Dual-targeting anti-angiogenic cyclic peptides as potential drug leads for cancer therapy

Peptide analogues derived from bioactive hormones such as somatostatin or certain growth factors have great potential as angiogenesis inhibitors for cancer applications. In an attempt to combat emerging drug resistance many FDA-approved anti-angiogenesis therapies are co-administered with cytotoxic drugs as a combination therapy to target multiple signaling pathways of cancers. However, cancer therapies often encounter limiting factors such as high toxicities and side effects. Here, we combined two anti-angiogenic epitopes that act on different pathways of angiogenesis into a single non-toxic cyclic peptide framework, namely MCoTI-II (Momordica cochinchinensis trypsin inhibitor-II), and subsequently assessed the anti-angiogenic activity of the novel compound. We hypothesized that the combination of these two epitopes would elicit a synergistic effect by targeting different angiogenesis pathways and result in improved potency, compared to that of a single epitope. This novel approach has resulted in the development of a potent, non-toxic, stable and cyclic analogue with nanomolar potency inhibition in in vitro endothelial cell migration and in vivo chorioallantoic membrane angiogenesis assays. This is the first report to use the MCoTI-II framework to develop a 2-in-1 anti-angiogenic peptide, which has the potential to be used as a form of combination therapy for targeting a wide range of cancers.

Structural analysis using NMR spectroscopy. One-dimensional spectra were recorded on all of the grafted peptides and showed well dispersed peaks in the amide region (Supplementary Figs S1 and S2) consistent with β -sheet containing peptides. Analysis of the secondary chemical shifts indicates that the grafted cyclic peptides have similar secondary structure to the native peptides. For example, the similarity in the secondary shifts of SFTI-SST-01 and SFTI-PEDF with the secondary shifts of native SFTI-1 indicates that both grafted cyclic peptides have similar structures to their native framework at all conserved residues, except at the locations of the grafted anti-angiogenic epitope (Fig. 3A). This conservation of native structure was also observed in grafted cyclic peptides with the MCoTI-II framework (Fig. 3B). Although the same epitopes were inserted into different cyclic frameworks, a difference in the secondary shifts was observed and can most probably be attributed to the different structural constraints or flexibility in the loops where the anti-angiogenic epitopes were inserted. The proline residues in the grafted peptides had the same conformations as the native cyclic frameworks (SFTI-1 Pro 6-cis, Pro 7-trans; MCoTI-II Pro 2-trans and Pro 15-trans) based on analysis of NOEs. The trans proline residues displayed strong NOEs between the δ H's of the proline residues and the α H of the preceding residue, whereas the cis proline residue displayed a strong NOE between the α H of the proline and the α H of the preceding residue. In addition, the proline residue in PEDF adopted a trans conformation when inserted into both SFTI-1 and MCoTI-II frameworks. A chemical shifts table of MCoAA-02 is included as Supplementary Table S2 and deposited in the Biological Magnetic Resonance Data Bank with the accession number 26882 (http://www.bmrb.wisc.edu/). The two-dimensional TOCSY and NOESY spectra of MCoAA-02 are shown in Supplementary Fig. S3. Linear epitopes, native and grafted cyclic peptides are non-toxic in a range of human cells. All peptides and drug controls (cilengitide, octreotide, and sunitinib) were screened in both hemolytic (Fig. 4) and cell cytotoxicity assays (a 2-h incubation; data not shown), and the grafted peptides were found to be non-toxic to mammalian and human red blood cells.
