The Nanoparticle Delivery Gap in Immune Cell Engineering: From Uptake to Endosomal Escape

Abstract

Although therapeutic opportunities beyond the scope of either field alone are offered by the combination of immune cell treatment and nanotechnology, a dominant nanoparticle design for immune-cell engineering has not been yielded by two decades of research. Here, it is contended that this lack of advancement is the result of a misdirected focus: endosomal escape, rather than cellular uptake, is held to be the actual rate-limiting stage in cytosolic distribution, even though cellular uptake is the target of most optimization efforts. Adopting endosomal escape as a unifying framework, this review examines the mechanisms governing nanoparticle–immune cell interactions, adopting endosomal escape as a unifying framework, this review examines the mechanisms governing nanoparticle–immune cell interactions, from uptake routing and intracellular trafficking to immune activation, and derives cell-type-specific design principles, escape efficiency, and a decision framework for pairing escape mechanisms with target cell types. In addition, cell-type engineering is examined through the “delivery gap” that separates clinical success in solid tumors from failure in hematologic cancers. Rather than being classified neutrally, nanocarrier platforms are evaluated here according to translational viability. Manufacturing, regulatory, and safety considerations are synthesized into useful decision tools, and the review is concluded with a falsifiable ten-year plan and the main unanswered concerns in the field. Standardized endosomal escape and repeatable production are identified as the most manageable, high-impact short-term priorities by which nano-engineered immune cell treatments can be brought closer to the clinic.

Graphical Abstract

Introduction

Background and Rationale

Equipping immune cells with nucleic acids, proteins, and drugs rewiring them from within has moved to the very heart of next-generation cell therapy. Nanotechnology, the art of engineering materials at always the 1–100 nm scale, offers a non‑viral conduit for this cargo, with chemistries that can be tuned almost at will and a proven ability to traverse biological barriers.¹ Yet after two decades of intense, global effort, the field still lacks what every researcher craves: a single, dominant nanoparticle design that works across immune cell types.²

Why the impasse? We suggest that the research community may benefit from reframing the question. Most studies relentlessly optimize for cellular uptake, but immune cells especially professional phagocytes like macrophages and dendritic cells are already champions at swallowing nanoparticles. The cargo’s real bottleneck lies deeper: rapid routing to the lysosome, where it is destroyed before ever reaching the cytosol. This is not a matter of conjecture. Quantitative single-cell studies have repeatedly shown that only a small fraction of internalized cargo ever reaches the cytosol: using colloidal-gold electron microscopy and live imaging, Gilleron et al demonstrated that only 1–2% of internalized siRNA escapes the endosome,³ and recent super-resolution live-cell imaging has since confirmed and refined this picture, resolving cytosolic release as a rare and transient event among many endosomal-damage events.⁴ Framed within the sequential physiological barriers to delivery, escape is increasingly recognized not as one obstacle among many but as the terminal, still-unresolved barrier that remains once circulation, extravasation, and cellular uptake have already been solved.^5,6 Mechanistic work further shows that even successfully escaped cargo can be re-trapped by a secondary, slow-dissolving cytoplasmic aggregate through a vesicle budding-and-collapse pathway, concentrating the field’s major efficiency losses at and immediately after the escape step.⁷ The central challenge is not how to enter the cell, but how to escape the endosome. This review therefore turns the conventional framing on its head, placing endosomal escape as the organizing lens through which every platform and strategy is examined.

Nanotechnology finds its way into immune cell therapy through several elegant strategies. Nanocarriers deliver genetic and protein cargo into cells, sidestepping viral vectors. Nanoparticles are also tethered to cell surfaces, endowing the cells with new targeting or protective functions. And they are designed to release their payload controllably in vivo. Yet, however clever the architecture, all of these approaches converge on a single vulnerability: therapeutic impact lives or dies by whether the cargo can slip out of the endosome.

Immune Cell Therapy: Current Landscape

Over the last two decades, immune cell therapy has stepped out of the laboratory and into the clinic, earning FDA approvals for a growing list of malignancies. Adoptive cell transfer encompassing tumor-infiltrating lymphocytes, CAR-T cells, and TCR-engineered T cells has delivered complete remission rates exceeding 80% in certain B-cell malignancies and fueled multiple regulatory approvals since 2017.⁸

Beyond T cells lies a wider therapeutic repertoire: dendritic cell vaccines such as sipuleucel-T, natural killer cell therapies, macrophage polarization strategies that shift cells between M1 and M2 identities, and regulatory T cell approaches each operating through distinct mechanisms.^9–12 From a nanoparticle delivery standpoint, this diversity is both a challenge and a design imperative. These cell types have different endocytosis pathways. Macrophages and dendritic cells, the immune system’s professional eaters, gobble up particles with ease but then aggressively dismantle their contents. T cells and NK cells, by contrast, resist uptake almost entirely. The delivery problem, and therefore its solution, is exquisitely cell-type specific.

For all the clinical triumphs, formidable barriers still fence in the field. Manufacturing remains labyrinthine and expensive: producing a single CAR-T dose consumes weeks and costs can be expensive.¹³ Severe toxicities cytokine release syndrome and immune effector cell-associated neurotoxicity syndrome continue to limit treatment courses.¹⁴ In solid tumors, efficacy stumbles against poor immune infiltration and profoundly immunosuppressive microenvironments. And even when responses do occur, durability is threatened by T-cell exhaustion, antigen escape, and the failure of engineered cells to persist long-term. Importantly, nanoparticles are not a panacea; not every one of these obstacles can be solved at the nanoscale. Table 1 distinguishes the tractable from the intractable, anchoring this review firmly where nanoengineering can realistically make a difference.

Table 1.

Nanoparticle-Solvable versus Intractable Problems in Immune Cell Therapy with Reasoning for Solvability and Urgency. This Table Presents the Authors’ Conceptual Perspective on the Field; the Classifications and Ratings are Our Interpretation of the Cited Literature Rather Than Direct Experimental Results

Problem	Solvable versus	Performance	Ref
Poor tumor infiltration	Chemokine-releasing and ECM-degrading carriers build recruitment gradients and open dense stroma, directly attacking the physical barrier that keeps T cells out of solid tumours.	This is the single largest reason CAR-T fails in solid tumours, and it is mechanically addressable with existing carriers.	[15–17]
Cytokine release syndrome (CRS)	Macrophage-targeted anti-inflammatory nanoparticles dampen the myeloid cytokine cascade at its source rather than suppressing the whole immune system.	CRS is a leading dose-limiting toxicity; a targeted fix would widen the therapeutic window and improve safety quickly.	[18–20]
T-cell exhaustion	Metabolic-reprogramming cargo can delay exhaustion, but exhaustion is also driven by chronic antigen signaling and transcriptional programs that a carrier cannot fully reset.	Important for durability, but only one contributor among several, so the expected benefit is incremental rather than decisive.	[21–23]
Endosomal entrapment of cargo	This is fundamentally a materials problem escape-enhancing hybrid carriers (fusogenic lipids, proton-sponge polymers) release cargo into the cytosol before lysosomal degradation.	It is the rate-limiting step for every gene-delivery application, so solving it lifts the whole field at once.	[24,25]
Antigen escape	Tumours lose or downregulate the target antigen through biological adaptation; this is an evolutionary response, not a delivery failure, so a better carrier cannot prevent it.	Must be addressed by multi-antigen targeting or biological strategies, outside what nanoengineering controls.	[26,27]

Nanotechnology: An Enabling Platform for Advanced Therapeutics

A broad palette of nanomaterials lipid, polymeric, inorganic, and protein-based has been applied to immune-cell engineering.^28–30 Rather than catalog each in turn, we evaluate them against the criterion that matters most for cytosolic delivery.

Surface chemistry including PEGylation, targeting ligands, and pH- or enzyme-responsive coatings together with physical parameters such as size, shape, and charge, can be modulated to control circulation time, tissue targeting, and cellular uptake (Table 2).^31,32 Critically, however, these parameters govern delivery to the cell, not intracellular release. Composition, cargo, functionalization, targeting, and barrier navigation collectively define a multidimensional design space; the principal architectures are summarized in Figure 1.^33–35

Table 2.

Nanomaterial Platforms for Gene Delivery in Immunotherapy: Advantages, Limitations, and Clinical Applications

