A stapled chromogranin A-derived peptide homes in on tumors that express αvβ6 or αvβ8 integrins

Rationale: The αvβ6- and αvβ8-integrins, two cell-adhesion receptors upregulated in many tumors and involved in the activation of the latency associated peptide (LAP)/TGFβ complex, represent potential targets for tumor imaging and therapy. We investigated the tumor-homing properties of a chromogranin A-derived peptide containing an RGDL motif followed by a chemically stapled alpha-helix (called “5a”), which selectively recognizes the LAP/TGFβ complex-binding site of αvβ6 and αvβ8. Methods: Peptide 5a was labeled with IRDye 800CW (a near-infrared fluorescent dye) or with 18F-NOTA (a label for positron emission tomography (PET)); the integrin-binding properties of free peptide and conjugates were then investigated using purified αvβ6/αvβ8 integrins and various αvβ6/αvβ8 single - or double-positive cancer cells; tumor-homing, biodistribution and imaging properties of the conjugates were investigated in subcutaneous and orthotopic αvβ6-positive carcinomas of the pancreas, and in mice bearing subcutaneous αvβ8-positive prostate tumors. Results: In vitro studies showed that 5a can bind both integrins with high affinity and inhibits cell-mediated TGFβ activation. The 5a-IRDye and 5a-NOTA conjugates could bind purified αvβ6/αvβ8 integrins with no loss of affinity compared to free peptide, and selectively recognized various αvβ6/αvβ8 single- or double-positive cancer cells, including cells from pancreatic carcinoma, melanoma, oral mucosa, bladder and prostate cancer. In vivo static and dynamic optical near-infrared and PET/CT imaging and biodistribution studies, performed in mice with subcutaneous and orthotopic αvβ6-positive carcinomas of the pancreas, showed high target-specific uptake of fluorescence- and radio-labeled peptide by tumors and low non-specific uptake in other organs and tissues, except for excretory organs. Significant target-specific uptake of fluorescence-labeled peptide was also observed in mice bearing αvβ8-positive prostate tumors. Conclusions: The results indicate that 5a can home to αvβ6- and/or αvβ8-positive tumors, suggesting that this peptide can be exploited as a ligand for delivering imaging or anticancer agents to αvβ6/αvβ8 single- or double-positive tumors, or as a tumor-homing inhibitor of these TGFβ activators.

Considering the enhanced expression of these integrins by several types of cancer cells, compounds that target these integrins in tumors are of great experimental and clinical interest, as they could be used either for delivering imaging and therapeutic agents to tumors, or to modulate their activity.
According to this view, various ligands of αvβ6 have been developed, such as the foot-and-mouth disease virus-derived peptide A20FMDV2, the cyclic peptide c[RGDLATK] (Cycratide), the trimerized nonapeptide Trivehexin, the cysteine knot peptides, and the sunflower trypsin inhibitor-derived peptides [15]. These compounds, coupled to tumor imaging agents, are currently being tested in cancer patients [16][17][18][19][20]. Similarly, a radiolabeled trimerized peptide c[GLRGDLp(NMe)K], an αvβ8 ligand, has shown encouraging preclinical imaging results in a murine αvβ8-positive melanoma model [21]. Furthermore, considering that both αvβ6 and αvβ8 can activate TGFβ (a potent immunosuppressive cytokine) by interacting with the RGD sequence of the inactive latency-associated peptide (LAP)-TGFβ complex [1,7,11], compounds capable of targeting and blocking this integrin may have therapeutic activity [22]. This view is supported by the results of recent studies showing that anti-αvβ8 antibodies can inhibit TGFβ activation in murine tumors and elicit durable antitumor immunity [12,23], while anti-αvβ6 antibodies can inhibit the growth of αvβ6-positive tumors, again through a TGFβ-regulated mechanism [24].
Thus, the development of bi-specific tumorhoming compounds that target the active site of both αvβ6 and αvβ8 integrins in tumors may represent an important advance in this field. With this in mind, we have investigated the tumor-homing properties of a peptide that selectively binds the active site of both αvβ6 and αvβ8 with high affinity [25]. This peptide, derived from the region 39-63 of human chromogranin A (a neurosecretory protein), contains an RGDL motif, followed by an alpha-helix chemically stapled with a triazole bridge, which binds the RGD binding site of αvβ6 with interactions similar to those observed with the proTGFβ1/αvβ6 complex [25,26]. To assess its capability to home in on αvβ6-and/or αvβ8-positive tumors we have labeled this peptide with optical-and radio-imaging compounds, analyzed their capability of recognizing various αvβ6/αvβ8 double-or single-positive cancer cells, and investigated their tumor-homing and biodistribution properties in subcutaneous and orthotopic murine models of pancreatic cancer (αvβ6 + and αvβ8 -) and prostate cancer (αvβ6and αvβ8 + ). We show that this peptide binds αvβ6/αvβ8 single-or double-positive cancer cells in vitro and efficiently accumulates in αvβ6-or αvβ8-positive tumors through a receptor-mediated mechanism. Furthermore, we show that this peptide can inhibit TGFβ activation by cancer cells, suggesting that this compound is an inhibitor of αvβ6 and αvβ8 endowed of tumor-homing properties.

