Protein engineering to explore andimprove affinity ligands

Martin Linhult

Royal Institute of Technology

Department of Biotechnology

Stockholm 2003

ISBN 91-7283-596-6

Stockholm 2003 © Martin Linhult

Royal Institute of TechnologyAlbanova University CenterDepartment of BiotechnologySE-106 91 StockholmSweden

Printed at Författares BokmaskinS:t Eriksgatan 10Box 12071SE-102 22 Stockholm

Linhult M.2003. Protein engineering to explore and improve affinity ligands.Department of Biotechnology, Albanova University Center, Royal Institute ofTechnology, Stockholm, Sweden.

AbstractIn order to produce predictable and robust systems for protein purification and detection, wellcharacterized, small, folded domains descending from bacterial receptors have been used. Thesebacterial receptors, staphylococcal protein A (SPA) and streptococcal protein G (SPG), possess highaffinity to IgG and / or HSA. They are composed of repetitive units in which each one binds the ligandindependently. The domains fold independently and are very stable. Since the domains also have well-known three-dimensional structures and do not contain cysteine residues, they are very suitable asframeworks for further protein engineering.

Streptococcal protein G (SPG) is a multidomain protein present on the cell surface ofStreptococcus. X-ray crystallography has been used to determine the binding site of the Ig-bindingdomain. In this thesis the region responsible for the HSA affinity of ABD3 has been determined bydirected mutagenesis followed by functional and structural analysis. The analysis shows that the HSA-binding involves residues mainly in the second a-helix.

Most protein-based affinity chromatography media are very sensitive towards alkaline treatment,which is the preferred method for regeneration and removal of contaminants from the purificationdevices in industrial applications. Here, a protein engineering strategy has been used to improve thetolerance to alkaline conditions of different domains from protein G, ABD3 and C2. Amino acidsknown to be susceptible towards high pH were substituted for less alkali susceptible residues. The new,engineered variants of C2 and ABD shown higher stability towards alkaline pH. Also, very importantfor the potential use as affinity ligands, these mutated variants retained the secondary structure and theaffinity to HSA and IgG, respectively. Moreover, dimerization was performed to investigate whether ahigher binding capacity could be obtained by multivalency. For ABD, binding studies showed thatdivalent ligands coupled using non-directed chemistry demonstrated an increased molar bindingcapacity compared to monovalent ligands. In contrast, equal molar binding capacities were observedfor both types of ligands when using a directed ligand coupling chemistry involving the introductionand recruitment of a unique C-terminal cysteine residue.

The staphylococcal protein A-derived domain Z is also a well known and thoroughly characterizedfusion partner widely used in affinity chromatography systems. This domain is considered to berelatively tolerant towards alkaline conditions. Nevertheless, it is desirable to further improve thestability in order to enable an SPA-based affinity medium to withstand even longer exposure to theharsh conditions associated with cleaning in place (CIP) procedures. For this purpose a differentprotein engineering strategy was employed. Small changes in stability due to the mutations would bedifficult to assess. Hence, in order to enable detection of improvements regarding the alkalineresistance of the Z domain, a by-pass mutagenesis strategy was utilized, where a mutated structurallydestabilized variant, Z(F30A) was used as a surrogate framework. All eight asparagines in the domainwere exchanged one-by-one. The residues were all shown to have different impact on the alkalinetolerance of the domain. By exchanging asparagine 23 for a threonine we were able to remarkablyincrease the stability of the Z(F30A)-domain towards alkaline conditions. Also, when grafting theN23T mutation to the Z scaffold we were able to detect an increased tolerance towards alkalinetreatment compared to the native Z molecule. In all cases, the most sensitive asparagines were found tobe located in the loops region.

In summary, the work presented in this thesis shows the usefulness of protein engineeringstrategies, both to explore the importance of different amino acids regarding stability and functionalityand to improve the characteristics of a protein.

Keywords: binding, affinity, human serum albumin (HSA), albumin-binding domain(ABD), affinity chromatography, deamidation, protein A, stabilization, Z-domain,capacity, protein G, cleaning-in-place (CIP), protein engineering, C2 receptor.

© Martin Linhult

An intellectual is a man who says a simple thing in a difficult way;an artist is a man who says a difficult thing in a simple way.

Charles Bukowski, “Notes of a Dirty Old Man”

List of publicationsThis thesis is based on the following publications, referred to in the text by theirrespective roman numerals:

I. Linhult, M., Binz, K., Uhlén, M. and Hober, S. (2002) Mutational analysis ofthe interaction between Albumin-Binding Domain (ABD) from Streptococcalprotein G and Human Serum Albumin (HSA). Protein Sci. 11: 206-213.

II. Gülich, S., Linhult, M., Nygren, P-Å., Uhlén, M. and Hober, S. (2000)Stabilization of an affinity protein ligand towards Cleaning In Place (CIP)conditions. J Biotechnol. 80:169-78.

III. Linhult, M., Gülich S., Gräslund, T., Nygren, P-Å. and Hober, S. (2003)Evaluation of different linker regions for multimerization and the couplingchemistry for immobilization of a proteinaceous affinity ligand. ProteinEngineering In press.

IV. Gülich, S,. Linhult, M., Ståhl, S. and Hober. S. (2002) EngineeringStreptococcal protein G for increased alkaline stability. Protein Engineering15(10):835-42.

V. Linhult, M., Gülich, S., Gräslund, T., Simon, A., Karlsson, M.,Sjöberg, A., Nord, K. and Hober, S. (2003) Improving the tolerance of proteinA to repeated alkaline exposures using a by-pass mutagenesis approach.Proteins: Struct., Funct. and Genet. In press.

Table of contentsIntroduction 10

Protein structure 11

Protein engineering 13

Site directed mutagenesis 13

Combinatorial approaches 15

Protein stability 17

Rational design 20

Tolerance to alkaline conditions 22

Protein-protein interactions 24

Kinetic studies 24

Measuring protein-protein interactions 26

Bacterial surface receptors 27

Staphylococcal protein A 28

Streptococcal protein G 30

HSA binding region 30

IgG binding region 32

Affinity chromatography purification 33

Present investigation 36

HSA binding domain 36

Mutational analysis of an ABD 36

Stabilization of ABD towards alkaline conditions 40

IgG binding domains 49

Stabilization of C2 towards alkaline conditions 49

Stabilization of Z towards alkaline conditions 53

Concluding remarks 58

Acknowledgements 59

Abbreviations 60

References 61

10

IntroductionFifty years have now passed since the structure of the DNA double helix was

discovered (Watson and Crick 1953). In this time, great efforts have been made in the

field of life sciences, leading to advance across a range of important and emerging

areas. A continued challenge for scientists in the coming years will be to further

deepen our understanding of biology at the molecular level, including all classes of

biomolecules involved in life processes. An increased knowledge about protein

function in particular has become especially important in the so-called post genomic

era, where many unknown gene products are currently being characterized for

localization, expression levels, interactions and ultimately function. Somewhat

simplified, proteins are linear biomolecules of different lengths which are synthesized

from 20 different amino acid building blocks by the cellular machinery according to a

blueprint stored as genetic information. Following synthesis, the linear molecule

undergoes complex folding processes forming various secondary structure elements

such as a-helices, b-sheets and different turn/loops. Subsequent arrangement of these

structure elements results in the final three-dimensional structure of a protein. Despite

the limited number of building blocks, proteins can be very diverse with respect to for

example their functions, and the environments in which they act. Fundamental

questions to answer in this respect are for example how proteins have evolved to

interact with other proteins and to be function in radically different environments.

Protein engineering is one approach that can be used to shed some light onto these

questions, and increase the utility of the proteins in different biotechnological and

industrial applications. Protein engineering techniques have proven fruitful for

resolving questions in the field of biochemistry since the late 1970s; pioneered by the

work done on site-directed mutagenesis on Tyrosyl-transfer RNA synthetase (TyrS)

by Michael Smith and co-workers in Vancouver (Hutchison et al. 1978; Winter et al.

1982). In recent years, protein engineering has evolved from studies involving the

substitution, deletion or insertion of amino acids one-by-one, to studies involving the

manipulation of larger parts of proteins. This has become possible through the

development of a range of new techniques, for instance the polymerase chain reaction

(PCR) facilitating genetic work, gene shuffling for generation of diversity and

different in vitro selection techniques, such as phage display. At its inception, protein

engineering was used as a tool for understanding more about protein structure and

11

function. More recently, a number of different applications have come into existence,

proving the success of protein engineering outside the lab. For instance, proteases and

lipases have been engineered to work well in modern washing machines, and

antibodies are now being engineered for clinical use in therapeutic applications

(Holliger and Bohlen 1999; McCafferty and Glover 2000).

Protein structure

The building blocks of proteins and peptides are the amino acids linked together with

planar peptide bonds, described by Pauling as early as 1960 (Pauling 1960). The 20

natural amino acids differ in their respective properties, such as hydrophobicity, size,

shape, and electrostatics. By altering the combinations and numbers of these amino

acids, nature can build an immense number of more or less complex protein

structures, of different stabilities and functions.

The different properties of amino acids make them more or less suited to be

present in different types of secondary structure elements (usually referred to as

having different helix or b-sheet propensities). A number of studies have been

performed to explore such different amino acid propensities in a-helices and b-sheets

(Lyu et al. 1990; O'Neil and DeGrado 1990; Horovitz et al. 1992; Blaber et al. 1993;

Munoz and Serrano 1994; Smith et al. 1994; Smith and Regan 1995; Myers et al.

1997). In addition, also loops and turns have been investigated in relation to amino

acid propensities (Williams et al. 1987; Wojcik et al. 1999; Crasto and Feng 2001).

These results can be beneficial in protein engineering, giving advice on which amino

acids could be introduced in what structure elements without disturbing the overall

structure. A popular strategy to investigate the importance of a particular amino acid

position for a given trait is to replace the native amino acid with an alanine. The

characteristics of this amino acid (relative small, pH-independent properties, high a-

helix and b-sheet propensities) make it a good first choice in site-directed

mutagenesis approaches (Fersht 1999).

The a-helix consists of a right-handed coil wherein the n residue forms a hydrogen

bond with the n+4 residue, resulting in 3.6 residues per turn. Hence, neither the first

nor the last residues in the a-helix can make the intra-helical hydrogen bond between

the backbone C=O groups of one turn and the NH groups of residues in the next

(Fersht, 1999). This requires that for stabilization of the helix, hydrogen bonds must

12

be made with either other groups in the protein or the solvent. Capping boxes, termed

N- and C-caps are often occurring in the N- and C-terminal to stabilize the ends of the

helices. These residues form side chain hydrogen bonds to the backbone of the

polypeptide chain inside the helix (Richardson and Richardson 1988; Doig 2002). The

alignment of the dipoles in the polypeptide backbone causes a net dipole moment

along the helix. The N-terminal is positively polarized while the C-terminal is

negatively polarized Thus, there is a preference for negatively charged N-terminal

amino acids, and consequently a preference for positively charged C-terminal amino

acids (Hol 1985; Chakrabartty et al. 1993). Attempts to stabilize a-helices via

favoring the electrostatics has been successful, for example, the introduction of

negatively charged amino acid in the N-terminus of an a-helix of the T4 Lysozyme

protein resulted in a stabilization as determined by (Nicholson et al. 1988).