Grafted cyclic peptides showed specificity in inhibiting cell proliferation of HUVECs and cancer cells. Cell proliferation is one of the major processes to occur during tumor angiogenesis prior to cell migration 21 ; thus, a reduction of cell proliferation during this time should inhibit new vascular growth to nearby tumor cells by stopping oxygen and nutrient supply. We determined the cell proliferation inhibitory activities of the linear epitopes, native and grafted cyclic peptides, and drug controls on HUVECs, MCF-7, PC-3, and HT-29 cells (Table 1). Cilengitide showed good potency in inhibiting the proliferation of HUVECs, consistent with this cell line previously being shown to have high levels of expression of integrins 30 . Octreotide was included in the Scientific RepoRts | 6:35347 | DOI: 10.1038/srep35347 study due to a previous report on its potent anti-angiogenic activity on HUVECs at concentrations within the range of 10 −10 -10 −6 M; however, it had no effect on the proliferation of the cell lines tested in this study 31 . Full dose-response curves for all peptides listed in Table 1 are included in Supplementary Fig. S4. Neither of the native cyclic frameworks (SFTI-1 and MCoTI-II) inhibited cell proliferation in any of the cell lines at concentrations > 100 μ M. The linear epitopes and grafted peptides had varying degrees of inhibitory activity, likely due to the different types of receptors expressed within each cell line (the greater the number of types, the greater the number of potential targets), or the level at which the receptors are expressed (affecting the number of possible binding sites) 30,31 . Several grafted cyclic peptides showed better inhibition of proliferation compared to their linear counterparts, potentially due to the greater stability conferred upon grafting into a cyclic framework. However, there are also examples where the grafted cyclic peptides had less inhibition on some cell lines in the proliferation assay compared to the linear form. It is possible the small peptides are restrained in the cyclic peptides in such a way as to inhibit binding to the appropriate target (Table 1).
Screening of peptides in VEGF-mediated HUVEC migration assay. Although the grafted cyclic peptides were inhibitory in the cell proliferation study, the ability to inhibit cell migration also plays an important role in tumor angiogenesis. Consequently, all peptides, including cilengitide and octreotide, were tested in a VEGF-mediated HUVEC migration assay. VEGF was included as a positive control as it is a potent angiogenic growth factor that stimulates HUVEC migration (taken as 100% in the assay). All first-generation peptides The PEDF sequence comprises residues Tyr 388 to Phe 394 from a human PEDF protein (Uniprot ID: P36955), two somatostatin mimetic sequences (SST-01 and SST-02) were derived from a human somatostatin receptor comprising residues Phe 109 to Thr 112, and an anti-VEGF mimetic (polyR) was derived from a phage display library 27 . The first tryptophan in the somatostatin epitopes was designed with a d-amino acid conformation, as a previous study showed this change is vital for its β-turn formation to selectively target SST2 and SST5 receptors 34 . (B) Anti-angiogenic epitopes that gave potent activity among these first-generation cyclic peptides were selected, and further grafted into the MCoTI-II framework. (C) The second-generation peptides then underwent similar screening as the first-generation peptides. (D) Only the best analogues, along with corresponding single-targeting counterparts, were tested in in vivo assays and further characterized by NMR. The location of epitope insertion into SFTI-1 and MCoTI-II are indicated with red arrows. Different epitope insertions into loops of the SFTI-1 and MCoTI-II frameworks are showed using different colored circles (i.e. red (represents polyR), blue (represents PEDF), pink (represents SST-01), and green (represents SST-02)).
showed better inhibition of HUVEC migration compared to their linear epitopes. Neither MCoTI-II nor SFTI-1 showed any effect on HUVEC migration and were subsequently treated as negative controls. The MCoTI-II grafted peptides containing the SST-01 and PEDF epitopes showed 50% inhibition on HUVEC migration at 0.005 μ M and between 1 to 5 μ M, respectively. The most potent SFTI-1 graft contained the PEDF epitope, which displayed 50% inhibition on HUVEC migration at 0.05 μ M, as shown in Fig. 5.
Design and characterization of dual-targeting grafted analogues. Second-generation analogues, containing two grafted epitopes, were designed based on the activity observed in cell proliferation and migration assays, as well as the NMR characterization studies. MCoTI-II was chosen as the second-generation framework for dual-epitope insertion because the larger number of loops and loops of different sizes offer greater scope for epitope insertion compared to SFTI-1. For the second-generation peptides (MCoAA-01 and MCoAA-02), the SST-01 epitope was grafted into loop 5, consistent with the potent first-generation peptide, and the polyR and PEDF epitopes were grafted into loop 1 and loop 6, respectively. The locations of the polyR and PEDF epitopes were primarily based on the first-generation cell migration assay results. The polyR and PEDF epitopes were flanked by glycine residues to assist in peptide folding. The addition of the glycine residues maintained the original loop size of the native cyclic framework and were unlikely to have steric clashes during the folding process. Figure 1 illustrates the stages in the development of a potent dual-targeting cyclic peptide. A full list of peptide molecular masses is shown in Supplementary Table S1. All peptides were purified to > 95% before being used in the assays. An example of purity check and mass spectrometry data of MCoAA-02 is provided in Supplementary Fig. S5.
