Most of the mechanism explainers on this site have described what happens after a peptide meets its target. The melanocortin system, the GH secretagogue axis, the HPG axis, the SNARE complex — in each case the target sits on the outside of a cell, or in a compartment the ligand reaches by simply arriving in the extracellular fluid. Binding is the event, and everything downstream follows from it.
But a meaningful fraction of this catalog does not work that way. FOXO4-DRI disrupts a protein-protein interaction in the nucleus. MOTS-c's best-characterised binding partner is a cytosolic kinase. TB-500's parent protein buffers monomeric actin inside the cell. KPV inhibits an intracellular signalling cascade. Adipotide's payload is inert until it is internalised. For every one of these, there is a question that never comes up for a GLP-1 receptor agonist: how does the molecule get to where the target is?
This is not a footnote. It is a second, independent hurdle, and it fails far more often than binding does.
The Membrane Is Not a Sieve, and Peptides Are Badly Suited to Crossing It
The plasma membrane is a lipid bilayer with a hydrophobic core. Passive diffusion across it favours molecules that are small, uncharged and reasonably lipophilic. Peptides are typically none of those things. A 15-mer carries multiple ionisable groups, a polar backbone with an amide bond every three atoms, and a hydrated shell that has to be stripped before anything can enter a lipid phase.
Worse, the outer face of a mammalian cell is anionic — sialylated glycoproteins and heparan sulfate proteoglycans give it a net negative surface charge. Cationic peptides stick to it avidly. That is the first thing that happens to an arginine- or lysine-rich sequence in a culture well, and it is emphatically not the same thing as entering the cell. Distinguishing surface-bound from internalised material turns out to be the whole methodological problem in this field.
The same physical argument appeared in the cosmetic class primer about the stratum corneum, and it generalises: for a hydrophilic, charged molecule, size is rarely the obstacle — polarity and charge are.
Route One: A Transporter Built for the Job
The cleanest way in is a carrier protein that already moves molecules of that shape.
KPV is the shelf's best example. Dalmasso and colleagues (Gastroenterology 2008;134:166-178) showed that the tripeptide Lys-Pro-Val is taken up by PepT1, the di/tripeptide transporter of the intestinal epithelium, which is induced in colonic tissue during inflammation. They demonstrated this properly — cold KPV competed with a radiolabelled PepT1 substrate, and tritiated KPV gave measurable uptake kinetics. Inside, KPV inhibited NF-κB and MAP kinase signalling at nanomolar assay concentrations, and the effect did not require melanocortin receptor signalling, despite KPV being the C-terminal fragment of α-MSH.
Note what this route buys and what it costs. It is saturable, characterised, and tissue-specific. It is also structurally unavailable to almost everything else in the catalog, because PepT1 handles di- and tripeptides. A tetrapeptide is already too long. This is not a general solution; it is a narrow architectural coincidence that KPV happens to fit.
Route Two: Endocytosis, and the Bottleneck Nobody Advertises
The dominant route for cationic and amphipathic peptides is endocytosis — the cell engulfs a patch of membrane with the peptide adsorbed to it. Clathrin-mediated endocytosis, caveolae-dependent uptake and macropinocytosis have all been implicated, and which one dominates varies with cell type, peptide and concentration.
Here is the problem: being endocytosed is not the same as being inside the cell in any useful sense. The peptide is now in a membrane-bounded vesicle — topologically still outside the cytosol — on a conveyor belt toward the lysosome, where acidification and cathepsins await. To reach a cytosolic or nuclear target it must escape the endosome first.
Endosomal escape is spectacularly inefficient. The most-cited quantification comes from a different modality but sets the scale: Gilleron and colleagues (Nature Biotechnology 2013;31:638-646) tracked gold-labelled siRNA delivered by lipid nanoparticles and estimated that escape into the cytosol occurred at 1-2% efficiency, and only during a narrow window when the particle occupied a compartment with mixed early- and late-endosomal character. Everything else was degraded or recycled.
That figure is about lipid nanoparticles carrying nucleic acid, not about peptides — but it explains why the entire delivery literature, including the 2026 work on peptide-mediated siRNA platforms and engineered penetrating scaffolds, is organised around escape rather than uptake. Uptake was never the hard part.
The practical consequence: a fluorescence image showing peptide "inside the cell" may be showing peptide inside endosomes, which is compatible with zero target engagement.
Route Three: Direct Translocation, and Why It Is Contested
Some peptides do appear to cross the bilayer directly, without a vesicle — probably via transient pore-like or inverted-micelle intermediates, driven by the membrane potential and by charge neutralisation of the cationic side chains against membrane counter-anions. Direct translocation tends to become detectable at higher peptide concentrations and is generally reported alongside, not instead of, endocytosis.
Membrane-active peptides blur this boundary further. LL-37 is not designed to deliver anything; its cationic amphipathic helix forms on contact with an anionic membrane and disrupts it. Membrane permeabilisation and membrane crossing sit on the same physical continuum, which is one reason cell-penetrating and antimicrobial peptide sequences overlap so heavily — and one reason "it gets in" and "it damages the membrane" can be difficult to separate experimentally.
A 2025 review by Vedekhina and colleagues (International Journal of Molecular Sciences 2025;26(22):11015) surveys monomeric penetrating peptides alongside peptide coacervates and fibrils, and its conclusion is worth quoting in spirit: internalisation mechanisms still require further verification and detailed characterisation before rational design is possible. That is the state of the art in 2026 — not a solved pathway.
