Optimized serum stability and specificity of an αvβ6 integrin-binding peptide for tumor targeting

Abstract

The integrin αvβ6 is an antigen expressed at low levels in healthy tissue but upregulated during tumorigenesis, which makes it a promising target for cancer imaging and therapy. A20FMDV2 is a 20-mer peptide derived from the foot-and-mouth disease virus that exhibits nanomolar and selective affinity for αvβ6 versus other integrins. Despite this selectivity, A20FMDV2 has had limited success in imaging and treating αvβ6⁺ tumors in vivo because of its poor serum stability. Here, we explore the cyclization and modification of the A20FMDV2 peptide to improve its serum stability without sacrificing its affinity and specificity for αvβ6. Using cysteine amino acid substitutions and cyclization by perfluoroarylation with decafluorobiphenyl, we synthesized six cyclized A20FMDV2 variants and discovered that two retained binding to αvβ6 with modestly improved serum stability. Further d-amino acid substitutions and C-terminal sequence optimization outside the cyclized region greatly prolonged peptide serum stability without reducing binding affinity. While the cyclized A20FMDV2 variants exhibited increased nonspecific integrin binding compared with the original peptide, additional modifications with the non-natural amino acids citrulline, hydroxyproline, and d-alanine were found to restore binding specificity, with some modifications leading to greater αvβ6 integrin selectivity than the original A20FMDV2 peptide. The peptide modifications detailed herein greatly improve the potential of utilizing A20FMDV2 to target αvβ6 in vivo, expanding opportunities for cancer targeting and therapy.

Keywords: A20FMDV2; amino acid substitution; cancer therapy; chemical modification; cyclization; enzymatic stability; integrin; molecular imaging; peptides; αvβ6.

Figure 1

Site-specific cyclization of A20FMDV2 via perfluoroarylation retains peptide binding to αvβ6⁺cancer cells.A, 3D model of A20FMDV2 peptide predicted by PEP-FOLD3 computational framework (68). Amino acids are listed, and positions substituted with cysteines for cyclization are shown in red. B, chemical structure of decafluorobiphenyl (DFBP) molecular linker used for cyclization. C, amino-acid sequences of A20FMDV2 peptide and DFBP-cyclized variants. Cysteine substitutions for cyclization by perfluoroarylation are shown in red. The RGDLXXL motif that is important for αvβ6 recognition is underlined in all sequences. D, flow cytometry binding curves of A20FMDV2 peptide and DFBP-cyclized variants to K562 and K562 αvβ6:mCherry cells, normalized to 400 nM A20FMDV2 binding to K562 αvβ6:mCherry cells. The curves represent a nonlinear regression of one independent experiment in which binding data are fitted to a Hill equation. K_D values are not shown here and will be reported for promising peptides in a later figure with triplicate datasets. SA-AF647, streptavidin Alexa Fluor 647.

Figure 2

Partial decafluorobiphenyl (DFBP)-cyclized A20FMDV2 candidates display moderate serum stability with exocyclic C-terminal degradation.A–C, MALDI-TOF spectra of DFBP-cyclized C1C18, C1C20, and C2C18 peptide variants incubated in normal mouse serum for 0, 2, 4, and 6 h at 37 °C. Molecular weights of prominent peaks are shown. No peptide-fragment peaks were observed for C1C20 DFBP at the 4 and 6 h time points. Bottom: predicted amino acid sequences of degradation products based on measured molecular weights.

Figure 3

Further modifications to decafluorobiphenyl (DFBP)-cyclized A20FMDV2 peptides do not impact binding αvβ6⁺cancer cells.A, schematic of modifications made to the sequences of C1C18 DFBP and C2C18 DFBP to further improve their serum stability. Chemical structures of the original (black) and modified (blue) amino acids are shown at positions of incorporation. The resulting modified peptide sequences are also listed, with cysteine substitutions for DFBP cyclization shown in red and amino acid modifications shown in blue. The RGDLXXL motif that is important for αvβ6 recognition is underlined in all sequences. B, flow cytometry binding curves of A20FMDV2, C1C18 DFBP, C2C18 DFBP, and additionally modified peptides to K562 and K562 αvβ6:mCherry cells, normalized to 400 nM A20FMDV2 binding to K562 αvβ6:mCherry cells. The curves represent a nonlinear regression of at least three independent experiments in which binding data are fitted to a Hill equation. K_D values were calculated by averaging the individual regression values of the independent experiments. Data points, error bars, and K_D values represent the mean ± SD; n = 3 to 8 independent experiments. K_D values of cyclized and modified peptides were not statistically different than that of the original peptide (p > 0.05, one-way ANOVA with Dunnett's test) and each other (p > 0.05, one-way ANOVA with Tukey's test). SA-AF647, streptavidin Alexa Fluor 647.

Figure 4

Further modified decafluorobiphenyl (DFBP)-cyclized A20FMDV2 peptides exhibit prolonged serum stability with slow internal arginine cleavage.A–D, MALDI-TOF spectra of DFBP-cyclized C2C18 R_DTKA_D, C2C18 CitTKA_D, C2C18 P_h R_DTKA_D, and C1C18 A_D R_DTK_DA_D peptides incubated in normal mouse serum for 0, 4, 8, 12, and 24 h at 37 °C. Molecular weights of prominent peaks are shown. Bottom: predicted amino acid sequences of degradation products based on measured molecular weights. For some molecular weights, multiple predictions are listed.

Figure 5

Further modified decafluorobiphenyl (DFBP)-cyclized A20FMDV2 peptides have long and comparable serum half-lives. Stability of DFBP-cyclized C2C18 R_DTKA_D, C2C18 CitTKA_D, C2C18 P_h R_DTKA_D, and C1C18 A_D R_DTK_DA_D peptides over a 24-h incubation in normal mouse serum, as measured by LC–MS. Values are normalized to the 0 h time point. Curves represent a nonlinear regression of one independent experiment assuming one-phase exponential decay.

Figure 6

Citrulline, hydroxyproline, andd-alanine substitutions reduced nonspecific integrin binding of decafluorobiphenyl (DFBP)-cyclized A20FMDV2 peptides. Nonspecific binding of DFBP-cyclized and modified peptides to A375P cells by flow cytometry, normalized to 125, 250, 500, and 1000 nM A20FMDV2 binding. Values that fall above the dotted red line represent increased nonspecific binding compared with the original A2FMDV2 peptide, whereas those that fall below represent decreased nonspecific binding. The green dots represent the data from individual experiments. Columns and error bars represent the mean ± SD; n = 3 independent experiments. For statistical testing, values from independent experiments and all four concentrations were pooled together for each peptide before comparison to the original peptide (n = 12, p > 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, paired one-way ANOVA with Dunnett's test). SA-AF647, streptavidin Alexa Fluor 647.