Designer protein delivery: From natural to engineered affinity-controlled release systems

Science  18 Mar 2016:
Vol. 351, Issue 6279,
DOI: 10.1126/science.aac4750

You are currently viewing the abstract.

View Full Text

Temporarily caught in a bind

When an atom or molecular fragment is covalently bonded, its release is controlled by the rate of degradation of the bond or by the rate of degradation of the matrix or shell material that keeps it trapped. When molecules are held by noncovalent interactions, the affinity of the molecule to its binding site can be tuned by a number of parameters, such as changing pH, temperature, or salt concentration. This makes it possible to finely tune the release profile. Pakulska et al. review our increased understanding of these interactions as found in nature, and their increased use for the delivery of molecular agents and therapeutics.

Science, this issue p. 10.1126/science.aac4750

Structured Abstract


Protein therapeutics constitute a multibillion-dollar market, yet their formulation and sustained delivery still pose a substantial challenge. Controlled release strategies developed for small-molecule drugs, such as microparticle encapsulation, typically involve organic solvents and harsh processing conditions that are detrimental to protein structure and function. Affinity-controlled release has emerged as an alternative strategy for the sustained and tunable release of protein therapeutics in a neutral aqueous environment, thus reducing protein loss and improving loading.

Affinity-controlled release depends on a preferred noncovalent interaction between a protein therapeutic and a binding ligand. This binding ligand can be another protein, a peptide, or an oligonucleotide. Typically, the binding ligand is covalently linked to a polymer matrix, such as a hydrogel. Soluble protein is added, and equilibrium is established between free protein and ligand-bound protein. Whereas free protein is able to diffuse from the system, bound protein cannot. This equilibrium is dynamic and changes in response to local conditions. The rate of protein release from the system is therefore governed not only by protein diffusivity and the concentration gradient, but also by the concentration of the binding ligand, the strength of the interaction, and the binding kinetics. The challenge lies in finding binding ligands that afford the desired release profiles.


The earliest affinity-controlled release systems mimicked the extracellular matrix by using heparin to reversibly bind and control the release of various growth factors. Other natural interactions have since been used for affinity-controlled release, including albumin with small-molecule therapeutics and antibodies with cognate antigens. These systems have allowed for sustained release of protein therapeutics while maintaining protein activity; however, naturally occurring interactions are inherently limited in terms of available targets and binding strengths. In vitro selection and directed evolution are established techniques for isolation and engineering of binding partners against virtually any protein target. Harnessing these techniques for affinity-controlled release applications is now underway and has resulted in novel peptide-, protein-, and oligonucleotide-based binders for the sustained release of several growth factors.


Many opportunities exist for the discovery or design of binding ligands for affinity-controlled release. Computational techniques can help to identify protein backbones that have geometric and electrostatic complementarity to a target, reducing the screening required to isolate lead variants. Selection conditions can be tailored to isolate intermediate-strength binders, or iterative rounds of in vitro evolution can provide a series of related variants with a spectrum of affinities for a target. Competition selections can ensure selectivity for simultaneous yet independent release of multiple proteins from their corresponding binding ligands. On-demand affinity-controlled release has yet to be explored, but structure-switching aptamers and computational design of allosteric regulator sites show potential. These techniques, coupled with concurrent advances in accurate high-throughput measurement of binding constants, will allow for the creation of libraries of binding partners with various affinities for each target therapeutic. Such a standardized yet versatile controlled release strategy has the potential to improve reproducibility and accelerate optimization of protein delivery systems.

Discovery of binding ligands for affinity-controlled release.

Ligands (peptides, proteins, or oligonucleotides; shown as polygons) that bind a protein therapeutic (triangles) through noncovalent interactions (ionic, hydrophobic, van der Waals, or hydrogen bonding; indicated in bold) can be discovered in nature, selected from a library, derived through in vitro evolution, or designed computationally. By choosing ligands with different affinities, represented by the size of the equilibrium arrows (bottom), and covalently linking them to a polymer matrix such as a hydrogel, one can control the diffusive release (dashed arrows) of a protein therapeutic.


Exploiting binding affinities between molecules is an established practice in many fields, including biochemical separations, diagnostics, and drug development; however, using these affinities to control biomolecule release is a more recent strategy. Affinity-controlled release takes advantage of the reversible nature of noncovalent interactions between a therapeutic protein and a binding partner to slow the diffusive release of the protein from a vehicle. This process, in contrast to degradation-controlled sustained-release formulations such as poly(lactic-co-glycolic acid) microspheres, is controlled through the strength of the binding interaction, the binding kinetics, and the concentration of binding partners. In the context of affinity-controlled release—and specifically the discovery or design of binding partners—we review advances in in vitro selection and directed evolution of proteins, peptides, and oligonucleotides (aptamers), aided by computational design.

View Full Text