Molecular Biomimetics: Linking Polypeptides to Inorganic 8 Structures

Candan Tamerler and Mehmet Sarikaya

Abstract Introduction In developing novel materials, Mother Na- Mother Nature has provided a high degree ture gave us enormous inspiration with its of sophistication in materials and systems already existing highly organized structures at the nanometer scale. Naturally occurring varying from macro to nano- and molecu- materials have remarkable functional prop- lar scales. Biological hard tissues are the ex- erties derived from their highly organized amples of composite hybrid materials hav- structures from the molecular to the nano-, ing both inorganic and organic phases that micro-, and macroscales, with intricate ar- exhibit excellent physical properties, all chitectures (Fig. 8.1). They are self-direct- based on their evolved architectural design. ed in their organization and formation, Biocomposites incorporate both structural operate in water environment, dynamic in macromolecules, such as proteins, lipids their interaction with the surroundings, and polysaccharides and minerals, such as complex in their structures and functions hydroxyapatite, silica, magnetite, and cal- self-healing in damage control. Yet, they are cite. Among these, proteins are the most not achievable in purely synthetic systems instrumental components for use in mate- under the same efficient energy conserving, rials fabrication because of their molecu- no waste delivering manner (Lowenstam, lar recognition, binding and self-assembly 1989; Sarikaya, 1999; Ball, 2001; Sanchez characteristics. Consequently, based on this et al., 2005). With the integration of recent premise, inorganic surface specific polypep- developments in molecular and nanoscale tides could be a key in the molecular engi- engineering in physical sciences, and the neering of biomimetic materials. Peptides advances in molecular biology, materials can now be selected by directed evolution, fabrication through biology, biomimetics, adapted from molecular biology, by using is now entering the molecular scale (Sari- combinatorial peptide libraries, analogous kaya et al., 1995; 2003). Utilizing closely to natural selection. Adapting genetic ap- controlled molecular, nano- and micro- proaches further allow to redesign, modify structures through molecular recognition, or engineer the selected first generation templating and self assembling properties peptides for their ultimate utilization in of Nature, molecular biomimetics is evolv- bionanotechnological applications as mo- ing from the true marriage of physical and lecular erectors, couplers, growth modifiers biological sciences (Niemeyer, 2001; Sari- and bracers. kaya et al., 2004). 192 | Tamerler and Sarikaya

Figure 8.1 Examples of biologically fabricated complex nanomaterials. (A) Layered nanocomposite: growth edge of nacre (pearl) of abalone (Haliotis rufescens): Aragonite platelets separated by a thin-film of organic matrix. (B) Nanomagnetics: magnetite (Fe3O4) particles in magnetotactic bacteria: Aquaspirillum magnetotacticum. (C) Hierarchical structure: 3D woven enamel rods of hydroxyapatite crystallites of mouse teeth. (D) Biofiber-optics: a layered siliceous spicular optical fiber of a sponge (Rosella) and its apex (inset), novel design of a lens, a light collector.

Biological hard tissues are the ex- modifiers, brazers and molecular erector amples of composite hybrid materials sets, for self assembly of materials with having both inorganic and organic phases controlled organization and desired func- and exhibiting excellent physical proper- tions. The realization of heterofunctional ties thereby creating ecological intakes for nanostructure materials and systems could the host organisms (Mann 1996; Mann et be at three levels, all occurring simulta- al., 1998; Ball, 2001). Biocomposites have neously feed backing each other as the incorporated both structural macromol- Mother Nature produces her materials and ecules such as proteins, lipids and polysac- components. The first is that the inorganic charides and minerals, such as hydroxyap- specific peptides are identified and pep- atite, silica, magnetite, and calcite (Berman tide/protein templates are designed at the et al., 1988; Ratner et al., 1996; Cha et al., molecular level through directed evolution 1999; Mayer et al., 2002). Among these, using the tools of molecular biology. This proteins are the most promising molecules ensures the molecular-scale up process- because of their recognition, binding and ing for nanostructural control at the low- self assembly characteristics. The advan- est dimensional scale possible. The second tage of a molecular biomimetic approach is that these peptide building blocks can to nanotechnology, therefore, is that in- be further engineered to tailor their rec- organic surface-specific proteins could be ognition and assembly properties similar used as couplers, growth initiators and to the Nature’s way of successive cycles of Tools for Bionanotechnology | 193 mutation and generation can lead to prog- tems, polypeptides are the major displayed eny with improved features eventually for molecules, which can be screened for the their utilization as couplers or molecular specific properties. erector sets to join synthetic entities, includ- In the following sections, we provide ing nanoparticles, functional polymers, an overview of molecular biomimetics ap- or other nanostructures onto molecular proaches to achieve the premises of nano- templates (molecular and nanoscale rec- technology and summarize its potentials ognition). Finally, the third is that the bio- and limitations. Then, we look into the logical molecules self- and coassemble into ways finding polypeptides that recognize ordered nanostructures. This ensures an inorganics, and describe the protocols energy efficient robust assembly process of combinatorial biology for identifying, for achieving complex nano-, and possibly characterizing and engineering peptides to hierarchical-structures, similar to those utilize them as molecular buildings blocks found in Nature (self-assembly) (Sarikaya of future bimimetic materials and systems. et al., 2004). Here we emphasize on the cell surface and There are different ways to obtain the phage display technologies that are well inorganic surface specific proteins such as adapted for the identification of inorganic extraction from hard tissue, designing them surface specific peptides, and to further tai- via theoretical approaches or utilizing the lor the characterized peptides using post- limited number of already existing ones selection engineering. We then discuss (Carlolou et al., 1988; Paine et al., 1996; the possible mechanisms through which a Schneider et al., 1998; Kroger et al., 1999; given protein might selectively bind to an Cha et al., 1999; Liou et al., 2000). Each of inorganic based on their thoroughly bind- these approaches has its own major limita- ing characterization. We present examples tions and may not be practical enough to of current achievements in utilizing engi- serve in all nanoscale-engineering applica- neered polypeptides are given to demon- tions. Inorganic surface specific peptides strate their potential use and, finally, we could be the key in the molecular engi- present future prospects of molecular bio- neering of bioinspired materials. However, mimetics in bio- to nanotechnologies. there are only a few polypeptides have been identified that specifically bind to the in- Potentials and limitations organics. With the recent developments in of nanotechnology recombinant DNA technology, these inor- The fundamental premise in the field of ganic surface specific proteins can now be nanotechnology has been that the length designed, modified or engineered for the scales, which characterize materials struc- production of nanostructured materials. ture and organization, predominantly de- During the last decades, combinatorial bi- termine their physical properties (Drexler, ology based molecular library systems have 1992; Schmid, 1994; Ferry et al., 1997; been developed for selecting substrate- Katz et al., 2004). Mechanical properties specific peptide units, mostly for medical of nanocomposites, light harvesting prop- applications but only recently they are ap- erties of nanocrystals, stain defender prop- plied for selecting short peptides for inor- erties of nanoparticles, magnetic properties ganic surfaces (Brown, 1997; Whaley et al., of single-domained particles, barrier prop- 2000; Gaskin et al., 2000; Naik et al., 2002; erties of nanoclays to extend the shelf lifes Sarikaya et al., 2004). In these library sys- of bottles, and solution properties of col- 194 | Tamerler and Sarikaya

loidal suspensions are all examples to show etry or functionality. One way to overcome that nanotechnology is not a futuristic tech- this problem is to combine “self-assembly” nology, it is already establishing its place in with more conventional “bottom-up tech- our daily life (Jackson et al., 2002; Shipway nology” to provide suitable functionalities et al., 2001; Hoenlein et al., 2003; Thayer with specific structures. However, there a et al., 2004). All of the given examples number of challenges to be overcome such correlate directly to the nanometer-scale as if the structures available from “self-as- structures that characterize these systems. sembly” technique can provide function- In building the nanometer scale structures, ality comparable to that realized by “bot- the approach is to design molecule by mol- tom-up” process, or if the architectures ecule with a purpose such as developing an can be built by the choice of material func- advanced nanoscale machine or assembler tionality rather than the availability of ma- or fabricator. Once, the materials are at the terials which are applicable to the system nanoscale then they present unique char- (Seeman et al., 2002; Sarikaya et al., 2004) acteristics based on physical phenomena, Self-assembled layers are often demon- and therefore, the physical and chemical strated using thiol-derived molecules on rules governing the macroscale materials gold or silanes on oxides. This is because might not reflect their displayed proper- the sulfur or hydroxyl atoms chemisorb to ties (Ferry et al., 1997; Muller, 2001). Re- the gold or silica surface, respectively. This cent experimental research in the field of is advantageous for the gold-or silica-based nanometer-scale electronics and photonics architectures. However, more practical ap- has confirmed theoretical predictions in proach to multifunctional materials is to molecular and nanometer-scale structures, use substrates other than gold or silica. e.g. organized quantum dots, and electrical Availability of new materials will extend transport in nanotubes and wires. In addi- current technology with biosorption in ad- tion, colloidal particles of metals, function- dition to traditional chemisorption. al ceramics, and semiconductors have po- The realization of the full potential of tentially useful electronic, optoelectronic nanotechnological systems has so far been and magnetic properties that derive from limited because of the difficulties in their their small size. These properties may lead controlled-synthesis and the subsequent to their application as chemical, biological, assembly into useful functional structures and optical sensors, spectroscopic enhanc- and devices. Most traditional approaches ers, nanoelectronics, and quantum struc- to synthesis of nanoscale materials are tures, among others (Harris, 1999; Gittins energy inefficient, require stringent syn- et al., 2000; Bachtold et al., 2001; Huang et thesis conditions, and often produce toxic al., 2001; McDonald et al., 2005). byproducts. These techniques still use Successful integration of nanoscale “top-down” approaches, and even the most materials to technology requires creation advanced microtechnology and recently of millions of these structures in parallel developed nanotechnology, such as self-as- (Glotzer, 2004). The conditions for con- sembly through chemistry, in-jet technol- trolled structures at nanometer scale can ogy, dip-pen lithography, and microcontact be obtained by promoting the self-assem- printing, require considerable external bly nature of the molecules through bal- manipulation that curtail the achievement ancing kinetic and thermodynamic forces, of complex 3D architectures and robust yet this does not provide a specific geom- scale-up, and, hence, limit the potential Tools for Bionanotechnology | 195 of nanoscale related physical properties sense the direction of gravity with many (Gooding et al., 2003; Quist et al., 2005). different morphologies depending on the Furthermore, the quantities produced are type of species, or it could be silica form- small and the resultant material is often ing the sponge spicule to serve as light col- highly irreproducible because of uncon- lector, or hydroxyapatite crystals in enamel trolled agglomeration. Even in the case of providing 3D woven enamel rod structure carbon nanotubes, one of the most suc- or the aragonite platelets in abalone shell cessful nanotechnological materials, there to present the microarchitecture (Fig. 8.1) are still some practical limitations to their (Fong et al., 2000; Sarikaya et al., 2001; widespread use, including uniformity, and 2003; 2004). They are simultaneously self- control of surface chemistry, and for two- organized, dynamic, complex, self-healing, and three-dimensional assembly (Harris, and multifunctional, and have characteris- 1999; Hoenlein et al., 2003). Despite all tics difficult to achieve in purely synthetic the promise of science and technology at systems even with the recently developed the nanoscale, the control of nanostruc- bottom up processes. Based on their tures and ordered assemblies of materials closely controlled nanostructures achieved in two- and three-dimensions remains not through molecular recognition, templat- fully accomplished. ing, and self-assembly, biological materials have properties of technological interest Inspiration from Nature that surpass synthetic systems with similar for realization of phase compositions. Under genetic control nanotechnology of the organisms, biological tissues are syn- Mother Nature has been an inspiration thesized in aqueous environments in mild in fully achieving the promises of nano- physiological conditions using biomac- technology. The key to nanotechnology is romolecules, primarily proteins but also primarily to understand how nature works carbohydrates and lipids. Proteins both at the highest level of sophistication with collect and transport raw materials, and efficient energy use without waste accu- consistently and uniformly self- and coas- mulating way (Sarikaya, 1999; Ball, 2000; semble subunits into short- and long-range Seeman et al., 2002). Biomaterials are ordered nuclei and substrates. Whether in highly organized from the molecular to controlling tissue formation, participation the nano-, micro-, and the macroscales, of- in its formation, or being integral part of ten in a hierarchical manner with intricate the tissue in its biological functions and nanoarchitectures that ultimately make physical performance, proteins are indis- up a myriad of different functional units, pensable part of the biological structures soft and hard tissues. Hard tissues such and systems. A simple conclusion is that as bones, dental tissues, spicules, shells, any future biomimetic system, whether for bacterial nanoparticles can be given as ex- biotechnology or nanotechnology, should amples which all have one or more protein include protein(s) in its assembly and, per- based components ( Carilolou et al., 1988; haps, in its final structure. Berman et al., 1988; Schultze et al., 1992; Engineering materials, containing one Kaplan et al., 1994; Paine et al., 1996; Fal- of more phases, are synthesized via a com- lini et al., 1996). The inorganic part could bination of approaches using, for example, be the magnetite (Fe3O4) nanoparticules melting and solidification processes that in the case of magnetotactic bacteria to are often followed by thermomechanical 196 | Tamerler and Sarikaya