Nanomaterial	Possible Cargos	Disease	Gene-Delivery Advantages	Gene-Delivery Limitations	Ref
Lipid Nanoparticles (LNPs)	mRNA, siRNA, antigens	Melanoma, lymphoma, infectious diseases	High transfection; strong endosomal escape; optimal electrostatic binding of nucleic acids; clinically validated for mRNA/siRNA.	Serum instability for long-term expression; potential inflammatory responses; sensitivity to lipid composition during loading.	[33,36,37]
Polymeric Nanoparticles	Peptides, DNA, siRNA	HIV, immunotherapy	Sustained release; nuclease protection; tunable physicochemical properties.	Lower transfection than lipids; slower cytosolic release; possible cytotoxicity at high doses.	[38–40]
Gold Nanoparticles (AuNPs)	Peptides, antibodies, siRNA	Solid tumors, GI cancers	Highly stable nucleic-acid conjugation; excellent thiol surface chemistry; minimal cargo degradation.	Poor endosomal escape; limited intracellular release; accumulation limits repeat dosing.	[41–44]
Iron Oxide (IONPs)	Peptides, antibodies, siRNA	Brain tumors, solid tumors	Magnetically guided delivery; good surface modification; concomitant imaging.	Low transfection alone; insufficient escape; potential oxidative stress.	[45,46]
Mesoporous Silica (MSNs)	Drugs, immunomodulators	Melanoma, solid tumors	High pore loading for plasmids/siRNA; stable enzymatic protection; controlled release.	Difficult high transfection; limited membrane fusion; slow escape.	[47,48]
Dendrimers	DNA, siRNA, small molecules	Leukemia, melanoma	Strong nucleic-acid interaction; excellent siRNA/miRNA complexation; efficient escape.	Cationic-charge cytotoxicity; aggregation; rapid clearance without modification.	[49,50]
Carbon Nanotubes (CNTs)	Antigens, drugs, DNA	Lung cancer, vaccines	High loading via π–π interactions; easy cell penetration.	Cargo denaturation on surface; poor release kinetics; biosafety concerns.	[51,52]
Graphene Oxide (GO)	DNA, siRNA	Ovarian cancer, melanoma	Strong nucleic-acid adsorption; high stability; large surface for multi-gene loading.	Limited controlled release; oxidative damage; variable biocompatibility.	[53,54]
Exosome-Mimetic NPs	siRNA	Leukemia, MS	Natural RNA/miRNA compatibility; high uptake; minimal immune clearance; good protection.	Low loading for large plasmids; complex manufacturing; batch inconsistency.	[55,56]
MXenes	CRISPR-Cas RNPs, DNA plasmids	Cancer, immunotherapy	High conductivity; large surface area; strong nucleic-acid binding.	Oxidation instability; requires polymer coating.	[57,58]
Virus-Like Particles (VLPs)	Antigens, DNA	HPV, hepatitis, cancer vaccines	Exceptional packaging; strong cell entry; naturally optimized transport.	Hard to engineer large payloads; immunogenicity limits repeat dosing.	[59–61]
Nanogels	Antibodies, drugs	Solid tumors	High encapsulation; pH/enzyme-responsive release; gentle cargo protection.	Lower transfection than lipids; circulation instability; limited escape.	[62,63]
Protein-based NPs	Proteins, drugs	Breast cancer, immune disorders	Protect nucleic acids; receptor-mediated endocytosis; good biocompatibility.	Limited large-plasmid capacity; weak escape; lower efficiency than lipids.	[64,65]
Carbon/Quantum Dots	Fluorescent markers	Immunotherapy monitoring	Bind nucleic acids for pathway imaging; high surface-to-volume ratio.	Inefficient for therapeutic delivery; heavy-metal interference; toxicity limits use.	[66,67]

Figure 1.

Polymeric nanocarrier platforms. Composition, functionalization strategies, cellular interactions, and therapeutic applications. (Center) Core platforms polyplex, polymersomes, dendrimer, micelles, core-shell, nanosphere, and lip polyplex built from PBAE, PDMAEMA, PEI, PLGA, lipid–polymer hybrids, and pH-buffering polymers. (Inner ring) Cargo payloads (DNA, mRNA, siRNA, miRNA, plasmids, proteins, peptides, CRISPR-Cas9 components, small molecules, fusogenic peptides, cytokines) and carrier architectures. (Middle ring) Surface functionalization (PEGylation, charge modification, peptide/aptamer/antibody conjugation, enzyme-sensitive linkers, pH-responsive polymers, click chemistry) and immune-cell targets (CAR-T, TCR-engineered T cells, NK, Treg, B, γδ T cells, dendritic cells, M1/M2 macrophages). (Outer rings) Endocytic routes (phagocytosis, clathrin- and caveolae-mediated and -independent endocytosis, pinocytosis, macropinocytosis, RES uptake), intracellular trafficking and escape mechanisms (pore formation, membrane fusion, charge reversal, pH-buffering, ROS-induced release), systemic delivery barriers (protein coronas, proton-sponge effect, MPS clearance, anti-PEG antibodies, ECM density, blood–brain barrier), and therapeutic applications spanning vaccines, autoimmune modulation, liquid and solid tumors, gene therapy, and cell-based immunotherapy.

This review provides a focused analysis of nanotechnology-enabled immune cell therapy, organized around endosomal escape as the rate-limiting step. This paper examines mechanisms of uptake and intracellular trafficking, nanoengineering strategies relevant to T cells, NK cells, dendritic cells, macrophages, and regulatory T cells, and the preclinical and clinical evidence supporting each. Also, further address manufacturing, quality control, tracking, safety, immunogenicity, and regulatory considerations.

Designer’s Summary: Platform Selection by Cell Type and Cargo

For T-cell mRNA delivery, lipid nanoparticles formulated with ionizable lipids (pKₐ 6.5–7.0) are the platform of choice, as they combine high transfection efficiency with genuine endosomal escape. For macrophage gene editing, lipid–polymer hybrids or viral vectors are strongly preferred. Polymeric nanoparticles (PNPs) alone typically achieve less than 10% functional delivery because aggressive phagolysosomal degradation outpaces cytosolic release. These platforms should therefore be paired with an active escape mechanism, such as proton-sponge polymers, fusogenic lipids, or photothermal triggers, rather than deployed in isolation. Inorganic nanoparticles like gold, iron oxide, and mesoporous silica offer attractive physical properties but are consistently limited by inefficient endosomal escape. It seems the heavy-metal toxicity and propensity to induce oxidative stress associated with quantum dots and graphene oxide currently outweigh their potential advantages as carriers for therapeutic gene delivery, and we consider them a lower priority than lipid- and polymer-based systems for this application. We note, however, that efforts to develop safer, surface-modified, or biodegradable formulations of these materials are ongoing, and future advances may alter this assessment. Altogether, no monolithic nanocarrier satisfies all requirements simultaneously. The next generation of immune-cell engineering must move toward structurally rationalized hybrid systems that jointly balance payload protection, cell-type-specific uptake, and above all efficient endosomal escape, without provoking off-target systemic inflammation.

Mechanisms of Nanoparticle–Immune Cell Interactions

Cellular Uptake Mechanisms

The internalization of cationic-polymer and lipid-nanoparticle (LNP) vectors comprises several parallel routes whose relative contribution is set by the physicochemical identity of the carrier and the endocytic repertoire of the target cell. No single pathway is general, and this contingency is the defining feature for immune-cell gene therapy, because the cells of therapeutic interest T cells, dendritic cells (DCs), macrophages and B cells differ enormously in their native endocytic and membrane remodeling capacity. A route that dominates in a hepatocyte or a HeLa cell may be nearly absent in a resting lymphocyte. The mechanisms below are grouped by the strength of their evidence: those that are firmly established, those that are emerging but incompletely validated, and those that remain genuinely contested. Throughout, the immune-cell literature shows that entry is only the first of a series of barriers. Crucially, these barriers are interdependent rather than strictly hierarchical: uptake, targeting, intracellular trafficking, and cargo unpacking interact with rather than are subordinate to endosomal escape, and in primary T and NK cells the upstream uptake step is itself a major constraint. Resting T cells are largely non-phagocytic and show low rates of macropinocytosis,⁶⁸ and uptake and escape are physically coupled, since the same formulation variables that govern endocytosis rate also shape endosomal acidification and thus escape.⁶⁹ Optimizing escape alone therefore cannot resolve delivery where uptake remains limiting.

Endocytic Pathway Diversity Across Immune Cell Types

The dominant route for most synthetic vectors is endocytosis, but its subdivision into clathrin-mediated, caveolae-mediated and clathrin/caveolae-independent pathways is cell-dependent in a way that undermines any general claim (Figure 2). Dos Santos et al systematically applied pharmacological transport inhibitors across multiple cell lines and reported that the same carboxylated polystyrene nanoparticle engages different endocytic machinery depending on the cell, with inhibitor effects varying substantially between lines a direct demonstration that uptake route is a property of the particle–cell pair rather than of the particle alone.⁷⁰ In immune cells this variability is stark. In a Jurkat T-cell line, Tanaka et al optimized the PEG-lipid and phospholipid (POPE) content of an LNP and obtained a 221-fold increase in luciferase expression over a prior formulation, reaching efficiencies comparable to electroporation; using 4 °C blockade together with Pitstop 2, Dynole 34–2, genistein and cytochalasin D, they assigned the productive route specifically to clathrin/dynamin-dependent endocytosis with early endosomal escape.⁷¹ The Antigen-presenting cells (APC) behave differently. Kraus et al screened fifteen uptake inhibitors against fluorescently labelled virus-like nanoparticles in THP-1 monocytic and DC 2.4 dendritic cell lines and in primary lung and spleen APC, and found that only the macropinocytosis inhibitors rottlerin and hyperosmolar sucrose blocked uptake without cytotoxicity, identifying macropinocytosis by alveolar macrophages and CD103+ DCs as the principal route.⁷² Macropinocytosis and phagocytosis thus dominate in myeloid immune cells, whereas the same carrier must exploit clathrin-dependent routes to enter a lymphocyte at all.

Figure 2.

Major endocytic pathways for nanocarrier internalization and intracellular trafficking. (Left to right) Phagocytosis extends actin-filled pseudopods to engulf large particles (>500 nm) into phagosomes that fuse with lysosomes. Macropinocytosis uses actin-driven membrane ruffling to capture large fluid volumes into macropinosomes. Clathrin-mediated endocytosis (CME) is receptor-dependent: ligand binding recruits clathrin coats and dynamin to form coated vesicles. Caveolae-mediated endocytosis proceeds via caveolin-rich lipid-raft invaginations and GPI-anchored proteins to form caveosomes that usually escape lysosomal degradation. Clathrin- and caveolae-independent endocytosis (CLIC/GEEC) operates via coat-independent mechanisms. Pinocytosis non-specifically internalizes small soluble molecules and fluid. Most pathways converge on the endolysosomal system (early endosomes pH 6.0–6.5; late endosomes pH 5.0–6.0; lysosomes pH 4.5–5.0), where endosomal escape is critical before lysosomal degradation.