Materials and methods
Full details of all methods, reagents and equipment used are presented in the Supplemental Materials.

Peptide synthesis, purification, and characterization
Peptides were prepared by chemical synthesis; their identity and purity were checked by mass spectrometry (MS) and reverse-phase HPLC analysis. The affinity of peptides for αvβ6-coated plates was determined by competitive binding assay using an isoDGR peptide (an RGD mimetic) labeled with peroxidase as a probe for the integrin-binding site [25]. Cell adhesion assays were carried out as described previously [26].
Binding of peptide-IRDye conjugates to αvβ6 and αvβ8 and to cultured integrin-expressing cells The affinity of peptide-IRDye conjugates for αvβ6 and αvβ8 integrins was determined by a direct binding assay using microtiter plates coated with αvβ6 or αvβ8 (Bio-Techne). After 1 h of incubation, the plates were washed; bound fluorescence was quantified by scanning the plates with an Odyssey CLx near-infrared fluorescence imaging system (LI-COR).
To assess the binding of peptide-IRDye conjugates to cells various amounts of conjugates (range 0.32-200 nM) were added to cultured cells and left to incubate for 1 h at 37 °C. After three washings, the cells were fixed and analyzed by scanning the plate with the Odyssey CLx.

In vivo studies in animal models
Procedures involving laboratory mice and their care were approved by the Ospedale San Raffaele Animal Care and Use Committee and approved by the Minister of Health. The study was performed at the San Raffaele Hospital (authorized organization) according to institutional guidelines and in compliance with national and international laws and guidelines.
The tumor-homing properties of the labeled peptides were evaluated using a) subcutaneous mouse tumor models of pancreatic and prostate cancer, and b) an orthotopic mouse model of pancreatic cancer (see Supplemental Materials).

Near-infrared imaging studies
Mice were injected with peptide-IRDye conjugate (~1 nmol) into the tail vein and imaged at various time points (0-72 h) using the IVIS Spectrum CT Imaging System (PerkinElmer) (see Supplemental Materials). For blocking experiments, mice were injected intravenously with unlabeled peptide 5a (128 nmol/mouse) 10 min before 5a-IRDye.

PET imaging and biodistribution studies
The 5a-NOTA conjugate was radiolabeled with 18 F using a modified Tracerlab FX-N automatic module (GE Healthcare). Full details are presented in the Supplemental Materials. The radiotracer uptake (~4 MBq/mice, in 100 μl of water containing <10% ethanol) was monitored by whole-body PET/CT using the preclinical β-cube® and X-cube® scanners (Molecubes), respectively. For ex vivo biodistribution, mice were euthanized, and tumor and selected organs were collected, rinsed, weighed, and analyzed for their radioactivity content using a γ-counter (LKB Compugamma CS 1282).

