PROTEIN POLYMERS Gene libraries open up

Protein polymers — engineered repetitive polypeptides, up to 500 or more amino acids in length — are a uniquely tunable family of functional biomaterials. They can mimic aspects of natural protein structure and function, or be designed de novo to solve problems that nature never contemplated^1–3. With 20 years of work, the palette of useable non-natural amino acids has grown in breadth and nuance, and the methods of production improved. Still, the number of biomolecular engineering research groups tinkering with protein polymers is relatively small. The spectrum of innovative users must broaden if protein polymers are to fulfil their considerable potential.

Design and discovery of useful protein polymers has been, until now, labour intensive and slow. Sequence domains (~30 amino acids or fewer) from proteins such as elastin are `polymerized' via genetic engineering. Tedious, low-yielding cloning work has been unavoidable in a quest for long, synthetic genes that Escherichia coli expresses well.

This is about to change. Writing in Nature Materials, Chilkoti and colleagues report a new way to achieve the rapid, parallel synthesis of long, repetitive genes, which they developed through a clever amalgamation of prior gene cloning and amplification methods. Obtaining new genes that encode long protein polymers (>500 amino acids) is made easier, faster and more reliable. In the present study, this method is demonstrated for the discovery of thermoresponsive proteins, as well as pharmaceutically relevant glucagon-like peptide-1 (GLP-1) polymers.

Protein polymer synthesis requires the generation of repeating DNA sequences encoding proteins of different lengths. A few cloning methods were previously used to achieve this goal. Gene concatemerization, although widely used and straightforward, is inefficient by virtue of the required enzymatic digestion and ligation steps, and offers low utility for the parallel creation of large gene libraries (Fig. 1a). A more efficient method, overlap-extension polymerase chain reaction (OEPCR), replaces ligation-based gene multimerization with just one PCR step (Fig. 1b). Although amenable to high-throughput projects, OEPCR has the disadvantage of low fidelity as a result of non-specific self-priming of short gene sequences, and denies access to the long genes that are often ideal for biomaterials⁴.

Figure 1.

A schematic of strategies to synthesize oligomer genes. a, Concatemerization. The double-stranded monomer genes are generated either by enzyme digestion from original plasmid or by annealing of two complementary single-stranded DNA fragments. Then, monomer fragments are ligated to form oligomers of different lengths. Type IIS restriction endonuclease can be used to eliminate sequence constraints and introduce direction control in oligomization. b, Overlap-extension PCR (OEPCR). Two ssDNA fragments anneal to each other through their overlapping regions, and are extended to longer sizes during chain extension steps. c, Rolling circle amplification (RCA). Cyclized DNA is used as the template in RCA. During the elongation cycle the polymerase rolls over the template and extends continuously to yield linear complementary single-stranded DNA fragments of various lengths.

The highly parallel method for gene synthesis reported by Chilkoti and colleagues, called `overlap-extension rolling circle amplification (OERCA)', combines RCA (Fig. 1c) with OEPCR, in one reaction tube<sup>5</sup>. Fast, high-throughput gene oligomerization can be accomplished with excellent retention of sequence fidelity. An RCA step replicates a circularized, single-stranded DNA (ssDNA) template, providing a library of repetitive nucleic acid polymers. Next, a linear gene-extension step creates complementary DNA fragments. Subsequent thermal cycling uses both forward and reverse primers. In later extension steps, ssDNA molecules with longer overlapping regions initiate overlap-extension reactions, thus generating long, repetitive DNA oligomers with high sequence fidelity. The size distribution of oligomers is tuned by varying the primer-to-template ratio or the thermal cycling protocol. Chilkoti and colleagues show that this approach generates diverse gene libraries, in a single reaction, by creating genes as long as 2,500 base pairs (bp) from 18-bp `monomers'. Prior cloning methods used in protein polymer production yield shorter genes, often with higher error rates.

Elastin-like protein polymers (ELPs) are of interest for their emergent biomaterial properties and especially for their stimuli-responsiveness. Depending on amino acid sequence and length, an ELP is soluble in aqueous media below a certain temperature, but partitions into a water-insoluble `coacervate' phase at higher temperatures. The ELPs are based on a pentameric consensus sequence, Val-Pro-Gly-Xaa-Gly, where Xaa is any amino acid, and have been engineered for a variety of applications. Amino acid sequence and total chain length affect the phase-transition temperature, as well as protein biodegradability⁶. The highly repetitive, guanine/cytosine-rich sequences of ELP-encoding genes have made it hard to synthesize ELPs with widely ranging compositions and lengths, hampering exploration of the relationships between phase-transition behaviour, protein sequence, chain length and coacervate secondary and tertiary structure. Using OERCA, Chilkoti and colleagues show the synthesis of an ELP gene library with large breadths of sequences and lengths. Using this library, they were able to investigate how variations of the Val-Pro-Gly-Xaa-Gly sequence motif affect ELP coacervation behaviour. Alanine insertion and substitution mutants of poly(Val-Pro-Gly-Xaa-Gly) sequences yielded proteins capable of new types of reversible stimuli-responsive phase behaviour. They compare the biophysical properties of the newly discovered variant ELPs with previously studied canonical ELPs, gaining novel insights into the influence of hydrophobic interactions on ELP phase-transition behaviour, which suggest that unordered and dynamic conformations of ELPs are the key to thermal reversibility. More thorough investigation of the sequence-related properties of ELPs should enable the creation of materials and hydrogels with easily tuned, fully reversible stimuli-responses, sharper phase transitions, and controlled biophysical structures.

To demonstrate the versatility of OERCA in generating repetitive genes, Chilkoti and colleagues synthesized another gene with a markedly different motif length and composition. By engineering proteins with weak, intermediate or strong thrombin protease cleavage sequences embedded between the repeating peptide units, a library of protease-sensitive polymers based on GLP-1 variants of differing potencies was generated and studied. Molecules of this type are under study for the treatment for type-2 diabetes, and may provide better control over blood glucose through slow in vivo release of GLP-1 peptides as the polymers are degraded by thrombin. The OERCA method could facilitate the screening of these and other pharmaceutically relevant protein polymers, and aid in the optimization of clinical efficacy by accessing a wide diversity of sequences and lengths.

Combining OEPCR and RCA to achieve rapid, highly parallel and well-controlled synthesis of highly repetitive genes is a significant advance in the preparation of protein polymers. Although traditional step-wise cloning strategies^7,8 are useful in the cases where precise control of protein polymer block location and distribution are required to achieve desired functions, OERCA offers advantages for screening new protein families. All cloning methods display some degree of sequence-dependence, however — the true robustness of OERCA, and its advantages and drawbacks, must be further explored with the synthesis of protein polymers that are based on a broader array of sequence motifs, which, thankfully, will be a relatively easy task through the use of this new method.