Discovery and design of self-assembling peptides

Abstract

Peptides are ubiquitous in nature and useful in many fields, from agriculture as pesticides, in medicine as antibacterial and antifungal drugs founded in the innate immune systems, to medicinal chemistry as hormones. However, the concept of peptides as materials was not recognized until 1990 when a self-assembling peptide as a repeating segment in a yeast protein was serendipitously discovered. Peptide materials are so called because they have bona fide materials property and are made from simple amino acids with well-ordered nanostructures under physiological conditions. These structures include well-ordered nanofibres, nanotubes and nanovesicles. These peptide materials have been used for: (i) three-dimensional tissue cell cultures of primary cells and stem cells, (ii) three-dimensional tissue printing, (iii) sustained releases of small molecules, growth factors, monoclonal antibody and siRNA, (iv) accelerated wound healing in reparative and regenerative medicine as well as tissue engineering, (v) used to stabilize membrane proteins including difficult G-protein coupled receptors and photosystem I for designing nanobiodevices, (vi) a few self-assembling peptides have been used in human clinical trials for accelerated wound healings in surgical uses and (vii) in human clinical trials for siRNA delivery for treatment of cancers. It is likely that these self-assembling peptides will open doors for more and more diverse uses. The field of self-assembling peptides is growing in a number of directions in areas of materials, synthetic biology, and clinical medicine and beyond.

Keywords: materials; peptides; self-assembling; sustained molecular releases; three-dimensional tissue cell culture.

Figure B1.

One of the Crete workshops on self-assembling peptides and proteins. I initiated and co-organized five workshops of ‘Self-assembly peptides and proteins’ in Capsis Hotel, Crete, Greece in summers of 1999, 2001, 2003, 2005 and passed the torch to Bill DeGrado and Joel Schneider in 2007. We invited people from diverse fields including peptide chemistry, protein science, materials science, mathematics, engineering, computer science, clinical medicine and more. The invited speakers include Alexander Rich, Nobel laureate Carleton Gajdusek, Mouad Lamrani, Benoit Mandelbrot, future Nobel laureate Martin Karplus, Alan Fersht, Chris Dobson, Susan Lindquist, Carol Robinson, Meir Wilchek, Carl Brädén, David Eisenberg, Joel Sussman, George Klein, Eva Klein, Peter Klein, Andrew Szent-Györgyi, Ingemar Ernberg, Xiaojun Zhao, Horst Vogel, Uwe Sleytr, Jeff Kelly, Bill DeGrado, Jonathan Weissman, Michael Hecht, Joseph Jacobson, Matthew Tirrell, Reza Ghadiri, Miheal Polymeropoulos, Alan Windle, Amalia Aggeli, Neville Boden, Anton Middleberg, Sheena Radford, Tim Deming, Vince Conticello, Stefan Wölfl, Anna Mitraki, Aphrodite Kapurniotu, Dek Woolfson, Tom McLeish, Ted Atkins, Alan Cooper, Nick Gay, Jianren Lu, Ehud Gazit, Sam Stupp, Joanna Aizenberg, Bruce Tidor, Phil Messersmith, Rutledge Ellis-Behnke, Amy Keating, Roger Kamm, Angie Belcher, Per Westmark, Gunilla Westermark, Joel Schnur, Yechiel Shai, Hilal Lashuel, Sylvie Blondelle, Donald Hilvert, Thomas Scheibel, David Kaplan, David Lynn, Dan Urry, Charlotte Hauser, Matt Tirrell, Dan Kirschner, Joel Schneider, Michal Dadlez, Lars Baltzer, Sandra Burkett, Kenji Yamamoto, Hisakazu Mihara, Shiroh Futaki, Tomi Sasaki, Itaru Hamachi, Susumu Yoshikawa, Margherita Morpurgo, Diego Mantovani, Georgios Archontis, Peter Butko, Yuan Luo, Eser Elcin, Murat Elcin, Thomas Zemb and Jack Aviv as well as many young and energetic students and postdocs including Rein Ulijn, Louise Serpell, Cait McPhee, Mark Krebs, Jesús Zurdo, Dominik Rünzler, Meital Reches, Brian Chow, Sawyer Fuller, Wonmuk Hwang, Steve Yang, Sloan Kulper, Davide Marini, Patrick Keilly, Andrea Lomander, Hidenori Yokoi, Beau Peelle, Andreas Mershin, Brian Cook, Liselotte Kasier and many more. These students and postdocs presented their research in front of the distinguished scientists. They exchanged ideas and received sound advice. These intimate workshops (approx. 70–80 people each time) mixed people from different disciplines and stimulated further interests in pursuing the new field. This self-assembling peptide material field is not only thriving and advancing at rapid pace but it also generates many novel products that have been commercialized in diverse fields.