Both MCoAA-01 and MCoAA-02 had native-like structures based on NMR analysis. For MCoAA-02, the secondary shifts of the inserted anti-angiogenic epitope in loops 5 and 6 ( Fig. 3B) were comparable to those grafted cyclic peptides with single anti-angiogenic epitopes (i.e., MCo-SST-01 and MCo-PEDF). Interestingly, MCoAA-01 and MCoAA-02 are capable of targeting particular cell lines with higher potency than peptides with a single epitope insertion. MCoAA-01 inhibited HUVEC proliferation with a higher potency than MCo-SST-01 and MCo-PEDF. Similarly, MCoAA-02 inhibited HT29 cell proliferation more potently than MCo-SST-01 and MCo-PEDF. However, there are exceptions in some cell lines where a single epitope insertion inhibited cell proliferation more potently compared to the dual-targeting peptides. These differences might be related to First-generation grafted cyclic peptides were designed with a single anti-angiogenic epitope inserted into a single loop. For the second-generation grafted cyclic peptides, two anti-angiogenic epitopes were inserted into two separate loops. Chosen anti-angiogenic epitopes are highlighted in bold. d-tryptophan is represented by 'w' .
variable expression levels of targeted receptors on these cell lines. MCoAA-02 was observed to be the most potent second-generation dual-targeting cyclic peptide, with 50% inhibition of HUVEC migration at 1 nM (Fig. 5T); for cilengitide, the same degree of inhibition required a concentration of 1 μ M (Fig. 5L). Images of the final step in the cell migration assay with MCoAA-02, prior to absorbance reading, are shown in Supplementary Fig. S7. Our data correlate well with that supplied by the manufacturer, with 10 μ M of cilengitide resulting in 100% inhibition of HUVEC attachment on vitronectin and fibronectin. Sunitinib, a potent multi-targeted tyrosine kinase small molecule inhibitor 32 , was also tested at 10 μ M and showed 100% inhibition of HUVEC migration ( Supplementary  Fig. S8A), which was similar to the effect of cilengitide and MCoAA-02 at this concentration. By contrast, octreotide showed a lower potency compared to cilengitide and MCoAA-02 between 1-10 μ M.
The effect of the potent dual-targeting cyclic peptide MCoAA-02 on blood vessel growth in the in vivo CAM model. Based on the potency of MCoAA-02 in the cell assays, it was also tested in a CAM assay to determine the effect on blood vessel growth in vivo (Fig. 6A). Octreotide, cilengitide and selected grafted cyclic peptides were included for comparative purposes, and VEGF (0.3 nM) was used as a positive control. To evaluate the inhibitory effect of these peptides on blood vessel growth, VEGF was added simultaneously to all peptides during the assay. Peptides were tested at 10 μ M, as most grafted cyclic peptides showed more than 50% inhibition in the cell migration assay at this concentration, with the exception of MCoAA-02, which was tested at concentrations ranging from 0.001 μ M to 10 μ M. MCoAA-02 more potently inhibited blood vessel growth compared to analogues with only single anti-angiogenic epitope insertions. An improvement in blood vessel growth inhibition, by approximately 42%, was observed for MCoAA-02 compared to MCo-PEDF. Although MCoAA-02 was not as potent as cilengitide in inhibiting blood vessel growth, it showed comparable results to octreotide at 10 μ M. In addition, MCoAA-02 was capable of inhibiting approximately 50% of blood vessel growth at 100 nM (Fig. 6A). Sunitinib was also tested and found to be more potent than cilengitide and octreotide at 10 μ M (Supplementary Disulfide bond connectivity is highlighted in yellow, and bold lines are used represent the cyclic nature of the peptides. Each cysteine is labeled with a Roman numeral and each loop is represented with the letter 'L' . The loop of insertion of an anti-angiogenic sequence is circled in red for both SFTI-1 and MCoTI-II structures. All spectra were assigned using CCPNMR 46 and each of the amino acid spin systems were specifically assigned based on Wuthrich et al. 47 . The α H secondary shifts were analyzed by subtracting the random coil 1 H NMR chemical shifts of Wishart et al. 48 from the experimental α H chemical shifts. The 3D molecular structure of SFTI-1 and MCoTI-II were illustrated using MOLMOL 49 . Peptide stability assessment in human serum. The stability of peptides