Tethering small molecules to a phage display library: discovery of a selective bivalent inhibitor of protein kinase A.

There is much current interest in developing fragment-based strategies for targeting protein surfaces and active sites.1 Toward this goal, we describe a new fragment-based bivalent ligand selection methodology that allows for the discovery of protein surface-targeted cyclic peptides that are steered by an active site binding small-molecule ligand (Figure 1). This approach allows for coupling high-affinity but promiscuous ligands to the large chemical space encompassed by biological libraries. Here, we demonstrate the feasibility of this approach by converting the nonselective kinase inhibitor, staurosporine, into a selective inhibitor of the cAMP-dependent protein kinase (PKA). With over 500 members in the human genome, protein kinases comprise an important class of enzymes, and their deregulation is often implicated in various disease states.2 Thus, selective inhibition of protein kinases has the potential to provide useful reagents and therapeutic leads.3 Most small-molecule inhibitors of kinases target the conserved ATP binding pocket, often displaying high affinity. While this has produced therapeutically useful inhibitors, they are often promiscuous.4 An elegant strategy for increasing affinity and selectivity is the development of bisubstrate analogue inhibitors,5a where a peptide analogue of the kinase’s protein substrate is covalently attached to an ATP-competitive small molecule.5b-f Cole and co-workers demonstrated the feasibility of this approach against the insulin receptor protein tyrosine kinase (IRK).5c In another example, Schepartz and co-workers5f improved selectivity by utilizing structurally constrained peptide substrates conjugated to the staurosporine-like natural product, K252a. Though elegant, a potential drawback of the aforementioned bisubstrate design approaches is that they necessarily rely on structural information as well as prior knowledge of a specific peptide substrate for the targeted kinase. Many kinases of biological or pharmacological interest do not satisfy the above criteria and therefore lie outside of the scope of these design strategies. We envisioned that an in vitro bivalent selection approach that combined small-molecule targeting with biological selection would have the potential to complement design strategies and provide selective bivalent inhibitors without prior structural or peptide substrate information. Two previous selection efforts have coupled large biologically derived peptide libraries to synthetic molecules utilizing the orthogonal chemistry of cysteine residues.6 This chemistry, though very promising, can have low yields and potentially suffers from nonspecific reactivity in the presence of the library platform (such as phage particles or ribosomes) as well as cysteine residues within the library being interrogated. With this in mind, we describe a noncovalent tethering strategy that would select for cyclic peptides from a phage display library7 directed toward the active site of a protein kinase by an ATP-competitive small molecule. The small molecule would be functionally tethered to the library by noncovalent self-assembly mediated by a coiled-coil heterodimer that has been previously utilized for numerous designs8 (Figure 1). This warhead-guided phage display selection would potentially afford cyclic peptide motifs with affinity for the kinase surface adjacent to the active site. The selected peptides could be subsequently conjugated to the small-molecule warhead to afford a bivalent inhibitor with the potential for increased affinity and enhanced selectivity for the targeted kinase. To test this new approach, we chose the much studied ATPcompetitive kinase inhibitor, staurosporine 1, as our promiscuous kinase targeting warhead. The carboxylated derivative of staurosporine 2 (Figure 2A) with a reported IC50 value 47-fold weaker than staurosporine9 1 was tethered to Jun (Figure 1B) and retained its ability to inhibit PKA (Supporting Information, Figure S1). Next, a phage-displayed peptide library7 was constructed adjacent to the Fos domain, which heterodimerizes with Jun. The library incorporated two conserved Cys residues flanking the six positions of diversity, each incorporating all 20 amino acids (Figure 1C).7b With the library in hand, selection was carried out against PKA. In each round of selection, the 2-Jun conjugate was mixed with the phage-displayed library conjugated to Fos and exposed to immobilized PKA. After six rounds of selection, several peptides were isolated and characterized for inhibiting PKA. The cyclic peptide CTFRVFGC was the most abundant sequence identified and was also found to be the most active in initial assays against PKA, with IC50 values in the mid-micromolar regime (Supporting Information, Figure S5). The successful selection of the cyclic peptides highlights the important distinction from previous approaches in that our bivalent tethering approach required no prior knowledge of the structure of the kinase or the chemical structure of the protein substrate.5 Having established that active-site-directed selection was indeed possible, we turned toward testing our central hypothesis: that our strategy could be used to discover covalently linked bivalent Figure 1. Bivalent inhibitor selection strategy: (A) Noncovalent tethering of staurosporine with a phage-display peptide library through a coiled-coil heterodimer for targeting a kinase, PKA. The selected peptide can be subsequently conjugated to staurosporine to provide a bivalent kinase inhibitor. (B) Staurosporine derivative 2 conjugated to the Jun. (C) Phagedisplayed cyclic peptide library attached to Fos (X ) any of the 20 natural amino acids). (D) Selected cyclic peptide 3. Published on Web 10/18/2007