PerspectiveDrug Delivery

Tumor-Penetrating Peptides: A Shift from Magic Bullets to Magic Guns

See allHide authors and affiliations

Science Translational Medicine  02 Jun 2010:
Vol. 2, Issue 34, pp. 34ps26
DOI: 10.1126/scitranslmed.3001174


The targeted delivery of drugs and imaging agents to tumor vessels is an attractive strategy to enhance anticancer therapy and tumor detection, but such targeting does not mean efficient distribution into the tumor. Two consecutive papers, one in Cancer Cell and one in Science, report that a single peptide has the potential to selectively deliver a large variety of therapeutic agents and diagnostics to a tumor site and then to ensure their distribution deep in the tumor parenchyma. This peptide has the capacity to bind specific αV integrins through an arginine-glycine-aspartate motif and, after local proteolysis reveals a cryptic arginine/lysine-X-X-arginine/lysine motif, to bind the neuropilin-1 receptor and thereby increase tumor vascular permeability. Remarkably, this penetrating peptide works not only when it is conjugated to the payload, but also when it is coadministered with small molecules, nanoparticles, or monoclonal antibodies.


Tumor targeting is a long-standing goal of anticancer therapy and will remain a challenge for the next generation of anticancer modalities. Whereas treatment outcomes with conventional cytotoxic drugs may benefit from preferential delivery, mostly because of reduced systemic side effects, outcomes from so-called intelligent new anticancer drugs, designed to target some aspect of tumor biology specifically, may instead improve because of an enhancement of their efficacy through a more extensive penetration of the tumor. The search for ligands that could be used to address therapeutic drugs to tumors has logically pinpointed targets expressed at the surface of the endothelial cells that line tumor blood vessels, for two major reasons. First, the position of the endothelium at the interface between blood and tumor cells makes receptors and proteins expressed at the surface of endothelial cells obvious targets for systemically administered drugs. Second, the tumor vasculature results from neoangiogenesis, a biological process supporting a phenotype with obvious differences from the quiescent status of endothelial cells in most healthy organs in adults.

An interest in integrins naturally emerged from this double rationale. Integrins are heterodimeric membrane glycoproteins composed of noncovalently associated α and β subunits that activate signaling pathways by coclustering with kinases and scaffolds in focal adhesion complexes (1, 2). Some integrins are abundantly expressed at the surface of angiogenic endothelial cells and play active roles in different steps of the angiogenic process (3). The major impetus for targeting integrins, however, came from the identification of a three–amino-acid sequence—arginine-glycine-aspartate (RGD)—as a fundamental recognition site for cell attachment to different extracellular matrix (ECM) proteins, including fibronectin and vitronectin (4, 5). The identification of integrins as the receptors for these RGD-containing ECM proteins, as well as the discovery of integrin receptors on other, adjacent cell types, has strongly stimulated the field of tumor targeting of diagnostics and therapeutic strategies.


Two recent papers by Ruoslahti and collaborators, one in Cancer Cell (6) and one in Science (7), substantially extend this model by documenting that a peptide that combines the RGD sequence with a peptidic ligand for neuropilin-1 (NRP1), a transmembrane receptor, may not only give rise to selective tumor vascular targeting but can also facilitate the penetration of drugs into the tumor. These two papers provide some answers to the long-underestimated question: What is the fate of a drug selectively addressed to the tumor vasculature? The therapeutic efficacy in animal tumor models, as determined by tumor growth delay or at best by a reduction in tumor burden, is often used as the final proof of the targeting capacity of ligands coupled to therapeutic agents. Optimal clinical translation of targeting strategies, however, requires an understanding of whether the payload delivered through a tumor vascular targeting route acts locally as an antiangiogenic drug or penetrates into the tumor parenchyma (the tumor mass) to exert direct antitumor effects.

Ruoslahti and colleagues addressed this question by using a peptide containing the RGD sequence and the peptidic motif arginine/lysine-X-X-arginine/lysine (R/KXXR/K) (where X represents ambiguous residues), a sequence also named CendR because it is active only when occupying the C-terminal position in a peptide (Fig. 1). They previously documented that the CendR peptide binds NRP1 and that internalization of this receptor may stimulate cell penetration (8). NRP1 is a cell-surface (co-) receptor for both class 3 semaphorins (which function as chemorepellents during neuronal and vascular development) and vascular endothelial growth factor165 (VEGF165) (a key regulator of angiogenesis) (9). In neurons, endocytosis of NRP1 was reported to control growth cone collapse during axon guidance (10). In endothelial cells, VEGF165 competes with Sema3A for binding to NRP1, and consecutive internalization is associated with cell migration (11). Although the endocytic pathway of VEGF involves clathrin-coated pits and that of semaphorin is lipid raft–dependent, the intracellular trafficking of NRP1 is essential for both types of signaling (12). That NRP1 binding is associated with internalization is further documented by reports describing how viruses such as human T cell lymphotropic virus type 1 might enter into cells through a CendR motif (KPXR) (13) present in the viral capsid proteins (14).