Protein Corona-Mediated Targeting

For the clinically dominant LNP platform, uptake is often not intrinsic to the particle but conferred by the serum protein corona. Akinc et al established that apolipoprotein E (ApoE) adsorbs onto ionizable but not permanently cationic LNPs and acts as an endogenous targeting ligand directing the particle to the low-density-lipoprotein receptor (LDLR); using genetic knockouts they showed expression was abolished in Ape- and LDLR-deficient mice and restored in vitro by recombinant Ape.⁷³ This Ape/LDLR corona is the mechanistic basis of hepatocyte tropism for Nonpatrol-generation carriers and, by the same logic, a barrier to immune-cell delivery because unmodified ionizable LNPs are routed to the liver, reaching lymphocytes generally requires overriding the endogenous corona with an active targeting ligand for example antibody-conjugated LNPs against a lymphocyte surface marker rather than relying on it.⁷⁴

Receptor-Mediated Surface Capture and Filopodial Transport

Surface capture precedes and biases this uptake, and it is an active receptor-mediated event rather than passive electrostatic adhesion. Rehman et al used real-time live-cell imaging to show that lipoplexes and polyplexes bind the heparan-sulfate proteoglycan (HSPG) syndical on filopodia and that this binding triggers actin retrograde flow conveying the carrier to the cell body, with actin- or myosin-II inhibition reducing transfection by 50–90%.⁷⁵ Because cationic-polyplex capture depends on surface HSPG engagement and downstream actin coupling, immune cells with sparse proteoglycan display or low endocytic activity resting T cells foremost are poorly captured, one reason primary human T cells are notoriously refractory to polyplex transfection.⁷⁶ Carrier physicochemical properties gate this capture further. Manzanares and Ceña, reviewing the physicochemical determinants of endocytosis, describe how nanoparticle size, shape and surface charge jointly select the entry route and its efficiency, so that loading nucleic-acid cargo which enlarges the carrier can shift a particle out of its optimal uptake window.⁷⁷

Building on the syndecan-capture model, filopodial surfing has emerged as a distinct pre-uptake step in which the carrier is not merely bound but actively transported along the filopodial shaft to the cell body by actin retrograde flow before endocytosis occurs. Rehman et al placed this actin-driven transport upstream of, and mechanistically separable from, the internalization event itself.⁷⁵ This reframes the earliest stage of uptake as an exploitation of native cytoskeletal machinery relevant to immune cells, who’s motile, filopodia-rich surfaces (as in migrating DCs) may present distinctive capture kinetics not captured by adherent cell-line models.

Endosomal Escape: Lipid Phase Behavior and Nanostructure

Once internalized, a well-documented structural transition governs whether cargo escapes the endosome. Koltover et al used synchrotron X-ray diffraction to show that cationic-lipid–DNA complexes adopting the inverted hexagonal (H_II) phase transfect more efficiently than lamellar complexes, and the mechanistically decisive point that H_II complexes are unstable when mixed with anionic model membranes whereas lamellar complexes bind them stably.⁷⁸ Instability toward the target bilayer, not phase geometry as such, predicts delivery, and this lamellar-to-H_II transition remains the canonical account of lipid-mediated endosomal escape in immune and non-immune cells alike. Lu and Sun, in a 2025 mechanistic account, argue that conventional bilayer-forming amino lipids escaping only via the lamellar-to-inverted-hexagonal route give suboptimal cytosolic delivery, and that non-lamellar amino lipids bearing large wedge-shaped tails which do not form stable bilayers and access inverse-hexagonal and cubic mesophases achieve more efficient escape.⁷⁹ They further describe pH-triggered amphiphilicity as an emerging design principle: such ionizable lipids remain neutral, non-amphiphilic or minimally amphiphilic at physiological pH 7.4 but become amphiphilic upon protonation in the acidic endosome (pH ~6.5–5.4), whereupon electrostatic interaction with the anionic endosomal membrane, combined with their wedge geometry, disrupts the bilayer and drives escape.⁷⁹

Distinct from phase behavior, LNP internal nanostructure and topology have been shown to govern escape independently of lipid composition: bicontinuous-cubic and inverse-hexagonal architectures escape endosomes markedly better than compositionally identical lamellar lipoplexes, establishing nanostructure as a tunable handle for fusogenicity.²⁴ Cooperative multi-particle penetration, in which several particles jointly perturb a membrane region to enable entry, has likewise been proposed largely on the basis of simulation and awaits experimental confirmation in immune-cell contexts.⁸⁰

Cargo–Carrier Dissociation

A second body of work relocates the rate-limiting event from membrane crossing to the cargo–carrier dissociation step that follows. Xu and Szoka established that when a cationic-lipid–DNA complex reaches the endosome, anionic lipids from the cytoplasmic-facing leaflet flip-flop into the complex, laterally diffuse and form charge-neutral ion pairs with the cationic lipid, releasing DNA near a 1:1 charge ratio.⁸¹ This single mechanism couple’s bilayer perturbation and cargo release, and it frames dissociation as a discrete barrier separable from uptake a distinction that matters acutely in immune cells, where a carrier can be internalized efficiently yet still fail to release functional cargo.

The oldest dispute concerns whether the proton-sponge effect osmotic rupture of the endosome driven by the buffering capacity of polyamines such as PEI actually operates. The affirmative case rests on measurements of endosomal chloride accumulation and swelling, but Benjaminsen et al tested the hypothesis’s central prediction using a nanoparticle pH sensor and found that PEI did not raise lysosomal pH, with PEI-containing lysosomes acidifying essentially normally to about pH 5.5, directly challenging the osmotic-rupture model.⁸² The competing membrane-destabilization account holds that, rather than bursting the vesicle, the escape-active species directly perturbs the endosomal bilayer. Upon acidification, protonated ionizable lipids form inverse-cone-shaped ion pairs with anionic endosomal lipids, generating local negative curvature that thins and disrupts the membrane at a defect site rather than rupturing the whole vesicle.⁷⁹ The two models make different predictions osmotic rupture releases the entire complex into the cytosol, whereas local destabilization causes cargo-only leakage through a transient defect while much of the carrier remains membrane-associated and current escape assays do not cleanly separate them. Indeed, current assay methods fall into fundamentally different, non-interchangeable categories: fluorescent-labelling, leakage, membrane-lysis, and transfection-based assays, each with distinct limitations, so escape-efficiency values reported by different methods are frequently not directly comparable.⁸³

Another related and unresolved claim is that the caveolae route allows carriers to avoid lysosomal degradation, in contrast to the clathrin route that traffics to lysosomes. This is frequently invoked to rationalize why caveolae-preferring formulations transfect better, but there is no clean primary demonstration isolating the route from confounding formulation differences, and it should be treated as a plausible but contested correlation. Similarly, the identity of the compartment from which LNPs actually escape early versus late endosome remains without consensus, and quantitative single-cell studies suggest escape is a rare, spatially restricted event rather than a bulk process. Underlying much of this uncertainty is a methodological problem the field increasingly acknowledges. Beyond the fragility of inhibitor-based pathway assignment, the field lacks a standardized, validated, cross-comparable assay for endosomal escape itself, and this absence is a foundational constraint rather than a minor caveat. A comprehensive analysis concludes that the available escape data are limited and contradicting, that there is no consensus on the compartment of escape or the cause of inefficient escape, and that the field currently lacks a robust method to detect escape.⁸⁴ Escape has accordingly been described as inherently context-dependent and difficult to characterize, varying across cell types, tissues, and disease settings,²⁵ and the absence of standardized, quantitative metrics has been argued to prevent meaningful cross-study comparison and to contribute to irreproducibility and publication bias, with physicochemical proxies (size, zeta potential, encapsulation efficiency) routinely over-interpreted as surrogates for cytosolic delivery.⁸⁵ Even recent computational approaches developed specifically to close this gap acknowledge that existing methods still lack scalability and accuracy.⁸⁶ This limitation applies to every quantitative escape value discussed in this review including the values underlying our own decision framework which should therefore be interpreted within, and compared only across, matched assay, cargo, cell-type, and platform categories. inhibitor-based pathway assignment is fragile, because commonly used pharmacological inhibitors have substantial off-target effects and cell-line-dependent activity, so conclusions about “the” uptake route of a given carrier in a given immune cell should be regarded as provisional unless corroborated by orthogonal genetic approaches.⁸⁷

Carrier internalization is determined not only by size, charge, and shape but also by how the payload and cargo features are packed within the particle, which can shift the dominant entry route. Working with porphyrin-labelled, siRNA-loaded LNPs across a range of N/P ratios, Mo et al showed that the way lipid and siRNA are organized within the particle controls both the extent of LNP–membrane association and the uptake pathway engaged (Figure 3).⁸⁸ Increasing the N/P ratio progressively strengthened binding of the LNP lipids to the plasma membrane, and pharmacological blockade with EIPA (a macropinocytosis inhibitor) and bafilomycin A1 (an endosomal-acidification inhibitor) revealed that macropinocytosis and endosomal processing contribute differently depending on formulation. This reinforces a central theme of this section: for nucleic-acid carriers, the entry route is a formulation-dependent variable a property that can be engineered through internal particle organization rather than an intrinsic constant of the lipid chemistry alone.

Figure 3.

Lipid–siRNA packing organization governs the cellular uptake pathway of siRNA-loaded lipid nanoparticles. (a) Confocal microscopy of PC3-Luc6 cells after 6 h of uptake of porphyrin-labelled LNPs (green, FAM-siRNA; magenta, porphyrin-lipid) formulated at increasing N/P ratios; the progressive accumulation of lipid signal at the outer cell membrane with rising N/P ratio (red arrows) indicates stronger LNP–membrane association. (b) Magnified view of the boxed region in (a), where bright-field imaging resolves vesicular structures at the plasma membrane. (c) Uptake under the same conditions in the presence of the macropinocytosis inhibitor EIPA (50 µM) and (d) the endosomal-acidification inhibitor bafilomycin A1 (1 µM), showing how each treatment alters LNP internalization and intracellular distribution across N/P ratios. Scale bars, 40 µm (whole-field panels) and 10 µm (magnified panels). Reprinted with permission from ref.88, Copyright (2025) American Chemical Society.⁸⁸