treatments, or solution/vacuum deposition it may ultimately be possible to construct and growth processes and finally annealing. a “molecular erector set” in which different Chemical recognition and synthetic self- types of proteins, each designed to bind to assembly processes are a step beyond these a specific inorganic surface, could assemble traditional approaches. Many examples into intricate, hybrid structures composed have been shown in the last decade show- of inorganics and proteins. Below we dem- ing these processes can produce highly or- onstrate the general approaches with some dered and predictable structures, including, examples. for example, mesoporous systems based on surfactant/ceramic precursor molecules; Combinatorial biology self-assembled monolayers, and hybrid approach in selecting macromolecules. In many cases, however, inorganic-specific the final product is a result of a balance of peptides interactions, dictated by the kinetics and Combinatorial strategies in chemistry and thermodynamics of the system, that are biology have attracted great interest to often achieved through “heat-and-beat” search and generate active molecular com- approaches of traditional materials science pounds for various applications over the and engineering (Sarikaya et al., 1982; last decades. The discovery of the counter DeGarmo et al., 1988). In biological sys- parting active molecule from a series or a tems, the same balance is achieved through large number of mixtures has been revolu- evolutionary selection processes that result tionizing idea brought by the combinato- in the emergence of a specific molecular rial methods. First, combinatorial chem- recognition. For example, in antigen/anti- istry-based methods started with large body interactions, lock-and-key is one of peptide libraries following the establish- the main mechanisms by which two mol- ment of solid phase synthesis of peptides. ecules specifically recognize each other. In Optimization and rapid development of the new field of molecular biomimetics, parallel syntheses and automation resulted hybrid materials could be assembled from in screening large number of compounds the molecular level using the recognition for a particular pharmaceutically interest- properties of the proteins that specifically ing property such as increasing selectivity, bind to the inorganics. Using the peptide activity or lowering toxicity (Beck-Sick- based molecular approaches, new genera- inger et al., 2002). Later, the integration tion of binding agents, couplers, or molec- of combinatorial methods into biological ular erector sets could be designed for self selection strategies have brought advan- assembly of materials with controlled or- tages over the chemistry based ones since ganization and specific functions. The cur- they utilize the production capacities of rent state of protein folding prediction and the living systems, phages and cells. Over surface binding chemistry do not provide the last 10 years, various methods using sufficiently detailed information to perform organisms have been established to pro- rational design of these hierarchical struc- duce large libraries of peptides, proteins or tures. To circumvent this problem, massive nucleic acids. These libraries were gener- libraries of randomly generated peptides ated both in vivo environments where or- can be screened for binding activity to inor- ganisms such as cells and viruses have been ganic surfaces via the use of phage and cell utilized, and in vitro environments where surface display techniques. In either case, biological molecules key to synthesis, have Tools for Bionanotechnology | 197 been added into the reaction mixtures. The building blocks with more diverse func- generated molecules are directed towards tion. Since the invention of phage display a certain target interaction where they are nearly two decades ago, display technolo- enriched and identified according to their gies have proven to be an extraordinarily desired property. powerful tool for a various biotechnological The idea of searching for its own active and biological applications (Smith, 1985; molecule is not new considering biological Dani, 2001; Benhar, 2001; Ma et al., 2001; interactions, such as receptors on cell sur- Mrskich, 2002; Wernerus, 2004). Mainly faces recognize their ligands among many protein–protein interactions were studied different molecules or antibodies detect the in a variety of contexts including charac- certain fragments of bacterial or viral sur- terization of receptor and antibody bind- face proteins. These high-throughput strat- ing sites, ligand specificities, and the isola- egies are only bringing us closer to Mother tion and evolution of proteins or enzymes Nature’s ways in developing new molecular exhibiting improved or otherwise altered tools. Nature’s building blocks indeed are binding characteristics for their ligands. based on simple elements; ribonucleic acid Biological libraries composed of pep- (RNA) or deoxyribonucleic acid (DNA) tides, antibodies, or proteins can be dis- are formed through the polymerization of played by using either in vivo or in vitro four different nucleotides and peptides or display technologies. Regardless of the proteins are formed through the condensa- display technologies, there are three major tion reaction of 20 different natural amino components in the system: displayed mol- acids. For a nucleic acid that is 300 base ecule, its genetic code and a common linker pair in length or composed of 100 codons, to the displayed system. Following the in- there would be a possibility of translating teraction of the library with the counter- into 20100 different proteins. In combina- part molecule, high-throughput selection torial biology based library systems, poly- of the desired molecules with the possible peptides have been the major biological specificity and affinity towards the coun- compounds because of their very efficient terpart molecule is carried out. All tech- recognition properties at the nanometer nologies are based on the common theme scales. In this section, we will go over the of cloned gene and its encoded protein is basic principles of display technologies physically linked; therefore, the genetic in- and discuss phage and cell surface display formation translating to the protein with methods in detail, starting from their use the desired phenotypic character could as a tool for protein–protein interactions, be accessed easily. In in vivo display, stable then their adaptation for protein–inorgan- genes are expressed either through trans- ics interaction over viewing both advan- fection or introduction of foreign DNA tages and drawbacks of the each display into the cells, whereas in vitro systems cell system. free extracts will transcribe the cloned template (Hoess, 2001, Samuelson, 2002). Principle of display technologies Consequently, in vitro systems are not lim- Display technologies, in general as a rou- ited by the transformation efficiency of a tine tool, refers to a collection of meth- cellular host (Dower et al., 2002; Lipovsek ods for creating libraries of biomolecules et al., 2004). In in vivo display technologies, that can be screened for desired or novel biological host can be phage, such as very properties, by optimizing the assembly of well established filamentous bacteriphage 198 | Tamerler and Sarikaya

M13, or alternative ones as λ, T4 or T7 compatible environments (Takahashi, et phage, or cells, including prokaryotes and al., 2003). eukaryotes. In phage display, the coat pro- tein genes are used to display the molecu- Phage display lar library, whereas cell wall or periplasmic Phage display has been the most com- display systems are successfully applied to monly practiced combinatorial peptide li- expression of molecular libraries. Outer brary display method since Smith (1985) membrane proteins, lipoproteins, fimbria, first showed the linkage between pheno- and flagellar proteins can be used for heter- type and genotype in filamentous bacte- ologous surface display on bacteria. With riophage. In phage display, the majority of the data from human genome project, now the research has been performed mainly determining the functions of the proteins by bacteriophages M13, fd and f1 which is an important task especially for diag- are closely related to each other, with 98% nosis and therapeutics. Mammalian cell sequence identity. These phages require surface expressions of receptor or trans- fertility plasmid (F-plasmid) in bacteria membrane proteins have been attempted for infection and are called Ff type phages. for being displayed on the current systems, Phage has two main components, genome however low efficiency of cloned gene de- and the protein coat composed of several livery is a major problem. Many viral ex- different proteins. They are commonly pression-cloning systems have been still used as vectors in recombinant DNA ap- continually searched for displaying mam- plications. Common feature of vectors is malian proteins interactions in their in to accommodate the foreign DNA so that vivo environment. Among eukaryotic cells, when they replicate within the host, foreign yeast two hybrid systems have been very insert also replicates for the production of promising for expressing eukaryotic pro- desired molecule. Some of the phage coat teins (Ueda, 2004). However all of these proteins tolerate additional domains cod- systems still carry the limitations living ing for different biomolecules (e.g. anti- cells such as rather restricted library size bodies, small peptides or proteins) at spe- or suppression of certain mutant by the cific sequence positions (Kay et al., 1996; molecular machinery of the host or even Beck-Sickinger et al., 2002). In general, sometimes they are not correctly folded or DNA fragments coding for the random transported not contributing to the library library population are ligated into phage diversity. Ribosome, mRNA and DNA genome such that the encoded biomol- display technologies are developed as cell ecule is expressed as a fusion product to free protein synthesis systems to overcome the coat protein permitting the addition of the limitations brought by the transforma- these new segment. Once the recombinant tion efficiency of a cellular host, therefore phage DNA is introduced into E. coli cells, they can be operated in the absence of a heterogeneous mixture of recombinant living cell (Fitzgerald, 2000; Amstatz et phages will be produced, each displaying al., 2001). Consequently, they can present a different member of the library on their high library size and also unique applica- surface (Benhar, 2001). Once a phage with bility to directed evolution of proteins, the desired phenotype is selected, DNA since selection with these systems can also sequence of the specific insert will give the be performed under the biologically in- amino acid sequence of the peptide. Tools for Bionanotechnology | 199

The most widely used display system domain which initiates translocation of has been based on E. coli specific filamen- the viral DNA into E. coli during infection tous phage, M13, mostly because its life cy- through binding to To1A receptor while cle is studied in detail and relatively easy to the second domain confers host cell recog- work with. It is in a flexible rod shape with nition by binding to the tip of F-pilus of helically arranged molecules about 1 µm the E. coli surfaces, finally C-terminal do- in length but less than 10 nm in diameter main interacts with the other coat proteins (Fig. 8.2). A coat composed of five differ- for structural integrity, an important fac- ent proteins surrounds its single stranded tor in viral morphogenesis (Riechmann et DNA core. Four of the coat proteins are al., 1997; Rodi et al., 1999; Sidhu, 2001; present at about five copies per phage, P7 Hoess, 2001). In display of proteins, P3 and P9 cap one end of the virion while P3 has been the primary scaffolds (Fig. 8.2), and P6 cap the other end. Fifth one is pres- although P7 and P9 have been shown to ent at ~2700 copies per phage and covers tolerate fusion on their amino terminal the length of the phage. Among all the coat and P6 has been shown to display fusions proteins, P3 is the largest (42 kD) and the at its C terminus. M13 genome is single most complex one. It is responsible for host stranded however it is converted to dou- recognition and infection. P3 has three ble stranded plasmid like replicative form distinct domains connected by glycine rich (RF) following the infection of its host to linker regions. First domain is N-terminal serve as template for the production of vi-