Fig. 1. Multistep binding and internalization of iRGD and associated payload.

The iRGD cyclic peptide is either coadministered with or conjugated to (as shown here) a given drug. (A) Step 1: The RGD motif of the iRGD cyclic peptide interacts with αV integrin at the surface of endothelial cells lining tumor blood vessels; this interaction might also occur in tumor cells expressing αV integrins (Fig. 2B). (B) Step 2: A so-far-unidentified protease cleaves iRGD and exposes the cryptic CendR motif RGDK at the C-terminal end of the peptide. (C) Step 3: The presence of a CendR motif and the reduced affinity for αV integrin as a result of the loss of rigidity of the cleaved peptide favor the interaction with the NRP receptor NRP1. (D) Step 4: The peptide together with the drug, either associated through a disulfide bond or present in the direct neighborhood, are internalized through an active mechanism of endocytosis; a further break in the peptide then occurs through a currently unknown mechanism, releasing the N-terminal four residues of the cleaved iRGD peptide. The sequential requirement for RGD and R/KXXR/K binding to integrins and NRP1, respectively, is demonstrated by the blockade of step 1 by antibody (Ab) to αV (but not antibody to NRP1) or an excess of free cyclic RGD peptides and the blockade of step 3 by antibody to NRP1 (but not antibody to αV) or an excess of the free CendR-motif peptides CRGDK and RPARPAR.


The rationale in the technology detailed in the two papers by Ruoslahti and his colleagues is therefore to bring such an NRP1 ligand (R/KXXR/K) to a tumor by restricting the nature of the two ambiguous residues (XX) to GD to integrate the RGD consensus sequence into the final peptide (Fig. 1). The generic sequence of the cyclic peptide used in their studies is CRGDKGPDC, also called iRGD for internalizing RGD. The interest of exploiting a CendR motif to penetrate a tumor makes sense only if the payload is delivered to tumors and not to healthy organs. Indeed, without the tumor vascular targeting component of iRGD as provided by the RGD motif, nonselective delivery of chemotherapy combined with improved tissue penetration could considerably increase harmful side effects. In addition, the iRGD cyclic peptide also contains residues that confer sensitivity to cell surface–associated protease(s). The cryptic CendR element is thus made available for binding to NRP1 only after proteolysis, further protecting the cargo from nonselective distribution (Fig. 1).

Interestingly, whereas the research described in the first paper (6) examined drugs that were chemically linked to iRGD, the Science paper (7) demonstrates that the therapeutic agent does not need to be conjugated to iRGD to be influenced by this peptide’s stimulatory effect on penetration into the tumor parenchyma. A slight but reproducible advantage of the unlinked combination (versus the conjugated modality) was even observed when the effects of chemotherapy on the growth of human tumor xenografts in mice were evaluated. Also, as for the conjugated modality, a gain in selective tumor penetration was observed for drugs ranging in size from low-molecular-weight chemicals to nanoparticles and antibodies. These data open obvious possibilities for the use of therapeutic or diagnostic compounds for which chemical coupling could interfere with their activity. This feature might also considerably accelerate clinical trials that aim to evaluate the efficacy of coadministration of the iRGD peptide or derivatives thereof with compounds of interest. Still, this observation requires further investigation. Indeed, if the NRP endocytic pathway is also the mode of entry for nonconjugated drugs, it means that they must be very close to the peptide when internalization is initiated. Coadministered free drugs are unlikely to reach the same local concentration as iRGD-conjugated drugs. However, similar therapeutic effects are observed in both cases, suggesting that other mechanisms contribute to drug penetration in tumors from animals co-injected with free iRGD peptide. More generally, although in the work of Ruoslahti and colleagues, active internalization of iRGD was convincingly documented in vitro and ex vivo (using nonperfused, surgically removed tumor xenografts), it does not preclude the activation of signaling pathways in response to integrin/NRP binding (by free iRGD and iRGD conjugated to drugs) that could influence drug penetration and/or a role for passive drug accumulation in tumors, which could work additively, if not synergistically, with the endocytosis pathway in vivo.


