Many protein kinases contain two or more independent protein-binding domains. This includes the best-known example, Src, which contains SH1 (active site), SH2, and SH3 domains. Others, such as the LIM-kinase (LIM, PDZ, and active site regions) likewise contain multiple domains (Stanyon and Bernard 1999). The twin issues of potency and specificity can potentially be addressed by simultaneously targeting two protein interaction domains on the same enzyme. First, a single bivalent peptide that concomitantly interacts with two binding domains should display an enhanced affinity for the protein kinase target relative to either monovalent peptide alone. Second, although the sequence homology within closely related domains can be very high, the relative three-dimensional disposition of two protein interaction domains in the intact protein may differ substantially from one protein kinase to the next. Consequently, the linker that connects the two monovalent consensus sequences could play an important role in conveying enzyme selectivity. At this point in time, however, the potential of bivalent (or multi-valent) peptides that can interact with two or more protein interaction domains is at an early stage in development and has not yet been fully realized.
One of the first papers to describe the preparation of bivalent ligands for signaling proteins dealt with the SH-PTP2 protein tyrosine phosphatase (Pluskey et al. 1995). The latter contains two SH2 domains. Occupancy of these domains is known to stimulate phosphatase activity. Previous studies had identified peptides SLNpYIDLDLVK and LSTpYASINFQK that specifically target the N-terminal and C-terminal SH2 domains, respectively (Case et al. 1994; Sugimoto et al. 1994). These monovalent species were linked with one another via an aminocaproic acid bridge to create a consolidated derivative, LNpYIDLDLVK-(6-aminocaproic acid)4-LSTpYASINFQK, that can simultaneously interact with both SH2 domains. The latter does not act as an inhibitor, since it was designed to promote phosphatase activity. Nevertheless, it stimulates the phosphatase by 37-fold, compared with the 9- to 16-fold displayed by the monovalent consensus peptides. In addition, the heterodimeric peptide displays a 60- to 90-fold higher affinity for SH-PTP2 than either monomer peptide alone.
Cowburn, Barany, and their colleagues have described peptides that simultaneously target the SH2 and SH3 domains of Abl (Cowburn et al. 1995). The SH2-targeting sequence (PVpYENV-amide; KD=2.0±0.2 ^M) was attached to the SH3-targeting sequence (PPAYPPPPVP; KD=10.5±0.2 ^M) via a lysine side chain as illustrated in 25. The latter exhibits a KD of 249±5 nM for the SH2-SH3 domain construct, which is approximately 10- to 80-fold greater than that of individual monovalent consensus peptides. These investigators subsequently prepared a series of bivalent ligands containing the individual monomers in the four possible orientations illustrated in Scheme 1 25A-D (Xu et al. 1999). The highest affinity ligand, 25C, exhibits a KD of 190 nM for the Abl SH2-SH3 (where LINKER=Gly7K).
Roques, Garbay, and coworkers prepared peptides that interact with the two SH3 domains of Grb2 (Cussac et al. 1999). As with the bivalent analogs described in this section, the strategy employed coupling two monomer units via a linker. Two SH3-binding monomer units, Val-Pro-Pro-Val-Pro-Pro-Arg-Arg-Arg, were attached to each other via their C-termini using the lysine-based strategy described by Cowburn and Barany. The bivalent species exhibits an affinity (KD=40 nM) that is two to three orders of magnitude greater than that exhibited by the corresponding monomer (N-terminal SH3 domain: KD=2.6 |iM; C-terminal SH3 domain: KD=40 |iM). These investigators demonstrated that the peptide dimer blocks Grb2-Sos complex formation in cell lysates, can selectively pull down Grb2 from lysate versus other double SH3 domain-containing proteins, and interferes with neurite formation in nerve growth factor-treated PC12 cells.
Miller and his coworkers prepared a series of SH2 domain-assisted substrates for the Abl tyrosine kinase (Pellicena et al. 1998). These investigators found that the presence of an SH2 domain binding sequence appended to the active site-directed sequence enhances substrate efficacy, specifically via a tenfold reduction in Km. For example, a peptide of the general structure (active site-directed peptide)-(SH2-directed peptide) displays a Km of 69±11 |iM and a Vmax of 3.0±0.1 |imol/min-mg. The key residue in the SH2-directed component is phosphoTyr, which is essential for SH2 recognition. When a Phe replaces this critical residue, the peptide displays a significantly larger Km of 680±90 |iM, yet a Vmax (3.0±0.1 |imol/min-mg) that is essentially unaltered. A peptide containing an inversed orientation, namely (SH2-di-rected peptide)-(active site-directed peptide), likewise displays a relatively low Km value (72±5 |iM), which is presumably a reflection of enhanced affinity via coordination to the SH2 domain.
Lawrence and his collaborators employed the multidomain targeting approach to create combined active site/SH2 domain-directed inhibitors of Src (Profit et al. 1999; Profit et al. 2001). Peptide-based inhibitors of tyrosine kinases must confront at least two challenges. First, peptides that target the ac tive site tend to be exceedingly poor inhibitory agents. Second, although peptides that bind to the SH2 domain are of reasonably high affinity (~low |iM) they have the unintended consequence of activating the kinase. Indeed, by analogy, Shoelson's and Walsh's bivalent ligand, which binds to the two SH2 domains of SH-PTP2, dramatically enhances phosphatase activity. However, an inhibitor that combines SH2 and active site-binding properties should simultaneously display enhanced affinity and inhibitory potency. In addition, bivalent analogs can furnish an assessment of the distance and spatial relationship between the protein interaction domains on the protein under evaluation. Indeed, the Lawrence team referred to their small library (16 compounds) as "molecular rulers." In a manner analogous to Barany and Cowburn, a series of differentially oriented SH2-directed and active site-directed bivalent ligands were prepared. g-Aminobutyric acid (GABA) residues were employed in the linker region to connect the monovalent ligands. The active site-directed fragment, -Glu-Glu-Leu-Leu-(F5Phe)-, contains pentafluorophenylalanine, which had been previously identified as a nonphosphorylatable tyrosine surrogate (vide supra) (Niu and Lawrence 1997a). The SH2 sequence, -pTyr-Glu-Glu-Ile-, was based on the well-known preference of the SH2 domain from Src for a sequence motif containing pTyr followed by at hydrophobic amino acid at the P+3 position (Songyang et al. 1993).
Of the four possible relative orientations between the SH2- and active site-directed fragments (Scheme 2), only three were prepared (A-C) since the fourth (D) was subsequently ruled out as suboptimal (vide infra). The orientation illustrated in C requires that the peptide chains in the two monovalent units run anti-parallel to one another. This reversal of chain polarity was achieved via insertion of a glutaric anhydride (Ga) residue between the SH2-directed sequence and the GABA linker. Orientation B furnished the most effective inhibitors possessing the shortest linkers (optimal: GABAn= GABA3). In this particular case, the tyrosine surrogate F5Phe (occupying the active site) lies only three residues from the pTyr moiety which resides in the SH2 domain (see 26B).
This suggests that the active site region and the SH2 domain are situated close to one another in the active form of the enzyme. The series 26D was not prepared since, even in the absence of a GABA linker, the key F5Phe and pTyr moieties are positioned at least five residues apart. The latter is suboptimal relative to the relationship in 26B. The lead bivalent analog 26B (where GABAn=GABA3) displays an enhanced 120-fold inhibitory potency relative to the simple active site-directed monomer (IC50=13±1 |M versus IC50= 1590±170 |M, respectively).
Was this article helpful?