Peptide-mediated gene delivery to the CNS is one method of non-invasive, intravenous (IV) administration, which can leverage receptor-mediated transcytosis to transport DNA, mRNA, or CRISPR RNPs across the blood–brain barrier. Their small size (<5 kDa) and amenability to solid-phase chemistry enables rapid iteration and optimisation to various receptors on the brain endothelium (TfR, LRP1, nAChR). However, such peptides must find the right balance of affinity vs release, stability vs flexibility, and BBB transcytosis vs parenchymal distribution in order to effectively mediate their cargo to the target cells. If successful, these peptides can deliver and concentrate their gene cargo to neurons or glioma cells with little off-organ targeting; but if not, they are often sequestered in perivascular cuffs and produce strong imaging signals but little to no therapeutic effect.
Delivery to the CNS is an added challenge since >98% of systemically administered macromolecules are rejected by the blood–brain barrier (BBB). In addition, once the payload crosses the vessel wall, diffusion is limited by the brain's high interstitial pressure and dense extracellular matrix. In this context, potency in vitro is of little relevance if the gene delivery construct cannot (i) engage a luminal transporter, (ii) survive transcytosis, (iii) escape from endosomes, and (iv) diffuse far enough to reach dispersed neuronal populations. Peptide design thus becomes the central engineering lever because the same sequence must mediate receptor binding, membrane translocation, payload release, and tissue penetration within a single, renal-clearable chain.
Fig. 1 Key physicochemical parameters influencing nanoparticle transport across the BBB. 1,5
Tight junctions, efflux pumps (P-gp), and low transcytotic activity preclude paracellular and vesicular entry. Peptides must instead bind luminal receptors (TfR, LRP1, IR) but avoid recycling back to blood or expulsion by efflux transporters. Receptor occupancy must not be so high as to risk saturating it, given that endothelial receptor pools are many orders of magnitude smaller than hepatocyte ASGPR pools. Abluminal, the payload encounters a dense mesh of hyaluronan, chondroitin sulfate and astrocytic end-feet that limits convection. Peptides<5 nm can diffuse several millimeters, but larger RNAs or protein cargos need secondary neuronal targeting motifs or protease-cleavable spacers that reduce hydrodynamic radius post-crossing.
In contrast to AAVs, LNPs and polyplexes often show high gene editing or expression potency in vitro. However, upon i.v. injection, they usually accumulate in the liver or spleen. The CNS receives<1 % of the injected dose under these conditions. To date, developers have been forced to choose between toxic systemic exposure and invasive intracerebral injection. Peptide conjugates avoid this tradeoff by binding to BBB transporters at nanomolar concentrations, trapping the payload in brain capillaries before it can be cleared by peripheral pathways. Focused ultrasound or osmotic mannitol can also increase paracellular permeability but they are non-selective, admitting toxins and immune cells along with the therapeutic that can cause neuroinflammation. Peptide-based strategies are much more selective but they also have a narrow safety margin: overstimulation of receptors can down-regulate the transporters themselves that the brain depends on for homeostasis. For this reason, developers are willing to accept partial brain distribution in exchange for preserving the barrier and they iterate the peptide dose and cleavage kinetics to achieve the highest parenchymal gene expression possible without chronic barrier disruption.
Table 1 Barrier vs Peptide Design Response for CNS Gene Therapy
| Barrier Feature | Peptide-Level Countermeasure | Translational Risk |
| Tight junctions | Receptor-mediated transcytosis | Receptor saturation |
| Efflux pumps | Cleavable linker post-transit | Premature plasma cleavage |
| Dense ECM | <5 nm radius + cleavable spacer | Reduced neuronal retention |
| Low transcytosis | Moderate KD + bivalent avidity | Peripheral sink effect |
Peptides are versatile multifunctional nanoscale interfaces that combine targeting, trafficking, and packaging into a single tunable string. They do not perform a single function but rather cycle through vascular binding, transcytosis, and post-BBB retention to transport DNA, mRNA, or CRISPR RNPs safely across the vasculature and BBB and deliver them to target cells in an intact form.