Peptide 5a binds αvβ6 and αvβ8 with high affinity and inhibits cell-mediated TGFβ activation
We previously developed a chromogranin A-derived peptide (FETLRGDLRILSILRX 1 QNLX 2 KELQD, peptide 5), capable of recognizing αvβ6 and αvβ8 with high affinity and selectivity (Ki=0.6 nM and 3.2 nM, respectively) [25]. This peptide contains the RGDLXXL integrin recognition motif followed by an amphipathic alpha-helix chemically stabilized by a triazole bridge between the propargylglycine (X1) and azidolysine (X 2 ) residues.
To couple peptide 5 to imaging compounds containing maleimide groups, we fused a cysteine residue to its N-terminus (peptide 5a, see Figure S1). The capability of peptide 5, 5a and 2a (the latter containing RGE instead of RGD) ( Table 1) to inhibit the interaction of αvβ6 with an isoDGR-peroxidase conjugate (a probe for the RGD-binding site of integrins [25]) was then analyzed. Peptide 5 and 5a, but not 2a, could bind αvβ6 with similar affinities (Table 1 and Figure S2), pointing to a crucial role of RGD for αvβ6 recognition. Peptide 5a, but not 2a, inhibited the adhesion of TRAMP-C2 prostate cancer cells (αvβ6and αvβ8 + ) to a neutralizing anti-αvβ8 antibody (Fig. 1A), whereas it promoted cell-adhesion when adsorbed onto the solid-phase ( Figure 1B and C); furthermore 5a, but not 2a, inhibited the capability of these cells to activate TGFβ ( Figure 1D). Overall, these results suggest that a) the cysteine residue added to peptide 5 does not impair its ability to bind αvβ6 and αvβ8 and b) peptide 5a can inhibit cell-mediated TGFβ activation.

5a-IRDye binds to αvβ6/αvβ8 single-or double-positive cancer cells
To evaluate the ability of 5a-IRDye to recognize αvβ6 on the cell surface, we then analyzed the interaction of 5a-, 2a-or Cys-IRDye with αvβ6-positive and -negative cells, including human BxPC-3 cells (αvβ6 + /αvβ8 -), murine 5M7101 pancreatic adenocarcinoma cells (αvβ6 + and αvβ8 -), and human umbilical vein endothelial cells (HUVECs) (αvβ6and αvβ8 -) ( Figure S5A and S6A). As expected, 5a-IRDye could bind BxPC-3 and 5M7101 cells, but not HUVECs, while little or no binding of 2a-IRDye and Cys-IRDye (control conjugates) to all cells tested was observed ( Figure  S5B and S6B). Binding of 5a-IRDye to BxPC-3 cells was inhibited by the αvβ6-blocking antibody 10D5 (but not by an isotype-matched control antibody), and by an excess of 5a or A20FMDV2 (a known ligand of αvβ6), but not by 2a (Figure S5C and D). These results confirm the hypothesis that 5a-IRDye binding to cells is mediated αvβ6 and that its RGD sequence is critical for binding. Adhesion of TRAMP-C2 cells to PVC-microplates coated with a 5a-human serum albumin conjugate (5a-HSA), or with a control conjugate made without peptide (*HSA), after 2 h of incubation. Non-adherent cells were removed; adherent cells were stained with crystal violet and quantified by spectrophotometric analysis (A570nm) (mean±SE, n=3). Representative photomicrographs of wells coated with 5a-HSA or *HSA (16.6 µg/ml) are shown. (C) Effect of 5a and 2a on the adhesion of TRAMPC-2 cells to PVC-microplates coated with 5a-HSA (10 µg/ml). Adherent cells were stained with crystal violet and analyzed spectrophotometrically (mean±SE, n=2-3). (D) Effect of 5a and 2a on TGFβ activation by TRAMP-C2 cells. Cells were seeded on cell culture microplates, left to adhere for 3 h and treated with 5a or 2a for 16 h. The amount of active TGFβ in the supernatant was quantified using a bioassay based on HEK-Blue™ TGFβ cells. Cumulative results of two independent experiments are shown (mean±SE, n=4-6 wells). *, P<0.05; **, P<0.01; ***P <0.001 by two-tail t-test. Table 1. Binding affinity of chromogranin A-derived peptides and A20FMDV2 for human and murine αvβ6 and αvβ8 integrin.

5a-IRDye homes in on subcutaneous αvβ6 + -pancreatic adenocarcinomas
The ability of 5a-IRDye to bind αvβ6 in vivo was assessed using a subcutaneous xenograft model of pancreatic adenocarcinoma based on human BxPC-3 αvβ6 + /αvβ8cells implanted in NGS, and a subcutaneous syngeneic model based on murine 5M7101 cells implanted in C57BL/6N mice. In vivo NIR-fluorescence analysis of BxPC-3 tumors showed maximal 5a-IRDye uptake 1 h after injection, with a tumor-to-background ratio (TBR) of about 7; significant signal was still visible 24 h later (TBR: ~4) ( Figure S8A and B). Ex vivo analysis of tumor, pancreas, and kidney showed that the uptake was higher in tumor and kidney than in pancreas ( Figure  S8C). Of note, the tumor-to-pancreas ratio was ~12, indicating that 5a-IRDye accumulated better in pancreatic tumors than in normal pancreas.
A TBR of ~8 was observed in the 5M7101 model at 24 h ( Figure S8D). These results suggest that 5a-IRDye can home to human and murine αvβ6 + pancreatic adenocarcinomas.
To verify the specificity of 5a-IRDye, we compared the uptake of this conjugate and two controls made with Cys or a scrambled sequence in place of 5a (Cys-IRDye 5a-Scr-IRDye, see Table 1) by BxPC-3 tumors implanted in nude mice, a hairless model. As expected, 5a-IRDye accumulated in tumors more efficiently than Cys-IRDye and 5a-Scr-IRDye (Figure 2A and B). The uptake of 5a-IRDye was significantly inhibited by prior administration of an excess of 5a ( Figure 2B), suggesting it was mediated by the 5a moiety.