Figure 8.2 A schematic illustration of phage display and cell surface (flagellar) protocols adapted from molecular biology in this research for selecting polypeptide sequences with binding affinity to inorganic substrates and the substrate domains screened during the screen. 200 | Tamerler and Sarikaya

ral proteins and the single stranded DNA replication origin to replicate in host cell progeny. Phage progeny is assembled by and when the host cell infected with the encapsulating single stranded DNA with helper phage, phage replication protein the viral coat proteins and extruded from acts both on the phagemid and the helper the host cell into the medium. When the phage DNA (Smith et al., 1997; Rodi et al., phages reproduce themselves gene fusion 2002). products pass along their properties to next generation (Benhar, 2001; Hoess, 2001; Cell surface display Sidhu, 2001). Phagemid vectors are also Cell surface has been developed as an alter- commonly applied in phage display tech- native to phage display only over a decade nologies. Phagemids are hybrids of phage ago to display peptides and proteins on the and plasmid vectors and they are designed surface of microbial cells. First examples of to contain the origins of replications both cell surface display of foreign proteins were for M13 and E. coli. In addition to gene III, reported in 1986 (Charbit et al., 1986; they contain appropriate multiple cloning Freudl et al., 1986) and since then it has sites and an antibiotic resistance gene how- been utilized for various biotechnological ever, they lack all the other gene products and biological applications. The size of for- required for generating a complete phage. eign protein to be displayed on the surface Therefore they can be grown as plasmids of the phage is rather limited; however, cell in E. coli or alternatively coinfection with walls or periplasmic membranes can be a helper phage results in the production of utilized for fusion of larger proteins. Early recombinant phage (Kay et al., 1996). examples of this approach were the display In the construction of libraries, either of gene fusion products on the outer sur- all copies of the phage coat protein contain face of recombinant E. coli by insertion of the insert (polyvalent display) or only a short gene fragments into the outer mem- fraction of the coat protein contains the in- brane proteins LamB, OmpA and PhoE. sert (monovalent display). In the first type (Charbit et al., 1986; Freudl et al., 1986, of display, the foreign DNA is inserted Agterberg, 1987, Stahl et al., 1997). In ad- into the phage chromosome eg. to encode dition to outer membrane proteins, fim- a single type of P3 protein, therefore the bria and flagella proteins, lipoproteins have insert is displayed in all of the five copies of been applied for surface display on both P3 molecule. Whereas in the second type Gram-positive and Gram-negative bacteria of display, one has to either include a sec- and yeast (Fig. 8.2) (Lu et al., 1995; Stahl ond gene for the protein or use phagemid et al., 1997; Scembri et al., 1999; Wittrup, vectors to incorporate with the virus dur- 2001). As a general principle, the displayed ing morphogenesis. If the insert carrying peptides or proteins (the target protein) gene and the native gene are both in phage are fused with anchoring motif (carried genome, then the phage genome encodes protein). Fusion mechanism such as N- for two different types of P3 molecule as terminal, C-terminal and the sandwich fu- being recombinant and wildtype. The in- sions, is one of the most important factors sert carrying gene and the native gene are affecting the efficiency and the stability of contained in a phagemid and in a helper the display (Beck-Sickinger et al., 2002). phage, respectively. Phagemid carries both First step is to identify the best available fu- the plasmid and the filamentous phage sion site, which will allow the protein to be Tools for Bionanotechnology | 201 displayed facing the surrounding environ- Selection of peptides ment, the target. Each anchoring motif will Both phage and cell surface display rely on have a different characteristic, which can the use of chimaeric proteins that consist of be utilized for specific applications such as a target sequence fused within (or to) a pro- membrane proteins can be useful for im- tein that naturally localizes on the surface of munostimulation purposes. The next im- a bacteriophage (a bacterial virus) or a cell portant parameter is the displayed protein, to achieve display. Using standard molecu- its amino acid sequence, transportation lar biology techniques, the DNA sequence through different locations in the host cell of the target region can be randomized to and its folding structure. Next is the host create a library of phages or cells, each of cell as its easiness to work with, compat- which will synthesize a different version of ibility with the displayed figures. Among the chimera on its surface. Contacting the Gram-negative bacteria, E. coli is the com- library with an immobilized ligand, wash- mon host system because of very well es- ing out weak or non-binders and repeating tablished genetic protocols, although its the process to enrich for tight binders can outer membrane is relatively fragile and it select a subset of the original library ex- has high protease activities. Gram positive hibiting the ability to tightly interact with bacteria have more rigid cell wall and have the desired ligand. This process is known only one membrane to achieve the secre- as biopanning, which is an affinity, based tion in contrast to two membranes existing method for a specific substrate. Generally in Gram negative ones. Bacillus has been in early rounds, low affinity binders can be the common host systems used in Gram- caught if the selection is performed under positive bacteria. mild conditions. In the later rounds, as the Among the microbial display methods, conditions gets harsher, tight binders gets yeast display systems are promising and recovered. Because the chimera is encoded have many advantages over the bacteria within the phage genome or on a plasmid mainly because yeast has a eukaryotic pro- carried by the cell, the identity of the se- tein folding and secretion mechanisms. In lected sequences (e.g. their amino acid yeast display, mammalian origin proteins compositions) can be deduced by DNA can be displayed whereas it is rather hard to sequencing (Fig. 8.2) (Benhar, 2001). display them in phage or bacterial systems In both of the selection processes, the because of the lack of post-transcriptional stringency i.e. the degree of peptide’s fit- modifications. Especially Saccharomyces ness, and the yield, i.e. the fraction of the cerevisiae strain is recently used increas- surviving clones, are the two important ingly for this purpose since it is safe and parameters affecting the efficiency of the fermentation as well as genomic charac- selection process. Ultimate aim is to have ters are very well known (Ueda, 2004). In mild conditions in the early round to am- general, microbial cell surface display has plify the fittest clones, which may have a a wide range of biotechnological applica- very low yield at the first round of selec- tions such as live bioadsorbents for biore- tion. Whole optimization process during mediation purposes, antibody production, the rounds of selection will depend upon whole cell biocatalysts, biosensor technol- the relation between the stringency and ogy, live vaccine development, cell sorting the yield for the library population (Smith etc (Samuelson et al., 2002, Wernerus et et al., 1997; Hoess, 2001). Since the short al., 2004, Feldhaus et al., 2004). peptide sequence might have difficulty to 202 | Tamerler and Sarikaya

fold, structural constraints can also be in- process can be limited by the transforma- cluded in the encoded peptides. The sim- tion efficiency of the host. In general, in plest covalent constraint is the disulfide vivo display systems (~1010) contain small- bond that becomes the nucleus of a mi- er number of different library members croprotein. Bacterial disulpite bonds also than the in vitro display systems (~1015). formed if the insert contains non-paired However, a theoretical number of decapep- cysteine residues in random libraries be- tides that can be generated with the all pos- cause of dimerization through the forma- sible permutations of the 20 amino acids is tion of an intermolecular disulphide bond 2010 or about 1013. This number or even in solution. higher size can be reached easily in in vitro display systems, but not in in vivo display Advantages and drawbacks of display systems. The greater numbers of individual technologies members presented in the library will cer- Working with alternative display systems tainly increase the probability of identify- always helps to overcome the biases that ing a high affinity and a specificity binder might intrinsically exist in a particular dis- for the desired substrate. The functional play system, such as limited library size or expression of all the library members such expression problems. Therefore, it is neces- as inefficient folding or expression biases sary to integrate different display systems, associated with the living cells are among which could be dictated by the character of other factors affecting the library diversity the displaying molecule according to the (Sarikaya et al., 2004). Major expression desired application area. For example, yeast biases comes from the fact that each amino display has certain advantages for display- acid is represented by different number of ing proteins originating from mammalian multiple codons, such that while leucine cells; however this may also create false is specified by six codons, methionine is positives due to auto activating bait fusion specified only by one codon. Consequently, proteins. In phage display, it is rather dif- there might be a great variation in the ficult to display mammalian or any other representation of each amino acid in the larger proteins; however this system can library, when the sequences are generated tolerate harsher conditions and a low copy randomly. Another expression bias de- number display can be easily achieved. De- pends on the origin of the cells; the host spite the easy applicability of phage display, system might synthesize different levels of phage requires bacterial cells for its ampli- transfer RNAs, which result in using cer- fication through infecting the cells. During tain codons preferentially. Transformation the screening, some level of phage progeny efficiency of the each codon might not be might exhibit poor or no infectivity, which the same, then e.g. six codons specifying will result in their irretrievably lost from leucine might be represented by different the library pool. Conversely, efficiently pro- percentage. duced phages may yield a larger progeny In spite of the above limitations, and identical sequences may be selected phage and cell surface display remain the at multiple occasions during subsequent techniques of choice in many applications rounds of biopanning (Hoess, 2001; Sari- because of their well established features, kaya et al., 2004). applicability, ease of use and commercial A critical parameter affecting the li- availability. One of the most extensively brary size is the introduction of the library used system is the Ph.D. PD kit from within a cell or a phage genome, since this New England Biolabs (Beverly, MA) Tools for Bionanotechnology | 203 which relies on the exposure of a random proved binding affinity and catalytic activ- hepta- or dodecapeptide on the surface of ity of existing proteins through directed filamentous phage M13 by virtue of its fu- evolution (Giver et al., 1998; Petrouna et al., sion to the minor coat protein pIII (Fig. 2000, Antikainen et al., 2005). Nature uses 8.2). Heptapeptide libraries are available unique characteristics that are desirable for in both linear and disulphite constrained material synthesis including high selectivity, forms. In the constrained library, a pair of nanoscale self-assembly, and precise struc- cysteine residues forms a disulphite bond ture control and follows the route of com- under oxidizing conditions and provides bining simple building blocks to form mol- the display of the heptapeptide as a loop. ecules with millions of different properties. The necessary items for construction of Here, the molecular machinery is mostly custom-made libraries are also available. based on the protein–material interactions The complexities of the libraries are at the and, therefore, proteins could be the bio- level of two billion independent clones molecules in the self assembly processes for a possible sequence space of 4 × 1015 through their unique molecular recogni- (Parmley et al., 1988; Smith et al., 1993). tion and binding capabilities. The prem- In our group, we have selected the inorgan- ise is that, in addition to protein–protein ic binding sequences via in vivo systems, in- interactions, polypeptides can recognize cludes M13 phage display system as well as relatively simple and repetitive structure of cell surface display method utilizing both inorganic surfaces and even short peptide outer membrane and flagella protein based sequences can discriminate between two selection methods (Sarikaya et al., 2004). closely related inorganic surfaces. These In flagella display system, we have used molecules, therefore, offer large variety of FliTrx cell surface display system commer- chemical and structural characteristics; cialized by Invitrogen Life Technologies with many advantages compared with the (Carlsbad, CA). The library has random existing thiol and silane-based molecular sequences of 12 amino acids as disulfide- systems that have been the hallmark of the constrained loops within Thioredoxin 1 self assembly in chemistry for the last two (Trx), which is itself inserted into FliC, the decades. With engineered polypeptides, major E. coli flagellar protein and it has an there is now a possibility of producing mil- estimated diversity of about 1.8 × 108 (Lu lions of new potential linker molecules, et al., 1995). The resulting fusion proteins each being specific to an inorganic surface are exported to the cell surface where they and allowing fusion to create further tai- assemble into flagella, which are extended loring for multifunctionality. With this in surface features used for cell motility. The mind, molecular library systems have re- display on flagella offers a unique feature as cently attracted great interest to generate it provides the easy recovery of the bound molecular templates for diversified mate- cells following the physical breakage of the rial synthesis and assembly. When we refer flagella from the substrate upon application to combinatorial approaches, we mean the shear stress such as vortexing (Fig. 8.2). screening, synthesis, examination of mil- lions of similar molecules, which are dif- Application of phage and cell surface display ferent in their compositions of amino acids methods to the selection of inorganic binding and their arrangements. Here, the mo- peptides lecular templates can be generated for di- There has been a rapid accumulation of versified material synthesis by altering the efforts on protein engineering towards im- protein’s properties. These systems allow 204 | Tamerler and Sarikaya

