Synthesis of analogues of LXW7

Peptides were synthesized using Fmoc chemistry. Peptide cyclization was completed in the 50 mM NH₄HCO₃ buffer with activated charcoal. The crude product was purified by RP-HPLC. The final products were characterized by MALDI-TOF-MS and the purity was determined by analytical HPLC (Supplementary data Table S1).

Synthesis of Cyclic LXW7 Derivatives with N-Terminal Modification

The synthesis of linear peptide cGRGDdvc was done on Rink amide resin using Fmoc chemistry. After Fmoc deprotection, the resin were split into 14 portions, 14 different carboxylic acids were coupled to each portion of resin using DIC/HOBt in DMF. The reaction mixture was shaken overnight at room temperature. The resin was then completely washed and dried under vacuum. After cleavage from the beads, the crude products were cyclized and purified to give peptide 18–31 (Supplementary data Table S1).

Synthesis of Focused “One-Bead-One-Compound” Peptide Libraries

The OBOC libraries were synthesized on TentaGel S NH₂ resin using the bilayer bead encoding strategy and “split-and-pool” method. A biphasic solvent and topologic segregation approach was used to create bilayer beads. The library compounds are displayed on the outer layer (4% down substitution) and the coding peptide tags on the bead interior. The two focused OBOC peptide libraries were constructed on bilayer beads using Fmoc chemistry. The synthesis details are presented in the supplementary data “synthesis of focused libraries”, Table S2, S3 and S4.

On-Bead Whole Cell Screening Assay

U-87 MG cells adherent to the bottom of a T75 flask were trypsinized with 0.05% trypsin-EDTA and neutralized with culture medium. Floating cells were collected, spun down, and resuspended in 10 mL of culture medium in a 10 cm Petri dish. The library beads were washed sequentially with ethanol, water, and PBS. The beads were then incubated with suspended U-87 MG cells, and the entire dish was swirled at a speed of 40 rpm inside an incubator at 37°C under 5% CO₂. The plate was then examined under an inverted microscope every 15 min.

Flow Cytometry. K562 and K562/αvβ3+ cells were cultured in RPMI 1640 medium

To demonstrate the peptides binding affinity, the cells (3 × 10⁵) in each sample were incubated with 0.5μM biotinylated peptides in 50 μL of PBS containing 10% FBS and 1 mM MnCl₂ for 30 min on ice. Then each sample was washed three times with 1 mL PBS containing 1% FBS. Samples thereafter were incubated with a 1:500 dilution of streptavidin-PE (1 mg/mL) for 30 min on ice followed by a single wash with 1 mL of PBS containing 1% FBS. Finally, the samples were analyzed via flow cytometry (Coulter XL-MCL). To determine the IC₅₀ of the peptides, series of diluted peptides were incubated with 0.5 μM biotinylated LXW7 first, then the mixture was incubated with cells and followed by streptavidin-PE incubation. Samples were run with flow cytometry and Mean Fluorescence Intensity was decided for each individual sample. IC₅₀s were calculated based on the readings from each sample.

Substitution of D-Cys¹ and/or D-Cys⁸

The cyclic OBOC peptide libraries from which LXW7 was discovered utilized N- and C-terminal D-Cys to form the disulfide bridge (21). To determine the importance of the chirality at these Cys residues, we replaced each or both D-Cys at N- and/or C-termini with L-Cys. The replacement of one D-Cys at N- or C-termini (peptide 2 and peptide 3) caused 10- and 16-fold decrease in potency, respectively, than the parent compound. Substitution of D-Cys with L-Cys on both termini (peptide 4) resulted in a complete loss of binding affinity (Table 1).

We have conducted NMR study of LXW7 and peptide 4 (LXW11) to examine the conformational changes after replacement of the flanking D-Cys with the opposite enantiomers. The complete chemical shift assignment of proton and carbon resonances for LXW7 and peptide 4 are presented in supplementary data Table S5 and S6, using one- (¹H NMR and ¹³C NMR) and two- dimensional NMR experiments (HMQC, HMBC, TOCSY, and NOESY). Conformations of LXW7 and peptide 4 have been established on the basis of NOESY data (Figure 1, supplementary data Figure S1). In comparison of NOEs of LXW7 and peptide 4, proton correlations in LXW7 between γ CH of D-Val and α CH of Asp, and the β CH of D-Asp were absent in peptide 4. More inter-residue NOEs were observed in peptide 4. New proton correlations in peptide 4 were indicated between the γ CH of D-Val and β CH of Cys⁸, and the NH of Gly⁴, which suggested that they are located on the same side of the ring. New correlation was observed between β CH of D-Asp and NH of D-Val, exhibiting that the side chain of D-Asp was more close to the peptide backbone. Other new correlations in peptide 4 were shown between α CH of Cys⁸ and α CH of Cys¹, NH of Gly². The NOE intensity between two α CH of Cys¹ and Cys⁸ in peptide 4 is much stronger than the intensity between D-Cys¹ and D-Cys⁸ in LXW7, which should indicate that conformation of peptide 4 is more constrained than that of LXW7.