Polypeptide chains can also form complementary hydrogen bonds to a parallel

polypeptide chain. These parallel chains can be aligned in either the same or opposite

directions. This class of secondary structure elements is called the b-sheet. Because of

the nature of b, sheets-always forming a hydrogen bond with another polypeptide

chain, it is more difficult to predict b-sheet occurrence and amino acid propensities

for b-sheet than for a-helices (Minor and Kim 1994b; a). Interestingly, high quantities

of b-sheets occur in proteins prone to form arrays of insoluble fibrils, which are

involved in for instance prion diseases such as Creutzfeldt-Jakob Disease (CJD)

(Prusiner 1998).

Loop and turn regions connect the a-helices and b-sheets. The definitions of turns

and loops are not very distinct, with turns consisting of a few residues while loops are

longer and have less well-defined structures. Polar and charged residues quite often

occur on these surface exposed areas (Leszczynski and Rose 1986; Argos 1990). The

structures described above constitute secondary structure elements of a protein. The

tertiary structure of a protein is in turn defined by how these secondary structures of a

single polypeptide chain are grouped together. A number of proteins consist of two or

more polypeptide chains. The three dimensional arrangement of the different domains

is in turn called the quaternary structure.

Our understanding of the function/structure relationship of a protein is obviously

facilitated if a three-dimensional structure of the protein is available. To date (sept

2003) 20 413 protein, peptide and virus structures have become available in the

13

Brookhaven Protein Databank; 17 653 which have been solved by X-ray

crystallography and 2760 which have been solved by nuclear magnetic resonance

(NMR). Another technique used to investigate protein structure is Circular Dichroism

(CD) (Pace 1997a), a spectroscopy-based technique that measures the differential

absorption of left and right circularly polarized light. The CD technique quickly and

easily provides information about the secondary structure content of a protein sample

(Schmid 1997; Kelly and Price 2000). CD spectroscopy is widely used as a tool in

protein engineering, since it rapidly can assess the secondary structure of a range of

protein constructs. By superimposing the spectra of different mutants, it is possible to

verify structural differences. Another advantage with CD is that it is applicable in

different experimental conditions, allowing for the collection of spectra in for

example, different concentration of denaturants, different pHs and at different

temperatures which can be used to analyze the stability of a protein (Kelly and Price

2000). The main limitation of CD is that it only provides relatively low-resolution

structural information, but the conclusions can be enhanced through using secondary

structure prediction algorithms (Ribas De Pouplana et al. 1991; Unneberg et al. 2001).

Protein engineering

Site directed mutagenesis

Site-directed mutagenesis has proven to be a valuable tool for increasing our

understanding of different biomolecular principles, for instance enzyme mechanisms,

folding pathways and biomolecular recognition. By changing a limited number of

amino acids, and then investigating how these new amino acids influence the

inherited function, the contribution of the various building blocks can be mapped. A

high-resolution three-dimensional structure of the protein of interest (both the wild

type and the variants) will obviously facilitate the evaluation of the collected data.

However, even with the accumulated knowledge from numerous studies of different

proteins, it is still very difficult, if not impossible to make accurate predictions about

the effects, even from a single amino acid substitution in a protein. Therefore protein-

engineering strategies could turn out to be a rather laborious. However, when both the

three-dimensional structure is known, and the active part of the target protein is

mapped, then the capacity for predicting the effect is greater. The first example of site

14

directed mutagenesis was presented by Michael Smith and co-workers in 1978. In this

experiment, single-stranded viral DNA from an Escherichia coli (E. coli) specific

bacteriophage was used as a template for mutagenesis. A mismatch oligonucleotide

consisting of 12 bases was added and after hybridization, DNA polymerase I was

used to make the DNA double stranded (ds). After ligation using T4-DNA ligase, the

vector was transfected into E.coli cells resulting in the production of phages

exhibiting the desired mutation (Hutchison et al. 1978). However, since both strands

(wild type and mutation-containing strand) can serve as templates for replication,

several clones must often be screened before a “pure” mutant is found. Later, Kunkel

and coworker developed a more practical system (Kunkel et al. 1987), wherein the

template vector first is propagated inside a uracil n-glycsoylate deficient (ung-) strain.

This strain, to some extent incorporates uracil nucleotides instead of thymidines

nucleotides into newly synthesized DNA. The heteroduplex DNA containing DNA

with uracil in the template strand and newly synthesized DNA in the second strand is

used to transfect an ung+ strain. The strain will efficiently degrade the template DNA

strand, as it has incorporated uracil nucleotides. This can dramatically increase the

mutation frequency (Kunkel et al. 1987; Kunkel et al. 1991). In the mid 1980s Kary B

Mullis invented the PCR technique for enzymatic in vitro amplification of DNA.

Variants of this technique have later been developed for site-specific mutagenesis

purposes (Figure 1). In these methods PCR is performed with primers including the

mutations to be inserted resulting in a PCR products containing the desired mutation.

15

Figure 1. A PCR-based method for site-directed mutagenesis developed by Ho and

coworkers (Ho et al. 1989), where (A) shows two separate PCR reactions which are

performed with primers (a+b) and (c+d) respectively on the same template where c and b

includes the desired mutation. B shows the two PCR products which both contain the

inserted mutation. C shows the use of the previous produced PCR product as templates in a

second PCR reaction using primers a and b resulting in a full-length product harboring the

mutation (D).

Combinatorial approaches

An alternative way to change the properties of a protein would be to create a library

consisting of numerous member entities, and then carry out a selection or a screening

procedure to find a member with desirable properties. Over the last few years, several

techniques to create protein libraries have been developed providing links between

genotype and phenotype, thereby facilitating the identification of the selected

members by sequencing of the encoding gene. Phage display technology has to date

been the most commonly used system for selection (Clackson and Wells 1994), which

is described in the next paragraph. Recently a number of in vitro systems have been

developed, for instance, ribosomal display (Hanes and Pluckthun 1997) and water-in-

oil emulsion selection (Tawfik and Griffiths 1998). One significant advantage with

these systems is that they have overcome the library size-limiting step of

A

B

a

b

c

d

C

d

a

D

16

transformation in that cell-free extract rather than intact cells are used for the

biosynthesis of the library members.

The first phase when using protein-library methods are to create a gene library

encoding for the different members. Several different methods have lately been

reported in the literature for the generation of diversity in a library. One strategy is to

employ synthetic oligonucleotides to code for a part of the gene. Here parts of the

oligonucleotide are randomized to create diversity. Error-prone PCR is another

widespread technique in which a thermostable DNA polymerase is forced to

misincorporate nucleotides under certain conditions, such as the addition of Mn2+

(Miyazaki and Arnold 1999). In 1994 Stemmer described another strategy for creating

diversity involving DNA shuffling. In this method, homologous genes are digested

and reassembled using PCR to yield a pool of new genes (Stemmer 1994).

Once the library containing the different gene constructs is established, it needs to

be analyzed to find a member with the desired properties. This procedure can be

performed using a screening or selection procedure. In the screening procedure, each

member in the library is analyzed at the time. One elegant screening method which

have been developed by Phillips and co-workers use the in vivo folding properties to

screen for variants with an increased thermodynamic stability. The system is based on

detection of protein folding in vivo and makes use of the distance-dependent

fluorescence resonance energy transfer (FRET) from the green fluorescent protein

(GFP) to the blue fluorescent protein (BFP). FRET only occurs when BFP and GFP

are in close proximity. The system is based on the construction of a ternary fusion

protein in which a protein of interest (X) is fused between GFP and BFP. If protein X

is unfolded and degraded by cellular proteases FRET between the donor BFP and the

acceptor GFP does not occur (Philipps et al. 2003).

Working with selection, the optimal goal is to design a selection procedure in

which only the variants with desired properties are isolated. Phage display is a

powerful selection method applicable to both proteins and peptide libraries. The

principle was first described in 1985 for the display of EcoRI endonuclease fragments

on the surface of the E. coli filamentous phage M13 (Smith 1985). Phage display

enables a physical link to be established between the protein of interest and the

encoding genetic material. A foreign gene sequence is fused to the gene for one of the

phage coat proteins, in most cases pIII or pVIII, making a hybrid fusion protein

displayed on the surface of the phage. Thus when constructing a combinatorial library

17

of the protein or peptide of interest, each member exhibiting a unique property will be

accessible on the surface of the phage enabling “fishing out” of the protein with the

desired properties. Any method capable of separating clones with a desired trait from

their background can be used as a selection method. In principle, most selection

methods are based on affinity for a certain target molecule. After binding the phages

to the target, non-specific binders are washed away. The phages that remain bound are

then eluted and subsequently amplified. The amplified phages are again presented for

the target, while non-specific binders are washed away. This routine is then repeated

four to six times to find the desired proteins linked to the encoding DNA by phage

particles (Clackson and Wells 1994).

Protein stability

Over the last few years, stabilization of proteins has been of great interest to scientists

for academic as well as economic reasons. When discussing protein stability one

should bare in mind that different environment will affect the protein of interest

differently. Proteins that have high thermal stability are not necessarily stable in high

concentration of different unfolding agents, such as Urea, GdnHCl or alkaline

solutions. Many attempts have been made to understand the structural and chemical

reasons for the higher stability of enzymes isolated from extremozymes, in terms of

their three-dimensional (3D) structure, than enzymes adapted to physiological

environments. A structural comparison between mesophilic and extremophilic

enzymes could illuminate ways for identifying mutations that could lead to stabilized

variants. In some cases this has indeed been possible (Eijsink et al. 1995; Perl et al.

2000; Delbruck et al. 2001). Notable structural differences between mesophilic and

thermophilic enzymes could for instance include increased compactness, shorter

surface loops, smaller internal cavities and large ion-pair networks, as in the case for

the enzyme citrate synthase (Danson and Hough 1998). However, using this

knowledge in stabilizing proteins has proven to be more difficult than expected. The

notable differences, which could be used as rules to stabilize mesophilic proteins,

have been shown to not be generally applicable, and are therefore definitely not

universal (Van den burg and Eijsink 2002).

In an evolutionary context of adaptations to high temperatures, it can be kept in

mind that the hyperthermophilic archaea represent the shortest phylogenetic lineage to

18

the root of the universal phylogenetic tree derived from small subunit (SSU) rRNA

sequence data (Pace 1997b) (see Figure 2). Therefore stability, or at least

thermostability, could be regarded as a starting point for evolution. Hence, by

studying proteins from this source it might be possible to adapt numerous proteins for

high temperature environment.

Figure 2. Universal phylogenetic tree based on SSU rRNA sequences. The scale bar

corresponds to 0.1 changes per nucleotide. Reproduced, with permission from Pace (1997).

Thermostabilization of proteins would be advantageous for a number of reasons; for

instance, an increased thermal stability would be beneficial for many industrial

applications. A process running at a higher temperature would be accompanied by

19

higher reaction rates, higher solubility of reactants and a lower risk of microbial

contaminations (van den Burg and Eijsink 2002).