to form the self-assembly of nanostructure (such as silica and hydroxyapatite) that components with distinct functionalities were selected by using either phage display especially in the areas of sensing, electron- (specifically, M13) or cell surface display ics and catalysis (Klaus et al., 1999; Naik et (specifically, flagellar display, FliTrx) (Fig. al., 2002, Seeman et al., 2002, Koneracka et 8.3) (Sarikaya et al., 2004). There are also a al., 2002; Sarikaya et al., 2003). number sequences selected for various ma- A genetically engineered polypeptide terials by other groups either via cell sur- for inorganics (GEPI), selected through face display, including, iron oxide (Brown, the display protocols as described in Fig. 1992), zeolite (Nygaard et al., 2002), gold 8.2, normally defines a sequence of amino (Brown, 1997) and zinc oxide (Kjærgaard acids that specifically and selectively binds et al., 2000), and phage display, including to an inorganic surface (Sarikaya et al., gallium arsenide (Whaley et al., 2000), sil- 2004). The inorganic surface could be well ica (Naik et al., 2002a), silver (Naik et al., defined, such as a single crystal or a nano- 2002b), zinc sulfide (Lee et al., 2002), cal- structure; it might be rough, or totally non- cite (Li et al., 2002), and cadmium sulfide descriptive, such as a powder (Fig. 8.2). (Mao et al., 2003). The applications of the One of the first set of GEPIs selected were selected binders varied from the assembly gold binding proteins that were isolated as of inorganic particles (Whaley et al., 2000; extracellular loops of maltoporin which Lee et al., 2002; Mao et al., 2003) to con- were subsequently fused to the amino ter- trol the nucleation of the compounds they minus of the alkaline phosphotase with re- were selected for (Brown et al., 2002; Li et tention of gold-binding activity (Brown et al., 2002; Naik et al., 2002a; 2002b). al., 2002). Most of our initial work focused Inorganic materials are very different on using the 14 amino-acid binding motif, substrates than proteinaceous or general MHGKTQATSGTIQS, called GBP-1. biomolecular ligands and, therefore, adapt- The motif does not contain cysteine that is ing display technologies developed with known to form a covalent thiol linkage to biology in mind to the realm of materials gold, similar to thiolated molecules in self- science would require a new set of condi- assembled monolayers (SAM). To increase tions and protocols although this has not the binding activity, several repeats of the been widely discussed in the literature so same sequence have been engineered and far. Inorganic compounds come in a vari- strong binding activity required at least ety of forms, from polydisperse and mor- three repeats. phologically uncharacterized powders Similar to the protocols we followed of various particle sizes to single crystals in our initial collaboration with Brown with crystallographically defined flat - sur (2002), our recent research focused on faces. The chemical of physical nature of using materials that could be synthesized the inorganic substrate may disqualify a in aqueous environments at ambient con- particular display technology. For instance, ditions (biocompatible) and could have phage display is suitable for work with fairly stable surface structures and compo- powders even if a gradient centrifugation sitions in water. These include noble met- step is used to harvest complexes between als (such as Pt and Pd), oxide and nitride binding phages and particles. On the other semiconductors (e.g. Cu2O, ZnO, GaN, hand the FliTrx CSD system would not be TiO2, etc.), minerals (such as mica, cal- amenable to such an enrichment process cite, sapphire) or biocompatible substrates since centrifugal forces would shear off the Tools for Bionanotechnology | 205

Figure 8.3 Examples of inorganic binding sequences selected via phage and cell surface display methods that could be utilized as molecular toolbox for the next generation of peptide based materials and systems. flagella from the cell. Similarly, while both inorganic surfaces before and after pan- phage and cell surface display are theoreti- ning using various characterization tech- cally suitable for panning on single crystals, niques such as XPS, Raman spectroscopy tightly bound cells or phages may be very and AFM (Dai et al., 2004, Sarikaya et al., difficult to elute out from the material, 2004). It may also be useful to monitor thereby leading to the loss of high affinity wash or elution buffers (e.g. using atomic clones. In such cases, the use of the FliTrx adsorption spectroscopy to detect metals system may be advantageous since all bind- and metalloids). If any evidence of surface ers have an equal likelihood to be recovered modification or deterioration is obtained, following flagellar breakage. Another im- buffer conditions should be optimized to portant factor affecting the efficiency of the guarantee compatibility with the target in- display system, and consequently the selec- organic. Another important parameter is tion of the right binder is to follow the sta- that inorganic compounds come in a vari- bility of the inorganic nature (Sarikaya et ety of forms, from polydisperse powders to al., 2004). Many materials rapidly develop single crystals (Fig. 8.2). Peptides may ad- an oxide layer on their surface, expose dif- just themselves to recognize different sub- ferent crystallographic faces to the solvent, strates and, a new binding sequence could and may become chemically or physically be obtained for a new morphology dictated modified when incubated in the biological by the application area. media used during the panning process. In traditional biological applications To avoid becoming a victim of the first law of peptide libraries (e.g. antibody epitope of directed evolution, “you get what you characterization, mapping of protein–pro- screen for” (Schmidt-Dannert et al., 1999); tein contacts and the identification of pep- it is therefore imperative to characterize tide mimics of non-peptide ligands), 3–4 206 | Tamerler and Sarikaya

biopanning cycles are usually performed in eration of libraries. This is very similar to PD while 4–5 are carried out in CSD. After the evolution process where recursive cycle these cycles of enrichment, the selected se- of mutation and selection resulting in the quences typically converge towards a con- progeny with the improved features. The sensus consisting of identical sequences. information obtained from e.g. phage dis- Such consensus sequences reflect precise play libraries can be integrated into succes- interactions between the side chains of sive generations of phage display libraries. the protein under study and those of the In the recent years, this type of approach selected polypeptides. However, current led to 10–100 fold improvements in bind- experience indicates that this rule does not ing affinities in biological entities such as for the case of inorganic binding sequences mapping of binding sites of proteins, eg an- where similarities rather than a strict con- tigenic epitodes, or searching for inhibitors sensus are generally observed. This pre- for a specific function (Forrer et al., 1999; sumably reflects the heterogeneity of the Schmidt-Bannert et al., 1999; Petrouna et inorganic substrate at the atomic, crystal- al., 2000; Hoess, 2001). lographic and morphological levels and the The ultimate aim of the protein engi- fact that there are probably multiple solu- neering is to enhance the desired property tions to the problem of inorganic binding. of the proteins such as stability and cata- One could, for example, envision binding lytic activity via different methodologies. strategies relying on shape complementari- While in traditional biotechnology, avail- ties, electrostatic or stereochemical interac- able proteins and other macromolecules tions, van der Waal’s interactions or vari- with known functionalities are used via top ous combinations of these mechanisms. down approach (rational engineering), in Clearly, a better understanding of the rules genetic engineering, proteins and peptides that govern the binding of polypeptides are manipulated to tailor their structures to inorganic compounds is needed to un- and functions via the bottom up approach derstand the nature of specificity, predict (directed evolution). Directed evolution cross-specificity and affinity and, ultimate- and the rational engineering have been the ly, for the rational use of inorganic binding two main methodologies utilized greatly peptides for the design of hybrid materials for the improvement of enzymes. In direct- exhibiting controlled chemistry, structure ed evolution, the variant libraries are cre- and organization. ated through random mutagenesis, which could be directed over the entire gene or Post selection engineering on a relatively small region of the gene. of inorganic binding Mutations could be introduced by DNA peptides shuffling which mimics natural recombi- Up to now, we have discussed how the in- nation or by error-prone PCR at random organic surface specific polypeptides can without significant sequence bias. Here, be screened and identified using selection the important requirements are to be able techniques based on display technologies. to make the libraries complex and large Genetic engineering techniques can be fur- enough to contain a mutant having the ther applied to the selected peptides not desired property and also develop a rapid only to investigate the mechanism of bind- screen or selection protocol, as in the case ing through identification of core amino of peptide libraries. These approaches have acids and but also to create second gen- been very useful in biotechnological appli- Tools for Bionanotechnology | 207 cations where more stable, active and better inorganic specific peptides by using cell performing enzymes needed such as work- surface display. Here, gold binding affinity ing in non-aqueous solvents or extreme of peptides was improved by constructing pH or temperature or accepting different semirandom second generation peptide substrates. However, through randomiza- libraries expressed on the outer surface of tion and including the right pressure for E. coli as part of the maltodextrin porin, selection, completely novel property pos- LamB. From the results of the earlier find- sessing proteins are obtained via directed ings of two defined sequences contribut- evolution (Forrer et al., 1999; Hoess, 2001; ing to gold binding were combined in the Joern et al., 2002). Whereas rational engi- second generation libraries with repeat- neering or the top-down approach requires ing polypeptide populations to search for a knowledge-based study in which pro- altered rate of gold appearance which in teins are designed rationally using site-di- turn altered the morphology of the gold rected mutagenesis and generally coupled crystals. Genetic engineering tools can with computer aided molecular modeling be further utilized for the generation of to specify the possible substrate interaction multiple repeats of the already engineered locations. Site-directed mutagenesis, where sequences to incorporate the structural each amino acid is substituted by another, properties of the inorganic surfaces, e.g. could be an invaluable tool in studying the optimizing the length of the sequence or effect of individual amino acids and in re- shape of the sequence to achieve the de- designing the proteins. This method could sired effect (Brown et al., 2002). We have be perfectly well coupled with the combi- recently applied multiple-repeat based natorial methods, which together would strategy on phage display-selected pep- cover large and complex interactions to tides, specifically, platinum binders. Based obtain the engineered sequences with im- on the strong binding affinity of Pt-bind- proved controlled properties. Second gen- ers, examined via immunofluorescence mi- eration libraries offers the right platform croscopy, we genetically designed different for the convergence of these two main pro- number of repeats in both constrained and tein-engineering approaches to obtain the linear forms. Not surprisingly, depending improved progeny for the fully realization on the number of repeats, or the molecu- of directed evolution similar to natural lar architecture of the peptides, each GEPI evolution. presented different affinity and selectivity Second generation of phage and or cell for their substrates even if the basic amino surface display libraries, where the semir- acid sequence was the same in all (N. Gul- andom proposed alignments of peptides Karaguler et al., unpublished results). are designed for specific application pur- poses, could be also integrated in search for Understanding the nature improved GEPIs. These libraries could be of inorganic binding used to characterize the structural basis of In the design and assembly of functional binding by modifying the fragment expect- inorganics, it is crucial to understand the ed to be responsible of binding. The main nature of polypeptide recognition and benefit ofsecond generation display libraries; binding onto inorganic materials for engi- is the possibility of improved affinity and neered utility of the GEPI. Although con- binding specificities of the already selected siderable research has been directed in the peptides. This has already been applied in literature on how proteins recognize each 208 | Tamerler and Sarikaya