Stimulated permeability and retention effect. Some components of the enhanced permeability and retention (EPR) effect (15) could account for the necessary concomitant accumulation of unconjugated drugs in the tumor parenchyma to take advantage of the free iRGD–driven endocytosis process. The clinically validated EPR concept describes the selective accumulation of macromolecular drugs or liposomal/micellar nanocarriers in tumors. Open fenestrations in the immature tumor endothelium, poor clearance because of a lack of functional lymphatics, and altered intratumor hemodynamics are at the origin of this phenomenon. According to the strict definition of the EPR effect, it is a largely passive mechanism that is restricted to macromolecules, the tumor concentration of which increases with time because of the conjunction of the parameters above.

There are, however, numerous examples of pharmacological and physical treatments that lead to an increase in small-drug delivery into tumors through a modulation of the tumor vascular reactivity [see (16) for a review]. Fick’s law describes how low-molecular-weight drugs are transported across capillary walls. The equation states that the diffusion rate depends on the drug concentration gradient and is proportional to the area of the surface across which the drug will diffuse (the surface of exchange). In theory, any increase in the vessel diameter should lead to a π-fold–larger increase in vessel surface and associated changes in intracapillary pressure. Nitric oxide (17) and endothelin (18) were previously reported to influence the extent of drug delivery in tumors directly by causing changes in tumor microvascular pressure and perfusion. Bradykinin and angiotensin II are other good examples of vasoreactive peptides that promote the delivery of both small drugs and macromolecules through several means, including increasing the pore size of the tumor vasculature, increasing the total tumor vascular surface area, and reducing the interstitial fluid pressure within the tumor (1922); the expression “stimulated permeability and retention” (SPR) could be coined to describe these multiple effects. The speed of iRGD’s effects on drug penetration observed by Ruoslahti and colleagues is compatible with such a SPR effect (Fig. 2A). In other words, the increase in transvascular transport is likely to be at least in part dependent on signaling pathways activated by integrin and/or NRP1 binding by iRGD.

Fig. 2. Rationale for the stimulated tumor penetration of drugs coadministered with iRGD.

Besides the well-characterized iRGD-driven endocytic pathway in endothelial cells (Fig. 1), other mechanisms are likely to contribute to the delivery of drugs deep into the tumor parenchyma in the presence of iRGD. (A) NRP1 stimulates tumor vascular permeability in a VEGF-like manner; both the paracellular pathway and transcytosis can account for this improved transport of drugs across the endothelial cell barrier. This improved transport may be associated with changes in the surface of exchange, thereby also promoting the transport of small drugs. (B) iRGD binding to αV integrins expressed at the surface of tumor cells may exert direct anticancer effects. If NRP1 is also expressed in tumor cells, toxicity may also arise from endocytosis of the payload, as described for endothelial cells (Fig. 1). (C) Double blockade of αV integrins and NRP1 in endothelial cells can be associated with antiangiogenic effects, possibly leading to the normalization of the tumor vasculature and a consecutive improvement of drug delivery into the tumor parenchyma.


A role for NRP1 in driving the effects of VEGF on vascular permeability was revealed in experiments in which NRP1 and VEGF receptor 2 (VEGFR2) were cotransfected (23), and was recently confirmed with the use of NRP1 inhibitors (24). The capacity of VEGF to promote tumor vessel leakiness (25) is actually the first property that led to the identification of this factor [initially described as the vascular permeability factor (26)]. A VEGF-like effect of CendR through its binding to NRP1 could therefore account for an increase in the tumor vessel surface of exchange and for a global elevation in vascular permeability, which is beneficial for the penetration of small drugs and macromolecules. The limitation of the number and capacity of receptors for targeted ligands is also less acute, because generic drug transport pathways are made available through paracellular permeability (passage between endothelial cells that involves localized disruption of cell junctions) or transcytosis (transport through the endothelial cell body) (Fig. 2A); the latter could be on the continuum of the endocytic pathway through the formation of vesiculo-vacuolar organelles (27). Recent data also document that Sema3A, which inhibits VEGF-mediated angiogenesis, can function additively with VEGF to increase microvascular leakiness via the activation of NRP1 (28). Through a similar mechanism, the stimulatory effects of iRGD on vascular permeability could therefore be dissociated from possible antiangiogenic effects.