Brain-capillary endothelium constitutively cycles transferrin, insulin and LDL receptors; peptides that present miniature consensus loops outside the native ligand pocket hijack these pathways without competing against serum proteins. Head-to-tail cyclisation or D-amino-acid incorporation protect the epitope from luminal aminopeptidases during the 8–12 s micro-vascular transit time, ensuring that the ligand remains intact until clathrin recruitment occurs. Once engaged, the peptide–cargo complex is ferried across the cytosol inside recycling endosomes; inclusion of three to four histidines lowers the pKa of the vector, inducing vesicle leakage before lysosomal fusion. Because the same peptide can be lipidated or thiolated without altering the binding loop, one sequence can be ported across lipidoid, polymeric or inorganic carriers, simplifying scale-up while preserving modular adjustability for DNA, siRNA or mRNA cargos.
In addition to a perivascular epitope, peptides after abluminal release can present a second, cryptic epitope that may be a modified RGD or IKVAV motif. The alternative epitope can be masked by an endosomal spacer and tethered to integrins that are more enriched on the desired target cell (such as neurons, or NG2-expressing oligodendrocyte precursors). Cleavage of the spacer then unmasks this motif and delivers the payload away from perivascular astrocytes and towards the desired parenchymal cell type. The varying receptor densities in cortex, hippocampus and striatum allows moderate-affinity ligands to exploit these natural gradients to enable region-biased gene expression, without stereotactic injection. Pediatric gliomas or diffuse white-matter diseases can down-regulate the transferrin receptor; targeting entry through this low-density target requires ultra-high affinity and causes endothelial trapping. Moderate-affinity peptides (with Kd of 50–200 nM) can still sufficiently engage these residual receptors while still detaching before being trapped, allowing distribution throughout the brain and more homogeneous transgene expression between healthy and diseased tissue.
Arginine-rich helices condense onto siRNA or mRNA, neutralizing phosphate charges and forming 70–120 nm particles. These particles are stable to RNase A and physical shear forces during intra-carotid infusion. Disulfide cyclisation of the peptide backbone precludes unfolding in serum. A short PEG tail grafted onto the C-terminus affords steric stabilization against salt-induced aggregation, resulting in a nanocomplex that is small enough to avoid renal filtration. Electrostatic complexation is readily reversed by lowering pH or introducing cytosolic glutathione which reduces disulfide bridges, releasing naked nucleic acids within minutes of entry into the cytosol. Because the same peptide can be tuned to condense DNA at pH 7.4 and trigger its release at pH 6.0, this vector acts as a molecular spring, packaging the payload during circulation but unloading it precisely where histidine protonation begins, providing temporal coordination between protection and on-demand release.
Table 2 Functional roles of peptides in CNS gene delivery
| Functional role | Peptide design element | Outcome |
| BBB crossing | TfR/LRP1 mimic loop | Trans-endothelial transport |
| Cell redirect | Masked RGD/IKVAV | Neuron vs glia selectivity |
| Cargo protection | Arginine-rich helix | RNase resistance |
| Endosomal escape | Histidine cluster | pH-triggered release |
| Serum stability | D-cysteine, cyclisation | Protease evasion |
The optimization of a peptide-based brain vector is a complex challenge, because several parameters have to be optimized in parallel. A certain affinity and avidity for the luminal endothelial receptor of the peptide is required to initiate transcytosis. However, if the peptide binds with too high affinity, it might not dissociate fast enough on the abluminal side, which could cause vascular trapping. At the same time, the peptide should be small enough to fit through the 0.8–1.5 nm cleft between endothelial cells. It should also be sufficiently stable to withstand serum and brain-capillary proteases, but flexible enough to undergo conformational changes for receptor clustering and endosomal escape. Since the same molecule later on functions as a cell-specific anchor and as a protective shield of the payload, all the amino-acid choices influence the entire delivery process. This makes the optimization process iterative instead of based on single parameters.