other or other biomacromolecules, it is not strongly to the material B with a different completely readily possible to design a pep- composition having similar structure (e.g. tide for desired functionality. Therefore, two different perovskite structures). There- the studies on the understanding of the fore, if one seeks highly specific binders, the nature of peptides recognition of inorganic physical and chemical characteristics of the surfaces are in its infancy. The specificity of material must be known. Binders selected a protein for a surface may originate from for a specific size, morphology, crystallog- both chemical (e.g. H-bonding, polarity raphy or stereochemistry would constitute and charge effects) and structural (stereo- smaller number of binders. chemistry, size and morphology) recogni- Many sequences specific to different tion mechanisms (Izrailev et. al.,1997; Dai inorganics have been identified up to now et al., 2000; Evans, 2003). These two key via application of display technologies us- mechanisms may both be important and ing molecular peptide libraries. In the next work collectively to create a complemen- step, characterization and understanding tarity and compatibility in molecular rec- of binding by examining both its chemical ognition followed by self assembly. Weak, and physical basis providing the informa- non-covalent, chemical bonds can result in tion for the evaluation of polypeptide-in- a stable structures, when many bonds act organic surface affinities and selectivities, collectively. In addition, inorganic prop- essential for the realization of protein/in- erties also affect binding as their surface organic hybrid materials (Niemeyer, 2001; could be well defined, such as a single crys- Ball, 2001; Sakiyama et al., 2001; Sarikaya tal or a nanostructure or they might be et al., 2003) As discussed below, surface as- rough and, or totally non-descriptive such sessment of inorganics could be carried out as a powder. In the latter case, the sequence by suing both experimental and theoretical space is the largest as powders represent peptide/inorganic interactions character- many possible topologies, sizes, and chem- ization tools. istries. Because of the availability and non- specificity of non-descriptive surfaces on Experimental approaches to powders, selected polypeptides, often have study the protein adsorption and little or no consensus sequences. Even so, binding these selected (first generation) sequences Among the experimental approaches to may contain binding domains with similar monitor protein adsorption on and bind- functionalities. Binders selected for a given ing to inorganics, fluorescence microscopy size, morphology, crystallography or ste- is a routine and practical tool as the first reochemistry may share a higher degree of step in the semiquantitative evaluation of homology. For example, a GEPI binding to these sequences with respect to their affin- a material of a certain size may also bind ity and selectivity (Fig. 8.4). Although the to a smaller particle of the same material, FM experiments are an essential part of but less strongly. Similarly, a GEPI bind- the screening protocols, they do not pro- ing strongly to a specific crystallographic vide quantitative adsorption and binding surface may bind with an altered affinity to mechanism. In FM, immunofluorescence another surface of the same material, e.g. labeling detection using monoclonal anti- (111) versus (100) of α-alumina (unpub- body conjugated with secondary antibody lished). Finally, a GEPI strongly binding to fragments results in a powerful binding a material of composition A may bind less detection system (Brown, 1992; Whaley Tools for Bionanotechnology | 209

Figure 8.4 Examples of characterization tools for studying peptide binding. (A) Initial semiquantitative evaluation of binding affinity and specificity via immunolabelling fluorescent microscopy analysis. (B) The analysis of kinetics of binding via quartz crystal microbalance and surface plasmon resonance spectroscopy. et al., 2000; Naik et al., 2002; Dai et al., linity and surface conditions (Czenderna 2004; Sarikaya et al., 2004). Scanning et al., 1984; Homola et al., 1999; Sarikaya probe microscopy (SPM) protocols have et al., 2004). Conventional spectroscopy been also integrated for inorganic sample techniques, such as X-ray photoelectron surface preparation and peptide bind- spectroscopy (XPS) and time-of-flight- ing/self-assembly. Here the tip-design and secondary ion mass spectroscopy (TOF- engineering, observation conditions, data SIMS) techniques also recently proven analysis and interpretations are all part to be viable techniques in providing the of these new protocols to study specific fingerprint of peptide adsorption onto the polypeptide binding onto inorganic sur- surfaces (Coen et al., 2001). Although, dif- faces (Whitesides et al., 1991). While the ficult and time-consuming, solid and liquid SPM, specifically atomic force microscopy state nuclear magnetic resonance (NMR) (AFM) and scanning tunnel microscopy spectroscopy supplies the essential ex- (STM) techniques have been utilized to perimental evidence of peptide molecular obtain structural information of inor- structures and provides clues towards un- ganic binding, quartz crystal microbalance derstanding of mechanism of polypeptide (QCM) and surface plasmon resonance binding on inorganics (Long et al., 2001; (SPR) ispectroscopy are used to quanti- Evans et al., 2004). All of these techniques tatively analyze protein adsorption kinet- are being utilized in our on-going research, ics and thermodynamics of binding and and some will be briefly discussed below as stability under various protein concentra- examples. tions, solution properties such as pH, sa- 210 | Tamerler and Sarikaya

The QCM has been an established analyzed via QCM, but second generation mechanical measurement tool in studying engineered peptides can be also compared the adsorption of proteins on the metal with the first generation peptides to asses surfaces (Murray et al., 2000; Bailey et al., their improved binding. 2002). Here we use it to evaluate the kinet- The QCM, being a mass detection ics of binding of GBP on gold surface. In system is a good tool for determination of QCM, quartz crystal disk mounted with kinetics of binding but it does not provide e.g. gold electrode, sense the resonance detailed information on the mechanism of vibration behavior through the change of physical adsorption. The SPR, the other deposited molecular mass on the surface. hand, can provide more detailed adsorp- When worked in air, QCM can accurately tion data measuring very small amounts provide measurements even thick films al- of molecular adsorption as this causes low though the viscosity and the thickness of level of refractive index changes due to the the film could become important param- biological compounds bound on the sen- eters in the case of liquids measurement. sor surface. Reflected light intensity from In assessing quantitatively the nature the interface between metal substrate and of binding of GEPIs on inorganic surfaces, the analyte at a specific incident angle is we have examined many of the selected measured as a result of the optical excita- binders as many materials can be coated tion of surface plasmon waves (Jung et al., on a quartz crystal electrode (Fig. 8.4). 1998). The shift in SPR wavelength is re- Our first example was the GEPI-Au inter- lated to molecular adsorption on the metal action and, for this, we used several of the substrate (which is either metallic Ag or gold-binding polypeptide (GBP) series. Au). In our experiments, rapid assembly of Here, we demonstrated the adsorption be- GBP-1 onto sensor surface was observed havior of GBP1 on gold (coated on quartz) followed by a sharp increase in SPR shift as well as platinum (coated on quartz) by (Fig. 8.4). In the GBP-1 studies, we have using QCM (M. Duman et al., unpub- observed two different regions as a result lished). We showed both the kinetics of of the kinetic evaluation compared with adsorption as well as strength of binding, QCM results. The SPR results suggest and thereby, demonstrated the substrate that following the binding of the GP-1 on specificity-based kinetic parameters. The the gold surface, reorganization of the mol- GBP-1 was found to be stronger binder ecules takes place towards a more stable and more stable on the gold surface com- and two-dimensional ordered molecular pared with the platinum surface. Kinetic film formation. In fact, studying these self- analysis of the experimental data followed assembled surfaces with AFM revealed Langmuir monolayer adsorption. A 20- ordered molecular domains formation of fold differences between the equilibrium GBP-1 on atomically flat Au (111) surfac- constants and fourfold frequency differ- es. It is assumed that the monomolecular ences were obtained when GBP-1 bound ordered GEPI formation on the inorganic to Au and Pt surfaces, respectively. We also substrate in this case is a result of a com- examined other selected metal (e.g. Pd) bination of forces involving intermolecular and metal oxide (e.g. Cu2O, ZnO, SiO2) interaction as well as the lattice recognition binders by QCM system, and found out of Au by GBP-1. To be able to study GEPI their binding affinity and selectivity differ- adsorption and as well as kinetic and ther- ences. Not only selected peptides can be modynamic parameters on selected surfac- Tools for Bionanotechnology | 211 es other than Au and Au, we are presently the substrate through atomic grooves in- developing a model for generating SPR sig- trinsically present on this {112} crystallo- nal from any surface (other metals, oxides, graphic surface. and semiconductors). This new approach We have also extended our preliminary will extend the applicability of SPR analy- computational approaches on understand- sis of the peptide adsorption on any given ing of molecular recognition of inorganic- interesting material and provide invaluable binding polypeptides to phage display information not possible otherwise (L. Yee, selected septapeptides with constrained unpublished). molecular structure. Following their affin- ity characterization based on immunofluo- Modeling tools on polypeptide- rescent microscopy to metallic platinum, inorganic surface interactions we have investigated their crystallographic The experimentally determined binding surface recognition by conformational information using the approaches dis- analysis via HyperChem 7.5 molecular cussed above combined with the simula- modeling system. We used CHARMM tion based structural results could give a force field parameters and showed that the coherent understanding of GEPI-inorgan- selected septapeptides conform into certain ic surface interactions. These will include molecular architecture containing multiple the characteristics of the side chains that protrusions that spatially match with the are capable of interacting the surface atoms crystallographic metal surfaces (i.e. (111), on various materials and crystallographies. (110), and (100)). While the physical Unfortunately, however, studies on molec- recognition may originate from how well ular modeling of protein binding to inor- the molecular polypods spatially match a ganic surfaces have been very limited. Our given crystallographic surface, the reactive first attempt to analyze this interaction groups on the protrusions through which was performed by gold binding protein in the peptide binds to the inorganic surface a collaborative work (Braun et al., 2001). dictates the strength of the binding. Our Different tandem repeats of gold binding molecular modeling analysis is found to peptide was compared with all known pro- be consistent with the experimental bind- tein structures using FASTA searches and ing characteristics of the platinum binders also traced if a particular secondary struc- presenting different affinities (Oren et al., ture preference exists via Chou-Fasman 2005). and Holley/Karplus secondary structure Conventional spectroscopic tech- prediction algorithms. Initial findings sug- niques along with molecular dynamics gest that on Au{111} surface, GBP1 forms studies will shed more light into the chemi- an anti-parallel β pleated sheet conforma- cal specificity of inorganic surface-specific tion. This constraint, places OH groups polypeptides. Contribution of side chain from serine and threonine residues into a of amino acids is appearing as one of the regular lattice based on energy minimized major interactions. Up until now among in vacuo using X-PLOR. Again, based on the studied noble metals, namely, gold, these simulations, we also found that the silver, platinum and palladium binding binding of GBP-1 on {112} is not as it is on polypeptide sequences showed similarities Au{111} surface because of the decoupling in terms of conserved hydrophobicity and caused by the presence of water molecule polarity. A similar trend was also observed that sweeps under the binding protein on in terms of conservation of small amino 212 | Tamerler and Sarikaya