tested in the CAM assay was examined against human serum for 24 h. All peptides were stable when subjected to human serum except PEDF, a short linear peptide, which was degraded within 4 h, as expected. Interestingly, SST-01, which contains a d-amino acid, was found to be more stable than PEDF. The stability of MCoAA-02 was compared to sunitinib Melittin was used as a positive control with 100% hemolysis. Drug controls (cilengitide, octreotide, and sunitinib) were also included in the assay. All data are shown as mean ± SD (n = 3). and MCoTI-II using the method described by Ji and colleagues 19 . Although MCoAA-02 was found to be slightly less stable than its native framework (MCoTI-II), more than 75% of this peptide remained in the serum over 24-h, Error bars indicate ± SD (n ≥ 6). The dotted line indicates ~50% inhibition of blood vessels. A one-way ANOVA with Dunnett's post-test using a multiple comparison test was used for statistical analysis. In addition, unpaired t-test was used to test the significance of MCoAA-02 against MCo-SST-01 and MCo-PEDF. ****p ≤ 0.0001 and ***p ≤ 0.05. All peptides were compared to 0.3 nM VEGF (highlighted in grey), which is represented as 100% blood vessel growth. (B) This diagram shows the blood vessel growth of MCoAA-02 at various concentrations compared to the cyclic frameworks SFTI-1 and MCoTI-II in the CAM assay. VEGF was used as the positive control, and octreotide and cilengitide as the negative controls. All images were taken with an original magnification of x16 on an Olympus SZX12 dissecting microscope with a light box. DP capture and DP manager software packages were used during image acquisition. (C) This graph illustrates the percentage of peptide remaining over 24 h in the serum stability assay. All compounds showed better stability than the linear PEDF peptide (highlighted with red dashed lines). All data are represented as mean ± SD and were recorded in triplicate. Peptides labeled with an asterisk (*) were tested using the same method except with an additional step -dissolving the centrifuged pellet with 8 M guanidinium chloride before RP-HPLC analysis. which was approximately 20% higher than the percentage of sunitinib that remained after the same amount of time (Fig. 6C). Overall, results from this assay showed the native cyclic frameworks and all grafted cyclic peptides to be remarkably stable.

Discussion
Peptide therapeutics are known to have excellent selectivity and efficacy, low toxicity, and lower production costs than antibodies; however, they are prone to many challenges, including proteolytic instability and short in vivo half-lives 33 . These challenges can potentially be overcome by a range of chemical modifications to improve their physicochemical properties 33 . Despite these approaches for improving stability, many FDA-approved drugs are either small molecules or antibodies in the anti-angiogenic field, with only two cyclic peptide-based drugs -cilengitide and octreotide -currently in late-stage clinical trials 2 or on the market 34 , respectively. These developments suggest that peptide-based anti-angiogenic therapeutics is an area of potential growth. One approach for improving stability of bioactive peptides is to conjugate them with more stable molecules. However, when short bioactive epitopes are conjugated externally on a molecule, they are more susceptible to degradation, as they are more readily accessible by enzymes and typically require further chemical modification to achieve acceptable pharmacokinetic properties 35 . Based on this limitation, our study aimed to explore how short anti-angiogenic epitopes can be cyclized within a single framework to improve stability and bioactivity. In addition, a dual-targeting strategy was also explored to examine the feasibility of incorporating two short anti-angiogenic epitopes into a single framework. As in the recent use of a dual-targeting strategy to target an epidermal growth factor receptor mutation in non-small-cell lung cancer patients by combining two individual drugs (i.e., afatinib and cetuximab), resulting in dramatic inhibition of tumor growth. Cetuximab alone showed significant suppression of tumors with the L858R mutation but only a slight response to tumors with the T790M mutation, highlighting the advantage of using the combination therapy 36 .