It is possible to increase the tolerance for a protein by adding extrinsic factor, such

as glycine betaine. Glycine betaine has been found to stabilize a firefly luciferase used

in bioluminometric assays at elevated temperature (Eriksson et al. 2003). Extrinsic

factor seems also play a role in nature for different thermophilic bacteria. There is

increasing evidence that hyperthermophilic archaea (archaebacteria) produce organic

solutes such as 2-O-b-mannosylglycerate at high temperature (Hensel 1988; Martins

1997). Studies have also actually shown that these factors could have a direct effect

on enzyme thermostability (Ramos 1997).

In the field of protein stabilization through protein engineering, three different

main strategies could be outlined, as random mutagenesis, consensus approaches and

rational design. Recently, several groups have shown via different combinatorial

techniques that it is possible to stabilize a protein by random mutagenesis. For

instance, directed evolution, error-prone PCR, and gene shuffling are all techniques

that have led to the creation of many new biocatalysts, including variants with

significantly improved stability (Gershenson and Arnold 2000; Arnold 2001;

Bornscheuer and Pohl 2001; Chen 2001). The common theme in these strategies is the

creation of a library from which a suitable variant could be selected. The selection

procedure must be high-throughput as well as sufficiently discriminative. Preferably,

assay conditions are such that only stabilized variants are active, thereby enabling

them to be detected. For example, selections for a protein that is essential for growth

and survival of the host can be carried out in a thermophile (Gershenson and Arnold

2000). Another elegant selection strategy which can be mentioned here is one

designed by Sieber with co-workers using a phage display system (Sieber et al. 1998).

The general concept for the system is that an unfolded polypeptide chain, positioned

C-terminally to one of the domains of protein III in a multidomain fusion protein

displayed on the phage particle, would be cleaved through proteolysis. Hence, phage

clones for which this occurs will loose the infectivity.

20

Rational design

Rational design strategies have proven to be successful in making a protein more

tolerant to different environments, such as increased temperature, extreme pHs and

proteolytic enzymes. Traditionally, the most common strategy for increasing the

structural stability for stabilizing a protein has been to change residues internally in

the hydrophobic core (Van den Burg and Eijsink 2002). As the knowledge about

protein structure and stability has increased, changing residues at the surface has

turned out to be more successful strategy, since the hydrophobic core in most cases

already is well packed. Difficulties in predicting the effects of one or more of the

changed amino acids in the structurally important hydrophobic core may result in a

less stable protein, or even a ruined scaffold. Furthermore, as the hydrophobic core is

most often very well packed, the possibility to improve the stability by surface

mutations, which instead influence the electrostatic interactions on the surface, is

higher (Van den Burg and Eijsink 2002; Grimsley et al. 1999; Martin et al. 2001). In

spite of the degree of unpredictability a number of rational approaches for structural

stabilization have proven successful for a variety of proteins. These approaches

include the improvement of the packing of the hydrophobic core, the stabilization of

a-helix dipoles, the engineering of surface salt bridges and point mutations aimed at

reducing the entropy of the unfolded state (Nicholson et al. 1988; Van den Burg et al.

1998; Grimsley et al. 1999; Wang et al. 1999; Olson et al. 2001).

By comparing the amino acid content between two highly homologous domains,

Serrano and co-researchers presented a rational design strategy for stabilizing a

protein (Serrano et al. 1993). In this approach, each individual amino acid that differs

between the two domains is mutated. Afterwards, each mutated protein is analyzed

and the mutations that increase the stability can be combined to construct a multiple

mutant. However, this strategy is quite tedious and has to be carried out carefully, as

probably only a few key mutations will contribute to the increased stability (Van den

Burg and Eijsink 2002).

Eijsink and co-workers outlined another interesting alternative strategy in which a

reduction of local unfolding could be of importance for increasing the thermal

stability. For a neutral protease from Bacillus, it was found that stabilizing mutations

were all clustered in a certain surface region of the protein (Van den Burg et al. 1998;

Vriend et al. 1998). Further stabilization of the protein was found to be quite

21

straightforward as this region had been localized (Mansfeld et al. 1997; Van den Burg

et al. 1998).

The consensus approach, which is a semi-rational approach developed by

Lehmann and co-researchers, can be regarded as an expansion of Serrano’s rational

design of protein stabilization (Serrano et al. 1993; and previous section). This

strategy is built around the idea that the amino acids contributing to the stability of the

protein acid is retained during evolution and thereby a consensus residue should

contribute more to the stability of a protein than a non-consensus amino acid.

Therefore, while comparing the amino acid sequences of different homologous

proteins, the most frequently occurring residue should be used (Lehmann et al. 2000;

Lehmann and Wyss 2001; Lehmann et al. 2002). This strategy has turned out to be

successful for obtaining a highly thermostable fungal phytases (Lehmann et al. 2000)

and a more thermostable a-helical tetratricopeptide repeat (TPR) protein than its

natural counter part (Main et al. 2003).

Increasing the stability of proteins to proteases is also of interest in industrial

applications of enzymes. If the proteolytic pattern is specific, it would be relatively

easy to pinpoint the sensitive regions of the protein, and thus by rational design, alter

the amino acid content and stabilize the protein. In some cases, while a higher

temperature will unfold the proteins, the proteolytic stability correlates with the

thermal stability due to the exposure of flexible amino acid sequences (McLendon

and Radany 1978; Daniel et al. 1982; Parsell and Sauer 1989; Akasako et al. 1995).

However, even folded proteins could be attacked by proteases if they contain regions

that are accessible and flexible (Price 1990; Hubbard 1998). There are two different

foci for altering amino acids to increase their tolerance to proteolysis; i.e decrease the

conformational flexibility of the attacked structural region or changing the primary

structure and introducing amino acids that are disadvantageous for the protease

(Frenken et al. 1993; Markert et al. 2001).

22

Tolerance to alkaline conditions

Changing the primary structure by introducing amino acids that could increase the

tolerance of the protein towards harsh environment, such as temperature, pH or

proteases, is a general concept for some of the work presented in this thesis.

Asparagines are known to be susceptible to high pH, through covalent modifications

such as deamidation (see path a in Figure 3) or backbone cleavage (see path b in

Figure 3). Since NaOH is a common agent used to clean chromatographic media in

large-scale processes (for a more detailed description, see the section Affinity

chromatography) it is of importance for affinity ligands suitable for coupling to

chromatographic matrices to be tolerant to high pH. Glutamine is also susceptible,

though it is modified to a lesser extent than asparagine. These reactions are

spontaneous and may also occur at physiological solvent conditions, often resulting in

the loss of activity of the protein or peptide (Geiger and Clarke 1987). The extent of

modification of the different residues is highly sequence and conformation dependent

(Kossiakoff 1988; Lura and Schirch 1988; Kosky et al. 1999). The deamidation

reaction involves the main chain peptide nitrogen succeeding the asparagine. The

nitrogen functions as the nucleophile and attacks the side-chain carbonyl of

asparagine, resulting in a succinimide intermediate. This succinimide intermediate

may open at either of the two C-N bonds to form aspartic acid or isoaspartic acid,

resulting in an addition of a negative charge. These two isomers may occur in their L

or D forms (see path a in Figure 3).

Cleavage of the peptide chain could also occur at the asparagine residues. In these

reactions the mechanism proposed involves the side-chain amide nitrogen attacking

the Ca, forming a C-terminal succinimide, as illustrated in path b in Figure 3. Both

reactions are believed to occur simultaneously, although in general the deamidation

reaction is faster (Tyler-Cross and Schirch 1991). An interesting hypothesis has been

proposed by Robinson, in which through the reaction mechanisms presented above,

glutaminyl and asparaginyl residues in peptides and proteins act as molecular timers

of biological events such as protein turnover, development, and aging (Robinson et al.

1970; Robinson 1974; Robinson and Rudd 1974; Robinson and Lubke 1978; Tyler-

Cross and Schirch 1991).

23

Figure 3. Reaction mechanism for deamidation and backbone cleavage of asparagine

residues in peptides and proteins via the formation of a succinimide intermediate; where in a

the five-member succinimide intermediate is formed via a nucleophilic attack on the side-

chain carbonyl carbon of asparagine by the backbone nitrogen of the ensuing amino acid,

releasing NH3. The succinimide ring is labile to bases and readily opens, yielding either

aspartic acid or isoaspartic acid. In b, peptide cleavage could also occur through a

succinimide intermediate by attacking the side-chain amide of asparagine on the backbone

peptide carbonyl of asparagine.

C

NH2

NHRNH

H2C

C

H2C

O

O

RN+1

C

NH2

RNH

H2C

C

H2C

O

O

NRN+1

C

NH

RNH

H2C

C

H2C

O

O

RN+1

C

O

NHRNH

H2C

C

H2C

O

O

RN+1

O

O

C

NH

RNH

H2C

C

H2C

RN+1

O

a bAsparagine

SuccinimideSuccinimide

Aspartate isoAspartate

24

Protein-protein interactions

Many biological processes are regulated through protein interactions involving non-

covalent associations, for example signal transduction, cell regulation, immune

response, and regulation of enzymatic activities. Apart from basic science aspects, a

deeper understanding of the underlying principles of such biomolecular interactions

would be helpful for the design of improved or novel affinity proteins. In addition,

accurate prediction of binding surfaces, or hotspots from a three-dimensional structure

(Ma et al. 2001; DeLano 2002) would be beneficial for the design of either agonist or

antagagonist compounds for therapeutic applications. In general, hotspots residues in

protein-protein interaction are defined as residues that upon mutation cause a large

shift in binding affinity.

Kinetic studies

A specific, rapid association of protein complexes is essential for a range of diverseprocesses. The rate of association of a protein complex is diffusion controlled, and in

a random collision theory, the rate constant could be up to approximately 109 M-1 s-1

(Schreiber and Fersht 1996; Schreiber 2002). Most determined association rates are

some thousands times lower, but association rates close to the diffusion limit have

been measured (Fersht, 1999). In these cases, the speed of the process is functionallyimportant, and introduction of additional electrostatic forces has been used to enhance

the rate of association (Schreiber and Fersht 1996). The association of a pair ofproteins can be described as a two-step reaction, A+B ‹=› [AB]* ‹=› AB, where A and

B are the free proteins, [AB]* is the encounter complex, and AB is the final complex.

The two proteins, A and B tumble randomly around until they reach an areadesignated the steering region. These collisions create a transition state that

consequently produces the energetically favoured complex AB. The mechanismsbehind such complex formation have been a matter of considerable research

(Northrup and Erickson 1992; Janin 1997; Gabdoulline and Wade 1999; Selzer and

Schreiber 2001).

Many different strategies can be used to increase our understanding of the kinetics

of protein-protein interactions, for instance, adding cosolvents, introducing mutations

and changing the temperature or pH. The surroundings of the molecules are of great

importance for the relative association rate; for example a simple relationship between

25

kon and viscosity with a slope of ~1 has been demonstrated (Raman et al. 1992; Stone

and Hermans 1995). This means that the relative association rate (koon/k

von) in the

absence and presence of a viscous agent is linearly dependent on the relative viscosity

(h/ho). In addition, ionic strength and pH are of obvious importance for the

association of two biomolecules and different studies have shown a relationship

between kon and ionic strength which follow the Debye-Hückel equation

(Vijayakumar et al. 1998; Selzer and Schreiber 1999).