acid molecules. These results might give We define molecular toolbox as a data us a perspective in terms of the existence bank containing fully characterized GEPIs of common binding domain for similar that could be picked up and utilized for a materials. The correlation between how wide range of applications. Just like what functional groups are distributed within proteins can do in organisms, inorganic- the polypeptide depending on the inorgan- binding polypeptides could also be used ic lattice structure or the topology of the as molecular synthesizers, linkers, brac- substrate while conserving their binding ers, erectors, and assemblers in materials domain would be a major contribution to- science and biotechnology. Furthermore, wards tailoring engineered linkers in many GEPI could also be useful when fused to application areas. another protein, macromolecule, or DNA- binding protein as a functional ligand. Practical applications by Alternatively, a GEPI could also be chemi- utilizing GEPIs cally conjugated, onto a synthetic polymer, Nanometer-sized particles and nanostruc- to create multifunctional hybrid polymeric tured inorganics are now fundamental structures. When used in biological and building blocks for technological materi- synthetic macromolecules, GEPI’s role is als and devices both in nanotechnology to create a heterofunctional polymer. The and biotechnology as they are easy to pre- examples below give some areas of the pare, store and manipulate. To make any diverse applications of inorganic binding nanoparticle significantly useful, in addi- peptides for materials engineering and bio- tion to its intrinsic properties (magnetic, technology. photonic, semiconducting), it also needs to be chemically modified so that it can be Morphogenesis of inorganics via assembled in an efficient and controlled engineered polypeptides manner and can be manipulated. This In biomineralization, a significant aspect of is possible if the physical and chemical biological control over materials formation characteristics of the particles can be con- is via protein/inorganic interaction, such trolled and assembly is directed. Therefore, as in bone, dentin, mollusk shells, bacterial numerous challenges must be addressed and algal particle formation. In traditional before nanoscience could be implemented biomineralization, the studies in the search successfully into working systems. Some of for the nature of proteins’ affects on inor- these challenges include synthesizing nano- ganic formation has traditionally focused structures (e.g. particles, rods, and tubes) on templating, nucleation, and enzymatic with uniform size and shape, control their reactions. These studies were mostly carried mineralogy, surface structures and chemis- out the protein(s) extracted, isolated, and try, and organizing them in 2- and 3D with purified from a given hard tissue and the predictable spatial distribution. If biology effects were studied during the reconstruc- is to be a guide, some of these challenges tion of the specific inorganic(s) associated could be overcome utilizing the unique op- with it. Although instructive in pinpoint- portunities offered by the biomimetic ap- ing specific effects in biomineralization, i.e. proach at the molecular scale provided by mineral selection, habit modification, and the engineered inorganic binding polypep- enzymatic effects, these studies, besides be- tides discussed herein. ing extremely slow in progress, have been Tools for Bionanotechnology | 213 limited in their ability to study only a few the gold-colloid (from pale yellow to a red inorganics that these hard tissues are as- colloid) which was related to altered rate of sociated with. Also, neither temporal nor crystallization. Fifty mutants were tested spatial distributions of these proteins or this way, and the sequence analysis showed their fragments have been quantitatively that two separate mutants that accelerated associated with the hierarchical architec- the crystal growth also changed the par- ture of the tissues. ticle morphology from cubo-octahedral With the emergence of combinato- (the usual shape of the gold particles un- rial biology selected inorganic-proteins der equilibrium growth conditions) to flat, discussed herein, a natural first step is to triangular or pseudo-hexagonal, particles examine how and which of these polypep- (Fig. 8.5a). This new observation is inter- tide sequences affect inorganic formation. esting from the point of enzymatic effect of Since done under controlled conditions, protein in crystal growth rather than tra- these studies could provide an opportunity ditionally assumed templating effect. The to investigate various affects in mineraliza- polypeptides, in spite of being slightly ba- tion, including nucleation, growth, mor- sic, may have caused the formation of gold phogenesis, and enzymatic effects. These crystals similar to those formed in acidic studies have necessarily been carried out condition. This suggests that the role of under aqueous environments amenable the polypeptides in gold crystallization is for biological functions and show the po- to act as an acid, a common mechanism in tential use of inorganic binding polypep- enzyme function. tides for biofabrication of a wide variety Our second example focuses on silica, of material synthesis in biological environ- which exhibit diverse and extraordinarily ments. Here we demonstrate the first such designed shapes and structures in Nature study in which the morphology of gold from diatom skeletons to sponge spicules particles were affected by GEPIs selected (Sarikaya, 1999). The proteins directing using the CSD route (Brown et al., 2000) silica synthesis in biological systems have (similar studies were also carried out in Ag been studied extensively. For example, si- formation using GEPIs selected via the laffins extracted from the cell wall ofa PhD route, Naik et al, 2002). Nanogold diatoms Cylindrotheca (Kruger et al., 2000) (monosize, 120-Å diameter) particles can or silicatein extracted from the sponge be formed at ambient conditions using the spicules of Terhya aurantia (Shimizu et well-known Faraday’s technique by reduc- al.,1998) are very good examples of these ing AuCl3 by Na-citrate (or other reducing proteins. Also, silica binding peptides also agents) (Turkovich, 1951). Reducing the identified for the amorphous silica precipi- gold concentration and temperature allow tate which was formed in the presence of particle formation at a slower rate, giving the peptide unit of natural silicatein pro- the protein time to interact with surfaces tein (Weaver et al., 2003). In our research, during the growth and provides conditions we used the selected phage peptide clones to examine the effect of gold-binding dur- and examined a silica precipitation assay. ing colloidal gold formation. In a collabora- We found that the peptides that were rich tive study, one of us conducted a search for in basic and hydroxyl amino acids exhibited mutants that modulated the morphology more silica precipitating activity. We iden- of gold crystallites, i.e. the selection of mu- tified silica binding peptides via 12-amino tants was based on the change of color of acid phage display library on a single-crys- 214 | Tamerler and Sarikaya

Figure 8.5 Examples of applications by combinatorial biology selected inorganic binding sequences. (A–C) Enzymatic effect of protein in crystal growth rather than traditionally-assumed templating effect, effect of gold binding polypeptides on the particle morphology. The effect of GBP (A), boiling (B), and acidic conditions (C). (D–E) Biosilica gel formation. An SEM image of the gel formed by silica-binding phage mutant and silica particles, a viable material (D). Schematics of the gel containing conjugates of phage mutant and silica particles in water (E). (F–H) Directed immobilization achieved via heterofunctionality introduced to alkaline phosphatase through genetic fusion of gold binding polypeptide. (F) Line height profile across ′a-a shown in the AFM image. (G) 3D visualization of the AFM image. (H) Schematical illustration of AP-6GBP on gold.

tal silica (quartz) substrate and compared freshly prepared tetra methyl ortho silicate the differences between single crystal silica (TMOS) solution for various time inter- substrate and silicatein directed synthe- vals at room temperature. In the first pro- sized silica particles. We carried out five cess, silica gels were prepared using silisic rounds of biopanning and 50 sequences acid precursors using alkoxides as a pre- were obtained presenting different affini- cursor in aqueous solutions using slightly ties for silica. Binding characterization of acidic solutions. It was found that the time the selected phage samples were examined for gellation, in this case, depends on the by AFM, immunofluorescence microscopy pH of the precursor solution, in which this experiments, and by SPR (D. Sahin et al., period being very long, sometimes days, unpublished). Some of the high silica-af- depending on the pH value. In the pres- finity presenting selected phage samples ence of silica-binding phage particles, this were then tested for their effect on silica gellation time is reduced to a less than an synthesis. hour, even a few minutes depending on For this, the samples containing 1013 phage concentrations or the strength of phage forming units was incubated in the binding affinity of the inorganic-bind- Tools for Bionanotechnology | 215 ing polypeptide. In addition, post-gellation structs were produced by E. coli cultures processing, e.g. thin-film formation, can be by transforming the cells with the plasmid controlled due to the opportunities provid- having the desired number of repeat units ed for shaping and drying steps in the ma- of gold binding polypeptide insert. By fol- terialization procedures. The gellation of lowing the induction of the E. coli cultures, the TEOS alone takes about 30 h to fully the constructs in the periplasmic fraction gel while precursor/phage mixture can be was isolated and concentrated fraction fol- readily shaped into thin film (tape casting, lowed by purification using DEAE column slip casting, screen printing) or cast into chromatography. Subsequently, the eluted various architectural forms for specific and fractions showing ALKP activity was con- wide variety of applications (Fig. 8.5d). centrated again and passed through gel The solid material can be created through filtration column. The enzymatic activity controlled drying and densification pro- of ALKP was examined by monitoring cesses following the well-established, and the changes in the absorbance spectra of simple, ceramic processing methodologies the solution containing the ALKP-GBP (D. Sahin et al, unpublished). construct and p-nitrophenylphosphate as the substrate in an assay buffer in 500 mM GEPI based directed Tris-Cl (pH:8,0), 1mM MgCl2. To ob- immobilization serve if enzyme is catalytically active in Gold binding polypeptides selected by the presence of gold particles, we tested cell surface display are one of the first ex- the ALKP-GBP constructs following an amples of engineered polypeptides for in- overnight incubation with gold particles at organic surfaces. As discussed before, the 37ºC. The ALKP activity profile for both cell surface displayed selected gold binding wildtype and recombinant ALKP with polypeptide sequences were screened via gold particles were shown no differences. random peptide libraries expressed on the ALKP-GBP constructs were also tested outer surface of E. coli as part of the malto- for stability of their fused gold binding dextrin porin, LamB (Brown, 1997). Once polypeptide inserts in the main protein via they identified, GBPs were then fused to SDS-PAGE gel electrophoresis. The gel the amino terminus of the alkaline phos- analysis showed both the wildtype ALKP photase (ALKP) with their retained gold and the recombinant ALKP-GBP con- binding activity. One of the identified se- struct with an increased molecular weight quences, GBP-1, has been studied in detail corresponding to the number of repeat in our research group. Using the mutant unit of GBP-1. strains of the bacteria supplied to us by In the next step, we examined the pos- Brown, we have incorporated further ge- sible use of gold binding polypeptide as a netic engineering strategies for increasing molecular linker when it is part of ALKP. binding affinity to gold surface via -inser Here, the protein solutions having either tion of multiple repeats of GBP-1 into wildtype or ALKP-GBP were left for over- ALKP as 5 (5R-ALKP), 6 (6R-ALKP), 7 night binding in the presence of the gold (7R-ALKP) and 9 (9R-ALKP). The idea substrate and then the substrate was taken here is to use the GBP-1 as molecular erec- out of solution and the remaining super- tor, and introduce mutifunctionality to the natant was checked for the ALKP catalytic hybrid construct, i.e. enzymatic and inor- activity. While the activity of the wildtype ganic binding simultaneously. ALKP con- protein solution was retained fairly well, no 216 | Tamerler and Sarikaya

Figure 8.6 Potential Application areas of GEPI in the new field of molecular biomimetics.