Cell-Penetrating Peptides for Immune Cell Delivery

In recent years use of Cell-penetrating peptides (CPPs) improve cellular uptake of nanocarriers, typically cationic or amphipathic sequences of roughly 5–30 residues offer an alternative to lipid and polymer carriers by translocating cargo across the plasma membrane directly, and they are of particular interest for lymphocytes, which resist most non-viral vectors. The prototypical CPPs, the HIV-1 TAT peptide and Antennapedia-derived penetratin, were shown to carry peptides and proteins into the cytoplasm of antigen-presenting cells, and antigens delivered this way generate antigen-specific cytotoxic and helper T-cell responses in mice with protection from tumor challenge, establishing CPPs as functional immune-delivery agents rather than mere uptake markers.⁸⁹ The most delivery-relevant primary demonstration in the immune context is the work of Lim et al on the peptide dNP2, which addresses the central difficulty that primary T cells are refractory to most carriers. They showed by flow cytometry that dNP2-conjugated EGFP was delivered efficiently into both mouse and primary human T cells, and, uniquely, that the peptide crossed the blood–brain barrier into brain tissue after systemic administration.⁹⁰ Crucially, the delivery was functional and not merely an uptake artefact. When the peptide was fused to the cytoplasmic domain of CTLA-4 (dNP2-ctCTLA-4) and applied to anti-CD3/CD28-activated splenocytes, it suppressed the effector cytokines IL-2, IFN-γ and IL-17A measured by ELISA, and in mice it ameliorated experimental autoimmune encephalomyelitis in both preventive and therapeutic dosing regimens, reducing demyelination and the infiltration of pathogenic Th1 and Th17 cells into the central nervous system.⁹⁰ This couples a quantitative cellular-delivery read-out to a disease-level functional outcome in the exact immune population the T cell that is hardest to reach. For nucleic-acid rather than protein cargo, the challenge shifts to endosomal entrapment, which the first-generation TAT and penetratin peptides did not solve; the engineered CPP PepFect6, which incorporates a chloroquine-analogue moiety to promote endosomal escape, was shown by Andaloussi et al to form non-covalent nanoparticles with siRNA that achieved efficient RNA-interference-mediated gene knockdown in a range of hard-to-transfect cells, including primary cells, without the toxicity or immunogenicity that limits cationic carriers, and remained active after systemic administration in vivo.⁹¹ Building further toward T-cell specificity, Kim et al characterized a nine-residue cysteine-containing peptide, AP, that delivered EGFP efficiently into non-phagocytic human T cells and, when fused to the CTLA-4 cytoplasmic domain, modulated T-cell function reinforcing that short engineered peptides can reach the lymphocyte cytosol where bulkier carriers fail.⁹² The recurring limitation across these studies is that CPP selectivity is modest; the same cationic character that drives translocation into T cells also drives uptake by many other cell types so CPPs are most compelling for immune-cell delivery when combined with a targeting element or an endosomal-escape enhancer, converting an efficient but indiscriminate translocator into a cell-selective one; this is the direction the more recent immune-focused CPP designs pursue.

In this regard, one holds that premature cargo release frustrates delivery dissociation that occurs too early, before the carrier reaches a productive compartment, wastes the payload while the other treats vigorous endosomal rupture as the goal. These are not trivially compatible maximizing early membrane destabilization to force escape may also promote premature release, and the optimal balance is likely cell-type-specific. For immune cells, whose slow and incomplete endosomal acidification alters both the timing of pH-triggered destabilization and the kinetics of dissociation, resolving which of these events is truly rate-limiting is the central open question, and it is the barrier this review argues matters more than uptake itself.

Proposed Design Rule

Two carrier properties size and surface charge function as the most direct engineering levers for biasing uptake, and both must be understood before pathway-switching ligands can be rationally designed, because they modulate the same routes those ligands would target. This is especially consequential in immune cells, where the productive entry routes differ sharply between lymphocytes and myeloid cells, and where the endocytic compartment a carrier enters largely determines whether cargo survives to function.

Particle Size as a Determinant of Immune Cell Uptake and Trafficking

The size dependence of uptake is best demonstrated in the immune compartment itself, where distinct cell populations sample distinct size ranges. Foged et al quantified this directly in human monocyte-derived dendritic cells, incubating immature DCs with fluorescent polystyrene particles spanning 0.04–15 µm and measuring internalization by flow cytometry with confocal confirmation. Although particles across the whole range interacted with DCs in a time- and concentration-dependent manner, the optimal diameter for rapid and efficient acquisition by a substantial fraction of the DC population was 0.5 µm and below, establishing a practical sub-micron ceiling for efficient DC loading.⁹³ Manolova et al extended this to the in vivo trafficking level using fluorescence-labelled polystyrene beads of 20–2000 nm injected subcutaneously in mice. They found that nanoparticle delivery to draining lymph nodes is strictly size-dependent small particles (20–200 nm) and 30-nm virus-like particles drained freely to the node and were captured by lymph-node-resident DCs and macrophages, whereas large particles (500–2000 nm) reached the node only after being carried there by DCs from the injection site. In DC-depleted animals only the 20-nm particles reached the lymph node, formally proving that large-particle delivery is DC-dependent while small particles drain passively.⁹⁴ Blank et al resolved this size selectivity at the level of individual APC subsets in the lung, instilling 20–1000-nm polystyrene particles intranasally and analyzing airway, parenchymal and lymph-node APC populations by flow cytometry and confocal microscopy. Alveolar macrophages internalized particles largely independently of size, whereas dendritic-cell subpopulations selectively captured the smallest (20-nm) particles, and the largest 1000-nm particles were transported to lung-draining lymph nodes by migratory CD11b^low DCs and deposited near CD3⁺ T cells.⁹⁵ Together these studies show that size does not merely set uptake efficiency but partitions a carrier among immune-cell types and trafficking routes.

Because size also gates which endocytic route dominates sub-200-nm carriers favoring clathrin- and receptor-mediated uptake while larger constructs recruit macropinocytosis and, in professional phagocytes, phagocytosis (which for spherical particles is most efficient in the low-micron range) the enlargement that accompanies nucleic-acid loading is itself a pathway-shifting event a carrier optimized for size at the point of formulation may cross a route boundary once cargo is added.⁹⁶ For immune-cell delivery this coupling is a design constraint rather than a footnote. Myeloid cells such as dendritic cells and macrophages, whose constitutive macropinocytic and phagocytic activity accommodates larger constructs, are comparatively tolerant of the size increase that cargo loading imposes; resting T cells, which depend on the more size-restricted clathrin route and have low baseline endocytic activity, against a polyplex or lipid nanoparticle (LNP) that enters a lymphocyte efficiently when empty may fall outside the affecting window once condensed with mRNA or plasmid.

The route dependence of geometry can be visualized directly by colocalizing each particle with markers of specific entry pathways. Co-incubating spherical and worm-shaped silica nanoparticles with transferrin (clathrin-mediated endocytosis) and dextran (fluid-phase endocytosis) in RAW 264.7 macrophages, Herd et al observed markedly greater overlap of both markers with spherical than with worm-shaped particles, indicating that spheres are preferentially internalized through clathrin- and fluid-phase-mediated routes. Colocalization was nonetheless detectable for both geometries, confirming that these pathways contribute to uptake across shapes while their relative weighting shifts with particle geometry.⁹⁷

Surface Charge as a Modulator of Uptake Rate and Cell-Type Selectivity

Surface charge is the second lever, though its mechanism is more contested than commonly assumed. Fröhlich, synthesizing the charge literature, reported that nonphagocytic cells preferentially internalize cationic nanoparticles while phagocytic cells preferentially take up anionic ones, but that critically cells do not, as a rule, employ different endocytic routes for cationic versus anionic particles; charge changes the rate and the cell-type selectivity of uptake more than it switches the pathway itself.⁹⁸ The immune-cell corollary is directing the phagocytic myeloid compartment (macrophages, immature dendritic cells) is intrinsically biased toward anionic carriers, whereas cationic carriers are internalized more avidly by nonphagocytic cells including lymphocytes but, in both cases, charge sets how much and how selectively, not by which route.

Harush-Frenkel et al tested this directly in polarized MDCK epithelial cells using cationic and anionic PEG–polylactide nanoparticles of comparable size, tracking entry with pharmacological inhibitors and confocal colocalization. They found that both cationic and anionic nanoparticles were targeted mainly to the clathrin machinery, with a fraction of each entering through a macropinocytosis-dependent route; the charge-dependent difference lay not in the entry pathway but in the intracellular fate, as some anionic but not cationic particles transited the degradative lysosomal pathway, while cationic particles showed roughly two-fold greater overall uptake.⁹⁹ This dissociation of route from fate matters for immune-cell engineering, because it means that selecting charge to boost uptake and selecting charge to avoid lysosomal degradation are not the same decision, and can pull in opposite directions.

The internal phase structure of a lipid carrier is a further determinant of how much material a cell takes up, independent of overall size or surface charge. Comparing lipid nanoparticles with distinct internal architectures lamellar liposomes versus non-lamellar cubosomes, hexosomes, and micellar cubosomes Yap et al used label-free Raman microscopy to quantify internalized lipid and found that the non-lamellar structures were taken up substantially more efficiently, delivering roughly two-and-a-half to three times the lipid content of conventional liposomes (Figure 4). Because these formulations were matched in composition and differed principally in internal nanostructure, the result isolates internal architecture as an uptake-determining variable in its own right. For immune-cell delivery this widens the design space beyond the familiar size and charge levers: the way lipids are packed within a carrier can be tuned to bias both the amount internalized and, as the same study showed through pathway-marker colocalization, the endocytic route engaged.¹⁰⁰

Figure 4.

