guided Nanocarriers

Maleimide-thiol click post-functionalization of lipid or polymeric nanoparticles with cysteine-terminated peptides results in 5–15 ligands per particle. The peptide corona masks the cationic surface of the underlying nanocarrier, thus reducing complement activation and liver accumulation while still allowing for receptor-mediated uptake. Steric hindrance is minimized by PEGylation of the peptide itself at a distal lysine, preserving colloidal stability and receptor accessibility under shear flow. Multi-valent display enhances functional avidity, allowing moderate-affinity peptides (50–200 nM Kd) to target low-density receptors like CD169 or DC-SIGN that would otherwise be ignored. The peptide is displayed on a flexible PEG spacer, keeping the binding loop accessible even when the particle is coated with serum proteins, thus preserving targeting efficiency in blood-rich tissues such as spleen or inflamed joint. Additionally, incorporating a second, cryptic peptide that is unveiled after proteolytic cleavage inside the target endosome creates a Boolean AND gate, further sharpening specificity without sacrificing uptake speed.

Table 2 Comparative safety features of peptide-enabled immunotherapy architectures

Fig. 2 A peptide drug can be conjugated to various molecules to enhance properties such as enzymatic stability, plasma half-life, and target specificity.^2,5

Key Limitations and Risks of Using Targeting Peptides for Safety Improvement

Peptide guidance can reduce systemic exposure, but almost never with complete cell exclusivity. Receptor overlap, protease-cleaved fragments and shear-activated adsorption can all deliver payload to by-stander organs, resulting in a "safety floor" that is below that of antibody or liposome platforms, but still above zero. Because renal clearance is rapid, any partial miss is transient; however, the fleeting pulse can still activate innate pattern-recognition receptors if the sequence or cargo mimics microbial motifs, serving as a reminder that "peptide" does not inherently mean "harmless".

Partial Targeting and Residual Systemic Effects

CD11b, PD-L1 or ICAM-1 are common to both protective and pathogenic subsets; even an optimally designed peptide enriches a heterogeneous population. Boolean dual-motif constructs diminish but do not obviate this background noise, which can be exacerbated when inflammation broadly up-regulates the first epitope. The practical consequence is a "partial precision" reality: while systemic TNF-α or IL-6 may be diminished, regulatory T cells or tissue-resident macrophages will also be affected, and some off-target toxicity is thus retained. Proteolytic nicking can unmask cationic patches that activate TLR2/4 or complement, which is an issue when dosing is driven to a higher level to overcome the constraint of low receptor density. Since the peptide itself is<3 kDa, this activation is transient; however, the resulting cytokine burst can still induce flu-like symptoms or endothelial leak, and this highlights that peptide-guided delivery reduces but does not remove the safety floor.

Dynamic Immune Cell States

CD69, PD-L1 or CD25 are upregulated within hours and downregulated as corticosteroids or biologics work; a peptide optimized for maximum abundance may therefore bind weakly during resolution and risk under-dosing at the very moment that immunosuppression is needed most. pH-sensitive histidines or shear-activated anchors create a tissue rather than epitope gating, but these modules are also triggered by healthy exercise or infection, reintroducing a temporal element of unpredictability. Chronic lesions can also down-regulate the same receptors that were abundant in an acute flare; continued dosing then preferentially binds to healthy tissue expressing basal levels, recreating systemic exposure. As peptide half-life is short, this drift is picked up early, but it still requires frequent biomarker re-validation or dual-receptor switching, adding clinical complexity that can erode the original safety lead.

Table 3 Risk matrix for peptide-guided immunotherapy

When Targeting Peptides Are Likely to Improve Immunotherapy Safety

Peptide-guided immunotherapy is best when the pathogenic immune population express a lineage-restricted, recycling receptor, sufficiently abundant for >12 h to allow moderate-affinity ligands to concentrate payloads without saturating the pathway. Peptides can be cyclized or D-amino-acid stabilized in days, so they are tailored to localized, compartmentalized inflammation (joint, gut mucosa, tumor bed) where vascular leak is moderate but not chaotic, and where receptor expression is disease-associated rather than acute-phase.