activity was observed in the supernatant of delivery of ALKP to the desired locations the recombinant construct (Kacer et al., consequently promoting mineralization. unpublished observation). This means that Our results showed that combinato- most of the molecular construct was suc- rially selected polypeptides could be ad- cessfully immobilized on the gold substrate dressably immobilized and self-assembled via the GBP. The result is significant from on inorganic surfaces. Realizing the fact the point that that gold binding property of that chemical linkers, such as thiol and si- the recombinant alkaline phosphatase was lane linkages are the other two major mo- retained simultaneously with the activity of lecular linkers for noble metal and oxide ALKP, creating a portable sensor kit (Fig. (silica) surfaces, respectively, it is naturally 8.5f ). Alkaline phosphatase is an essential expected that self assembled GEPI mono- enzyme for regulating (preventing or en- layers—SA(GEPI)M—as “molecular hancing) biomineralization via the control erector sets” will open up new avenues for of extracellular phosphate concentration designing and engineering new and novel by catalyzing the pyrophosphate degra- functional surfaces for a wide variety of dation. For example, in the dental pulp, nano- and biotechnology applications, in- ALKP is responsible for dentin matrix cluding chemical and biological sensors, formation. The ALKP-GBP bifunctional nanobiotechnology, and proteomics. construct, therefore, could be promising probe especially in periodontal regenera- Future prospects tion. In this application, the new construct For development of materials and systems provides excellent probing properties to architectures at molecular or nanoscale follow the mechanism via the gold bind- levels, proteins and polypeptides could be ing and enzymatic properties and, hence, used as molecular erectors with controlled may be utilized for providing a controlled binding to and assembly on inorganics. Tools for Bionanotechnology | 217

developments and applications. Curr. Opin. The GEPIs represent a class of geneti- Biotechnol. 12, 400–405. cally engineered polypeptides that can be Antikainen, N.M., and Martin, S.F. (2005). tailored on technological demand through Altering protein specificity: techniques and direction evolution for their affinity and applications, Bioorg. Med. Chem. 13, 2701– 2716. specificity. The GEPIs could be utilized as Bachtold, A., Hadley, P., Nakanishi, T. and Dekker, part of hybrid molecular constructs (Fig. C. (2001). Logic circuits with carbon nano- 8.6). The ordered assembly of GEPI on tubes transistors. Science 294, 1317–1320. inorganic surfaces could have a significant Bailey, L.E., Kambhampati, D., Kanazawa, K.K., Knoll, W., and Frank, C.W. (2002). Using sur- impact on molecular biotechnology appli- face plasmon resonance and the quartz crystal cations offering several novel immediate microbalance t monitor in situ the interfacial practical advantages. The attachment of behaviour of thin organic films. Langmuir 18, biomolecules, in particular proteins, onto 479–89. Ball, P. (2001).Life’s lessons in design. Nature 409, solid supports is fundamental in the devel- 413–416. opment of advanced biosensors, bioreac- Beck-Sickinger, A. and Weber, P. (2002). tors, affinity chromatographic separation Combinatorial Strategies in Biology and materials, and many diagnostics such as Chemistry, New York: John Willey and Sons Press. those used in cancer therapeutics (Sarikaya Benhar, I. (2001).Biotechnology applications of et al., 2004: Sanchez et al., 2005). Protein phage and cell surface display. Biotechnol. Adv. adsorption and macromolecular interac- 19, 1–33. tions at solid surfaces play key roles in the Berman, A., Addadi, L. and Weiner, S. (1988). Interactions of sea-urchin skeleton macro- performance of implants and hard-tissue molecules with growing calcite crystals: A engineering (Ranter et al., 1996). Proteins study of intracrystalline proteins. Nature 331, adsorbed specifically onto probe substrates 546–548. are used to build protein microarrays suit- Brown, S. (1992). Engineering iron oxide-adhe- sion mutants of Escherichia coli phage λ re- able for modern proteomics (Cutler, 2003; ceptor. Proc. Natl. Acad. Sci. USA 89, 8651– Chicurel et al., 2002). 8655. Brown, S. (1997). Metal recognition by repeat- Acknowledgments ing polypeptides. Nature Biotechnol. 15, 269–272. We thank our colleagues F. Baneyx, D. Brown, S. (2001).Protein-mediated particle as- Schwartz, F. Ohuchi, and B. Traxler for sembly. Nanoletters 1, 391–394. invaluable discussions, and E.E. Oren, T. Brown, S., Sarikaya, M., and Johnson, E. (2002). Kacar, M. Gungormus, D. Sahin, C. So, E. Genetic analysis of crystal growth. J. Mol. Biol. 299, 725–732. Venkatamaburian, S. Dincer, M. Duman, Cariolou, M.A., and Morse, D.E. (1988). D. Heidel, R. Braun, M.H. Zareie, and H. Purification and characterization of calcium- Fong for technical help. This research was binding conchiolin shell peptides from the sponsored by DURINT/ARO and MR- mollusk, Haliotis-rufescens, as a function of development. J. Comp. Physiol.B. 157, 717– SEC/NSF programs. 729. Cha JN, Shimizu K, Zhou Y, Christiansen SC, References Chmelka BF, et al. (1999). Silicatein filaments Agterberg, M., Adriaanse, H., Tommassen, J. and subunits from a marine sponge direct the (1987). Use of outer membrane protein Phoe polymerization of silica and silicones in vitro. as a carrier for the transport of a foreign an- Proc. Natl. Acad. Sci. USA 96, 361–365. tigenic determinant to the cell surface of Charbit, A., Boulain, J.C., Ryter, A., Hofnung, Escherichia coli K-12. Gene 59(1), 145–150. M. (1986). Probing the topology of a bacte- Amstatz, P., Forrer, P., Zahnd, C. and Plückthun, rial-membrane protein by genetic insertion of A. (2001). In: vitro display technologies: novel a foreign epitope expression at the cell surface. EMBO J. 5(11), 3029–3037. 218 | Tamerler and Sarikaya

Chicurel, M.E., and Dalma-Weiszhausz, D.D. ture, folding and enzymatic activity in proteins. (2002).Microarrays in pharmagenomics—ad- Curr. Opin. Struct. Biol. 9, 514–520. vances and future promise. Pharmacogenomics Freudl, R., Macintyres, S., Degen, M., Henning, 3, 589–601. U. (1986). Cell-surface exposure of the outer Coen, M.C., Lehman, R., Groning, P., Bielmann, membrane protein OMPA of Escherichia coli M., Galli, C., Schlapbach, L., (2001). K-12. J. Mol. Biol. 188(3), 491–494. Adsorption and bioactivity of protein A on Gaskin, D. J.H., Strack, K., Vulfson, E.N. (2000). silicon surfaces studied by AFM and XPS. J. Identification of inorganic crystal-specific se- Colloid Interface Sci. 233, 180–9. quences using phage display combinatorial Cutler, P. (2003). Protein Arrays: The current library of short peptides: A feasibility study. state-of-the-art. Proteomics 3, 3–18. Biotechnol. Lett. 22, 1211–1216. Czenderna, A.W., Lu, C. (1984). Applications of Gittins, D.I., Bethell, D., Schiffrin, D. J., and Piezoelectric Quartz Crystal Microbalances, Michols, R. J. (2000). A nanometer-scale Methods and Phenomena. (New York: USA, electronic switch consisting of a metal cluster Elsevier) and redox-addressable groups, Nature 408, Dai, H.X., Thai, C.K., Sarikaya, M., Baneyx, F., 67–69. Schwartz, D.T. (2004). Through-mask an- Giver, L., and Arnold, F.H. (1998). Combinatorial odic patterning of copper surfaces and film protein design by in vitro recombination, Curr. stability in biological media. Langmuir 20(8), Opin. Chem. Biol. 2, 335–338. 3483–3486. Glotzer, S.C. (2004). Some assembly required, Dani, M. (2001). Peptide display libraries: design Science 306, 419–420. and construction. J. Recept. Signal Transduct. Gooding, J. J., Mearns, F., Yang, W.R., Liu, J.Q. Res. 21, 469–88. (2003). Self-assembled monolayers into the DeGarmo, E.P., Black, J.T. and Kohner, R.A. 21st century:recent advances and applications, (1988). Materials and processes in manufac- Electroanalysis 15(2), 81–96. turing (New York: McMillan, Publ. Co.). Harris, P. J. (1999). Carbon Nanotubes and Dower, W. J., Mattheakis, L.C. (2002) In: vitro se- Related Structures (Cambridge, UK: lection as a powerful tool for the applied evo- Cambridge Univ. Press). lution of proteins and peptides. Curr. Opin. Hoenlein, W. Kreupl, F., Duesberg, G.S., Graham, Chem. Biol. 6, 390–8. A.P., Liebau, M., Seidel, R., Unger, E. (2003). Drexler, K.E. (1992). Nanosystems. New York: Carbon nanotbues for microelectronics:status Wiley Interscience. and future prospects, 23, 663–669. Evans, J.S. (2003). ‘Apples’ and ‘oranges’: Hoess, R.H. (2001). Protein design and phage Comparing the structural aspects of biomin- display. Chem. Rev. 101, 3205–3218. eral- and ice-interaction proteins. Curr. Opin. Homola, J., Sinclair, S.Y., Gauglitz, G., (1999). Coll Interf. Sci. 8, 48–54. Surface plasmon resonance sensors: a review. Falini, G., Albeck, S., Weiner, S., and Addadi, L. Sensors Actuators B 54, 3–15. (1996). Control of aragonite and calcite pol- Huang, Y., Duan, X., Cui, Y., Lauhon, L. J., Kim, morphism by mollusk shell macromolecules. K.H., Lieber, M.L. (2001). Logic gates and Science 271, 67–69. computation from assembled nanowires. Feldhaus, M. J., Siegel, R.M. (2004). Yeasy Science 294, 1313–1317 Display of antibody fragments:a discovery and Ishli, D., Kinbara, K., Ishida, Y., Ishli, N., characterization platform. J. Immunol. Met., Okochi, M., Yohda, M., and Aida, T. (2003). 290, 69–80. Chaperorin-mediated stabilization and ATP- Ferry, D.K., and Goodnick, S.M. (1997). triggered release of semiconductor nanopar- Transport in Nanostructures, Cambridge, ticles. Nature 423, 628–231. UK: Cambridge Univ. Press. Izrailev, S., Stepaniants, S., Balsera, M., Oono, Y., Fitzgerald, K. (2000). In: vitro display tech- and Schulten, K. (1997). Molecular dynamics nologies: new tools for drug discovery. Drug study of unbinding of avidin–biotin complex. Discovery Today 5(6), 253–258. Biophys. J. 72, 1568–1581. Fong, H., Sarikaya, M., White, S., and Snead, Jackson, J.B., Westcott, S.L., Hirsch, L.R., West, M.L. (2000). Nanomechanical properties J.L., and Halas, N. J. (2002). Controlling the profiles across DEJ in human incisor teeth, J. surface enhanced RAMAN effect via the Mater. Sci. Eng.– C 7, 119–128. nanoshell geometry. Appl. Phys. Lett. 82, Forrer, P., Jung, S., Plunckthun, A. (1999). Beyond 257–259. binding:using phage display to select for struc- Tools for Bionanotechnology | 219