Figure B2.

The timeline of developing process and diverse applications of designer chiral self-assembling peptides. It spans over 20 years from the curiosity-driven research and serendipitous discovery of the first self-assembling peptide EAK16-II in 1990 to successful human clinical trials in 2011. In 1993, shortly after I discovered the self-assembling peptides, I met Marvin Caruthers at a conference and mentioned to him about my discovery. I was concerned about the expensive peptide materials for applications. He gave me a sound advice: ‘Find as many applications as possible, the economics will take care of itself’. Marvin Caruthers invented the phosphate amide nucleic acid synthesis chemistry, and now all nucleic acid synthesis uses his invention. In the 1980s, each nucleotide base cost $25, and to make 10 base oligonucleotides cost $250 in an academic laboratory! Now each base costs $0.05 and 10 base oligonucleotides cost $0.50. It is a reduction by 500 fold!.

Figure B3.

The timeline of self-assembling peptides discovery and development: from a serendipitous discovery to benefit of society. It took over 20 years from the initial discovery in 1990 to successful human clinical trials in 2011. There have been a lot of ups and downs. After gaining full understanding and knowledge of various detailed aspects of amino acid chemistry, peptide structure properties, dynamic molecular self-assembly behaviours of the self-assembling peptides, we started to design new peptides with active biological functions that further enhance their usefulness for a wide range of applications. Seizing the opportunity, I took the license and co-founded a startup biotech company 3DMatrix to translate the technology into new economy because a multi-billion-dollar US polymer company failed to move it forward, 6 years after it took the license and paid over a million dollars licensing fee. Currently, the self-assembling peptide nanofibre scaffold hydrogel has been used for surgical applications, slow releases, three-dimensional tissue printing and accelerated wound healing. To sum up the path from the discovery to a new medical technology: (i) to stimulate, encourage and support curiosity-driven scientific research in order to gain scientific knowledge, (ii) to take risk and not be afraid of failure to translate scientific knowledge and research into enabling technologies, and (iii) be undeterred and be persistent to develop the knowledge-based economy for the benefit of mankind.

Figure 1.

The simple molecular models of the designer amphiphilic self-assembling peptides. These peptides have two distinctive sides, one hydrophobic and the other hydrophilic. The hydrophobic side forms a double sheet inside of the fibre and the hydrophilic side forms the outside of the nanofibres that interact with water molecules, forming an extremely high water content hydrogel that contains as high as 99.9% water. At least three types of molecules can be made, with −, +, −/+ on the hydrophilic side. The individual self-assembling peptide molecules are approximately 6 nm long. The first such peptide, EAK16-II, was discovered from a yeast protein, zuotin [–2]. This peptide inspired us to design a large class of self-assembling peptide designer motifs. When they are dissolved in water in the presence of salt, they spontaneously assemble into well-ordered nanofibres and then further into nanofibre scaffold hydrogels. The right panel shows the EM image of yeast cells. Yeast cells are approximately 5–10 µm in size depending their cell phases, resting, growing and dividing phases; resting phase, cells are smaller and in the growing and dividing phases, cells are larger.

Figure 2.

Self-assembling peptide RADA16-I nanofibre scaffold hydrogel. (Upper panel) Amino acid sequence of RADA16-I, molecular model of a single RADA16-I peptide, the calculated peptide dimensions are approximately 6 nm long depending on end capping, 1.3 nm wide and 0.8 nm thick; hundreds of thousands or millions of individual peptides self-assemble into a nanofibre depending on the fibre length as revealed by the SEM image of RADA16-I nanofibre scaffold. Note the scale bar, 0.5 μm or 500 nm (SEM image courtesy of Fabrizio Gelain). RADA16-I peptide forms nanofibres in aqueous solution that further form hydrogel with extremely high water content (99.5–99.9% w/v water).

Figure 3.