5a-IRDye homes in on orthotopic αvβ6 + -pancreatic adenocarcinomas (BxPC-3)
The ability of 5a-IRDye to home in on pancreatic tumors was also investigated in an orthotopic model of pancreatic adenocarcinoma based on surgical implantation of BxPC-3 cells, suspended in Matrigel, into the head of the pancreas of nude mice. Mice implanted with Matrigel (without cells) served as controls. NIR fluorescence analysis of the surgically exposed pancreas of mice, 24 h after administration, showed higher accumulation of 5a-IRDye in the pancreas of mice with BxPC-3 tumors than in the pancreas of control mice ( Figure 3A). NIR fluorescence analysis of the excised pancreas showed that the ratio of tumor-to-normal pancreas was ~4 ( Figure 3B, C and Figure S9). Noteworthy, larger tumor lesions accumulated more dye than small lesions ( Figure 3B). Analysis of other explanted organs showed that 5a-IRDye accumulated also in kidney, lung, and liver, and little or not at all in heart, brain, muscle, bone, spleen, intestine, and stomach ( Figure 3C and Figure   S9). The higher accumulation of 5a-IRDye in kidney and urine (not shown), compared with the other organs, is likely related to renal clearance.

PET imaging of subcutaneous αvβ6 + -pancreatic adenocarcinomas (BxPC-3) with 5a-NOTA-18 F
To further evaluate the ability of 5a to recognize αvβ6 in vivo and to home in on αvβ6-positive tumors, we coupled this peptide with maleimide-NOTA (5a-NOTA) to allow radiolabeling with 18 F (Table S1, S2 and Scheme S2). The 2a-NOTA conjugate was also prepared (negative control). RP-HPLC and MS analysis of the conjugates showed that they were homogeneous and with the expected molecular weight (Figure S10A and  B).
Whole-body PET/CT scan of mice bearing subcutaneous BxPC-3 tumors, performed 1, 2, and 4 h after 5a-NOTA-18 F administration, showed radiotracer accumulation in tumors and kidneys, but not in muscles or femurs, i.e., tissues that do not express αvβ6 ( Figure S11A). Notably, the high and progressive retention of the radiotracer in tumors compared to muscles, but not or much less in femurs, suggest a specific mechanism of uptake ( Figure S11B). Accordingly, tumor uptake of 5a-NOTA-18 F, 2 h post-injection, was almost completely inhibited by prior administration of an excess of 5a ( Figure 4A and B), pointing to a specific mechanism involving ligand-receptor interactions.  Ex vivo biodistribution data (obtained 2 h post-injection) confirmed that the radiotracer accumulates in tumors in a specific manner, as shown by the marked decrease in tumor uptake (from 3.5% to less than 0.5% of the injected dose (ID)/g of tissue) in mice pretreated with an excess of 5a ( Figure 4C).
Lower, albeit specific, accumulation of 5a-NOTA- 18 F was also observed in lung (1.2% ID/g). Uptake in brain, heart, spleen, blood, and muscle was less than 0.5% ID/g and was not displaced by the free peptide. In addition, some accumulation (about 2% of ID/g) was also observed in intestine, femur, liver, and stomach. In this case, however, no significant reduction was caused by unlabeled 5a, arguing against a peptide-mediated mechanism of accumulation in these organs. Finally, high radiotracer levels were also observed in the kidneys (about 80% ID/g), probably due to renal clearance of the conjugate.