Ultra-high-affinity peptides (sub-nanomolar Kd) enter the slow endocytic pathway, with long-lived receptor complexes that are sorted to late endosomes and removed from the transport pool. Reducing affinity by truncating the binding loop or inserting a single glycine disruptor accelerates dissociation, tipping the equilibrium toward abluminal release, without compromising luminal capture under shear stress. A convenient tradeoff is to design bivalent architectures that present two moderate-affinity ligands in close apposition; avidity increases the functional affinity during the brief luminal transit time, while each individual epitope can still disengage once the vesicle acidifies, restoring transport efficiency. The peptide must also contain a second motif, either a masked cell-penetrating sequence or a reducible disulfide, which is only exposed following abluminal release. This time switch repurposes the peptide from BBB anchor to brain-cell director, reducing the risk that the therapeutic payload gets sequestered in the vessel wall instead of reaching the target neuropil. Because the same peptide can be lipidated or thiolated without modifying the binding loop, one sequence can be translated across lipidoid, polymeric or inorganic carriers, which simplifies scale-up but maintains the modular adjustability of DNA, siRNA or mRNA cargos.
The pores of most receptor-mediated transcytosis have an upper size limit of ～100 nm. Peptides longer than 30 residues or peptides with a rigid helical bundle fold exceed the physical size limit and are diverted to degradative pathways. Constraining the sequence to 12–24 amino acids but folding the peptide into a more compact shape with a disulfide or thioether staple can preserve receptor binding without sterically inhibiting vesicle budding. The constraint also limits conformational entropy, pre-organizing the binding interface and reducing the entropic penalty upon receptor binding to enhance transcytosis without adding molecular weight. Rigid macrocycles are resistant to endosomal proteases, but their increased bulk becomes a limiting factor for transcytosis if they are too large to enter the narrow neck of clathrin coated vesicles, thereby slowing vesicle budding and reducing the flux. By contrast, highly flexible linear peptides enter into vesicles with ease but are rapidly degraded prior to exocytosis. An intermediate between these two states is a constrained β-hairpin or α-helix that is sufficiently small to enter vesicles but sufficiently pre-organized to be protected from luminal peptidases, with the kinetics of vesicular transport matching those of stability. As the peptide must be soluble in plasma and still insert into endosomal membranes, the architecture of these peptides requires a balance between hydrophilic receptor recognition and a short hydrophobic segment, which is an ambivalent design that replaces both the antibody and the lipidoid in a single metabolically degradable chain.
Brain-capillary aminopeptidases and serum proteases are known to cleave linear peptides within minutes. Backbone N-methylation, thioether staples, or D-amino acid substitutions provide protease-resistant scaffolds that are stable to survive the systemic circulation but which can be cleaved within neurons by reductive conditions for safety post-delivery. The peptide should remain intact long enough to have an opportunity to bind to receptors throughout the entire cerebral vascular bed, but not so long that it accumulates in non-pathological brain regions. Achieving this goal involves a combination of a high affinity binding domain with a cleavable, hydrophilic C-terminal tag that will speed renal clearance once the cargo has been released, providing an automatic off-switch to limit off-target exposure.
The major strategy to convert systemically administered gene therapeutics into brain-penetrating, cell-specific therapeutics is to co-opt the vector with peptide-based architectures that incorporate targeting, condensation and release functions in a single synthetic entity. The modular chemical design of these architectures allows for three main delivery approaches: (i) chemical conjugation to pre-existing viral or non-viral vectors, (ii) surface modification of nanocarriers, and (iii) peptide–gene self-assembly. Each strategy presents a different balance of ease of production, targeting ability and cargo load. However, the common driving principle underlying all these vectors is receptor-mediated transcytosis across the BBB, followed by conditional release of the cargo within target neurons or glia.