The NMR-derived structures of LXW7 and peptide 4 were superimposed and final ensembles of 15 acceptable structures for each peptide are shown in supplementary data Figure S2 and Table S7. The energy minimized average structures of LXW7 and peptide 4 are illustrated in Figure 2. The two peptides adopt entirely different structures. LXW7 forms a bowl-like conformation, in which side chains of Arg, Asp, D-Asp and D-Val point to outside of the peptide ring. This might enhance the interaction of LXW7 with the αvβ3 integrin compared to peptide 4. The structure of peptide 4 was more compact than that of LXW7 (Figure 2). The bowl-like structure of LXW7 was twisted in peptide 4, which resulted in a drastic change of amino acids side chains orientation. The side chain of D-Val and Asp in peptide 4 was buried inside. The side chain of D-Asp approached to the peptide ring from the other side of the ring in comparison to LXW7. Therefore amino acids side chains cannot interact with αvβ3 integrin efficiently. These explained the complete loss of binding affinity of peptide 4 to K562/αvβ3+ cells.

Open in a separate window

Figure 2

Ribbon representation of the energy minimized average NMR structures of peptide 1 and peptide 4. Side-chain atoms of key residues are shown as sticks. Color code: green, carbon; blue, nitrogen; red, oxygen; yellow, sulfur.

Introducing bulky penicillamine (Pen or β, β-dimethylcysteine, the structure in Table 1C) can often increase peptide proteolytic stability. It is also often used in investigation of disulfide bridge role in peptide biological activity (22–24). D-Pen was adopted to replace D-Cys¹ and D-Cys⁸ singularly or simultaneously (peptide 5, peptide 6 and peptide 7, Table 1). However the replacements caused 4 to 10 times loss of binding affinity. This may be attributed to the constraint exerted by the bulky β dimethyl groups of the two D-Pen side chains. The above results indicated that disulfide bond formed between two D-Cys was required for maintaining the integrin-binding 3D conformation of LXW7.

Substitution of Gly²

The RGD containing peptide sequence, GRGDSP was initially reported by Pierschbacher et.al., which had better inhibition effect of cell attachment to fibronectin than the peptide RGDS, indicating the flanking sequence of RGD also affected peptide activity (25). Cyclo(GRGDSPA) was subsequently reported to be more effective at inhibiting cell attachment to vitronection than to fibronectin (26). For the design of cyclic RGD containing peptides, Gly or Ala was added as spacer flanking the RGD sequence. Our previous OBOC library screening effort for αvβ3 integrin ligands with 20% down-substituted library (cX₂RGDX₆X₇c) showed 100% preference for Gly at the X₂ position (21). In this study, replacement of Gly² of LXW7 with D-Ala yielded peptide 8 (D amino acids were used in this position for our previous library design) and led to a 7.8 time loss of binding affinity to K562/αvβ3+ cell (Table 1). Gly lacks side chain and it provides conformational flexibility to the peptide backbone and affords the appropriate spatial orientation, resulting in binding affinity enhancement (27, 28). Introducing D-Ala limits backbone flexibility and results in interruption of the proton correlations between α CH of Gly and NH of Arg (Figure 1). Interestingly, replacement of Gly² with β-alanine (peptide 9) also caused 11 times reduction of binding affinity to αvβ3-K562 cells (Table 1). The additional -CH₂- of β-Alanine changed the peptide ring size and might bring too much flexibility to the peptide backbone. Together, the above results indicate that Gly in the N-terminal side of the RGD motif is crucial in LXW7 for cell binding. Even slight change in this position is not acceptable.

Substitution of the RGD triad

The previous docking stimulation of LXW7 suggested that the orientation of the Arg side-chain with Asp-218 (in α-subunit of integrin αvβ3) might not be optimal (21). In order to clarify this point, and also to increase the peptide stability toward trypsin digestion at Arg side chain, we replaced Arg with three arginine homologues, HoArg, Agb, and Agp (structures in Table 2), to afford peptides 10–12 (Table 1). However, replacement of Arg with any of these three different arginine homologues resulted in a drastic loss of binding affinity. The optimization of Arg side chain interaction with integrin αvβ3 was not achieved using singular replacement of Arg. Once again, it indicates the dominant effect of RGD motif in LXW7 binding affinity to αvβ3 integrin. Therefore we did not make any further modification on Gly⁴ and Asp⁵.

Table 2

Structures of building blocks used at SAR study of peptide 1 (LXW7).

Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window
R1	R2	R3	R4	R5	R6	R7	R8
Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window
R9	R10	R11	R12	R13	R14	D-Pen	Sar
Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window	Open in a separate window
HoArg	Agb	Agp	DAbu	DBug	DAgl	DPra	β-Alanine

Open in a separate window

Substitution of D-Asp⁶

The previous OBOC screening study (21) showed that D-Asp and D-Glu appeared with higher frequency (63%) at X₆ position in the library, indicating that acidic residues are favorable at this position. The prior docking simulation study on LXW7 pointed out that D-Asp⁶ was able to form a salt bridge with Arg-214 in the β-subunit of integrin αvβ3. This additional interaction probably contributes to the specific binding of LXW7. Therefore we did not do further SAR study at this position. However in the highly focused OBOC library design for further optimization of LXW7, twenty-three acidic and polar amino acids were chosen as building blocks at the X₆ position (supplementary data Table S2).