Many new opportunities for the development of peptide therapeutics are likely to arise from multifunctional peptides, and this is an area that needs further exploration 33 . Limited studies have used the dual-targeting strategy with cyclic, disulfide-rich frameworks; indeed, there are only a few examples of bi-functional peptides using cyclic frameworks 37,38 . Previous examples in this area have demonstrated the insertion of two similar epitopes targeting an identical receptor, as summarized in Fig. 7A. Although an increase in activity was observed compared to their corresponding singly grafted peptides, they did not show comparable or better effects compared to drug controls. In the current study, we demonstrated the rational design of second-generation dual-targeting cyclic anti-angiogenic peptides with two different grafted epitopes. Both grafted peptides produced nanomolar potency in both in vitro and in vivo studies, and showed comparable effects to cyclic peptide drug controls and higher stability in human serum than the orally active sunitinib.
All peptides used in this study were non-toxic to either cancer or normal cells, which allowed further investigation on the bioactivity of these peptides in a range of assays. An increase in the inhibitory activity of cell migration studies was observed on grafted peptides when compared to the activity of their corresponding anti-angiogenic epitopes. This increase in potency is probably due to the increase in stability from grafting the epitopes into a cyclic framework. In this study, we have also demonstrated grafting epitopes into different locations of the same or different frameworks resulted in different biological effects. For example, differences in potency were observed for the polyR epitope when inserted into different loops of the same framework. This suggests that to gauge the best location of epitope insertion, a thorough understanding of the native framework structure is essential and screening using a range of assays to identify the location that results in the most significant increase in activity is important for the design of dual-targeting cyclic peptides. Among the grafted cyclic peptides, the dual-targeting MCoAA-01 and MCoAA-02 peptides were the most potent in the cell proliferation studies, with inhibition observed for several cell lines similar to the inhibitory effects in drug controls such as cilengitide and sunitinib. In particular, MCoAA-01 and MCoAA-02 most potently affected HUVECs and HT-29 cells, respectively. This can be explained by the expression of SST and PEDF receptors in these cell lines, both of which are required for these dual-targeting peptides 39 . When these peptides were examined in the cell migration assay, MCoAA-02 exhibited greater potency than MCoAA-01. At 50% inhibition of HUVEC migration, MCoAA-02 exerted 1000-fold higher potency than cilengitide; hence, it was further characterized in in vivo studies.
Given that MCoTI-II did not inhibit blood vessel growth in the CAM assay, it appears the anti-angiogenic activity of MCoAA-02 is derived solely from the grafted epitopes and not the cyclic framework. Furthermore, MCoAA-02 was more active in the CAM assay than MCo-SST-01 or MCo-PEDF, suggesting the inhibitory effect of blood vessels growth on MCoAA-02 could either be due to an additive or synergistic effect. More studies will be required in future to determine the specific effect of this peptide. Angiogenesis inhibition and tumor growth regulation events are both indirect effects of somatostatin. Therefore, MCoAA-02 could potentially have an indirect effect in regulating VEGF on endothelial or tumor cells by inhibiting signaling events downstream of the SST and PEDF receptors, which results in the suppression of cell proliferation and migration. This hypothesis is supported by studies by Craword 40 and Rai and colleagues 41 , who suggested that the anti-angiogenic effect caused by PEDF could be due to the suppression of pro-angiogenic miRNAs (e.g., miR-126, miR-23b and miR-27b). They suggest that blocking of these miRNAs leads to inhibition of endothelial cell proliferation, and the blocking of the SST receptor activates downstream signaling cascades such as MAPK or ERK pathways, and protein tyrosine phosphatases, which could lead to cell cycle arrest.