Peptide RADA16-I. (a) Amino acid sequence and molecular model of RADA16-I, the dimensions are approximately 5 nm long, 1.3 nm wide and 0.8 nm thick. Atomic force microscopy images of RADA16-I nanofibre scaffold are shown here (b–d). The sizes of the entire images are: (b) 8 µm², (c) 2 µm² and (d) 0.5 µm². Note the different height of the nanofibre, approximately 1.3 nm in (d) thus suggesting a double layer structure. Photographs of RADA16-I hydrogel at various conditions: (e) 0.5 wt% (pH 7.5), (f) 0.1 wt% (pH 7.5, Tris–HCl), (g) 0.1 wt% (pH 7.5, PBS) before sonication and (h) re-assembled RADA16-I hydrogel after four times of sonication (image courtesy of Hidenori Yokoi) [13].

Figure 4.

AFM images of RADA16-I nanofibre at various time points after sonication. The observations were made using AFM immediately after sample preparation. (a) 1 min after sonication, (b) 2 min, (c) 4 min, (d) 8 min, (e) 16 min, (f) 32 min, (g) 64 min, (h) 2 h, (i) 4 h and (j) 24 h. Note the elongation and reassembly of the peptide nanofibres over time. By approximately 1–2 h, these self-assembling peptide nanofibres have nearly fully reassembled (image courtesy of Hidenori Yokoi, PNAS) [13]. The process (sonication and re-assembly) can be repeated many times, thus truly demonstrating the power of self-assembly.

Figure 5.

Based on the experimental observations (figures 3 and 4), we proposed molecular sliding diffusion model for dynamic reassembly of a single-peptide nanofibre consisting of thousands of individual peptides. When the peptides self-assemble into stable β-sheets in water, they form intermolecular hydrogen bonds along the peptide backbones. The β-sheet structure has two distinctive sides, one hydrophobic with an array of alanines and the other with negatively charged and positively charged amino acids. These peptides form anti-parallel β-sheet structures. The alanines form overlap packed hydrophobic interactions in water, a structure that is found in silk fibroin from silkworm and spiders. On the charged sides, both positive and negative charges are packed together through intermolecular ionic interactions in a checkerboard-like manner. When the fragments of nanofibre first meet, the hydrophobic sides may not fit perfectly but with gaps. However, the non-specific hydrophobic interactions permit the nanofibre to diffuse in a sliding manner along the fibre in either direction that minimizes the exposure of hydrophobic alanines and eventually fill the gaps. The sliding diffusion phenomenon was also proposed for nucleic acids of polyA and polyU in 1956 [14,15]. For clarity, these β-sheets are not presented as twisted strands. Colour code: green, alanines; red, negatively charged amino acids; blue, positively charged amino acids (image courtesy of Hidenori Yokoi) [13].

Figure 6.

Molecular and schematic models of the designer peptides and of the scaffolds. Direct extension of the self-assembling peptide sequence by adding different functional motifs. Light turquoise cylinders represent the self-assembling backbone and the yellow, pink and tan lines represent various functional peptide motifs. Molecular model of a self-assembling peptide nanofibre with functional motifs flagging from both sides of the double β-sheet nanofibres. Either few or more functionalized and active peptide can be mixed at the same time. The density of these functionalized peptides can be easily adjusted by simply mixing them in various ratios, 1 : 1–1 000 000 or more before the assembling step. They then will be part of the self-assembled scaffold [–43]. The left panel is the enlargements of a small part of the peptide nanofibres in the right panel. The enlarged parts show more details of co-assembly of peptides with biologically active motifs, either different peptides (upper left panel), or different proteins (lower left panel: A–C).

Figure 7.

SEM images of Matrigel and various designer peptide nanofibre scaffolds. (a) Matrigel, (b) RADA16, (c) RADA16-BMHP1, (d) RADA16-BMHP2 nanofibre scaffolds assembled in PBS solutions. Matrigel nanostructures are comparable in size to nanofibres found after self-assembly of the designer peptides. Clusters and aggregates of the unidentified naturally derived proteins in Matrigel (a) are absent in the pure peptide scaffolds shown in (b–d). The interwoven nanofibres are approximately 10 nm in diameter in each of the peptide scaffolds with approximately 5–200 nm pores. The appended functional motifs did not prevent peptide self-assembly [37,38].

Figure 8.

SEM image of adult mouse NSCs embedded in designer peptide nanofibre scaffold RADA16-BMHP1 (1% v/w) after 14 day in vitro cultures. Cluster of three visible mouse NSCs embedded in three-dimensional self-assembling RADA16-BMHP1. It is important to point out that the nanoscales of peptide scaffold and the extracellular matrix made by cells are indistinguishable. This is totally unlike the many processed biopolymer microfibres that are often 10–50 µm in diameter, which is 1000–5000 larger (many more orders of magnitude larger in three dimensions) than the self-assembling peptide nanofibres [37,38].