Measuring the effect of mutation is a powerful tool that can be used to interpret

protein-protein interactions. Extensive mutagenesis of the surface residues of different

binding partners, such as TEM1-BLIP, barnase-barstar, interferon-receptor, growth

hormone-receptor and interleukin-4-receptor, has demonstrated that mutations

involving charged residues significantly affect the kon, whereas mutations of

uncharged residues do not (Schreiber and Fersht 1995; Wang et al. 1997; Clackson et

al. 1998; Albeck and Schreiber 1999; Piehler et al. 2000; Selzer et al. 2000). When

introducing a substitution into a given protein, careful consideration has to be paid to

the structural changes the new residue will generate. Hence, it is essential to carry out

a structural comparison between the wild-type protein and the new mutated version.

To ascertain whether the change in affinity upon mutation of two different residues is

energetically independent on each other, employing a double mutant cycle is an

effective strategy. Double mutant cycles were introduced in 1984 by Carter and co-

researchers to study structural changes (Carter et al. 1984) changes. Later studies have

shown that this strategy also is suitable for investigating protein-protein interactions

(Schreiber and Fersht 1995; Frisch et al. 1997). By comparing the difference change

in the free energy (DDG) upon complex formation between the single mutants (A and

B) and HSA and the double mutant (C) and HSA (DDG=DGC-DGA-DGB), a

synergistic effect introduced by the mutations can be detected. If the effects of the two

single mutations are independent (which means DDG = 0), the residues do not effect

each other; otherwise DDG ≠ 0, which means that the mutations are dependent on

each other.

One very interesting benefit arising from this knowledge is how the association

rate is dependent on the electrostatic interactions. Using rational design, it has been

shown possible to design faster association without affecting the dissociation (Selzer

et al. 2000). The method is based on increasing the electrostatic attraction between the

26

proteins by incorporating charged residues in the vicinity of the binding interface.

Improving the steering region, by changing the electrostatic interaction between the

molecules, the association phase changes from random diffusion to directional

movement towards the complex formation. This could be a more advantageous

approach for enhancing binding affinity than affinity maturation by creating a library

and going through a selection procedure.

How important is the association and dissociation for the biological activity in the

different events? A number of previous studies have indicated that the individual

kinetic constants are indeed important; for example the human growth hormone

receptor (hGH) requires an increase in dissociation of 30 fold compared to its wild

type in order to bring about a linear dependence in EC50 for cell proliferation (Pearce

et al. 1999). Similar results have also been observed for interleukin-4 and interferon

(Piehler et al 2000; Wang et al 1997).

Measuring protein-protein interactions

A number of techniques currently exist for investigating protein-protein interactions.

Determination of the binding region through structural analysis of co-complexes, by

for example X-ray crystallography and nuclear magnetic resonance (NMR), are two

examples (Deisenhofer 1981; Wider and Wüthrich 1999). Measurements of the

strengths of interactions between the two proteins could be done for example with

isothermal titration calometry (ITC), surface plasmon resonance (SPR) and optical

spectroscopy (Lakey and Raggett 1998). In combination with site-directed mutations,

all of these could provide an excellent platform for investigating interactions.

One technique for measuring protein-protein interactions that has also been greatly

relied on in this thesis is SPR-based biosensor technology (Biacore). This

instrumentation monitors the complex formation and dissociation between

macromolecules in real time by transducing the accumulation of mass of an analyte

molecule at the sensor surface that is coated with a ligand molecule. The instrument

has been used to investigate numerous different biomolecular systems, including:

cytokine growth-factor-receptor recognition (Cunningham et al. 1989; Cunningham

and Wells 1989; 1993), coagulation factor assembly (Fan et al. 1998) and virus-cell

docking (Rux et al. 1998; Willis et al. 1998). There are several advantageous with

using optical biosensors, such as the possibility of real-time measurements allowing

27

for dissection of the interaction into association rate and dissociation rate components,

dispensing the need for labeling the biomolecules, Therefore the technique is very

useful in the design of antagonists and ligands whose efficacy depends on their

binding dynamics (Myszka 1997; Canziani et al. 1999). Sometimes it may not be

necessary to conduct a complete kinetic study, but rather an affinity and rate ranking

of the analytes, to investigate the relative affinity of two biomolecules to the same

target. To facilitate the interpretation it is advantageous if these different binding

molecules have approximately the same molecular weights.

A number of research groups have shown that using optical biosensor in

combination with site-directed mutagenesis provides a powerful technique for

identifying structural elements that are key determinants of specificity and affinity

(Bass et al. 1991; Albeck and Schreiber 1999; Gårdsvoll et al. 1999).

Bacterial surface receptors

Several pathogenic Gram-positive bacteria interact with host proteins through

receptors expressed as anchored to the bacterial surface. The biological significance

of this phenomenon in relation to pathogenicity and the immunogenic response is not

fully clear, but most probably the pathogenic bacteria create a layer of host proteins

around themselves (Achari et al. 1992; Sauer-Eriksson et al. 1995; Starovasnik et al.

1996). Surface proteins that are capable of binding to host specific proteins include

for example streptococcal protein G (SPG) and staphylococcal protein A (SPA). SPA

and SPG have both high affinities for immunoglobulin G (IgG) from different species,

and SPG can also bind serum albumin from a variety of species. SPA and SPG have

been thoroughly investigated for various biotechnology purposes such as

immunological tools in a vast array of immunoassays, and purification of

immunoglobulins (Amersham Biosciences 1997). Interestingly, the HSA-binding

domains from SPG and the IgG binding domains from SPA share the same scaffold

despite the fact that SPG’s IgG binding domain has a different scaffold. The Fc

binding parts of SPA and SPG are example of convergent evolution, where two

proteins have found two different structural solutions for binding similar proteins with

similar affinity.

28

Staphylococcal protein A

SPA is a cell-wall associated receptor exposed on the surface of the Gram-positive

bacterium Staphylococcus aureus. SPA has high affinity to IgG from various species,

for instance human, mouse, rabbit and guinea pig. (Richman et al. 1982; Amersham

Biosciences 1997). This indicates that SPA facilitates the ability of the bacteria to also

infect animals other than humans. The gene encoding SPA was sequenced by Uhlén

and co-workers in 1984 (Uhlén et al. 1984). SPA consists of three different regions; S,

being the signal sequence that is processed during secretion (Abrahmsen et al. 1985),

five homologous IgG binding domains E, D, A, B and C (Moks et al. 1986) and a cell

anchoring region XM (Guss et al. 1984; Schneewind et al. 1995) (see Figure 4).

A number of features of the IgG-binding domains on SPA; such as being highly

soluble, proteolytically stable, and devoid of cysteins have made these suitable for use

as affinity gene fusion partners for production and purification of recombinant

proteins (Ståhl 1999). In addition to the most used expression host E. coli, SPA has

been successfully expressed also in yeasts, insect cells and mammalian cells (Ståhl

and Nygren 1997).

Figure 4. The staphylococcal protein A shown here is a cell-wall associated receptor

consisting of a signal sequence (S) processed during secretion, five homologous IgG-binding

domains (E, D, A, B and C), and a cell-wall attaching structure (XM). Also shown is the

commonly used Z domain, corresponding to an engineered version of the B domain of SPA

(Nilsson et al. 1987).

Staphylococcal protein A (SpA)

S DE A B C X M

Z

57 kDa

7 kDa

Immunoglobulin binding

29

Each of the five domains in SPA is arranged in an antiparallell three-a-helical bundle

of approximately 58 aa. The structure is stabilized via a hydrophobic core. Each

domain has high affinity for the Fc part of IgG1, IgG2 and IgG4, estimated to be

approximately Kaff =10-8 (M), but shows only weak interaction with IgG3 (Kronvall

and Williams 1969; Ankerst et al. 1974; Jendeberg et al. 1997). In addition, each

domain has high affinity for the Fab part of certain antibodies (Potter et al. 1996;

Jansson et al. 1998). The binding site for the Fc part of the IgG molecule has been

shown in a study of the B domain, to involve 11 residues of helix 1 and helix 2

(Deisenhofer 1981). In addition, another study using the D domain showed that the

Fab-binding part was located distinctly apart from the Fc-binding part. The 11

residues involved in the Fab interaction are situated on the second and third helices

(Graille et al. 2000). Domain B is closest to a hypothetical consensus of the five SPA

domains (Moks et al. 1986) and was therefore chosen for further improvements.

In order to increase domain B’s own tolerance towards site-specific chemical

cleavage of fusion proteins using the chemical cleavage agent hydroxylamine, the

sensitive Asn-Gly dipeptide in the B domain at residues 28-29 was changed to Asn-

Ala by site directed mutagenesis (Nilsson et al. 1987), resulting in an engineered

domain denoted Z. This allowed for development of bioprocesses where after an

initial purification step of ZZ-target fusion proteins with IgG-affinity purification,

hydroxylamine could be used to cleave off Z from the target protein via an Asn-Gly

dipeptide sequence genetically introduced between ZZ and the target protein. In a

second IgG affinity purification step, the cleaved target protein was collected in the

flow through, while Z-domain proteins were captured in the column (Moks et al.

1987).

While introducing the single mutations A29G to the B domain the Fab interaction

diminished, probably due to the fact that the Ala would perturb the interaction

between the two molecules (Jansson et al. 1998; Graille et al. 2000). For the

application as immobilized affinity ligand for capture of IgG, previous work indicate

that a head-to-tail dimer construct of the Z domain has a similar molar binding

capacity for IgG as the native SPA molecule, which has five repetitive domains

(Ljungquist et al. 1989). The Z scaffold has also been used in protein engineering to

introduce new properties. For example 13 surface exposed amino acids at the binding

site could be randomized in a phage display library and several new binders were

possible to select (Nord et al. 1995; Nord et al. 1997).

30

Streptococcal protein G

In 1984 Björck and Kronvall isolated and named the protein G receptor from

Streptococcus strain G148 (Björck and Kronvall 1984). SPG is exposed on the surface

of the bacteria and consists of a signal peptide that is processed during secretion and

repetitive regions capable of binding immunoglobulins and serum albumin from

different species, respectively. The albumin-binding region is separated from the

immunoglobulin-binding region by a spacer region and a cell-wall anchoring region

named W is located in the C-terminus (Olsson et al. 1987) (Figure 5).

Figure 5. Schematic presentation of SPG. The albumin and IgG- binding regions consist of

three albumin and IgG-binding domains respectively (Olsson et al 1987), with a spacer

region, S between these two regions. The signal peptide Ss is processed during secretion

and W is the cell-anchoring region. The albumin-binding region has been cloned in different

constructions denoted BB, ABP and ABD containing 2.5, 2 and 1 domains respectively.

HSA binding region

The HSA-binding region of SPG consists of three homologous albumin-binding

domains (ABDs) separated by linkers of approximately 30 residues each (Olsson et al.