Internal lipid nanostructure controls the extent of lipid nanoparticle uptake, imaged label-free by Raman microscopy. False-colour MCR-ALS component maps of cells after 6 h of incubation with structurally distinct lipid nanoparticles liposomes, cubosomes, hexosomes, and micellar cubosomes with the cellular region shown in green and lipid-rich regions in red. Quantification of the lipid-rich signal (integrated lipid intensity normalized to total cell area) confirmed that the non-lamellar carriers accumulated markedly more lipid than lamellar liposomes: liposomes gave the lowest signal (≈0.10 a.u), rising to ≈0.25 a.u. for cubosomes, ≈0.28 a.u. for hexosomes, and ≈0.29 a.u. for micellar cubosomes an approximately two and a half to threefold increase in internalized lipid for the non-lamellar structures relative to liposomes. Reprinted with permission from ref. 57, Copyright (2025) John Wiley and Sons.¹⁰⁰

Receptor-Mediated Pathway Switching via Scavenger Receptors

Where charge does redirect a pathway, it appears to do so through specific receptor engagement rather than bulk electrostatics. Choi et al showed that polyanionic spherical nucleic acids (SNAs) cores densely functionalized with oligonucleotides are internalized by binding class A scavenger receptors and entering through a lipid-raft-dependent, caveolae-mediated pathway. Using RNAi silencing, they demonstrated the mechanism directly: knockdown of the scavenger-receptor gene MSR1 produced the largest reduction in SNA uptake, and caveolin-1 knockdown also markedly reduced uptake, whereas disrupting the clathrin pathway had comparatively little effect identifying scavenger-receptor engagement and caveolae-mediated entry as the dominant uptake route.¹⁰¹ The route was thus determined by a defined ligand–receptor interaction the three-dimensional oligonucleotide architecture recognized by scavenger receptor A rather than by net charge alone. This is an important precedent for immune-cell design, because scavenger receptors are highly expressed on macrophages and dendritic cells, making scavenger-receptor engagement a rational handle for redirecting carriers into the less degradative caveolar route in precisely the myeloid populations that otherwise funnel anionic cargo toward lysosomes.

Taken together, these studies argue that future carriers for immune-cell gene delivery should incorporate cell-type-specific targeting ligands that actively redirect uptake away from degradative pathways (phagocytosis, clathrin-mediated endocytosis) toward less degradative ones (caveolae, macropinocytosis), rather than relying on bulk size or charge, which set rate and selectivity but rarely switch the route cleanly. For T cells specifically, the objective is a carrier that shifts from clathrin-mediated uptake which traffics to the lysosome toward caveolae-mediated uptake, which largely bypasses it; the Choi scavenger-receptor precedent shows this redirection is achievable through defined receptor engagement. Such pathway-switching ligands would convert uptake itself into the first line of cargo protection, rather than placing the entire burden on downstream endosomal escape.

Intracellular Trafficking and Cargo Release

Endosomal escape and lysosomal degradation are not the only fates that determine how much cargo reaches its intracellular target. A carrier that has been internalized and has evaded the degradative pathway can still be lost to the cytosolic delivery pool through active expulsion, and this route of loss is frequently omitted from mechanistic accounts even though it can dominate the intracellular mass balance. The scale of the problem is stark across the endo-lysosomal pathway, in some target cells fewer than 5% of internalized nanoparticles ever transfer their cargo to the cytosol, meaning the overwhelming majority is either degraded or returned to the extracellular space.¹⁰² This inefficiency is compounded in immune cells, who’s degradative and recycling machinery is highly active. The macrophages and dendritic cells are professionally equipped to sample, digest, and re-secrete internalized material, and even lymphocytes divert much of what they take up. Several distinct fates therefore compete for an internalized carrier lysosomal degradation, endosomal escape, size-dependent exocytosis of the intact carrier, and recycling-mediated expulsion of the cargo. Understanding delivery in immune cells means treating these as parallel, competing outcomes rather than a single linear pathway.

The Magnitude of the Endosomal Escape Bottleneck

The magnitude of the escape bottleneck was fixed by two landmark 2013 studies that reached the same conclusion by different methods. Gilleron et al combined quantitative electron microscopy with fluorescence imaging to track LNP-delivered siRNA in cells and determined that only a minor fraction 0177 roughly 1–2% of internalized siRNA ever escapes the endocytic system into the cytosol, and localized this release to a narrow window in the moderately acidic early-to-late endosome, before the cargo reaches the mature lysosome.³ In parallel, Sahay et al used automated high-throughput confocal microscopy with fluorescently labelled siRNA and systems-biology perturbation to map the same pathway, showing that LNPs enter primarily by macropinocytosis (requiring proton pumps, mTOR and cathepsins) and that escape competes against a dominant loss route.¹⁰³ Critically, this dominant competing route is not simple degradation but active expulsion: Sahay et al showed that approximately 70% of internalized siRNA is returned to the extracellular space through NPC1-mediated recycling and exocytosis, establishing endocytic recycling rather than escape failure alone as a distinct and quantitatively dominant bottleneck.¹⁰³ Together these define a delivery ceiling even a well-designed carrier delivers only a few percent of what it internalizes, and in immune cells where cargo such as CAR or antigen-encoding mRNA must reach the cytosol in functionally meaningful amounts this ceiling is the central engineering constraint. Importantly, the size of this bottleneck cannot be inferred from uptake measurements. Paramasivam et al demonstrated that total cellular uptake is not a sufficient predictor of productive cytosolic delivery, and that escape occurs preferentially from a narrow window of early/recycling compartments such that prolonged uptake can even be counterproductive by impairing endosomal acidification. Escape efficiency is therefore not a fixed property of a formulation but a function of trafficking kinetics that uptake-stage design choices also shape.¹⁰⁴ This reinforces the need to treat uptake and escape as coupled, rather than independent, determinants of delivery.

Wittrup et al resolved the timing and single-vesicle behavior of this rare escape event by correlating live-cell imaging of lipoplex- and LNP-formulated siRNA with knockdown of a GFP reporter. They found that cytosolic release occurred invariably from maturing endosomes within a narrow window of roughly 5–15 minutes after endocytosis, that escape happened from only a small subset of endosomes as a sudden burst rather than a gradual leak, and critically for immune-cell delivery that cytosolic galectins immediately recognized the damaged endosome and targeted it for autophagy.¹⁰⁵ This last observation converts galectin recognition from a passive read-out of damage into an active loss mechanism the same disruption that lets cargo out also flags the vesicle for degradation, so a slow or partial escape can be aborted by the cell’s own repair-and-clearance machinery before enough cargo has been released.

The first non-degradative loss route is expulsion of the intact carrier, which follows its own size law distinct from that governing uptake. Chithrani and Chan reported that transferrin-coated nanoparticles, having entered cells through clathrin-mediated endocytosis, were subsequently exocytosed in a near–linear relationship to particle size a dependence different from the non-monotonic size relationship that governs uptake and derived a predictive equation relating size to exocytosis rate across cell lines.¹⁰⁶ Because uptake and expulsion follow different size laws, intracellular accumulation is set by the balance between them rather than by uptake alone, so a particle optimized purely for entry may nonetheless be cleared rapidly. Single-particle tracking extended this to a general principle Jin et al followed individual carbon nanotubes entering and leaving cells and formulated a generic kinetic model in which cells continuously eject internalized material, establishing exocytosis as a general rather than a material-specific phenomenon.¹⁰⁷

In immune cells, this expulsion is not merely passive turnover but a receptor-regulated process. Cui et al showed that in RAW264.7 macrophages the exocytosis of internalized single-walled carbon nanotubes is mediated by the P2X7 receptor an ATP-gated cation channel highly expressed on macrophages and other immune cells so that macrophage expulsion of nanomaterial is under active signaling control, and shorter carriers were ejected faster than longer ones.¹⁰⁸ Sakhtianchi et al, reviewing the exocytosis literature, framed this as a general determinant of cellular retention because release competes with retention, the fraction of a dose that stays inside long enough to act is governed by exocytic rate as much as by uptake, a balance that is especially unfavorable in the professional phagocytes central to immune therapy.¹⁰⁹ For phagocytic cells, whose high membrane turnover makes them efficient at both engulfing and re-externalizing particulate material, a carrier that dwells too briefly is expelled before it can act.

These escape-stage constraints, moreover, do not act in isolation from the upstream uptake step, which is itself a serious barrier in certain immune cells. Jain et al showed that primary T cells exhibit marked resistance to nanoparticle transfection that is independent of any downstream escape step, underscoring that in these cell types uptake and escape represent distinct, sequential barriers and that optimizing escape alone will not resolve delivery where uptake itself remains limiting.^110,111 This interdependence is developed further in the cell-type-specific sections below.

Recycling-Mediated Loss of Nucleic Acid Cargo

For nucleic-acid carriers, the dominant non-lysosomal loss route is expulsion driven by endocytic recycling, and its magnitude is large enough to set an upper bound on delivery. Using high-throughput confocal imaging of fluorescently labelled siRNA, Sahay et al quantified this directly approximately 70% of internalized LNP–siRNA is exocytosed back to the extracellular space through egress of the particles from late endosomes and lysosomes, leaving only a minor fraction retained long enough for cytosolic release.¹⁰³ Critically, this expulsion is not a passive leak but an actively regulated pathway. The authors identified Niemann-Pick type C1 (NPC1) the same cholesterol-transport protein that Ebola virus exploits for entry as a key regulator of LNP recycling NPC1-deficient cells retained far more siRNA within late endosomes and, as a direct consequence, showed markedly stronger target-gene silencing, and silencing could be enhanced in wild-type cells simply by knocking NPC1 down. This establishes recycling-mediated exocytosis as a genetically tractable bottleneck rather than an immutable physical loss, and points to a target the recycling machinery itself that could be manipulated to raise retention in immune cells.

The mechanistic significance is that recycling and endosomal escape are competing fates for the same endosomal cargo, not separate loss channels acting on separate populations. A particle sorted from the macropinosome or late endosome into the recycling compartment whether returned directly to the plasma membrane, routed through the endoplasmic reticulum and Golgi, or expelled by fusion of multivesicular bodies as exosomes is removed from the pool before it can escape. Cargo returned to the surface before the escape-competent window closes is simply lost, which is why raising delivery efficiency depends as much on prolonging intracellular retention as on promoting membrane destabilization. For LNP–siRNA the expelled entity is the recycling-sorted carrier and its associated cargo, so the two loss routes differ in both mechanism and the cargo population they deplete.