Figure 9.

From designer peptide to scaffold to tissues. (a) Active synapses on the peptide surface. Primary rat hippocampal neurons form active synapses on peptide scaffolds. The confocal images show bright discrete green dot labelling indicative of synaptically active membranes after incubation of neurons with the fluorescent lipophilic probe FM-143. FM-143 can selectively trace synaptic vesicle turnover during the process of synaptic transmission. The active synapses on the peptide scaffold are fully functional, indicating that the peptide scaffold is a permissible material for neurite outgrowth and active synapse formation. (b) Adult mouse NSCs embedded in three-dimensional scaffold (image courtesy of Fabrizio Gelain). (c) Brain damage repair in hamster. The peptide scaffold was injected into the optical nerve area of brain that was first severed with a knife (image courtesy of Rutledge Ellis-Behnke). The gap was sealed by the migrating cells after a few days. A great number of neurons form synapses. (d) Peptide KLD12 (KLDLKLDLKLDL), chondrocytes in the peptide scaffold and cartilage. The chondrocytes are stained with TB showing abundant GAG production (left panel) and antibody to type II collagen demonstrating abundant type II collagen production (right panel). A piece of pre-moulded cartilage with encapsulated chondrocytes in the peptide nanofibre scaffold. The cartilage formed over a three to four week period after the initial seeding of the chondrocytes (image courtesy of John Kisiday). (e) Von Kossa staining showing transverse sections of primary osteoblast cells on HA-PHP-RADA16-I self-assembling peptide nanofibre scaffold. Scale bar, 0.1 mm. The intensely stained black areas represent bone nodules forming.

Figure 10.

Molecular representation of lysozyme, trypsin inhibitor, BSA and IgG as well as of the Ac-N-(RADA)4-CONH2 peptide monomer and of the peptide nanofibre. Colour scheme for proteins and peptides: positively charged (blue), negatively charged (red) and hydrophobic (light blue). Protein models were based on known crystal structures. Image courtesy of Sotirios Koutsopolous [56].

Figure 11.

The release profiles during the entire three-month period for IgG through hydrogels of different peptides and different peptide nanofibre densities. Hydrogels consisted of the self-assembling peptides (i) Ac-N(RADA)4-CONH2 with concentration 0.5% w/v (light blue), 1.0% w/v (blue) and 1.5% w/v (dark blue) and of (ii) ac-(KLDL)3-CONH2 with concentration 0.3% w/v (red) and 0.6% w/v (magenta). Release experiments were performed in PBS, pH 7.4 at room temperature. Data points represent the average of five samples. Image courtesy of Sotirios Koutsopolous [58].

Figure 12.

Cells are patterned on a peptide-coated surface. (a) Molecular models of the surface self-assembling peptides. This type of peptide has three distinct segments: a biologically active segment where it interacts with other proteins and cells; a linker segment that can not only be flexible or stiff, but also sets the distance from the surface; and an anchor for covalent attachment to the surface [9]. These peptides can be used as ink for an inkjet printer to directly print on a surface, instantly creating any arbitrary pattern, as shown here. (b) Mouse neural cells are seeded on the coated surface. The cells only attach where the adhesion surfaces and not the areas with non-adhesive substrate, such as oligoPEG. The entire frame of the cell pattern is 4 × 1 mm = 4 mm².

Figure 13.

The designer lipid-like peptides. These lipid-like peptides all have a hydrophilic head and a hydrophobic tail, much like lipids or detergents. They sequester their hydrophobic tail inside of micelles, vesicles or nanotube structures and their hydrophilic heads are exposed to water. At least three kinds of molecules can be made, with −, +, −/+ heads and in two orientations.

Figure 14.

Molecular models of peptide detergents at neutral pH. (a) Ac-AAAAAAD-COOH. (b) Ac-AAAAAAK-CONH2. (c) DAAAAAA-CONH2. (d) KAAAAAA-CONH2. (e) Ac-VVVD-COOH. (f) Ac-VVVK-CONH2. (g) Ac-IIID-COOH. (h) Ac-IIIK-CONH2. (i) Ac-LLLD-COOH. (j) Ac-LLLK-CONH2. (k) Ac-GAVILEE. (l) Ac-GAVILRR. Aspartic acid (D) is negatively charged and lysine (K) is positively charged. The hydrophobic tails of the peptide detergents consist of alanine (A), valine (V), isoleucine (I) and leucine (L). Each peptide is approximately 2–2.5 nm long, similar size to biological phospholipids. Colour code: teal, carbon; red, oxygen; blue, nitrogen; white, hydrogen.