A schematic diagram outlining our proposed mechanism of action of MCoAA-02 on endothelial and tumor cells is shown in Fig. 7B. We propose that MCoAA-02 could inhibit cell proliferation of cancer and endothelial cells, followed by the inhibition of endothelial cell migration to nearby tumor cells and ultimately block nutrient supply. Together, these events could result in the inhibition of blood vessel growth, consistent with the data from the CAM assay. Furthermore, MCoAA-02 could indirectly regulate VEGF on endothelial or tumor cells by inhibiting signaling events downstream of the SST and PEDF receptors, resulting in the suppression of cell Scientific RepoRts | 6:35347 | DOI: 10.1038/srep35347 proliferation, migration and blood vessel growth. In contrast, in the absence of MCoAA-02 during an angiogenesis event, VEGF acts as a potent angiogenesis growth factor by binding to its corresponding receptor, VEGFR and stimulating cell proliferation, migration and blood vessel growth, as shown in Fig. 7B.
Apart from the biological activities of these peptides, peptide stability is also an important factor when designing new peptide therapeutics, as it can enhance the probability of success in pharmacokinetic studies. Despite the significant number of residue changes in MCoAA-02 compared to MCoTI-II, both are stable in human serum over 24-h. This similarity in stability is presumably related to the similarities in structure reflected in the chemical shifts. Interestingly, MCoAA-02 was more stable in human serum than sunitinib. The stability of MCoAA-02 combined with its lack of toxicity to normal or cancerous cells highlights its potential as a candidate for the design of dual-or multi-targeting peptide therapeutics targeting cancer applications. PEDF and SST-01 epitopes have also been implicated in treating ocular vascular diseases such as diabetic retinopathy [40][41][42][43] , suggesting that MCoAA-02 might also have applications in this field.
Although there are examples of combinatory approaches based on somatostatin, which include the conjugation of somatostatin to different chemotherapy drugs such as camptothecin and methotrexate, resulting in improved efficacy with fewer side effects 39 , this is the first study combining a somatostatin epitope with another anti-angiogenic epitope in a single cyclic disulfide-rich framework. This study thus potentially opens up new possibilities of using cyclic disulfide-rich frameworks as an alternative strategy for the development of next-generation peptide-based combination therapies. A further understanding of the specific mechanism of action on endothelial and cancer cells, along with in vivo mouse model studies, could also enhance its relevance in a clinical setting. In addition, applying this strategy to other cyclic frameworks could lead to applications in a wide range of human diseases.

Materials and Methods
Peptide synthesis. Grafted peptides (all peptides listed in Table 1 except linear anti-angiogenic epitopes and MCoTI-II) and SFTI-1 were synthesized using solid-phase peptide synthesis on a CS Bio synthesizer using Boc chemistry. Peptides were constructed on Boc-Gly-PAM resin (Chem-Impex International) with S-tritylmercaptopropionic acid (Peptides International) as a linker using N,N,N′,N′-tetramethyl-O-(6-chloro-1H-benzotriazol-1-yl)uronium hexafluorophosphate (HCTU) for amino acid activation. Linear peptide analogues (polyR, PEDF, SST-01, SST-02) were synthesized using Fmoc chemistry on a Symphony microwave synthesizer using chlorotrityl chloride resin. Peptides made using Boc chemistry were cleaved from the resin using hydrogen fluoride with p-cresol as a scavenger at 0 to 5 °C for 1 h. Fmoc-synthesized peptides were cleaved using a mixture of 95% trifluoroacetic acid (TFA)/2.5% triisopropylsilane/2.5% H 2 O. The TFA was removed by rotary evaporation and the residue partitioned between 50% acetonitrile in water containing 0.1% TFA and cold diethyl ether. All grafted peptides and linear peptides were synthesized on a 0.5 mmol scale. The aqueous layer was lyophilized and the resulting crude peptides were purified using