Endosomal Escape as an Immune Danger Signal

A dimension absents from most trafficking accounts, but central to immune-cell delivery, is that endosomal escape does not go unnoticed by the cell. The membrane disruption that permits cargo release is detected by an endogenous surveillance system the galactoside-binding lectins known as galectins rapidly relocate to damaged endosomal membranes, a response that has been exploited to visualize and quantify escape events but that also constitutes a danger signal.¹¹² In immune cells this sensing has functional consequences: the endosomal membrane damage produced by LNP escape can trigger inflammatory signaling, so that the very event required for delivery may simultaneously activate the target cell’s innate immune machinery an effect shown in RAW 264.7 macrophages and MLE-12 epithelial cells that likely depends on carrier composition, cargo, and cell type. Importantly, this link is not fixed: ionizable lipids engineered to form smaller, ESCRT-reparable endosomal holes reduce inflammation while preserving delivery, showing that escape efficiency and tolerability can be decoupled through carrier design.¹¹³ For a hepatocyte this is largely irrelevant, but for a macrophage, dendritic cell, or T cell it means that maximizing escape can also maximize an unwanted inflammatory or activation response, creating a therapeutic-window problem unique to immune targets. Escape efficiency and escape tolerability are therefore distinct optimization targets in immune-cell delivery, a tension that does not arise in the classic hepatocyte model.

A loss route with no counterpart in the hepatocyte model is the direct hand-off of internalized carriers between immune cells. Because macrophages are professional scavengers that line the microvasculature, they sequester a large share of any systemic dose, and that material does not necessarily stay put. Franco et al showed, using mesoporous silica nanoparticles as a model, that macrophages transfer scavenged nanoparticles to neighboring cells along tunnelling nanotubes and other cytoplasmic bridges, with robust transfer from macrophages to cancer cells and rich nanoparticle exchange among immune cells in vivo.¹¹⁴ For a review focused on immune-cell delivery this is doubly important it is simultaneously a loss route (cargo intended for one macrophage is redistributed elsewhere) and a potential delivery strategy (macrophages as living carriers that relay cargo into a target tissue). Either way, it means the intracellular fate of a carrier in an immune cell cannot be treated as a closed system, because the cell can export intact carrier to its neighbors.

Design Strategies to Mitigate Intracellular Cargo Loss

Several design approaches can mitigate these losses, though each carries a trade-off. First, because exocytosis scales with size, tuning particle dimensions toward the range that maximizes retention rather than uptake can shift the balance, at the cost of potentially lower initial internalization.¹⁰⁵ Second, ligand-directed routing toward compartments less prone to recycling can extend intracellular residence, though this narrows the choice of targeting ligand. A concrete realization of receptor-directed design in the immune-cell context is the antigen-presenting-cell-mimetic LNP reported by Metzloff et al, who conjugated reduced anti-CD3 and anti-CD28 antibody fragments to the surface of an ionizable-lipid (C14-4/DOPE) LNP carrying CAR mRNA. The resulting activating LNPs performed one-step activation and transfection of primary human T cells without separate activation beads, generated anti-CD19 CAR T cells that killed target cells and proliferated ex vivo, and reduced tumor burden in a xenograft leukemia model; because the ionizable core is neutral at physiological pH but protonates in the acidic endosome to drive escape, the design couples uptake, activation, and cytosolic release in a single construct illustrating how surface engineering can bias both entry route and downstream fate rather than uptake alone.¹¹⁵ Third, endosomal-escape kinetics can be accelerated so that release outcompetes recycling a carrier engineered to destabilize the endosomal membrane early, before the recycling machinery returns its contents to the surface, converts the timing competition in favor of delivery. This is the most direct approach, because the recycling loss is fundamentally a race between escape and expulsion; the practical implication is that escape efficiency and escape speed are distinct optimization targets, and the field’s near-exclusive focus on the former has left the latter underexploited.¹⁰³ Finally, PEGylation and surface chemistry that reduce recognition by the recycling and exocytic machinery can prolong retention, but excessive shielding impairs the very membrane interactions required for escape the recurring tension between stealth and fusogenicity. No single approach resolves the loss problem; each redistributes it, and the appropriate balance is cell-type-specific and in immune cells, uniquely constrained by the additional requirement that escape not provoke a damaging inflammatory response.

Critical Evaluation of the Proton-Sponge Effect

The high cationic charge density that enables proton-sponge-driven endosomal escape is also the source of its dose-limiting toxicity.^116,117 Membrane destabilization is not selective for the endosome; it extends to the plasma and mitochondrial membranes and can provoke inflammatory signaling cascades.^113,118 This toxicity is substantially amplified in primary immune cells compared with immortalized cell lines precisely the comparison that most published studies omit.¹¹⁹ As a consequence, endosomal-escape efficiencies measured in robust cell lines systematically overstate the performance achievable in the clinically relevant primary T cells, NK cells, and macrophages (Table 3).³ The field therefore requires two advances a standardized assay for quantifying endosomal escape, and an explicit, decision-oriented mapping that pairs each escape mechanism with its appropriate cell type. its recommended that all future studies quantify endosomal-escape efficiency using a Galectin-8 recruitment reporter or a calibrated pH-sensitive fluorescence probe, and that these measurements be performed in primary immune cells, not cell lines alone (Table 4).¹²⁰ An escape efficiency below ~10% is insufficient to support therapeutic gene editing. Also, a minimum threshold of 30% escape efficiency as the benchmark for clinical relevance, reported alongside a matched viability measurement so that efficiency is never decoupled from toxicity.¹²¹

Table 3.

Endosomal-Escape and Cytosolic-Delivery Efficiency Across Nanocarrier Platforms

Platform	Cargo	Model/Cell Type	Assay/Method	Reported Escape Efficiency or Qualitative Level	Ref
LNP (DLin-MC3-DMA)	siRNA/mRNA	HeLa; hepatocyte	Quantitative EM; single-molecule fluorescence microscopy	~1–2.5% of internalized cargo reaches cytosol	[3]
LNP (SM-102/MC3/ALC-0315)	mRNA	HeLa	SNAPSwitch cytosolic sensor	~10%/~5%/~4% (clinical ionizable lipids)	[122]
LNP (optimized ionizable lipid, eg Lipid 5/SM-102 analogue)	mRNA	In vitro	Single-molecule fluorescence microscopy (FISH); deep-learning reporter (RNASCAPE)	~15% (best-in-class; ~6-fold higher than MC3); ~5–9% sterol-dependent	[79,123]
Polymeric NP (PBAE/PEI)	mRNA/siRNA	High-throughput screen	Galectin-8 fluorescent sensor	Low; escape inversely correlated with uptake, strongly formulation-dependent	[124]
Gold NP (AuNP)	siRNA/peptide	In vitro	Imaging; knockdown	Low, consistently escape-limited; requires added escape moiety	[25,41]
Iron oxide (IONP)	siRNA/antibody	In vitro	Imaging	Low escape alone; needs cationic or escape-active coating	[45]
Mesoporous silica (MSN, PEI/GuIL-functionalized)	siRNA/drug	MDA-MB-231	qPCR knockdown; flow cytometry + CLSM	Moderate, ~87% EGFR1 knockdown at 48 h with escape functionalization; escape shown by CLSM at 6 h	[125]
Dendrimer (PAMAM)	siRNA/DNA	In vitro	Gene knockdown	Moderate–high via proton-sponge; generation-dependent	[49]
Carbon nanotube (CNT)	DNA/siRNA	In vitro	Imaging	Variable; escape via needle-translocation, proton-sponge, or photothermal routes	[126]
Graphene oxide (GO)	DNA/drugs	In vitro	Imaging	Poor/variable escape; oxidative-stress confound	[53]
Exosome-mimetic NP	siRNA/protein	In vitro	Imaging	Membrane-fusion escape; efficiency not standardized-quantified	[55]
MXene	CRISPR/siRNA	In vitro	Functional editing readout	Emerging; escape viaresponsive/logic-gated release, not yet quantified	[57]
Virus-like particle (VLP)	mRNA/protein/editor	In vitro; in vivo	Functional delivery readout	High functional delivery via native viral membrane-penetration machinery	[59]
Nanogel	siRNA/protein	In vitro	Imaging; knockdown	Moderate; escape depends on pH-responsive/charged network	[62]
Protein-based NP	drug/protein	In vitro	Imaging	Generally, escape-limited without an added endosomolytic moiety	[65]
Carbon/quantum dots	siRNA/DNA	In vitro	Imaging	Low functional delivery; heavy-metal and oxidative toxicity confound assessment	[67]

Table 4.

Selecting an Endosomal-Escape Mechanism by Immune-Cell Type Efficiency, Toxicity, and Scalability Explained Rather Than Scored

Target Cell	Dominant Pathway	Preferred Escape Mechanism	Efficiency	Toxicity	Scalability	Ref
T cells (primary)	Clathrin-mediated (CME)	Ionizable-lipid LNPs/fusogenic lipids	High. ionizable lipids fuse with the endosomal membrane at low pH and release cargo efficiently in resting T cells.	Low. charge appears only transiently at acidic pH, so membrane damage is limited and viability is preserved.	High. The LNP supply chain is mature and clinically validated by mRNA vaccines.	[71,79,127]
NK cells (primary)	Clathrin-mediated (CME)	Ionizable LNPs (NK-optimized)	Moderate. works once the formulation is re-tuned, because NK membranes differ in lipid-raft composition from T cells.	Low. same transient-charge advantage as in T cells, with NK cells being somewhat more delivery-resistant.	High. Uses the same LNP manufacturing base, so scale-up is inherited rather than rebuilt.	[128,129]
Dendritic cells	Phagocytosis, macropinocytosis	Fusogenic peptides; pH-responsive polymers	Moderate. DCs internalize readily, but aggressive lysosomal processing competes with escape.	Moderate. Peptide- and polymer-driven disruption is less self-limiting than ionizable lipids, raising off-target membrane risk.	Moderate. Peptide synthesis and conjugation add cost and batch variability versus plain LNPs.	[72,93]
Macrophages (phagocytic)	Phagocytosis, macropinocytosis	Photochemical internalization; fusogenic peptides	High. light-triggered escape ruptures the phagosome on demand, overcoming the strongest degradative compartment in any immune cell.	Low. Toxicity is restricted to the illuminated region, so damage stays local.	Moderate. Requires a light source and controlled illumination, which complicates in-vivo and large-batch use.	[130,131]

Rather than proposing a single universal threshold, we emphasize standardized reporting, because reported escape efficiencies vary by more than an order of magnitude with cargo, platform, and cell type: siRNA-LNP ~1–2%,³ mRNA-LNP ~5–9% with other compositions reaching ~15%,⁸⁶ and protein cargo delivered via ZF5.3 reaching 47–72% and in some cases >80%.¹³² Escape has moreover been described as inherently stochastic, transient, and heterogeneous,⁸⁶ and dependent on uptake kinetics and endosomal-compartment residence time. We therefore recommend that escape efficiency be reported alongside a matched viability measurement so that efficiency is never decoupled from toxicity and compared only within matched cargo–cell–platform categories, rather than against any single fixed benchmark.¹²¹

Immune Cell Activation and Modulation

Nanoparticles do more than ferry cargo; they directly shape the phenotype of the immune cells they encounter. This dual role creates both an opportunity for rational immunomodulatory design and a substantial risk of unintended, off-target immunogenicity. Pattern-recognition receptors including Toll-like receptors, NOD-like receptors, and C-type lectins are tuned to sense pathogen- and damage-associated molecular patterns, and several nanomaterial classes inadvertently engage them. Cationic nanoparticles, for instance, can stimulate TLR4, while many PNPs activate the NLRP3 inflammasome, triggering inflammatory cascades that can confound therapeutic intent.^14,133 While useful for adjuvants, this is usually an unwanted side effect in gene-delivery contexts. This off-target activation is rarely characterized systematically, yet it is a leading cause of nanoparticle-associated inflammation and a likely determinant of clinical failure.