Peptides are either genetically fused to adeno-associated capsids at the VP3 loop or chemically conjugated to cationic lipidoids via thiol-maleimide linkages. The peptide serves as the homing signal while the vector provides the endosomal escape and nuclear import machineries. It represents a merger of the targeting specificity of a ligand with the transduction capability of a viral particle. Since the peptide is displayed on the surface, receptor engagement takes place before the capsid is routed to lysosomes, increasing the chances of productive transduction. Each additional targeting peptide adds another epitope that may be cleaved by serum proteases or recognized by pre-existing antibodies, increasing immunogenicity and lowering batch-to-batch reproducibility. Designers therefore restrict the surface density to one to two copies per capsid or use cleavable spacers that detach after the BBB has been crossed, thereby restoring the native capsid surface for the downstream infection while maintaining the initial targeting specificity.
Lipid nanoparticles are functionalised by post-insertion of cysteine-terminated peptides into pre-formed vesicles, whereas polymeric micelles utilise copper-free click chemistry to azide-modified coronas. Multi-dentate peptides bearing two or more thiols create irreversible anchors that resist ligand exchange in blood, ensuring that the targeting motif remains surface-exposed after nebulisation or long-term storage. Because the peptide itself can be PEGylated at a distal lysine, steric hindrance is minimised, preserving both colloidal stability and receptor accessibility. Peptide-guided nanocarriers<80 nm penetrate the narrow brain extracellular space, whereas larger or overly cationic constructs adhere to perivascular astrocyte end-feet, creating a sink that limits onward diffusion. Neutral zeta potential combined with a moderate hydrophobic moment (0.4–0.6) allows particles to travel several millimetres from the vessel wall, achieving hemisphere-scale distribution when convection-enhanced delivery or focused ultrasound is applied.
Amphiphilic peptides with both cationic and hydrophobic blocks self-condense siRNA or mRNA into 40–100 nm micelles without additional carriers. Since the same sequence can be synthesized on a solid-phase synthesizer, iterative alanine scanning or D-amino-acid substitution cycles can be completed within days to rapidly optimize for brain versus spinal cord tropism. The resulting one-component nanomedicine significantly simplifies the regulatory chemistry-manufacturing-controls package and eliminates the need for excipients that might trigger complement activation. Direct condensation is limited to nucleic acids<10 kDa as larger plasmids or CRISPR ribonucleoproteins exceed the charge-neutralization capacity of a 20-mer peptide and require helper lipids or polymers. Moreover, the high peptide-to-phosphate ratios needed to condense large genes will invert the zeta potential to strongly cationic values, promoting non-specific uptake by liver and spleen. Designers therefore insert cleavable spacers that halve the cationic charge after endosomal escape, restoring a near-neutral particle that can diffuse through brain tissue without creating a perivascular sink.
However, there is a trade-off with peptide-mediated gene delivery to the CNS. The more serum-resistant, the higher affinity, and the larger and more stable a gene delivery construct is, the more likely it is to become sequestered in endothelial cells, more susceptible to peripheral sequestration, and less likely to penetrate into deep brain regions. It is critical that the BBB-crossing peptide/drug also releases its payload upon entering neurons, so overall performance is optimized at less than maximum efficiency for each criterion.
Fig. 2 Challenges in designing peptide-decorated nanocarriers: navigating the maze.2,5
Plasmids >5 kbp, or Cas9-mRNA complexes, are too large to fit through the physical limit of receptor-mediated transcytosis pores (<100 nm) and are rerouted to degradative compartments even when the peptide crosses efficiently. On the other hand, siRNA or antisense oligos that are <10 kDa can fit nicely into 70–90 nm peptide micelles but have such a low information density that their therapeutic index is limited by the need for repeated dosing. Therefore, gene designers fragment larger genes into two or more overlapping AAV genomes or split-Cas9 systems to keep the nanoparticles compact at the cost of co-delivery complexity. Increasing the N/P ratio (peptide N to RNA P) will compact a larger payload, but it inverts the surface charge from weakly cationic to strongly cationic, favoring opsonization and lung trapping. Charge reversal by anionic glutamate residues restores circulation time, but it weakens endosomal escape, so the practical compromise has been to install a cleavable PEG layer that detaches after BBB crossing and endosomal re-acquires fusogenicity.