Figure 15.

High-resolution transmission electron microscopy images of Ac-G6D2 showing different structures and dynamic behaviours of these structures. (a) A pair of finger-like structures branching off from the stem. (b) Enlargement of the box in (a), the detail opening structures are clearly visible. (c) The openings (arrows) from the nanotube which may have resulted in the growth of the finger-like structures. Some nanovesicles are also visible. (d) The nanovesicles may undergo fission (arrows). Image courtesy of Dr Steve Yang [22].

Figure 16.

(a) Transmission electron microscopy image of nanotubes and nanovesicles of lipid-like peptide ac-VVVVVVD-OH in water. Micelles are also present (image courtesy of Dr Steve Yang) [21,22]. (b) AFM image of nanotubes of A₆K lipid-like self-assembling peptides [25]. When the solution pH is less than the lysine pKa of 10, the peptide bears a positive charge. The openings of peptide nanotubes are clearly visible [21,22]. These nanotube structures can also undergo structural changes depending on various conditions, particularly pH changes, ionic strength of salts, temperature and incubation time. The other sheet like materials are likely the un-assembled peptides at the time of the image being collected.

Figure 17.

Molecular modelling of cut-away structures formed from the peptides with negatively charged heads and non-polar tail. Peptide nanotube with an area sliced away. Peptide nanovesicle. Colour code: red, negatively charged aspartic acid heads; green, non-polar tail. The non-polar tails are packed inside of the bilayer away from water and the aspartic acids are exposed to water, much like other surfactants. The modelled dimension is 50–100 nm in diameter.

Figure 18.

A proposed scheme for how the designer lipid-like peptides stabilize membrane proteins. These simple designer self-assembling lipid-like peptides have been used to solubilize, stabilize and crystallize membrane proteins. These peptides have a hydrophilic head and a hydrophobic tail, much like other biological lipids. They use their tail to sequester the hydrophobic part of membrane proteins, and the hydrophilic heads are exposed to water. Thus, they make membrane proteins soluble and stable outside of their native cellular lipid milieu. These lipid-like peptides are very important for overcoming the barrier of high resolutions of molecular structure for challenging membrane proteins.

Figure 19.

Designer self-assembling peptides stabilize membrane proteins. (a) A proposed model of peptide surfactants. (b) Quick-freeze/deep-etch transmission electron microscopy image of peptide surfactant (V₆D) dissolved in water. (c) Stability kinetics of rhodopsin (Rho) in different surfactants; kinetics of bovine rhodopsin under different conditions. Stability of rhodopsin in the absence of OG at different temperatures. Half-life of rhodopsin was as follows: not available in 2.5 mM A₆D at 40°C, 50°C and 55°C (left); decay of A500 in de-lipid rhodopsin in the absence of OG. Half-life of rhodopsin was as follows: 122 min in 1.25 mM A₆D, 47 min in PBS and 27 min 1% OG (middle); stability of de-lipid rhodopsin at 40°C (right). (d) Lipid-like peptides stabilize functional photosystem I [67,69].

Figure 20.

Designer lipid-like peptides are used in cell-free systems. Tens of functional olfactory receptor purification and secondary structure studies have been achieved. Olfactory receptors are soluble in Brij-35 and peptide detergents. Each receptor was expressed in the presence of Brij-35 or a peptide detergent using a commercial E. coli cell-free expression system (Qiagen, RiNA and Invitrogen). (a) The presence of a detergent was necessary to solubilize the olfactory receptors, and all of the peptide detergents were able to solubilize four unique receptors. (b) The detergent peptides and Brij-35 were able to solubilize similar fractions of protein. Peptides that were positively charged or had longer tails tended to solubilize higher fractions of receptors. (c) Detergent peptides can yield milligram quantities of solubilized olfactory receptors, and the maximum yield of the monomeric form of all tested olfactory receptors expected in a 10 ml reaction. Only results from the most effective detergent peptide are shown. (d) Circular dichroism spectra of Brij-35 and peptide detergent-produced olfactory receptors to Olfr226, mOR174-4, mOR174-9 and mOR103-15 (courtesy of Karolina Corin) [71,72].