5a-IRDye accumulates on αvβ8-positive prostate tumors (TRAMP-C2)
Finally, the ability of 5a to home to αvβ8positive tumors was investigated in the TRAMP-C2 prostate tumor model (αvβ6 -/αvβ8 + ), implanted subcutaneously in immunodeficient mice. Uptake of 5a-IRDye by these tumors was significantly inhibited by an excess of 5a (Figure S12), suggesting a 5a-mediated mechanism of uptake. The low, albeit specific, uptake of 5a-IRDye in this model compared with αvβ6-positive pancreatic adenocarcinomas, is likely related to the lower αvβ8-expression of TRAMP-C2 cells compared with that of BxPC-3 cells, as observed by flow-cytometry analysis ( Figure S5A and Figure S7A). Nevertheless, these results lend further support to the hypothesis that 5a can recognize αvβ8 in vivo.

Discussion
This work demonstrates that 5a, a chromogranin A-derived peptide with high affinity for human αvβ6 and αvβ8, homes in on αvβ6 or αvβ8 integrin-positive tumors. In these experimental setups, 5a was coupled via cysteine to maleimide-IRDye® 800CW (a NIR fluorescent dye) or maleimide-NOTA (a macrocyclic chelating agent for radiolabeling with 18 F). The resulting conjugates, called 5a-IRDye and 5a-NOTA, could bind human and murine αvβ6 or αvβ8 with an affinity similar to that of 5a (0.6-3 nM), suggesting that the peptide retains its integrin-binding properties after conjugation. Peptide 5a could also recognize αvβ6 expressed on cell membranes, as suggested by the observation that 5a-IRDye could recognize αvβ6-positive cells (but not αvβ6-negative cells) in vitro, in a manner that was inhibited by a neutralizing anti-αvβ6 antibody. The binding to cells was also competed by an excess of 5a or A20FMDV2 (a known ligand of αvβ6), but not by the control peptide 2a (containing RGE instead of RGD), suggesting that 5a recognizes the RGD-binding site of αvβ6.
The uptake of 5a-IRDye by BxPC-3 PDAC tumors in vivo was inhibited by an excess of 5a, suggesting that also the tumor-homing properties of this compound depend on a receptor-mediated targeting mechanism. This hypothesis is further supported by the observation that a) tumor uptake of 5a-IRDye was higher than that of Cys-IRDye or 5a-Scr-IRDye, the latter consisting of a conjugate prepared with a scrambled sequence of 5a, and b) tumor uptake of 5a-NOTA-18 F was significantly inhibited by an excess of 5a.
On-and off-target peptide accumulation studies performed in the orthotopic model of αvβ6 + /αvβ8pancreatic cancer showed good uptake of 5a-IRDye in tumors but not in heart, brain, muscle, bone, spleen, intestine, and stomach. Some degree of dye uptake was also observed in liver, lung, and kidney. The uptake in the kidney might be due, in part, to a specific targeting mechanism, considering that mural mesangial cells can express αvβ8 [11] . However, the notion that αvβ6 expression in mouse kidney is negligible [27], and the observation that 5a-IRDye and control 5a-Scr-IRDye accumulated in this organ to similar extents suggests that most uptake in the kidney was related to renal clearance and not to a specific targeting mechanism. The uptake of 5a-IRDye by the lung was possibly related to the fact that this organ expresses αvβ6 and αvβ8 integrins [28,29].
Biodistribution studies performed with 5a-NOTA-18 F showed a pattern similar to that of 5a-IRDye, except for some additional uptake in the stomach and femur. Of note, uptake of 5a-NOTA-18 F in the lung and kidney, but not in the femur, was significantly inhibited by an excess of 5a. The fact that the uptake in the femur was not blocked by the free peptide is probably due to the presence of some Al 18 F in our preparation or to a partial release of 18 F from 5a-NOTA- 18 F, known to accumulate in the bone [30]. Interestingly, other investigators have shown that similar radiopeptides specific for αvβ6 accumulate in the kidney, stomach, and intestine [27,31]. In these studies, uptake in the stomach and intestine was attributed to the expression of αvβ6 in these organs [19,27], whereas the uptake in kidney was related to an αvβ6-independent mechanism [27,32], as also suggested by the observation that the accumulation in this organ was significantly blocked by administration of a scrambled sequence [27]. In summary, the on-and off-target accumulation studies of 5a in PDAC-bearing mice reveal specific accumulation in αvβ6-positive tumors and off-target accumulation in lung, liver, intestine, and kidney, with the last three organs likely mostly related to the excretory pathways of the peptide or its metabolites. On the other hand, the results of in vitro studies showing that 5a can specifically bind αvβ6 -/αvβ8 + cells (in an αvβ8dependent manner) and the results of in vivo studies showing specific accumulation of 5a-IRDye in αvβ6 -/αvβ8 + prostate tumors (TRAMP-C2 model) suggest that this peptide can also accumulate on cells expressing this integrin in tumors, such as carcinoma cells and infiltrating regulatory T cells [12,23,33].