Suitable Targets and Therapeutic Modalities

Targets of choice are CD11c on dendritic cells, CD163 on M2 macrophages or CXCR4 on regulatory T cells—epitopes that are either constitutively restricted or only up-regulated during pathologic inflammation. Cyclic peptides that mimic loops on natural ligands (RGD for integrins, mannose for CD206) bind outside the native pocket, obviating competition with serum ligands while still enabling clathrin recruitment. Because the same peptide can be lipidated or thiolated without changing the binding face, one sequence can be ported from liposomes to polyplexes or virus-like particles, enabling scale-up without losing modular adjustability for siRNA, STING agonists or CRISPR ribonucleoproteins. Diseases where a single immune subset is the key driver of pathology—M2 macrophages in tumour progression or autoreactive T cells in type 1 diabetes, for example—offer a clear mechanistic link between targeted delivery and therapeutic outcome. Peptides directed to these subsets can deliver STING agonists or STAT3 siRNA to reprogramme the niche without perturbing systemic immunity. Because the peptide itself is<3 kDa, renal clearance provides an in-built dosing ceiling that prevents the prolonged exposures that accompany antibody or liposomal platforms, therefore matching transient modulation to safety margins.

Scenarios Where Safety Gains May Be Limited

When systemic immune suppression is the intended effect (cytokine storm syndromes, fulminant autoimmune flares) the target cells are scattered in many organs and express the target receptor at significant levels throughout the body. For example, an antigen-specific peptide would be captured indiscriminately in lung, kidney and gut, replicating the broad tissue exposure it was designed to spare. In these situations, an untargeted corticosteroid or a wide-spectrum JAK inhibitor is still favored, as it will equilibrate throughout the body within minutes and incur the off-target penalty in the name of speed and ease of use. CRS or neurotoxicity during CAR-T therapy is mediated by rapidly escalating and self-amplifying cytokine cascades that peak in hours or less, a timescale faster than the circulation half-life of most peptide therapeutics. To intercept this storm with a receptor-targeted peptide, you would need to infuse it at impractically high rates to provide adequate temporal coverage, overwhelming renal clearance while still missing the front end of the cytokine spike. In these situations, non-targeted IL-6R blockade or Bruton's kinase inhibition can provide faster, global suppression without the formulation burden of a ligand-guided vector.

How Our Targeting Peptide Services Support Safer Immunotherapy Development

Safety remains one of the primary limiting factors in immunotherapy development. Many safety liabilities arise not from insufficient potency, but from excessive or poorly controlled immune engagement outside the intended cell population or tissue context. Targeting peptides offer a potential route to improve safety—but only when selectivity and immune biology are addressed with rigor. Our targeting peptide services support safer immunotherapy development by integrating selectivity-driven design, risk-aware optimization, and feasibility assessment focused specifically on immune safety rather than delivery efficiency alone.

Safety-Oriented Immune Targeting Design

Selectivity-Driven Peptide Engineering: In immunotherapy, improving safety requires restricting drug exposure to the intended immune cell population while minimizing engagement of off-target immune subsets. Our selectivity-driven peptide engineering approach prioritizes discrimination over maximal binding strength. Peptide sequences are designed around markers with the highest functional relevance and the lowest overlap across immune cell types. Affinity and binding kinetics are tuned to match physiological expression levels, reducing the risk of unintended engagement driven by high-dose or prolonged exposure. This approach helps ensure that immune modulation is guided by biological specificity rather than by non-specific accumulation.

Risk-Aware Optimization Strategies: Immune-targeting peptides can introduce risks if they trigger unintended signaling or immune recognition. Our risk-aware optimization strategies explicitly account for these factors during peptide design. Sequence features that may promote immune activation, aggregation, or off-target receptor engagement are identified and minimized early. Optimization balances stability and exposure control to avoid prolonged immune engagement that could amplify inflammatory or cytotoxic responses. This risk-aware framework is critical for improving the safety margin of immunotherapy programs without compromising therapeutic intent.

Feasibility Assessment Focused on Safety and Specificity

Marker Evaluation: Not all immune markers that are technically targetable are suitable for improving safety. Our feasibility assessment includes a detailed marker evaluation that examines expression breadth across immune cell subsets, activation states, and tissues. Markers with high overlap, inducible expression in non-target cells, or strong roles in immune signaling are flagged early, as targeting these markers may increase rather than reduce safety risks.

Early Toxicity Risk Analysis: Many immunotherapy safety issues can be anticipated at the targeting stage. We conduct early toxicity risk analysis to assess whether peptide-mediated targeting is likely to restrict exposure sufficiently to meaningfully reduce systemic immune activation. This analysis helps distinguish scenarios where targeting peptides can realistically improve safety from those where the mechanism of action itself is inherently systemic and unlikely to benefit from targeted delivery.

Discuss Strategies to Improve Immunotherapy Safety with Our Team

If your immunotherapy program is limited by dose-limiting toxicity, broad immune activation, or narrow safety margins, an early technical discussion can help clarify whether targeted delivery is likely to improve safety—or whether alternative strategies should be considered. Discuss strategies to improve immunotherapy safety with our team to evaluate targeting feasibility, assess specificity and toxicity risks, and define a peptide design approach aligned with your program's safety objectives.