1987). Different parts of the albumin-binding region have been cloned and

characterized, (BB, ABP and ABD) with 2.5, 2 and 1 domains respectively (Ståhl and

Nygren 1997) (Figure 5). Each ABD domain consists of approximately 46 amino

acids. The domain has no additional stabilizing features such as bound ligands, metal

ions or disulphide bridges (Kraulis et al. 1996). HSA is postulated to contain one

binding site for protein G, formed by loops 6-8 (Falkenberg et al. 1992). Protein G

Streptococcal protein G

Ss E A1 B1 A2 B2 A3 S C1 D1 C2 D2 C3 W

Serum albumin binding

63 kDa

BB

ABP

ABD

25 kDa

15 kDa

5 kDa

Immunoglobulin binding

214 aa

121 aa

46 aa

31

shows strong interaction with serum albumin from different species including rat,

mouse, rabbit and human, but very low interaction to bovine serum albumin (Nygren

et al. 1990; Falkenberg et al. 1992; Johansson et al. 2002). These results imply that

streptococcus strain 148 could also be pathogen for rat and mouse.

Many of the positive features associated with the Z-domain; for example, stability

to proteolysis, high-level production, ability to be secreted, and a solubilization of the

fused protein can also be related to ABD (Nygren et al. 1988; Ståhl et al. 1997).

The ability of the ABD to bind serum albumin is also of interest from a therapeutic

perspective. Various difficulties have been encountered when administering proteins

for therapy purposes, such as low efficacy due to a rapid clearance rate from the

blood. Because kidneys generally filter out molecules below 60 kDa, efforts to reduce

clearance rates have focused on increased molecular size through glycosylation or the

addition of polyethylene glycol polymers (PEG) (Dennis et al. 2002). Alternatively,

the HSA binding ABD has been investigated as a gene fusion partner to produce

ABD-tagged proteins capable of binding to circulating serum albumin after

administration. Serum albumin is a protein with a long half-life exemplified by HSA,

which has an in vivo T1/2 of 19 days in humans (Peters 1985) Hence the half-life of

ABD would be extended by the binding to serum albumin. By fusing the human

soluble complement receptor type 1 (sCR1) to the albumin binding part of the SPG

Makrides with co-workers were able to an extended half-life of the sCR1 in rats

(Makrides et al. 1996).

Furthermore, the albumin-binding region BB seems to have immunopotentiating

properties when genetically fused to an immunogen (Power et al. 1997; Sjölander et

al. 1997; Libon et al. 1999). This property has not been fully elucidated yet, but it has

been proposed that it might result from T-cell epitopes (Goetsch et al. 2003) or the

serum albumin-binding affinity (Makrides et al. 1996; Sjölander et al. 1997). In a

recent study, it was proposed that the region involved in serum albumin binding also

contained a T cell epitope (Goetsch et al. 2003).

32

IgG binding region

The IgG-binding region of SPG consists of three individual IgG-binding domains,

where each one has high affinity for the Fc and the Fab regions of IgG. In contrast to

protein A, SPG exhibits binding to all human subclasses of IgG, including IgG3. SPG

also has a high affinity for IgG from several animals, including mouse, rabbit and

sheep (Åkerström et al. 1985; Amersham Biosciences 1997). The domains consist of

about 55 amino acids and the structure is constituted of two b-hairpins that are

associated to form a four stranded mixed antiparallel/parallel b-sheet with a single a-

helix lying across one face of the sheet. No extra stabilizing features such as metal

ions or disulfide bridges is needed for the stability of the domains as in the case of the

three helical scaffold of SPG and SPA (Åkerström et al. 1985; Gronenborn et al.

1991). Interestingly, the IgG binding domains of SPG show clear structural

similarities with the immunoglobulin light chain binding protein L from

Peptostreptococcus magnus (Wikström et al. 1994), despite the fact that their

sequence homology is low. The overall folds of proteins L and G are similar, though

the two proteins also demonstrate significant differences, the orientation of the a-

helices in protein L run nearly parallel to the b-sheet, while the helices in protein G

run diagonally across the b-sheet (Wikström et al. 1995). The Fc binding part of the

SPG molecule is located mainly in charged and polar residues of the helix and the

loop connecting the helix with the third b-strand (Gronenborn and Clore 1993).

Protein G binds Fc at the region that connects the CH2 and CH3 domains, thus SPA

and SPG bind to overlapping sites of the Fc-molecule. Interestingly, the Fab binding

region is almost non-overlapping with the Fc binding part of the molecule, located

mainly in the second b-strand and in the loop between the first and second b-strand

(Lian et al. 1994).

In summary, the bacterial domains originating from the surface receptors of SPA and

SPG are small, easy to produce, stable, and can be secreted. In addition, several

studies have shown that it is possible to change a number of amino acids at the

surface and retain the native structure (Klemba et al. 1995; Nord et al. 1995; Gräslund

et al. 2000). These features make the domains very interesting for protein engineering.

33

Affinity chromatography purification

A number of different protein purification techniques exist today, including ion-

exchange chromatography, reverse-phase and gel filtration, which take advantage of

differences in electrostatic characteristics, hydrophicity or the size of proteins in order

to separate them. One very powerful technique for protein purification is based on

biospecific interactions, namely affinity purification. A number of different

interactions have proven successful in this area, including for example interactions

between enzyme and substrate, bacterial receptor and serum protein, and antigen and

antibody (Nilsson et al. 1997). Some of the most commonly used affinity partners

have been listed in Table 1.

Table 1. Commonly used affinity fusion systems, modified from Nilsson et al. 1997.

Affinity tag Ligand Elution ReferencesProtein A hIgG low pH Ståhl and Nygren. 1997Z hIgG low pH Nilsson et al. 1987

Albumin binding protein

(ABP)

HSA low pH Nygren et al. 1988

Glutathione S-tranferase

(GST)

Glutathione Reduced

glutathione

Smith and Johnson 1988

Polyhistidine tag Me2+/Chelator Imidazole/low

pH

Porath et al. 1975

Maltose binding protein Amylose Maltose di Guan et al. 1988

FLAG peptide mAb1, mAb2 EDTA/low pH Hopp 1988

These different systems all have various characteristics, which have to be taken into

consideration when selecting purification strategy. Common for all listed systems is

that the target protein needs to be modified to include the affinity gene fusion partner,

which, if a native target product is desired, calls for efficient means to release the

affinity tag after the purification. In other and more attractive formats of affinity

chromatography, native target proteins can be directly purified, which obviously

requires that ligands recognizing such native targets are available. By usingcombinatorial chemistry, it is possible to generate new specific binders that could be

used as affinity ligands (Nord et al. 1997). Affinity chromatography has a number of

advantages over other purification techniques currently available; for instance it is an

easy, fast and selective means of capturing biomolecules. This allows the introduction

of an affinity-purification step early in the purification chain, and the number of

successive unit operations can be reduced, which would be a benefit in industrial

purification (Harakas 1994; MacLennan 1995). Achieving the final product in large-

scale industrial applications often requires the overall purification procedure to

include several consecutive unit operations. Previously it has been shown to be

34

advantageous to conduct initial purification before the affinity chromatography step,

for example precipitation or ion-exchange chromatography (Cutler 1996) for

removing some of the major contaminants and thereby increasing the column lifetime.

Purification of plasma proteins, such as human serum albumin (HSA) or

Immunoglobulin (IgG) has for a long time been employed for therapeutic

applications. In the mid-1940s, ethanol fractionation was introduced and is still in use

today together with other separations techniques, such as ion-exchange and affinity

chromatography (Foster 1992). Recently, affinity purification has been applied in fine

and rapid capture of plasma proteins in large-scale production. Some of the proteins

produced and purified in this process include factors VIII, IX and XI, von Willebrand

factor, protein C and antithrombin III (Burnouf and Radosevich 2001). Using affinity

techniques has dramatically improved the purity of the products, and thus unwanted

side effects, such as hemolysis, hypotension and fever is considered to be very rare

(Burnouf and Radosevich 2001). Development of affinity chromatography for

purification of plasma products in large-scale production faces a number of different

issues, such as the selection and development of ligands, leaching problems, the

stability of ligands exposed to cleaning procedures, and the capacity of the column

(Burnouf and Radosevich 2001).

In an industrial plant, the cleaning in place CIP solution is routinely applied

throughout the process after each purification procedure. This is done to diminish

contamination between different processes or batches. A universal chemical agent that

has been shown to be successful in the inactivation of most microorganisms, such as

bacteria, viruses, yeasts and also destroys endotoxins is NaOH in solution (Girot et al.

1990; Burgoyne et al. 1993; Asplund et al. 2000). However, the harsh conditions

associated with such CIP procedures involving high pH will decrease the performance

of many proteinaceous ligands in an affinity-chromatography process. The most

sensitive amino acids to alkaline treatments in several studies have been shown to be

Asn and Glu (Geiger and Clarke 1987; Kossiakoff 1988; Lura and Schirch 1988). The

overall stability of the protein is also dependent on secondary structure elements

surrounding these amino acids and neighboring amino acids (Robinson and Robinson

1991; Wright 1991; Kosky et al. 1999; Xie et al. 2000; Robinson and Robinson

2001). A more detailed description about the susceptibly of asparagine residues in

alkaline conditions can be found in the section of the thesis entitled tolerance to

alkaline condition. Improving the stability of a protein in alkaline treatment will not

35

only improve its performance as an industrial affinity purification ligand, it might also

decrease the leaching problem. As a result, the costs of using a proteinaceous ligand

in a large-scale purification scheme will be reduced. The use of affinity-based

chromatography has as a consequence of the above issues, hitherto not been as

extensively used as predicted.

36

Present InvestigationThe research presented in this thesis covers a number of different applications of

protein engineering technology where site-directed mutagenesis has been used to

study and improve the performance of bacterial domains in biotechnological

applications.

In order to explore the binding surface of ABD originating from SPG, a number of

surface exposed amino acids were substituted and the different mutants were analyzed

with SPR technology. Thereby, the contribution from the different amino acid on the

affinity to HSA was discovered.

To improve the ABD for biotechnology applications, amino acids sensitive to

alkaline conditions were replaced to obtain an increased tolerance to alkaline

conditions. In addition, different connective linkers were analyzed, for the production

of divalent ligands. Since the stabilization strategy worked successfully in improving

the alkaline stability of ABD, a similar strategy was investigated for alkaline

stabilization of two other protein domains, namely the IgG-binding C2 domain

originating from SPG and the Z domain originating from SPA.

HSA binding domain (I)

Mutational analysis of an albumin binding domain

An alanine scanning procedure was carried out to localize the surface on ABD

responsible for the interaction with HSA. Initially, residues in positions E3, Y20, E32

and E40 were replaced. These amino acids were chosen as they were surface-exposed,

located on all three a-helices, and pointing in different directions. In order to ensure

an easy purification procedure, all ABD constructs were fused to the IgG- binding Z-

domain. Of the four first residues chosen, Y20A was determined to contribute most to

the interaction (see Table 2). To determine more precisely which amino acids were

involved in the interaction, further alanine substitutions were made in helix 2.