These results pave the way for several potential applications of peptide 5a, including: a) diagnostic molecular imaging of αvβ6 + tumors with 5a-NOTA-18 F, b) fluorescence-guided surgery of tumors with 5a-IRDye, and c) the delivery of other diagnostic, theranostic or therapeutic agents to tumors (such as antibodies, cytokines, drugs, nanoparticles, DNA complexes, and other compounds). Considering that different tumor types may express both αvβ6 and αvβ8 integrins (e.g., oral squamous cell carcinoma) the dual-receptor-targeting properties of 5a provide an advantage over other mono-targeting compounds developed to date, as this might improve the tumortargeting sensitivity and/or detection efficiency. This possibility should be further investigated in appropriate models. An additional advantage could be related to the fact that 5a, being derived from a human protein (chromogranin A), could be less immunogenic than other peptides carrying viral sequences or containing non-natural amino acids, previously described. Finally, the fact that 5a can recognize the LAP-TGFβ complex-binding site of both αvβ6 and αvβ8 integrins opens the way to another potential application of this peptide in cancer therapy, i.e., targeting TGFβ activation in the tumor microenvironment. Indeed, previous NMR and computational/ biochemical studies showed that the stapled CgA-derived peptide binds the RGD binding site of αvβ6 with receptor-ligand interactions similar to those observed for the proTGFβ1/αvβ6 complex [25], thereby suggesting that 5a can block the binding of the inactive LAP-TGFβ complex to this integrin and, consequently, its activation. This view is supported the results of the present study, showing that 5a can block the binding of αvβ6 and αvβ8 to isoDGR-peroxidase, a probe for the RGD-binding site of integrins [25], and inhibit TGFβ activation by αvβ8 + TRAMP-C2 cells. Thus, this peptide may represent a sort of bi-selective inhibitor of both αvβ6 and αvβ8 endowed of tumor-homing properties. Interestingly, recent studies have shown that regulatory T cells colonizing tumors express higher levels of αvβ8 than those isolated from lymphoid organs, and that these cells, by activating TGFβ in tumors, induce immunosuppression and favor tumor growth [23,33]. Other studies have shown that even αvβ8 + tumor cells can activate TGFβ and favor tumor growth in murine models [12]. Notably, anti-αvβ8 antibodies capable of blocking TGFβ activation can induce tumor regression [12,23]. Similarly, anti-αvβ6 antibodies can inhibit the growth of αvβ6-positive tumors in mice through a TGFβ-regulated mechanism [24]. Thus, considering that both αvβ6 and αvβ8 can activate TGFβ, both integrins may represent promising targets for cancer immunotherapy. Based on these concepts, a major novelty of the present study lies in the fact that 5a might be considered as an efficient bispecific "inhibitor" of both αvβ6 and αvβ8 with the peculiarity of being also a tumor-homing compound. However, this hypothesis needs to be demonstrated in appropriate animal models. Furthermore, considering the short plasma half-life of 5a-IRDye (~8 min) (Figure S13), the development of derivatives with longer half-life, e.g., PEGylated-5a, is likely necessary to enable sustained blockade of αvβ6/αvβ8 in tumors.
Of note, another dual αvβ6/αvβ8 peptide ligand has been recently described (RGD-Chg-(NMe)E]-CONH2) [34]. However, the tumor homing properties of this compound -which inhibits the binding of soluble integrin αvβ6 and αvβ8 to an immobilized ECM protein with an IC 50 of 1.6 and 60 nM, respectively -have not been demonstrated.

Conclusion
The results demonstrate that 5a can home to αvβ6-and αvβ8-positive tumors. This compound can be exploited as a tumor-homing ligand for delivering imaging and anticancer compounds to αvβ6/αvβ8 single-or double-positive tumors. Furthermore, considering that 5a can block the active site of αvβ6/αvβ8 and inhibit TGFβ activation (a potent immunosuppressive mechanism in tumors), our results suggest that this peptide is a peculiar tumor-homing bi-selective inhibitor of αvβ6 and αvβ8 that could be exploited for targeting this immunosuppressive mechanism in tumors.