Joern, J.M., Meinhold, P., Arnold, F. (2002). Lipovsek, D., and Pluckthun, A. (2004). In-vitro Analysis of shuffled gene librariers. J. Mol. protein evolution by ribosome display and Biol. 316, 63–656. mRNA display. J. Immunol. Methods 290, Jung, L.S., Campbell, C.T., Chinowsky, T.M., 51–67. Mar, M.N., and Yee, S.S. (1998). Quantitative Long, J.R., Shaw, W. J., Stayton, P.S., and Drobny, interpretation of the resonance of surface G.P. (2001). Structure and dynamics of hy- plasmon resonance sensors to adsorbed films. drated statherin on hydroxyapatite as deter- Langmuir 14, 5636–5648. mined by solid-state nmR. Biochemistry US Kaplan, D., Adams, W.W., Farmer, B., and Viney, 40, 15451–15455. C. (1994). Silk-Biology, Structure, Properties, Lowenstam, H.A., and Weiner, S. (1989). On and Genetics. Silk Polymers ACS Symposium Biomineralization. Oxford: Oxford University Series 544, 2–16. Press. Katz E, Willner I. (2004). Integrated nanopar- Lu, Z., Murray, K.S., VanCleave, V., LaVallie, ticle-biomolecule hybrid systems: synthesis, E.R., Stahl, M.L., McCoy, J.M. (1995). properties and applications. Angewandte Expression of thioredoxin random peptide Chemie 43, 6042–6108. libraries on the Escherichia coli cell surface as Kay B, Winter J, McCafferty, (1996). Phage functional fusions to flagellin: a system de- Display of Peptides and Proteins, A Laboratory signed for exploring protein–protein interac- Manual, California: Academic Press. tions. Biotechnology 13, 366–72. Kiargaard, K., Sorensen, J.K., Schembri, M.A., Ma, D., and Li, M. (2001). Application sof display and Klemm, P. (2000).Sequestration of zinc technologies to proteomic analyses. J. Cell. oxide by fimbrial designer chelators. Appl. Biochem. Suppl 37:34–41. Env. Microbiol. 66, 10–14. Mann, S. (1988). Molecular recognition in Klaus,T., Joerger, R., Olsson, E., and Granqvist,C- biomineralization. Nature 332, 119–123. G. (1999). Silver-based crystalline nanopar- Mann, S. (ed.) (1996). Biomimetic Materials ticles bacterially fabricated. Proc. Natl. Acad. Chemistry. New York: VCH Publ. Inc.. Sci. USA 96, 13611–13614. Mann, S., and Calvert, P. (1998). Synthesis and Koneracka, M. et al. (2002). Direct binding pro- biological composites formed by in-situ pre- cedures of proteins to fine magnetic particles. cipitation. J. Mater. Sci. 23, 3801–3815. J. Mol. Catal. B-Enzym. 18, 13–18. Mao, C., Flynn, C.E., Hayhurst, A., Sweeney, R., Kroger, N., Deutzman, R., and Sumper, M. Qi, J., Georgiou, G., Iverson, B., Belcher, A.M. (1999). Polycatyonic peptides from diatom (2003). Viral assembly of oriented quantum biosilica that direct silica nanosphere forma- dot nanowires. Proc. Natl. Acad. Sci. USA tion. Science 286, 1129–1132. 100, 6946–6951. Kruger, N., Deutzmann, R., Bergsdork, C., Mayer, G., and Sarikaya, M. (2002). Rigid biolog- Sumper, M. (2000). Species-specific poly- ical composite materials: Structural examples amines from diatoms control silica morphol- for biomimetic design. Exp. Mechanics 42, ogy. Proc Natl. Acad. Sci. USA 97, 14133– 395–403. 14138. McDonald, S.A., Konstantatos, G., Zhang, S.G., Kulp, J.L., Sarikaya, M., Evans, J.S. (2004). Cyr, P.W., Klem, E.J.D., Levina, L., Sargent, Molecular characterization of a prokaryotic E.H. (2005). Solution-based PbS quantum polypeptide sequence that catalyzes Au crys- infrared photodetectors and photovoltaics, tal formation. J. Mater. Chem. 14(14), 2325– Nature Mater. 4(2),138–144. 2332. Mrksich, M. (2002). What can surface do for cell Lee, W.S., Mao, C., Flynn, C.E., and Belcher, biology. Curr. Opin. Chem. Biol. 6, 794–797. A.M. (2002). Ordering quantum dots using Muller, B. (2001).Natural formation of nanostruc- genetically engineered viruses. Science 296, tures: from fundamentals in metal heteroepi- 892–895. taxy to applications in optics and biomaterials Li, C.M., Botsaris, G.D., and Kaplan, D.L. (2002). science. Surf. Rev. Lett. 8, 169–228. Selective in vitro effects of peptides on calcium Murray, B.S., Deshaires, C. (2000). Monitoring carbonate crystallization. Crystal Growth protein fouling of metal surfaces via a quartz Design 2, 387–393. crystal microbalance. J. Colloid Interface Sci. Liou, Y.C., Tocilj, A., Davies, P.L., and Jia, Z.C. 227, 32–41. (2000). Mimicry of ice structure by surface Naik, R.R., Brott, L., Carlson, S. J. and Stone, hydroxyls and water of a beta-helix antifreeze M.O. (2002). Silica precipitating peptides protein. Nature 406, 322–324. isolated from a combinatorial phage display libraries. J. Nanosci. Nanotechnol. 2, 1–6. 220 | Tamerler and Sarikaya

Naik, R.R., Stringer, S.J, Agarwal, G., Jones, S.E., Sanchez, C., Arribart, H., and Madeleine, G.G. and Stone, M.O. (2002). Biomimetic synthesis (2005). Biomimetism and bioinspiration as and patterning of silver nanoparticles. Nature tools for the design of innovative materials and Mater. 1, 169–172. systems, Nature Mater. 4, 277–288. Niemeyer, C.M. (2001). Nanoparticles, pro- Sarikaya, M, Tamerler, C., Schwartz, D.T., Baneyx teins, and nucleic acids: Biotechnology meets F. (2003). Molecular biomimetics: nanotech- materials science. Angew. Chem. Int. Ed. 40, nology through biology. Nature Mater. 2, 4128–4158. 577–85. Nygaard, S., Wandelbo, R., and Brown, S. (2002). Sarikaya, M. (1999). Biomimetics: Materials Surface-specific zeolite-binding proteins. Adv. Fabrication through biology. Proc. Natl. Acad. Mater. 14, 1853–1856. Sci. USA 96, 14183–14185. Oren, E.E., Tamerler, C., Sarikaya, M. (2005). Sarikaya, M. et al. (2001). Biomimetic model of Molecular recognition of septapeptides via a sponge-spicular optical fiber—mechanical molecular polypod architecture. Nanoletters. properties and structure. J. Mater. Res. 16, 5(3), 415–419. 1420–1428. Paine, M.L. and Snead, M.L. (1996). Protein in- Sarikaya, M. Steinberg, B., and Thomas, G. teractions during assembly of the enamel or- (1982). Optimization of Fe/Cr/C base struc- ganic extracellular matrix. J. Bone Miner. Res. tural steels for improved strength and tough- 12, 221–226. ness. Met. Trans. 13A, 2227–2237. Parmley, S.F., Smith, G.P. (1988). Antibody-se- Sarikaya, M., and Aksay, I.A. (eds.) (1995). lectable filamentous fd phage vectors: affinity Biomimetics: Design and Processing of purification of target genes. Gene. 73, 305– Materials. New York: American Institute of 18. Physics. Petrouna, I.P., and Arnold, F.H. (2000). Designed Sarikaya, M., Fong, H., Heidel, D., and Humbert, evolution of enzymatic properties. Curr. Opin. R. (1999). Biomimetic assembly of nano- Biotechnol. 11, 325–329. structured materials. Mater. Sci. Forum 293, Quist, A.P., Pavlovic, E., Oscarsson, S. (2005). 83–87. Recent Advances in microcontact printing. Sarikaya, M., Tamerler, C., Schwartz, D.T., Anal. Bioanal. Chem. 381(3), 591–600. Baneyx, F.O. (2004). Materials assembly and Ratner, B., Schoen, F., Hoffman, A., and Lemons, formation using engineered polypeptides. J. (1996). Biomaterials Science: Introduction Ann. Rev. Mater. Res. 34, 373–408. to Materials in Medicine. San Diego: USA, Scembri, M.A., Kjaergaard, K., and Klemm, P. Academic Press. (1999). Bioaccumulation of heavy metals by Ratner, B., Schoen, F., Hoffman, A., Lemons, J. fimbrial designer adhesions. FEMS Microbial. (1996). Biomaterials Science: Introduction Lett. 170, 363–371. to Materials in Medicine. San Diego, USA: Schmidt-Dannert, C., Arnold, F.H. (1999). Academic Press. Directed evolution of industrial enzymes. Reed, B.W., Zareie, M.H., Hong, M., Jen, A.K.Y., Trends. Biotechnol. 17, 135–6. and Sarikaya, M. (2002). Applications of a Schneider, G., and Wrede, P. (1998). Artificial simple topographic imaging model to submo- neural networks for computer-based molecu- lecular-resolution scanning probe microscopy. lar design. Biophys. Mol. Biol. 70, 175–222. Submitted to Ultramicroscopy. Schultze, S., Harauz, G., Beveridge, T.I. (1992). Riechmann, L., Holliger, P. (1997). The C-termi- Participation of a cyanobacterial S layer in nal domain of ToIA is the coreceptor for fila- fine-grain mineral formation. J. Bacteriol. 174, mentous phage infection of E. coli. Cell 90(2), 7971–7981. 351–360. Seeman, N.C., and Belcher, A.M. (2002). Rodi, D. J., Makowski, L. (1999). Phage display Emulating biology: building nanostructures technology-finding a needle in a vast molecu- from the bottom up. Proc. Natl. Acad. Sci. lar haystack. Curr. Opin. Biotechnol. 10(1), USA 99, 6452–6455. 87–93. Shimizu, K., Cha, J., Stucky, G.D., Morse, D.E. Rodi, D. J., Makowski, L., Kay, B.K. (2002). One (1998). Silicatein α: Cathepsin l-like protein column A and two column B: the benefits of in sponge biosilica. Proc. Natl. Acad. Sci. USA phage display in molecular-recognition stud- 95, 6234–6238. ies. Curr. Opin. Chem. Ciol. 6(1), 92–96. Shipway, A.N. and Willner, I. (2001). Samuelson, P., Gunneriusson, E., Nygren, P.Å., Nanoparticles as structural and functional Ståhl, S. (2002). Display of proteins on bacte- units in surface confined architectures, Chem. ria. J. Biotechnol. 96:129–54 Commun. 20, 2035–2045. Tools for Bionanotechnology | 221

Sickinger, A.B., and Weber, P. (2002). of processes in the synthesis of colloidal gold. Combinatorial Strategies in Biology and Trans. Faraday Soc. Disc. 11, 55–75. Chemistry. New York: John Wiley & Sons. Ueda, M. (2004). Future Direction of molecular Sidhu, S.S. (2000). Phage display in pharmaceu- display by yeast cell-cell surface engineering, tical biotechnology. Curr. Opin. Biotechnol. Journal of Molecular Catalysis B: Enzymatic, 11(16), 610–616. 28, 139–143. Slocik, J.M., Moore, J.T., and Wright, D.W. Weaver, J.C., Morse, D.E. (2003). Molecular biol- (2002). Monoclonal antibody recognition of ogy of demosponge axial filaments and their histidine-rich peptide capsulated nanoclus- roles in biosilicification. Micro. Res. Tech. 62, ters. Nano. Lett. 2, 169–173. 356–67. Smith, G.P. (1985). Filamentous fusion phage: Wernerus, H., and Stahl, S. (2004). Novel expression vectors that display cloned Biotechnological applications for surface en- antigens on the virion surface. Science 228, gineered bacteria. Biotechnol. Appl. Biochem, 1315–1317. 40, 209–228. Smith, G.P., and Petrenko, A. (1997). Phage dis- Whaley, S.R, English, D.S., Hu, E.L., Barbara, play. Chem. Rev. 97, 391–410. P.F., and Belcher, M.A. (2000). Selection of Smith, G.P., Scott, J.K. (1993). Libraries of pep- peptides with semiconducting binding speci- tides and proteins displayed on filamentous ficity for directed nanocrystal assembly. Nature phage. Methods Enzymol. 217, 228–57. 405, 665–668. Stahl, S. Uhlen, M. (1997). Bacterial surface dis- Whitesides, G.M., Mathias, J.P., and Seto, C.T. play: Trends and progress. Trends Biotechnol. (1991). Molecular self-assembly and nano- 15(5), 185–192 chemistry: a chemical strategy for the synthesis Takahashi,T.T., Austin,R. J., and Roberts,R.W. of nanostructures. Science 254, 1312–1319. (2003) mRNA display: ligand discovery, inter- Wittrup, K.D. (2001). Protein engineering by cell action analysis and beyond. Trends Biochem. surface display. Curr. Opin. Biotechnol. 12, Sci. 28, 159–165. 395–399. Thayer, G.E., Hannon, J.B., and Tromp R.M. Yonezawa, M., Doi, N., Kawahashi, Y., (2004). Shape and stability of self-assembled Higashinakagawa, T., and Yanagawa, H. surface domains, Nat. Mater. 3, 95–98. (2003). DNA display for in vitro selection of Turkevich, J., Stevenson, P.C., and Hillier, J. diverse peptide libraries. Nucleic Acids Res. (1951). A study of the nucleation and growth 31, 118