Residues in positions S18, D19, Y21, K22, N23, L24 and K29 were replaced by

alanine to determine their respective contributions to the interaction with HSA. To

analyze the effect of multiple mutations on structure and affinity, three additional

mutants were made, ABD(Y20A, E40A), ABD(Y20A, E32A) and ABD(S18, Y20A,

37

K22A). An extensive biosensor-based binding study was performed to analyze the

affinity of the different mutated protein domains to HSA.

The CD spectra of the mutated proteins were recorded to analyze the effect on the

secondary structure, as this has proven to be suitable for detecting structural changes

in a-helical proteins (Johnson 1990; Nord et al. 1995). The various mutants of ABD

were subjected to a subtractive CD spectroscopic analysis in which the signal

contribution from the Z domain was subtracted (Nord et al. 1997). The results from

the CD measurements show that the backbone configuration is similar for all

constructs. However, because the exact structures of the different mutants are

unknown, minor changes in the structure due to the mutations cannot be ruled out

(Clackson et al. 1998; Vaughan et al. 1999).

Table 2 presents the results of the kinetic study with the SPR technology (Biacore).

Residues in the second helix give a larger contribution than residues in the first and

third helices. In the same table, it can be seen that the two tyrosines at positions Y20

and Y21 located in the second helix, are the residues that have the largest effect on the

affinity between ABD and HSA. These two tyrosines are both located in the N-

terminal part of the second helix, and are also surface-exposed. Interestingly, residue

D19, which is close to both tyrosines, does not appear to contribute to the interaction.

Both HSA and ABD are negatively charged at the pH used in the binding studies.

This can explain why replacing a positively charged Lys with an Ala in position 29

decreased the association rate, and accordingly why replacing the negatively charged

Glu acids in positions 32 and 40 with an Ala increased the association rate.

In order to further verify the position of the binding surface, three mutants were

constructed through changing three neighbouring amino acids at a time. Alanine

substitutions were made in helices one and three, in positions where the amino acids

were pointing away from the postulated binding site. The resulting proteins were:

ABD*(E3A, V6A, L7A), ABD*(R10A, E11A, K14A) and ABD*(K35A, I38A,

D39A). More information about ABD* can be found in the section of the thesis

entitled Stabilization of ABD towards alkaline conditions. All three mutants were

shown to have similar or better affinity to HSA than the parental molecule. Hence, the

conclusion is that the main part of the binding site is located in the second helix, and

the two tyrosines have the greatest impact on the stability of the interaction between

ABD and HSA. Interestingly, Johansson and co-workers found in an NMR study that

a larger part of the ABD molecule was found to be affected while binding to HSA

38

(Johansson et al. 2002). This could be explained by the fact that in the NMR

experiment, interaction surfaces described by chemical shift perturbation methods are

usually larger than the region in direct contact (Spitzfaden et al. 1992; Foster et al.

1998). This illustrates the differences between using these two methods and how they

could complement each other.

Table 2. An overview of the kinetic study of the ABD variants carried out on the Biacore.

koff kon Kaff DDG

[vs wt]ZABD variant [10-3 s-1] [105 M-1s-1] [107M-1] [kcal/mol]

wt 0.6 (0.2) 1.4 (0.2) 25.9 0.0 (0.2)E3A 0.6 (0.3) 1.4 (0.7) 22.2 0.1 (0.1)

S18A 3.5 (1.6) 1.3 (0.3) 3.9 1.1 (0.1)

D19A 0.8 (0.3) 1.5 (0.3) 20.9 0.1 (0.3)

Y20A 5.1 (2.1) 0.9 (0.2) 1.9 1.5 (0.2)

Y21A 3.9 (2.4) 0.1 (0.07) 0.3 2.6 (0.2)

K22A 1.3 (0.7) 0.7 (0.2) 5.6 0.9 (0.2)

N23A 0.9 (0.3) 1.0 (0.1) 12.7 0.4 (0.1)

L24A 0.6 (0.4) 0.3 (0.02) 4.5 1.0 (0.2)

K29A 1.1 (0.4) 0.3 (0.1) 3.4 1.2 (0.6)

E32A 2.4 (0.9) 2.5 (0.3) 12.2 0.4 (0.3)

E40A 1.2 (0.5) 2.6 (1.5) 29.5 -0.1 (0.6)

Y20E32A 17 (11) 0.5 (0.2) 0.5 2.3 (0.1)

Y20E40A 6.8 (2.1) 1.4 (0.1) 2.1 1.5 (0.1)

S18Y20K22A 6 0.0001 0.0002 7

Standard deviations are given in brackets. Note that as the very low affinity of the triple

mutant leads to problems in the detection of the binding, no standard deviation is given for

that specific mutant.

Electrostatic interactions have been found to be an important tool for the association

of biomolecules (Selzer and Schreiber 1999; Selzer et al. 2000; Schreiber 2002). This

could also be seen in the interaction between ABD and HSA. Some of the mutations

that seem to affect the association could be explained by removal or addition of

charges, for example E32A and E40A. Interestingly, when exchanging a glutamatic

acid for an alanine in position 32, both the association rate and dissociation rate

increase (see figure 6). It has been shown in previous studies that a mutation can have

a specific effect on either kon or koff (Schreiber and Fersht 1995) giving an unchanged

39

affinity constant despite large changes in the on- and off-rate. In the third helix,

substitution both in position 32 and 40 results in a slightly increased kon.

The importance of the electrostatic interactions could also be seen when alanines

were introduced into the three different ABD* variants where the side chains are

located on the opposite side of the postulated binding site. The binding analyses show

that the off-rate is almost identical for all three mutants; indicating that the stability of

the ABD-HSA complex has not been affected. However, one of the mutants shows an

increased on-rate. This can be explained by the change of charge of the molecule

resulting from changing negatively charged aspartic acid to an uncharged alanine,

which might increase the electrostatic attraction between HSA and ABD, as both are

negatively charged at the pH used in the binding studies. Furthermore, in this triple

mutant, a positively charged lysine was also exchanged for an alanine. This might

help to direct the HSA-binding surface of ABD to the HSA molecule. These two

mutations could thereby increase the “steering region”, as described in the paper by

Seltzer and co-researchers (Selzer et al. 2000). The steering volume is defined as the

volume surrounding a molecule in which it through electrostatic interactions could

change its movement from random diffusion to directional movement towards the

complex formation.

40

Figure 6. The three-dimensional structure of ABD displaying the effect of different mutations,

where: (A) shows the change in kon by mutation, with orange indicating a decreased on-rate,

green an increased on-rate, and yellow an unchanged on-rate where changing a residue to

alanine; (B) shows the change in the off-rate when replacing certain residues for alanine, with

orange indicating that the off-rate increases upon mutation and yellow that the exchange to

alanine does not affect the koff

Stabilization of ABD towards alkaline conditions (II & III)

A chemically stable protein ligand with high affinity for albumin would be interesting

from a biotechnological point of view, due to the fact that HSA is the most abundant

protein in clinical use worldwide. HSA is used to restore colloidal osmotic pressure in

severe burn damages, and to treat the of loss of albumin in traumatic accidents. HSA

for industrial purposes is produced by fractionation from serum, though a

recombinant process in yeast is also available (Quirk et al. 1989).

The work presented in Papers II and III includes a protein engineering strategy to

stabilize and optimize proteinaceous ligands for large-scale applications, namely

ABD. The model chosen for this thesis was an ABD domain originating from SPG

(Kraulis et al. 1996). A more detailed description of this domain can be found in the

chapter of the thesis entitled Bacterial surface receptors. The asparagine residues that

are known to be sensitive to alkaline conditions (Geiger and Clarke 1987; Kossiakoff

1988) were replaced by comparing the ABD-domain to homologous sequences (see

Figure 7), i.e other albumin binding domains. This procedure suggested that,

asparagines in positions 9 and 27 could be replaced by leucine and lysine, while

A B

41

asparagine in positions 23 and 26 could be replaced with aspartic acids. The new

mutant ABD (N9L,23D,26K,27D) was denoted ABD*. The single domain was then

characterized with respect to its affinity, stability and function as an affinity ligand.

LAEAKVLANRELDKYGVSDYYKNLINNAKTVEGVKALIDEILAALP ABDwt--K--AD-LK-FN----------------------D-QAQVVESAK SPG ABD-1--------------------H-----K------IME-QAQVVES SPG ABD-2-DN--NA-LK-F-R------------K------IME-QAQVVES DG12 ABD-1-S---EM-I----AN----F--DK-DD-------V--K-L--NS DG12 ABD-2--------L-------------D--DK------------------- ABD*

Figure 7. The primary sequence of ABD (here denoted ABDwt) (Kraulis et al. 1996) and

ABD*. Shown are also homologous sequences of albumin binding domains derived from

other strains of Streptococcus as well as the homologous domains from SPG. These

homologous sequences were referred to when constructing ABD*. SPG is derived from

Streptococcus G148 (Olsson et al. 1987), DG12 from Streptococcus DG12 (Sjöbring 1992)

and PAB (ALB-strains) from Peptostreptococcus magnus (de Chateau and Björck 1994).

After substituting amino acids in a protein scaffold, it is important to characterize how

the mutations affect the protein. A kinetic study was carried out with the Biacore, and

the structure along with the thermal and chemical stabilities were analyzed with CD.

The kinetic parameters for binding ABDwt and ABD* to HSA is shown in Table 3.

The binding behavior of the mutated protein was similar to the binding behavior of

the native protein. The dissociation rate constant was almost identical, while a

decrease could be seen in the association rate constant. An explanation could be that

the introduced mutations affect the “steering region”, as has been shown by Selzer

and co-workers and is discussed in the chapter of the thesis entitled Protein-protein

interactions (Selzer et al. 2000; Selzer and Schreiber 2001).

42

ka

[M-1 s-1]

kd

[s-1]

KA

[M-1]

DDG

[kcal mol-1]

ABD 11x104 1.2x10-3 9.0x107 -

ABD* 2.5x104 1.3x10-3 1.9x107 0.9

Table 3. Parameters of the binding of ABD and ABD* to HSA. ka is an association rate

constant. kd is a dissociation rate constant. KA is the affinity constant. DDG is the difference

in binding free energy between ABDwt and ABD*.

The ABD* molecule also appears to have a similar secondary structure to ABDwt

according to the CD measurement. However the a-helical content seems a little

decreased in ABD* compared to ABDwt. The thermal and chemical stabilities were

analyzed with a CD spectropolarimeter. Both proteins show high stability to both

types of denaturation and consequently a defined unfolded state was hard to reach

during the measurements (see Figure 8). Therefore the midpoint values of temperature

or denaturant-induced transition have been used as estimations for melting points. The

Cm value for ABDwt, when denaturating with GdnHCl, was estimated to be 4.2M,

and more than 5 M for ABD*. When analyzing thermal stability, the mutant was

found to be more stable than the native molecule. The Tm value was estimated to be

70 ºC for ABDwt , and over 80 ºC for ABD*.

43

Figure 8. CD-analysis of ABD and ABD*, for a comparison of the structural stability. The

protein concentration was 0.5 mg ml-1 in a phosphate buffer of pH 7. The signals at 222 nm

have been transformed to MRE. (A) Stability towards chemical denaturation. The

temperature is 20°C and concentration of chemical denaturant, GdnHCl, is varied between 2

and 6 M. (B) Stability towards thermal denaturation. The temperature was varied between 40

and 95ºC.