A major reason for the cationic peptides' disappointing performance is that they bind avidly to anionic glycosaminoglycans of the normal endothelium and renal tubules to form a peripheral sink that depletes the plasma pool before it can reach the capillaries of the brain. Moreover, even when peptides are designed to be neutral at pH 7.4, they can reveal cationic patches again when cleaved by proteases, subjecting the now unprotected vectors to macrophage scavenging and further sequestration from BBB interaction. The sink effect is further exacerbated by saturation of the relevant receptors on the BBB: once the available surface TfR or LRP1 are saturated, more vector is essentially wasted in circulation, but increasing the dose merely increases the peripheral compartments. The overall result is a steep therapeutic window where brain exposure plateaus while systemic exposure (and hence potential for complement activation and acute hypersensitivity reactions) continues to rise.
In many peptide–gene conjugates that are successful in transcytosis across the endothelial monolayer, the products can become entrapped in the perivascular space through interactions with heparan-sulfate rich basement membranes or by endocytosis into astrocyte end-feet. This "peri-vascular sink" can then prevent diffusion away from these anatomical regions towards neurons or oligodendrocytes, so that although the total brain homogenate may be positive, the DNA/RNA is non-functional with respect to its therapeutic target. And even when delivery vectors have access to the parenchyma, they can become sequestered in late endosomes within the target cell itself if the peptide release module is insufficient or if the payload is cleaved too early from the targeting epitope. The result is histological evidence for brain entry with no measurable gene expression. In other words, passage across the BBB is only one of a series of "relay races" which occur intracellularly.
Peptide-mediated delivery to the CNS is most useful when the disease target is enriched in a well-characterized, rapidly endocytosed receptor (TfR, LRP1, IGF1R) and the gene payload is small enough to be condensed into or covalently tethered to a vector without surpassing the ～5 nm cutoff for renal filtration. If these criteria are met, a single 20-mer peptide can be used as a "cross-then-home" vector: binding to endothelial receptors to initiate transcytosis, then re-binding to the same receptor on neurons or glia to concentrate the mRNA/siRNA at the site of disease. This modular chemistry is especially attractive for repeat-dosing regimens (e.g. chronic siRNA for Huntington's disease) where viral immunity or LNP accumulation would be a concern. Suitability therefore depends on receptor abundance >10 000 copies/cell, payload molecular weight<30 kDa, and a therapeutic index that is broadened when the gene is selectively expressed only in receptor-positive cells.
The best targets recycle fast (TfR, LRP1) and are co-expressed in both luminal endothelium and target parenchymal cells to enable sequential targeting with one peptide. Bivalent peptides carrying angiopep-2 (BBB) and RGD (neuronal integrin) transduce neurons 5-fold better than angiopep-2 alone while pH-cleavable linkers unveil the neuronal ligand only after transcytosis and so avoid on-target off-tissue expression in capillaries. Target receptor density has been confirmed by quantitative immunohistochemistry in all relevant brain regions and when below 5 000 copies per cell, the dendrimeric peptides must be avidity-enhanced. At higher densities monovalent constructs are tolerated and so synthetic complexity can be ratcheted up or down depending on the biological challenge. siRNA (21 mer), antisense oligos (20 mer) and short, compact mRNA (<2 kb) are ideal because they can all be condensed into <40 nm particles that are stable in the endosomal acidification step but disassemble in the reducing cytosol. Self-amplifying RNA (9–12 kb) or CRISPR RNP complexes fall outside this sweet spot and so developers of these approaches have been forced into peptide-lipid hybrid carriers that retain entry through the BBB but lose the renal-clearance advantage. Gene modalities that work in the cytosol (siRNA, CRISPR guides) are further favored over payloads that have to get to the nucleus (AAV genomes) because nuclear pores are limited in mature neurons so endosomal escape is rate-limiting instead of nuclear import.