Nonvascular effects. Nonvascular binding of RGD and CendR certainly plays a role in the global effect of iRGD. RGD interacts mostly with αV (associated with one of several possible β subunits) and α5β1 integrins (2, 3). The expression of these integrins is not strictly limited to the tumor endothelium. A variety of tumor cells do express integrins (2), and the iRGD peptide may therefore stimulate the endocytosis of conjugated or coadministered cytotoxic payload into tumor cells (Fig. 2B). The possibility of a direct toxic effect of iRGD peptide (independent of the coadministered chemotherapy) is also supported by paradoxical results recently reported with RGD mimetics (Fig. 2B). Because αVβ3 and αVβ5 integrins regulate angiogenesis, inhibiting these proteins is predicted to slow angiogenesis. However, low doses of RGD-mimetic αVβ3 and αVβ5 inhibitors can actually promote VEGF-mediated angiogenesis by altering VEGFR2 trafficking (29). These data therefore suggest that the encouraging clinical results obtained with cilengitide (a cyclic RGD pentapeptide) could be attributed to direct cytotoxic effects on cancer cells, in particular in tumors such as glioblastomas, which contain a lot of vitronectin (the natural αVβ3 ligand) (30). NRP receptors also are not exclusively located in the tumor vasculature. For instance, NRP1 was reported to mediate interactions between dendritic cells and T cells (in particular regulatory T cells) and thereby to initiate a primary immune response (31). Whether iRGD promotes an immune response against tumor cells certainly warrants further investigation.

Antiangiogenic effects. Direct antiangiogenic effects of the CendR moiety can be anticipated. A recent study documented that combined inhibition of αVβ3 integrin and NRP1 might decrease VEGF-mediated angiogenic responses further than would individual inhibition of these receptors (32). These authors found that in the presence of αVβ3 integrin, NRP1 minimally contributes to VEGF-induced angiogenesis but that when β3 integrin is blocked with RGD-mimetic inhibitors, VEGF-mediated responses become NRP1-dependent. As emphasized above, although pro- and antiangiogenic effects may be obtained depending on the concentration of administered RGD peptides, the double targeting of integrins and NRP should tip the balance toward an inhibition of tumor vascular development (Fig. 2C). In the context of improved drug delivery, an interesting parallel can then be made with the normalizing effect of antibodies to VEGF and other antiangiogenic small molecules on the tumor vasculature (33, 34). Pruning of immature angiogenic blood vessels might indeed reduce the heterogeneities in tumor blood flow and contribute to a better distribution of drugs in the tumor parenchyma. The benefits of iRGD for the improved distribution of many types of drugs, including small molecules, antibodies, and chemotherapy-loaded liposomes (as well as imaging agents), support the existence of such a general phenomenon. Although iRGD alone had no effect on tumor growth (6), this result does not preclude the possibility that when coadministered with chemotherapy, it could improve delivery through a normalization process as reported for different antiangiogenic drugs.


The set of data obtained by Ruoslathi and colleagues will now need to be tested in the clinic. Obvious goals of the first human studies with iRGD will be to determine to what extent the less angiogenic nature of the endothelium in human versus rodent tumors (34) maintains the basis for the selectivity of the targeting, and more generally, whether toxicity is not exacerbated by nontumor expression of RGD-recognition sites and/or NRP receptors in humans. Also, although a search for the spread of potential metastases was made by Ruoslahti and colleagues in their more recent study (7) and strictly negative results were obtained, clinical studies must carefully evaluate tumor cell intravasation (migration from the tumor compartment into blood) as a possible side effect of a global increase in tumor vascular permeability. These issues were underestimated when antiangiogenic drugs were tested in clinics shortly after obtaining proof of concept in mice. Today, we know that these aspects need to be considered, as recently emphasized by reports documenting a hypoxia-driven increase in metastases after anti-VEGF treatments (35, 36). The challenge is major because it will determine whether strategies such as iRGD will contribute to shift the treatment model from magic bullet (as antiangiogenics were historically claimed to be) to a magic gun that can be be loaded with any kind of ammunition to efficiently treat or image cancer.



    1. Funding: O.F. is professor at the Université catholique de Louvain and Research Director of the Fonds de la Recherche Scientifique (FRS-FNRS). His work is supported by grants from the FRS-FNRS, the Télévie, the Belgian Fundation Against Cancer, the J. Maisin Foundation, an Action de Recherche Concertée (ARC 09/14-020) from the Communauté Française de Belgique, the Région Bruxelles-Capitale, and the Région Wallonne. Competing interests: The author has no competing interests.
    • Citation: O. Feron, Tumor-penetrating peptides: A shift from magic bullets to magic guns. Sci. Transl. Med. 2, 34ps26 (2010).

    View Abstract

    Stay Connected to
       Science Translational Medicine

    Navigate This Article