Stability to alkaline treatment was analyzed by three different approaches. In order to

analyze the functionality and stability of ABD* the protein was immobilized to a

standard affinity matrix by N-hydroxysuccinimide (NHS) amine coupling chemistry.

An analysis of the capacity of the new column was done by loading an excess of HSA

into the column then measuring the amount of eluted material to determine the total

capacity.

44

The selectivity of the mutated protein was analyzed and compared with the wild type

by mixing the HSA with protein extract from E. coli. A washing step was included

between each purification cycle, wherein a solution containing 0.5 M NaOH was

pumped through the column for two minutes. After 34 minutes of exposure to high

pH, the ABDwt column showed a 50% decrease in capacity, while the ABD* column

on the other hand, seemed to retain almost all capacity despite the harsh treatment.

After rounds 2, 8 and 15, the eluted material was analyzed with SDS-PAGE and

compared with pure HSA to examine how the specificity had been affected (see

Figure 9). The results indicated that the column with immobilized ABD* retained its

HSA selectivity.

Figure 9. Analysis of different protein fractions after affinity purification with an integrated

CIP-step using ABD* as affinity ligand coupled to the matrix. SDS-PAGE gel (4-20%

gradient) under reducing conditions. Lane 1 and 2, ABD and ABD* respectively; lane 3, a

molecular weight standard; lane 4, E. coli disintegrate spiked with HSA before column

loading. The purification procedure was repeated several times; lane 5, 6 and 7, show the

eluted material after round 2, 8 and 15 respectively; lane 8 shows HSA reference.

After the work with the single domain was done, an optimization approach was

undertaken to increase the capacity for HSA. In order to improve the ABD* domain

as an affinity ligand in HSA purification, a head-to-tail dimeric version was

constructed through genetic engineering. A divalent capturing ligand could potentially

contribute here by creating advantageous avidity effects in the binding between the

ligand and the target, possible binding of two HSA molecules, as well as by providing

a spacer between the solid phase and the solution. Hence, a larger ligand would have a

beneficial effect on the presentation of the ligand to the surrounding solution. The

1 2 3 4 5 6 7 8

94

kDa

67

43

30

2014

45

divalent ABD* motif was also equipped with a unique C-terminal cysteine residue,

providing a tool for directed coupling of the ligands to the solid chromatographic

resins.

Previous work done on linker regions connecting domains has shown that these

regions are important for the stability and folding of the protein (Robinson and Sauer

1998) and careful consideration has to be taken when designing them (Alfthan et al.

1995; Arai et al. 2001; George and Heringa 2002). These regions are highly flexible

and therefore susceptible towards proteases (Hubbard 1998) and CIP conditions

(Kosky et al. 1999). Three different linkers were investigated, VDANS, VDADS and

GGGSG (each given a single amino acid letter code), referred to as A, B and C

respectively. The first linker (A) was a result of the cloning procedure. In the second

linker (B), the asparagine was replaced by an aspartic acid according to the previously

reported fragility of Asn-Ser sequences (Geiger and Clark 1987). The third linker (C)

was composed of glycine and serine residues that in previous reports have been found

to be stable and highly flexible (Argos 1990; Alfthan et al. 1995; Takeda et al. 2001).

The length of these designed linkers are about five residues, which can be compared

to the naturally occurring linkers in SPG, which are about 15 to 30 residues (Olsson et

al. 1987). On the other hand, in SPA the linkers between the domains are

approximately 8 to 13 amino acids (Uhlén et al. 1984).

Binding studies were performed to elucidate differences in the binding capacity

between the monovalent original construct ABD* and the divalent variant, the ABD*

dimer A. Additionally, two different chemistries were used to examine the importance

of the coupling-chemistry when attaching the domains to a solid phase. The non-

specific amine coupling was compared to directed immobilization through the C-

terminal cysteine. As can be seen in Figure 10, the directed coupling gives a higher

overall binding capacity than when non-directed chemistry was used. The divalent

variant shows a higher proportion of active domains than the monomer does, when

using non-direct coupling chemistry. Interestingly using a directed-coupling

chemistry, the divalent variant shows the same proportion of active domains as the

monovalent variant.

46

Figure 10. Biosensor analysis of the interactions between injected HSA and different ABD

constructs. The monovalent ABD* and one divalent variant (ABD*dimerA) were immobilized

on the chip surface by (A) NHS-chemistry (non-directed) or (B) thiol-chemistry (directed),

respectively. The amount immobilized ABD*-domains was similar in all cases (for NHS-

chemistry: ABD* 970 RU, and ABD*dimerA 900 RU; for thiol-chemistry: ABD* 905 RU, and

ABD*dimerA 894 RU). A 300 nM HSA solution was injected over the different surfaces. Mean

values from duplicate samples are shown.

The different divalent ABD* variants as well as the monovalent ABD* domain were

coupled separately to activated sepharose by recruiting their C-terminal cysteines,resulting in directed coupling. The different affinity media were packed into columns

and used in affinity purification experiments for HSA-binding capacity studies. Theresults have been summarized in Table 4. These results were in agreement with the

biosensor studies that the divalent and monovalent proteins were equally accessible

for HSA molecules when directed-coupling chemistry was used for immobilization.Furthermore, the data indicate that it might be possibly to increase the capacity of an

0

200

400

600

800

1000

1200

1400

1600

0 200 400 600 800 1000 1200 1400

Time[sec]

Response[RU]

[RU]

ABD*dimerAABD*monomer

0

100

200

300

400

500

600

700

800

0 200 400 600 800 1000

1200

1400

Time[sec]

Responsnse[RU]

ABD*dimerA

ABD*monomer

47

affinity column by immobilizing a divalent ligand to increase the amount of coupled

ligands on the surface of the matrix. In the experiments conducted here, theABD*dimerB (13 mg) exhibits more than twice the capacity of the monovalent

variant ABD* (5 mg).

Table 4. Dynamic binding capacity of the different ABD* chromatography matrices.

ConstructLiganddensity[mg/ml]

Columnvolume

[ml]

CoupledABD*domain

[nmol]

HSA[nmol]

BoundHSA/

coupledligand

HSA[mg/ml]**

ABD* 3.0 1.06 440 71 0.16 5

ABD*dimerA 3.4 1.00 534 95 0.18 6

ABD*dimerB 5.1 0.82 656 159 0.24 13

ABD*dimerC 2.6 0.98 406 86 0.21 6

Mw for ABD* and ABD*dimer A, B, and C are 7221.3, 12753.7, 12051.7, and 12585.7

respectively.

** The values have been weighed to the total column volume, i.e. capacity of the

column/volume of the column.

Further, the different affinity media with their attached monovalent and divalent

affinity ligands were subjected to a stability study. The results of this study have been

summarized in Figure 11. All the ligands possess a high tolerance to alkaline

exposure, however the ligands containing the VDADS linker show the highest

tolerance. Even after as long exposure as 7 h in 0.1 M NaOH, the capacity of the

ABD*dimerB remained at about 95% compared to approximately 85% for all the

other ligands. It is interesting that a divalent ligand could be more tolerant to alkaline

conditions than the monovalent ABD* ligand. The reason for this remains elusive, but

it can be speculated that the two domains in a divalent ligand could exert a mutual

stabilizing effect relating to refolding after exposure to alkaline conditions (Robinson

and Sauer 1998; Wenk et al. 1998).

48

Figure 11. Influence of alkaline sanitation on the HSA binding capacity of an affinity column.

The monovalent ABD* ligand and the three divalent ABD* ligands were separately analyzed

in a column chromatography protocol in order to explore the tolerance to repeated alkaline

sanitization (0.1 M NaOH). After injection of HSA, the captured protein were eluted and the

binding capacity determined. In order to clean the affinity purification device, 0.1 M NaOH

was pumped though the column for 20 minutes between each purification round. The

ligands were immobilized using thioether chemistry by taking advantage of the C-terminal

cysteine. ◊;ABD*dimerA, o; ABD*dimerB, D; ABD*dimerC, x; ABD*monomer.

0,0

20,0

40,0

60,0

80,0

100,0

0 50 100 150 200 250 300 350 400 450 Time of exposure [min]

Capacity[%]

Linker BLinker ALinker CABDmonomer

49

IgG binding domains

Stabilization of C2 towards alkaline conditions (IV)

The same strategy that was applied to enhance ABD for industrial applications was

investigated for alkaline stabilization of the IgG-binding domain C2 originating from

SPG. The sensitivity of these domains to alkaline pH has earlier been reported

(Goward et al. 1991). Although this structure differs considerably from the ABD-

domain worked with earlier, the same stabilizing strategy could possibly still be

applied. First, the sensitive Asn residues were replaced and then a divalent variant

was constructed in order to increase the capacity of an affinity matrix for IgG capture.

Instead of replacing all the asparagines simultaneously, each asparagine was replaced

one by one, to investigate their contribution to the deactivation during alkaline

treatment (Stephenson and Clarke 1989; Wright 1991). In addition, the aspartates and

glutamines were replaced to investigate whether the tolerance to alkaline conditions

could be increased even further (Stephenson and Clarke 1989; Wright 1991;

Tomizawa et al. 1995). The amino acid composition of C2 is shown in figure 12, with

the sensitive asparagines highlighted.

Three single mutants were designed to resolve the contribution of each asparagine

to the deactivation rate, C2N7A,C2N34A and C2N36A. In addition, one variant was

constructed with two asparagines exchanged for alanine (denoted C2N7,36A), and one

constructed with three asparagines exchanged for alanine (denoted C2N7,34,36A). A

second generation of mutants was then generated using the most stable mutated

variant (C2N7,36A) as a scaffold. Two single mutants and one triple mutant were

designed yielding a protein where the aspartate residues were substituted for

glutamate residues in order to retain the overall charge, and these were designated:

C2N7,36AD35E, C2N7,36AD39E and C2N7,36AD21,45,46E respectively . In order to explore the

sensitivity of the single glutamine in the structure, C2N7,36A was used as a scaffold

again and the glutamine was substituted for alanine resulting in C2N7,36AQ31A.

50

Figure 12. The 55-residue sequence of the C2 domain (Olsson et al. 1987). The aspargagine

were replaced with alanine.

The different mutants of C2 were analyzed regarding affinity, structure and stability

using Biacore, CD and isoelectric focusing gel electrophoresis (IEF). In addition, an

affinity chromatography scheme was used to analyze their tolerance to alkaline

treatment.

The strategy involved the replacement of each asparagine residue on a one-by-one

basis to investigate and assess each individual asparagine’s contribution to function

and stability as an affinity ligand. Mutations N7A and N36A were found to be optimal

for increasing the tolerance of the IgG-binding domain C2 to alkaline treatment. In

addition, genetically fused dimers of this double mutant were designed and exhibited

same tolerance as the monovalent double mutant.

SPR technology was used to investigate the interaction between the Fc fragment of

IgG and the C2 variants. As can be seen in Table 5, C2, C2N7A, C2N36A and C2N7,36A

all had approximately the same kinetic properties.