siRNA (21 mer), antisense oligos (20 mer), and compact mRNA (<2 kb) are preferred as these are small enough to be condensed into <40 nm particles that can withstand endosomal acidification but will dissociate in the reducing cytosol. Self-amplifying RNA (9–12 kb) or CRISPR RNP complexes are larger than this window, requiring developers to use peptide-lipid hybrid carriers that are still capable of entering the BBB but lose the renal-clearance advantage. Gene modalities that are active in the cytosol (siRNA, CRISPR guides) are also favored over gene modalities that must enter the nucleus (AAV genomes) as the nuclear pores are limited in mature neurons, making endosomal escape the rate-limiting step instead of nuclear import. When the therapeutic gene is larger than 12 kb in size, or requires nuclear entry (e.g. AAV9 genomes), peptide condensation becomes inefficient and peptide-lipid hybrids reintroduce the liver-accumulation liability peptides were designed to obviate. In these cases, focused ultrasound or osmotic mannitol may be preferred despite the possibility of transient barrier disruption. On the other hand, for chronic, cell-type-specific gene modulation (e.g. allele-specific siRNA for Huntington's), the modular, repeat-dose nature of peptide conjugates may confer a safety and manufacturing advantage that compensates for the lower payload capacity.
CNS gene therapy presents one of the most demanding delivery challenges in drug development. Beyond crossing the blood-brain barrier (BBB), successful delivery requires precise control over transport pathways, payload integrity, and cell-type specificity within a highly heterogeneous brain environment. Many peptide-based approaches fail because they address only one of these dimensions in isolation. Our targeting peptide services support CNS gene therapy programs through CNS-oriented design and feasibility-first evaluation, focusing on whether peptide systems can enable functional gene delivery, not just CNS exposure.
BBB-Aware Sequence Engineering: For CNS gene therapies, peptide sequences must be compatible with BBB transport mechanisms rather than optimized solely for high-affinity binding. Our BBB-aware sequence engineering approach designs peptides to support receptor-mediated transcytosis while avoiding endothelial trapping or premature degradation. Sequence length, flexibility, and charge distribution are tuned to maintain stability during circulation and endothelial transit, while preserving the ability to disengage from BBB receptors once transport is initiated. This balance is critical for enabling forward transport of gene therapy payloads rather than accumulation at the barrier interface.
Cell-Type-Specific Targeting Strategies: Crossing the BBB does not guarantee effective gene delivery to relevant CNS cell populations. Our design strategy therefore extends to cell-type-specific targeting within the brain, addressing neurons, glial cells, or other defined CNS compartments as required by the therapeutic objective. Peptide targeting elements are selected and optimized based on differential receptor expression and internalization behavior within the CNS, helping guide gene payloads toward the intended cell types while minimizing off-target distribution. This layered targeting approach improves functional delivery efficiency and reduces dilution of gene expression across non-relevant tissues.
Target and Pathway Evaluation: Not all BBB transport receptors or CNS targets are suitable for peptide-mediated gene delivery. Our feasibility assessment begins with a target and pathway evaluation that examines receptor abundance, transport capacity, recycling behavior, and compatibility with gene therapy payload size and format. This evaluation helps determine whether a given pathway can realistically support meaningful gene delivery at practical doses, or whether biological constraints are likely to limit effectiveness regardless of peptide optimization.
Early Delivery Risk Identification: CNS gene delivery programs are particularly vulnerable to late-stage failure due to risks that can be identified early, such as low transcytosis efficiency, peripheral sequestration of peptide-gene complexes, or decoupling between BBB crossing and intracellular gene expression. We emphasize early delivery risk identification to flag these constraints before extensive optimization or scale-up efforts begin. By defining feasibility boundaries upfront, teams can refine design strategies, adjust expectations, or explore alternative delivery routes while development remains adaptable.
If your CNS gene therapy program shows limited brain distribution, inconsistent gene expression, or a disconnect between BBB crossing and functional delivery, an early technical discussion can help clarify whether peptide design or pathway selection is the limiting factor. Discuss your CNS gene delivery challenge with our scientists to evaluate feasibility, identify delivery bottlenecks, and define a peptide-based strategy aligned with your CNS gene therapy objectives.
References