Substitution of D-Val⁷

Various amino acids (hydrophobic amino acids, polar amino acids and negative charged amino acids) were found at D-Val position from the previous OBOC library screening effort (21). These results may be explained in two ways: first, D-Val position was not fully optimized and further optimization may increase binding affinity; second, D-Val position was not important for the peptide binding to cells, therefore diverse properties of amino acids appeared in this position. However our prior docking stimulation study of LXW7 indicated that side-chain of D-Val bound in an almost identical position as the D-Phe side-chain in cyclo(RGDfV). D-Val did contribute to the binding affinity of LXW7 to the integrin. Therefore we predicted that X₇ position is a potential site for LXW7 optimization. Based on D-Val structure, we introduced several D-form hydrophobic amino acids into this position, giving peptide 13–17 (Table 1). Replacement with D-Abu (LXW25, D-Abu structure in Table 2) has similar binding affinity with LXW7 (Table 1). The replacement with D-Bug (LXW18, D-Bug structure in Table 2) led to a slightly increased binding affinity (Table 1). Other three compounds (LXW16, 17 and 26, special amino acids structures in Table 2) have 4–5 times decrease in binding affinity (Table 1). The slight changes of amino acid structure at X₇ position caused variable binding affinity results. The above results indicated that it is a potential position for further optimization. Therefore, in the design of two highly focused OBOC libraries, we introduced seventy-one different natural and unnatural amino acids to this position (supplementary data Table S3 and S4).

Modification of the N-terminus

In previous studies, LXW7 binding to αvβ3 integrin was based on the structural feature that LXW7 had free N-terminus. The electrostatic interaction between integrin and basic N-terminus of peptide may play an important role in binding affinity. Since peptides with uncharged N-termini have increased cell permeability and more stable against aminopeptidase digestion, we explored the effect of N-capping on binding affinity. Fourteen analogues of LXW7 (peptide 18–31, structures of R₁ to R₁₄ in Table 2) were designed and synthesized (Table 1). Linear LXW7 was first constructed on Rink amide resin; then the resin was aliquot and reacted with various aliphatic and aromatic carboxylic acids. Finally cyclization was done in solution after peptide cleavage from resin with TFA cocktail. Capping of the N-terminus did not result in significant changes in binding affinity. Only 2 substitutions (LXW 77 and 78, Table 1) triggered a modest loss of potency. Six analogues (LXW80, 83, 84, 85, 86 and 89, Table1) have similar binding affinity as LXW7. The above results indicated that N-terminus doesn’t have any significant impact on the binding affinity. One major application of the integrin-binding ligand is to use it as vehicle for drug delivery. The well-known head-to-tail cyclic peptide cyclo(RGDfV), although has high binding affinity and specificity against αvβ3 integrin, lacks a handle for attachment of drug payload. Introducing a handle via replacement of D-Phe or Val by Lys or Glu, resulted in a 2–8 fold decrease in binding affinity (21). The LXW peptides such as LXW7 and LXW64, were discovered through OBOC combinatorial library screening utilizing an on-bead whole-cell binding assay. They all have a built-in C-terminus handle because these ligands were discovered while tethered to the bead. The result of the current N-capping experiment suggests that the N-terminus, if not capped, can also be used as a handle for attachment of drug payload. Therefore, unlike cyclo(RGDfV) which has no handle for tethering, the LXW peptides have two handles for functionalization with drugs and/or imaging agents.

N-methylation scan of LXW7

Incorporation of N-methylated amino acids into peptides changes its backbone conformation and increases peptide membrane permeability and proteolytic stability. It has been often used to affect peptide potency and selectivity to receptor subtypes (29, 30). N-methylation scan strategy was applied to LXW7, from X₂ position to X₇ position, resulting peptides 32–37 (Table 1). Their structures and physicochemical characteristics are given in Table 1 and supplementary data Table S1. Among N-methylated amino acids, D form of N-methylated aspartic acid was synthesized according to published method (31). N-methylated Arg and D-Val led to slight loss of binding affinity to αvβ3 integrin. The N-methylation of Gly² triggered a 4 folds loss of potency. This confirmed the importance of Gly² in LXW7, which has a big impact on the peptide potency. N-methylation at other position (peptide 34–36, table 1) caused drastic loss of binding affinity. The N-methylation scan did not yield any additional potent peptides in comparison to LXW7. Despite this, we included some N-methylated amino acids as building blocks in the focused OBOC libraries design in order to detect their combination effect with amino acids replacements at other positions.

Financial information: This work was supported by NIH Grants awarded to K.S. Lam: R01CA115483 and R01EB012569.

We thank Dr. Diana Lac for editorial assistance.