The observation that most cationic nanoparticles inadvertently activate TLR4 and the NLRP3 inflammasome and thereby provoke unwanted inflammation carries a clear design imperative future carriers must actively suppress innate immune sensing rather than simply ignoring it. Two complementary strategies hold particular promise. The first incorporates TLR-inhibitory motifs, such as elements derived from the lentiviral accessory protein Vpx. The second replaces cationic surfaces with zwitterionic polymer coatings that resist nonspecific protein adsorption and thereby avoid triggering pattern-recognition receptors. It is anticipated that the first broadly successful clinical nano-immunotherapy will be built not on a cationic platform, but on a deliberately TLR-silent carrier one in which immunological “stealth” is engineered with the same rigor as cargo protection.

Beyond avoiding unintended activation, nanoparticles can actively and productively direct immune-cell function. In dendritic cells, they enhance antigen presentation and drive maturation marked by upregulation of CD80, CD86, and CD40 through signaling cascades that converge on MyD88/TRIF, NF-κB, and the MAPK axis (ERK, JNK, p38). Targeting of C-type lectin receptors such as DEC-205 or DC-SIGN further sharpens specificity (Figure 5a).^35,134,135 Macrophage polarization can similarly be steered by cargo choice. Delivery of IFN-γ or TLR agonists biases cells toward a pro-inflammatory M1 phenotype via NF-κB and STAT1, while IL-4 or IL-13 drives a reparative M2 program through STAT6 and PI3K–Akt. These opposing fates find application in cancer immunotherapy and regenerative medicine, respectively (Figure 5b).^{127,136–138} For T cells, nanoparticles decorated with anti-CD3 and anti-CD28 function as artificial antigen-presenting cells. TCR engagement triggers CD3 phosphorylation, ZAP-70 activation, and LAT/SLP-76-dependent signaling that culminates in NFAT and AP-1; CD28 co‑stimulation engages PI3K–Akt; and co-delivery of cytokines such as IL-2 or IL-15 activates JAK–STAT5 to support clonal expansion and long-term persistence (Figure 5c).^139,140 In contrast, nanoparticles can also be configured for immunosuppression delivering anti-inflammatory drugs, siRNA targeting pro-inflammatory mediators, or Treg-inducing signals offering therapeutic avenues for autoimmune disease, graft rejection, and cytokine release syndrome.¹³⁹

Figure 5.

Nanocarrier-mediated modulation of immune-cell function and therapeutic applications. (a) Dendritic-cell activation via DC-SIGN and TLR recognition, triggering MyD88/TRIF signaling that activates NF-κB and MAPK, upregulates CD80/CD86/CD40, and induces IL-12 and TNF-α. (b) Macrophage polarization to pro-inflammatory M1 (NF-κB, STAT1; IFN-γ/TLR stimulation; TNF-α, IL-1β, IL-6) or anti-inflammatory M2 (STAT6, PI3K-Akt; IL-4/IL-13; IL-10). (c) T-cell activation by anti-CD3/anti-CD28 nanoparticle aAPCs engaging TCR and CD28 co-stimulation through ZAP-70, PI3K-Akt-mTOR, and JAK-STAT, with NFAT and AP-1 driving IL-2/IL-15 production and proliferation. (d) Lymph-node drainage of subcutaneously injected nanocarriers, where DCs process antigen via endosomal MHC-II loading (CD4+) or cytosolic escape and MHC-I cross-presentation (CD8+). (e) Tumor-targeted delivery to macrophages recruited by CCL2/CSF-1 gradients. (f) mRNA-nanocarrier reprogramming of macrophages into CAR-macrophages. (g) Metabolic re-education of Tregs (metformin, AMPK activators, mitochondrial antioxidants, FAO enhancers) switching glycolysis to oxidative phosphorylation to stabilize Foxp3. (h) Recruitment of Tregs to inflamed tissues via CCL22/CXCL10 gradients and CCR8 engagement.

Nanoparticle-Based Engineering of Therapeutic Immune Cells

In immune cell therapy, clinical success is largely dictated by geography. CAR-T cells achieve complete remissions in B-cell malignancies because their targets circulate in the blood and bone marrow compartments that infused cells can readily access. In solid tumors, that advantage vanishes the same cellular platforms confront poor infiltration, an immunosuppressive microenvironment, and limited persistence, all of which erode efficacy. Across every cell type examined below, the decisive question therefore shifts from whether a nanoparticle can modify a cell in vitro to whether it can close the specific delivery or functional gap that limits clinical impact. We use that criterion to rank strategies rather than simply enumerate them.

T Cell Engineering and Enhancement

T cells are the most clinically advanced target, with CAR-T therapy achieving remarkable success in hematological malignancies.^141,142 Non-viral delivery now offers several routes to engineer T cells without the cost, immunogenicity, and insertional-mutagenesis risks of viral vectors ionizable-lipid mRNA-LNPs give transient CAR expression with strong persistence and low toxicity;^143–145 antibody-targeted LNPs against CD3/CD5/CD7 enable in vivo CAR generation with up to 90% B-cell depletion;^146,147 DNA-LNPs with transposase achieve durable integration;¹⁴⁸ PBAE, dendrimer, and PLGA carriers offer lower-cytotoxicity alternatives;^149,150 and nanoparticle-delivered CRISPR RNPs enable multiplexed editing such as simultaneous PD-1/CTLA-4 knockout and CAR knock-in.^151,152 These methods are not equivalent in clinical readiness, and we say so explicitly rather than list them neutrally. Surface “backpacking” conjugates nanoparticles to the T-cell membrane without internalization, preserving viability while adding function IL-15 depots for persistence, local checkpoint-inhibitor delivery, tumor-targeting ligands, and protective coatings that scavenge adenosine, lactate, or reactive oxygen species in the tumor microenvironment (Table 5).^153–156

Table 5.

Cell Engineering Methods Compared by Efficiency, Safety, and Scalability, with an Author-Proposed Verdict for Each Platform

Method	Efficiency	Safety	Scalability	Verdict	Ref
Viral vectors	High, integrate and express transgenes very efficiently in T cells.	Medium, random genomic integration carries an insertional-mutagenesis risk.	Low, MP viral production is slow, costly, and capacity-constrained.	Keep for now: the proven, approved workhorse, but its cost and integration risk are the very problems non-viral methods aim to remove.	[157]
mRNA-LNPs	Medium, expression is strong but transient, since mRNA is not integrated.	High, no genomic integration and self-limiting charge mean a clean safety profile.	High, leverages the mature, vaccine-proven LNP supply chain.	Prioritize: the best balance of safety and manufacturability; transient expression is acceptable for many indications and enables in-vivo dosing.	[145,158]
Electroporation	High, physical pore formation forces cargo in very effectively.	Low, the electric pulse kills a large fraction of cells and stresses survivors.	Medium, works at clinical scale but throughput and cell loss limit it.	Avoid for sensitive cells: efficient but harsh acceptable for robust cells, poor for fragile primary NK or stem-like T cells.	[143,145]
Nanoparticle CRISPR RNPs	Medium, editing works but cytosolic and nuclear delivery efficiency is still being optimized.	Medium, transient RNP lowers off-target risk versus plasmids, but escape chemistry can be toxic.	Medium, formulation is reproducible but not yet standardized for primary cells.	Develop selectively: high-value for multiplex editing (eg, PD-1 knockout + CAR knock-in); back specific, well-characterised formulations rather than the category broadly.	[152,159]
Inorganic NP carriers	Low, poor endosomal escape leaves most cargo trapped and degraded.	High, chemically stable and generally well tolerated as materials.	Medium, synthesis scales, but accumulation limits repeat dosing.	Deprioritize: safe and stable but ineffective as stand-alone gene vectors; better confined to imaging or as an escape-assisted hybrid component.	[160,161]

Most importantly for the delivery gap, nanoparticles directly attack poor solid-tumor infiltration chemokine-releasing carriers (CXCL10, CCL5) build recruitment gradients; MMP-responsive and ECM-degrading carriers (collagenase, hyaluronidase) open dense stroma; deformability- and selectin-modulating particles ease extravasation; anti-VEGF carriers normalize vasculature; and TAM/CAF-reprogramming or STING/TLR-agonist carriers remodel the microenvironment, with magnetic guidance as a physical adjunct.^162–164 This trafficking toolkit is the most direct nanoparticle answer to why CAR-T fails in solid tumors and is the highest-value T-cell application to prioritize.^165–168