reverse-phase HPLC (RP-HPLC). MCoTI-II was isolated from M. cochinchinensis seed extract, as described by Chan et al. 44  Peptide purification and folding conditions. Crude peptides obtained from either plant extracts or chemical synthesis were purified using a series of Phenomenex C18 columns on RP-HPLC. Gradients of 1%/min of 0-80% solvent B (90% MeCN in 0.045% TFA in H 2 O) and solvent A (aqueous 0.05% TFA in H 2 O) were used and the eluent was monitored at 215 and 280 nm. Peptide purity was examined on a Nexera UHPLC (Shimadzu) with a flowrate of 0.4 mL/min on a 0.8 mL/min Agilent column using a 2% gradient of 0-50% solvent B and masses were determined by electrospray mass spectrometry. All peptides used in assays had > 95% purity. Folding trials were carried out using 1 mg/mL aliquots of peptide in various folding buffers of different pHs, as well as different concentrations of ammonium bicarbonate and dimethyl sulfoxide (DMSO). The optimal conditions for peptide folding were as follows: SFTI-1 grafted peptides were folded in two steps: cyclization with 0.1 M ammonium bicarbonate (pH 8. Cell culture. HUVECs (human umbilical vein endothelial cells) were cultured in EGM ™ -2 BulletKit ™ supplemented with SingleQuots ™ (supplements: growth factors, cytokines, antibiotics; Lonza) and 10% FBS (catalog no. FFBS-500; Scientifix). Both HT-29 and MCF-7 cells were cultured in 10% FBS/DMEM (Dulbecco's Modified Eagle Medium) with 1% penicillin-streptomycin (5000 U/mL; Life Technologies), whereas PC3 cells (prostate cancer cells) were cultured in 10% FBS/RPMI (Roswell Park Memorial Institute) 1640 with 1% penicillin-streptomycin (5000 U/mL; Life Technologies). All cells were maintained at 37 °C in 5% CO 2 .

Cell cytotoxicity and proliferation assays on HUVECs (non-cancerous) and cancerous cells.
Cell cytotoxicity and proliferation assays were performed using similar methods to those described by Chan et al. 45 . All cells were maintained using the media conditions described above, and passages two to ten were used for all cell lines. For all cell lines, 5.0 × 10 3 cells/well (100 μ L) and 1.5 × 10 3 cells/well (100 μ L) were used for cell cytotoxicity and cell proliferation assays, respectively. All peptides (10 μ L, at final concentrations ranging from 0.05-100 μ M) and drugs, which included cilengitide, octreotide and sunitinib (at final concentrations ranging from 0.00001-100 μ M), were used in both assays. Triton X-100 (0.1% (v/v); 10 μ L) was used as a positive control. Cells were treated with fresh media the day after plating, before the addition of peptides, followed by a subsequent 2-h and 48-h incubation for cell cytotoxicity and cell proliferation, respectively. After different incubation periods, 3-(4, 5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide (MTT) (10 μ L; 5 mg/mL in PBS) was added, and cells were incubated for a further 4 h. The supernatant was then removed and 100 μ L DMSO added to solubilize formazan salts. Experiments were performed in triplicate. Cell numbers were recorded at 600 nm. All data represent average mean ± SD.
Transwell migration assay using HUVECs. A cell migration assay was performed as described by Chan et al. 45 HUVECs between passages three and ten were used. Briefly, initial peptide concentrations of 10-500 μ M were prepared for all peptides except SFTI-SST-01, MCo-SST-01, SFTI-SST-02, MCo-SST-02, SFTI-PEDF, MCo-PEDF (0.01-500 μ M), cilengitide, octreotide, and MCoAA-02 (1 × 10 −7 M-100 μ M). To the apical well, 100 μ L of 1 × 10 5 cells/well were seeded and 600 μ L of medium supplemented with or without 0.3 nM VEGF (Sigma, catalogue no. V7259) was added to the basolateral well. The well containing cells with 0.1% FBS/EGM-2 basal medium was used as a negative control, and the well containing cells with 0.3 nM VEGF was used as a positive control. All peptides were stimulated with 0.3 nM VEGF. Experiments were done in triplicate. All data represent mean ± SD. Prism Version 6 software (GraphPad) was used for statistical analysis. A one-way analysis of variance (ANOVA) with Dunnett's multiple comparison test was applied in the analysis.