Route Four: Targeted Internalisation, and Route Five: Accumulation Gradients
Two catalog compounds solve the problem in ways that are neither of the above.
Adipotide is a chimera: a homing domain that binds prohibitin on white-adipose vasculature, fused to a proapoptotic D-amino-acid payload. The homing domain does not just find the tissue — receptor engagement is what drives internalisation, and the payload is only cytotoxic once inside, where it can reach mitochondrial membranes. Entry is a designed function of the targeting element, which is a genuinely different architecture from a generic cationic tag.
SS-31 is stranger still. Its alternating cationic/aromatic tetrapeptide motif produces large accumulation at the inner mitochondrial membrane, where it associates with cardiolipin. This was, as covered in the Dermorphin deep-dive, a serendipitous observation made while Schiller and Szeto were studying membrane crossing in an opioid-peptide series. The compound does not merely enter cells; it concentrates in a specific organelle, and that physicochemical partitioning is the mechanism.
The Two Artifacts That Reset the Field
This is the part of the story most worth carrying away, because it is a methodology lesson, not a pharmacology one.
Through the 1990s, cell-penetrating peptides were widely reported to enter cells by an energy-independent process — uptake was still visible at low temperature, still visible with metabolic inhibitors, and apparently faster than any known endocytic pathway. That observation shaped a decade of thinking.
It was wrong, and it was wrong for a sample-preparation reason. Richard and colleagues (Journal of Biological Chemistry 2003;278:585-590) re-examined Tat(48-60) and nona-arginine and showed that cell fixation — even under mild conditions — artifactually redistributes these peptides into the cell. They also showed that flow cytometry over-reads uptake unless a trypsin digestion step is included to strip membrane-adsorbed peptide from the surface. On live, unfixed cells the peptides showed characteristic endosomal distribution, uptake kinetics matching endocytosis, and inhibition by low temperature and ATP depletion.
The second artifact is more specific and, for this catalog, closer to home. Szeto, Schiller and colleagues (FASEB Journal 2005;19:118-120) took a single cell-penetrating tetrapeptide from the [Dmt¹]DALDA series and labelled it two different ways. The two fluorescent analogues gave two different intracellular localisations — one mitochondrial, one diffusely cytoplasmic — and further work indicated the labels were not passive reporters but were altering where the peptide went and what it did. On a tetrapeptide, a fluorophore is a large fraction of the molecule.
Put together: for a decade the field's headline mechanistic claim was an artifact of fixation, and the standard tool for locating a small peptide inside a cell can change the answer it reports. Neither finding says cell entry does not happen. Both say the evidence that a given peptide reaches a given compartment needs to be examined, not assumed — and that is a fair summary of where several intracellular-target compounds on this shelf stand.
What This Means for Reading the Literature
Three transfers, consistent with the evidence hierarchy framework:
A demonstrated intracellular interaction is a rung-1 finding about biochemistry, not a claim about access. Surface plasmon resonance showing MOTS-c binds CK2, or co-immunoprecipitation showing FOXO4-DRI disrupts a FOXO4-p53 complex, establishes that the interaction is possible. It says nothing about how much peptide reaches that compartment in an intact organism — and for MOTS-c, no entry route has been established at all.
Adding peptide to a culture well is not a delivery experiment. In a dish, the compound sits at a fixed concentration against the outer membrane indefinitely, with no clearance, no barriers and no distribution. Exogenous thymosin β4 is instructive here: Ho and colleagues (Investigative Ophthalmology & Visual Science 2007;48:27-33) found that internalisation was required for its antiapoptotic effect in corneal epithelial cells and that blocking actin dynamics with cytochalasin D prevented entry — an entry route that is itself a cell-state-dependent variable.
Cell entry is not a documentation question. A certificate of analysis reports identity, purity, mass accounting and, sometimes, endotoxin. There is no line on any COA — and no HPLC or mass spectrometry method — that reports whether a molecule reaches the cytosol. That is a biological property measured in an experiment, not a material property measured in a vial. See /quality/ for what a COA can and cannot establish.
FAQ
Does a cell-penetrating tag guarantee delivery? No. A cationic penetrating segment reliably increases uptake — association with, and endocytosis by, the cell. Whether cargo reaches the cytosol depends on endosomal escape, which is the inefficient step. "Cell-permeable" describes an intended design property, not a measured delivery fraction.
Why do some peptides not need any of this? Because their targets face outward. A GHS-R1a or MC4R agonist engages an extracellular binding pocket on a receptor embedded in the membrane; signal transduction crosses the membrane, the ligand does not. For those compounds, receptor pharmacology vocabulary applies cleanly — see the receptor pharmacology primer. For intracellular-target compounds, it does not apply at all, and access becomes a variable in its own right.
Does this change how a compound's mechanism should be described? It adds a required qualifier. For compounds with no established receptor and an intracellular proposed target, the honest formulation is that an interaction has been demonstrated in a system where access was not a limiting factor. Whether the same interaction occurs at meaningful levels in a whole organism is a separate, and generally open, question.
Browse the full compound library for mechanism summaries and documentation notes on individual research compounds.
This article is educational and for the laboratory research community. Trulogic Labs products are sold for laboratory and research use only and are not for human consumption.