Natural Killer Cell Nanotechnology

NK cells offer two defining advantages HLA-independent target recognition and the potential for off-the-shelf allogeneic use.^169,170 Nanotechnology has begun to support the NK-cell therapy pipeline at multiple points enabling expansion through nano-scale artificial antigen-presenting cells and K562-membrane paramagnetic particles that can drive ~1000-fold expansion; genetic engineering via LNP-delivered CAR mRNA or CRISPR-mediated knockout of inhibitory receptors such as KIR and NKG2A; enhanced tumor trafficking using chemokine-loaded and ECM-modifying carriers; the development of acellular, NK-derived extracellular vesicles; and amplified antibody-dependent cellular cytotoxicity through CD16 engagers and Fc-optimized nanocarriers.^128,171,172 Yet a decisive problem overshadows this progress nearly all NK delivery data have been generated in the NK-92 cell line and do not transfer reliably to primary NK cells. In this regard, It is assumed that primary NK cells, owing to their distinct membrane lipid-raft composition, likely require smaller nanoparticles (below 100 nm) and a higher fusogenic-lipid content than those optimized for T cells. A systematic optimization of LNP composition specifically for primary NK cells an effort that, at present, does not exist constitutes the single most urgent NK-specific need. Whether allogeneic NK products ultimately complement or partially replace autologous CAR-T cells will hinge on solving this delivery problem.^129,173

Dendritic Cell Vaccines and Nano Formulations

Dendritic cells sit at the apex of the antigen-presentation hierarchy, and nanotechnology now touches every stage of their therapeutic use: antigen loading, maturation, and in vivo targeting.^174,175 Size-tuned carriers (20–200 nm) and pH-responsive formulations enhance cross-presentation on MHC class I; co-delivery of antigen with adjuvant in PLGA particles bearing TLR agonists improves potency; and an expanding toolkit of in vivo DC-targeting strategies antibodies against DEC-205 or DC-SIGN, mannose-based ligands, and albumin-hitchhiking designs that drain to lymph nodes now makes it possible to bypass the cumbersome step of ex vivo loading altogether. DC-derived exosomes further offer a cell-free vaccine modality (Figure 5d).^176–179 These advances converge on a single strategic question: will ex vivo loading or in vivo targeting dominate the next generation of dendritic cell vaccines? Ex vivo loading is expensive and scales poorly. It seems that in vivo DC-targeting nanoparticles anti-DEC-205 LNPs being a leading candidate will largely displace ex vivo approaches within approximately next five years. However, that transition depends entirely on whether cross-presentation efficiency can be raised from its current ceiling of roughly 10–20% to above 50%. That efficiency threshold, not the targeting chemistry itself, remains the rate-limiting milestone.

Macrophage and Microglia Engineering

Macrophages sit at the crossroads of homeostasis, wound healing, and pathogen clearance, and their innate capacity to infiltrate tissues makes them compelling therapeutic vehicles.^180,181 Their functional identity spans a spectrum from pro-inflammatory M1 to reparative M2.^182,183 Nanoparticles can direct this polarization through cargo choice IFN-γ, TLR7/8 agonists such as R848 in PLGA, or PI3Kγ inhibitors push macrophages toward an anti-tumor M1 state, while IL-4, IL-13, dexamethasone, or resolving D1 steer them toward a tissue-repair M2 phenotype.^184,185 Metabolic programming glycolysis in M1 versus oxidative phosphorylation in M2 offers a complementary lever for reinforcing these fates.^186–188 Beyond polarization, macrophages can be loaded with therapeutic cargo and deployed as “Trojan-horse” carriers that actively home to tumors, infarcts, and sites of infection^189,190 (Figure 5e). The toolkit has further expanded to include CAR-macrophages and macrophage-membrane-coated biomimetic nanoparticles that exploit endogenous signals such as CD47 and integrin–selectin interactions, opening avenues in solid tumor therapy, sepsis detoxification, and tissue repair^181,191 (Figure 5f). This versatility derives from a singular advantage. Macrophages infiltrate tissues that largely exclude T cells, positioning them as the best-suited cell type to close delivery gaps in solid tumors and regenerative medicine. Yet driving macrophages toward an M1 phenotype carries a specific and under-addressed hazard: uncontrolled inflammation and cytokine storm. Macrophage-based therapies therefore demand inducible safety switches engineered directly into the construct. Future designs should co-deliver a suicide gene iCasp9, for example, enabling rapid drug-triggered apoptosis or a depletion marker such as CD20 that permits antibody-mediated clearance should severe inflammation arise. Without such a fail-safe, potent macrophage repolarization should not advance to the clinic.

Regulatory T Cells and Immune Tolerance

Regulatory T cells are the main cells that keep the immune system from attacking the body’s own tissues, and researchers have explored using them to treat autoimmune diseases, prevent transplant rejection, and control graft-versus-host disease.^192,193 The central therapeutic hurdles are Foxp3 instability and the challenge of generating sufficient cell numbers. Nanoparticle-based strategies address these from multiple angles sustained release of IL-2 and TGF-β, delivery of HDAC inhibitors and retinoic acid, and targeted demethylation of the TSDR locus all act to lock in a stable suppressive phenotype.^194,195 Tolerogenic nano-scale artificial antigen-presenting cells, signaling through PD-L1 and IDO, generate antigen-specific Tregs with far greater precision.^196,197 Metabolic carriers that reinforce the oxidative phosphorylation program sustaining Foxp3 offer a complementary layer of control,¹⁹⁸ while chemokine-releasing or tissue-targeted carriers localize Tregs to inflamed joints, transplanted organs, or pancreatic islets^199,200 (Figure 5g and h).

Yet these capabilities converge on a critical safety question the field can no longer defer: is systemic Treg therapy ever truly safe, given the inherent risk of immunosuppression and opportunistic infection? In our assessment, the safety profile of polyclonal Treg therapy for systemic autoimmune disease may not justify continued development relative to antigen-specific approaches. We recognize that polyclonal Treg therapies are still being actively evaluated in clinical trials and that some may prove safe and effective in defined settings;¹⁹³ our view reflects a weighing of the current risk–benefit balance rather than a definitive verdict. On balance, we consider antigen-specific Tregs whether generated by nanoparticle-delivered CAR or TCR constructs, or expanded on peptide-MHC nano-aAPCs to possess the more favorable therapeutic window, and we propose this antigen-specific pathway as a promising default roadmap for the field.

Conclusion

Two decades of nanoparticle-based immune cell engineering have optimized the wrong variable. This review has argued, mechanism by mechanism, that cellular uptake was never the bottleneck the field treated it as professional phagocytes swallow nanoparticles readily, and even reluctant uptake in T cells and NK cells can now be forced open through ionizable lipids, cell-penetrating peptides, and receptor-directed targeting. What none of these advances solve is what happens in the five to fifteen minutes after internalization: more than 95% of cargo that enters an endosome never reaches the cytosol, lost to lysosomal degradation, recycling-mediated exocytosis, or intact-carrier expulsion before escape can occur. Endosomal escape, not uptake, is the rate-limiting step, and every design decision in this field size, charge, lipid phase behavior, targeting ligand should be evaluated first against how it shifts that single number.

This reframing has concrete consequences for how the field should allocate effort over the next decade. First, cell-type specificity is not a caveat to be mentioned and set aside; it is the design constraint. A carrier optimized in an immortalized cell line routinely overstates the escape efficiency achievable in primary T cells, NK cells, or macrophages, because the toxicity of proton-sponge and membrane-destabilizing mechanisms is systematically amplified in primary cells and because immune cells’ slow, incomplete endosomal acidification changes the timing of every pH-triggered event. Reporting an escape percentage without a matched viability measurement, and without testing in primary cells, should no longer be considered sufficient evidence of translational readiness.

Second, the translational gap between hematologic and solid-tumor success is not primarily a cell-engineering problem; it is a delivery and trafficking problem. CAR-T’s remissions in circulating and marrow-resident malignancies reflect accessibility, not superior cell design. Closing the solid-tumor gap will depend less on new payloads and more on chemokine-releasing, ECM-degrading, and vasculature-normalizing carriers that solve infiltration directly this is the single highest-value nanoparticle application identified in this review, ahead of further incremental gains in transfection efficiency.

Third, several currently active lines of investigation should be deprioritized or redirected rather than pursued as stated. Quantum dots and graphene oxide carry heavy-metal and oxidative toxicity that outweighs any delivery advantage and should be abandoned as therapeutic gene vectors. Polyclonal Treg therapy for systemic autoimmune indications carries an immunosuppression risk that a delivery improvement cannot offset; only antigen-specific Treg engineering has a viable therapeutic window. Macrophage repolarization toward M1 phenotypes should not advance to clinical testing without a built-in safety switch (eg, inducible caspase-9 or an antibody-clearable surface marker), given the risk of uncontrolled cytokine storm in a cell type this potent.

Fourth, the field’s blind spot-on immunogenicity needs to close. Cationic carriers that inadvertently engage TLR4 and the NLRP3 inflammasome, and the newly recognized fact that endosomal membrane damage itself is sensed as a danger signal, mean that escape efficiency and escape tolerability are separate optimization targets in immune cells in a way they are not in a hepatocyte. We predict that the first broadly successful clinical nano-immunotherapy platform will be a deliberately TLR-silent carrier engineered with the same rigor applied to cargo protection not the cationic, immunostimulatory chemistries that currently dominate the literature. This is a falsifiable claim: within ten years, either such a platform reaches late-stage trials, or the field’s continued reliance on cationic escape mechanisms will have been vindicated by a solution to the inflammation problem we have not yet identified.

Standardizing endosomal escape measurement and solving reproducible, cell-type-tuned manufacturing are the two highest-impact, most tractable priorities available to the field today. Neither requires a new material class or a conceptual breakthrough both require the community to agree on what to measure, in which cells, and against what threshold. That agreement, more than any single carrier innovation, is what stands between the current state of the art and a nanoparticle platform that performs as promised across the immune cell types this review has surveyed.

Funding Statement

This work received funding from the Iran National Science Foundation under grant code 4043509.

Data Sharing Statement

No new datasets were generated or analyzed in this review. All data discussed are derived from previously published studies cited in the reference list.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors report no conflicts of interest in this work.