Table 5. Parameters of the interaction of C2 and C2 mutants to Fc from human IgG,

measured by SPR. All calculations have been done with Fc immobilized on the surface.

ka [M-1 s-1] kd [s

-1] KA [M-1]

C2 2.9x105 27.0x10-3 1.1x107

C2N7A 2.3x105 21.1x10-3 1.1x107

C2N36A 2.0x105 13.2x10-3 1.5x107

C2N7,36A 2.2x105 13.4x10-3 1.7x107

ka is the association rate constant, kd is the dissociation rate constant and KA is the affinity

constant (ka/kd).

Potentially, a divalent ligand also could beside the availability of two affinity sites

lead to advantageous avidity effects in the binding between the ligand and the target.

It could also provide a beneficial effect by effectively presenting the immobilized

ligand, by providing a spacer.

A

TYKLVINGKTLKGETTTEAVDAATAEKVFKQYANDNGVDGEWTYDDATKTFTVTE

AA

51

A CD study was carried out to investigate the structural impact the different mutations

had. These CD results indicated that all variants exhibited similar secondary structure

contents as the parental protein G fold, although the CD signal was slightly decreased

for some of the variants compared with the wild-type C2, indicating that only small

changes arose in the secondary structure on mutation. The variant C2N7,36A and the

wild-type C2 were subjected to thermal and chemical denaturation with GdnHCl in

order to analyze their structural stabilities. The change in signal upon denaturation

was measured at 217 nm (a typical minima for a b-sheet proteins). In both studies, all

proteins adopted clearly defined pre and post transitions, and subsequently the signals

could be transformed to Fapp. For the wild-type C2, the thermal Tm1/2

was 77 ºC and

Cm1/2 was 2.9 M; The mutant C2N7,36A had a slightly lower Tm

1/2 of 71 ºC, and also a

lower Cm1/2 of 2.5 M.

As has been shown elsewhere (Geiger and Clark 1987; Tyler-Cross and Schirch

1991) the Asn residues can be modified in alkaline conditions to Asp or isoAsp. This

results in an increased negative charge on the protein, which can be studied by

isoelectric focusing (IEF). The IEF study indeed showed that after two hours of

exposure in 0.1M NaOH, the C2 protein undergoes a transformation to a new more

negatively charged species. The double mutant on the other hand revealed no

additional charge indicating that the deamidation-susceptible residues were

substituted.

The ability of C2 and all mutated-C2-variants to function as affinity ligands in an

ordinary purification scheme for IgG antibodies was investigated by immobilization

of the different domains on HiTrap‘ columns (Amersham Biosciences). Between the

different cycles, a two-minute CIP step consisting of 0.1 M NaOH was integrated.

The results of the capacity study presented in figure 13 show that after a total of 18

minutes of exposure to NaOH, the C2 column had only about 30% capacity left. The

C2N7A column showed a similar loss of binding capacity, while the C2N34A column

lost capacity even faster. The C2N36A column showed a significantly higher stability;

as did the double mutant C2N7,N36A column. In an attempt to increase the stability even

further, several mutants were designed in which the electrostatic nature of the domain

was intentionally retained through replacement of aspartate with a glutamate. The

double mutant C2N7,36A was used as a scaffold for introducing an alanine to the single

glutamine residue substitution. However, the different aspartate and glutamine

52

substitutions revealed no further tolerance to the CIP treatment (see Figure 13). In

conclusion, it was possible to stabilize one of the IgG-binding domains from

streptococcal protein G in respect to alkaline conditions, by employing a protein

engineering strategy.

Figure 13. Comparison of capacity after repeated CIP-treatment following a standard affinity

chromatography protocol on C2 and all the mutated variants C2-variants as immobilized

ligands. The cleaning agent was 0.1 M sodium hydroxide.

0

20

40

60

80

100

120

2 4 6 8 10 12 14 16 18 20 22 24 26 28

Capacity[%]

Time of exposure [min]

C2N7,36AdimC2N36AC2N7,36AC2N7,36AD35EC2N736AD39EC2N7,36AD21,45,46EC2N7,36AQ31A

C2dimC2C2N7A

C2N34A

53

Stabilization of Z towards alkaline conditions (V)

SPA is one of the most widely used affinity ligands in affinity chromatography.

Previous studies have shown that SPA-based chromatography media are relatively

stable in CIP conditions. A loss of about 1% in binding capacity with exposure to 0.5

M NaOH for 15 minutes has been reported (Hale et al. 1994). However for large-scale

applications it would it be valuable to have an even more stable ligand. As a starting

scaffold for further stabilization, the Z domain originating from the B domain in SPA

was used. The three-dimensional structure of the Z domain has been illustrated in

Figure 14, indicating all the asparagines and the different substitutions made in this

study. The Z domain consists of 58 amino acids, which includes eight asparagines in

positions: 3, 6, 11, 21, 23, 28, 43 and 52 (Nilsson et al. 1987).

Figure 14. The three-dimensional structure of the Z domain (Tashiro et al. 1997). All

asparagines and the different substitutions are indicated.

In an initial trial, phage display technology was used as a means to select Z variants

with retained IgG binding capacity after substitutions. The library was designed with

randomizations at positions 21, 29 and 52. Position 29 was chosen to investigatewhether it was possible to retain the Fab interaction with an increased alkaline

stability. The selection procedure consisted of three rounds of panning with IgG as

bait. After the selection, a number of colonies were sequenced and analyzed. Since

N3N6A

N11S

N21A

N23T

N28AN43E

N52A

54

the Z-domain already is significantly stable to alkaline conditions it was not possible

to introduce alkaline treatment in the selection procedure. Therefore none of theselected binders achieved an increased alkaline stability compared to the native Z

domain.

An alternative strategy to phage display is to use a by-pass mutagenesis method

(Kotsuka et al. 1996). This strategy involves the use of a domain analogue as starting

template, into which a destabilizing mutation has been introduced to simplify the

search for variants of increased stability. Secondary mutations are then introduced and

the effects analyzed. Mutations that improve stability of the destabilized molecule are

then grafted into the original protein. Previous studies on the Z domain have shown

that a mutation of phenylalanine 30 to an alanine, destabilize the structure about 3.5

kcal/mol (Cedergren et al. 1993). The original phenylalanines side chain contributes

to the stability by being part of the hydrophobic core.

Z MKAIFVL GTAD NKFNKEQQNAFYEILHLPNLNEEQRNAFIQSLKDDPSQSANLLAEAKKLNDAQAPKZ(F30) -------NAQHDEAVD ---------------------------A----------------------------

E AQHDEA------QV-NM----AD---G--------------V-G--Q----S----D ADAQQ-N---D--S------NM---------G----------------G------ES----A AD -N-------------NM---------G----------------S------ES----B AD --------------------------G-----------------------------C AD --------------------T-----G----------V-KEI--------------

Figure 15. Amino acid alignment of the Z, Z(F30A) and five homologous domains (E,D,A,B,

and C). The horizontal lines indicate amino acid identity. The replaced asparagines and

glycine in the B domain are underlined. Z(F30A) and all its mutants include the same N-

terminal as Z(F30A). Z(N23T) was constructed with the same N-terminal as Z.

A comparison of the amino acid sequence in the different homologous domains in

SPA, suggests that asparagines on position 11, 23 and 43 could be replaced by serin,

threonin and glutamic acid respectively. The asparagines in position 6, 21 and 28

were replaced by alanines, as no obvious alternative amino acid exists in the

homologous domains. As can be seen in Figure 15, there is an asparagine in position

3. As this is in the flexible N-terminal end of the domain and outside the ordered part

of the structure, it was reasoned that this would probably not contribute to the alkaline

tolerance of the entire domain, and was therefore excluded from the study. However

55

this position will still be very important to investigate if multimeric constructs are

desired.

When the single mutants had been constructed, a study was undertaken to

investigate how the introduced point mutations affected the function of the Z variants

as affinity ligands. Before this investigation could proceed, a careful study was

necessary to characterize how the single mutations affected structure and affinity.

Structure analysis was performed using CD, and the affinity was elucidated with a

Biacore where different Z domains were compared to the native Z molecule.

The structural analyses using CD showed that all constructs exhibited CD spectra

similar to the native Z molecule. All spectra had typical a-helical characteristics with

a minimum of 208 and 222 nm, in combination with a maximum of around 195 nm.

However Z(F30A, N52A) seemed to have a somewhat lower a-helical content than

their parental Z molecule, and the other mutants thereof. This might be due to the

postulated hydrogen bond between dO of N52 and the backbone amide proton of

N21, and that the side chain of N21 is hydrogen postulated to bind the carbonyl

oxygen of N52 (Starovasnik et al. 1996).

Biospecific affinity interaction analysis demonstrated no dramatic effect in the

affinity of any of the variants (see Table 6). As can been seen in the same table, the

suppressor mutation F30A did not seem to decrease the affinity between the Z domain

and IgG. Rather the opposite can be seen, which is in accordance with earlier results

reporting that a small increase in affinity could be measured (Cedergren et al. 1993;

Jendeberg et al. 1995).

56

Table 6. Overview of the kinetic study on the different Z domains carried out using the

Biacore.

Mutant ka kd KA ∆∆G ∆∆G[vs Z] [vs Z[F30A]]

[105 M-1 s-1] [10-3 s-1] [107 M-1] [kcal/mol] [kcal/mol]

Z 1.5 3.7 4.0 0Z(N23T) 2.7 3.9 7.0 -0.3

Z(F30A) 1.9 4.2 4.5 -0.1 0

Z(F30A,N6A) 7.0 21 3.3 0.1 0.2

Z(F30A,N11S) 1.6 4.9 3.2 0.1 0.2

Z(F30A,N21A) 1.0 3.8 2.6 0.3 0.3

Z(F30A,N23T) 2.1 3.8 5.6 -0.2 -0.1

Z(F30A,N28A) 3.1 9.9 3.2 0.1 0.2

Z(F30A,N43E) 1.3 5.1 2.6 0.3 0.3

Z(F30A 1.5 4.9 3.0 0.2 0.2

Z(F30A,N23T,N43E) 0.8 3.8 2.0 0.4 0.5

Z was used as an internal standard during the different measurements. The differences in

free binding energy were calculated relative to Z and Z(F30A) respectively.

The aim of the study was to investigate the effect of the various different single

mutations on the alkaline resistance of the Z domains. The results of the affinity

chromatography experiments using amine coupled monomeric domains are shown in

figure 16. The suppressor mutated ZF30A was still quite stable to alkaline conditions.

Even after 7.5 hours of exposure to 0.5 M NaOH, the column remained functional,

though with reduced capacity. The most important mutation was shown to be the

asparagine to threonin substitution in the loop between helices one and two, leading to

ZF30AN23T. This mutation increased the alkaline tolerance of Z(F30A) remarkably.

Also, the mutation was able to stabilize the native Z molecule when it was grafted

into the parental Z-domain. This agrees with earlier results showing that asparagines

located in unstructured regions are more susceptible to covalent modifications than

those located in structurally more inflexible regions (Kosky et al. 1999; paper IV).

57

Figure 16. Comparison of capacity after repeated CIP-treatment with a 0.5 M NaOH

cleaning agent, following an ordinary affinity chromatography scheme. The protocol was run

16 times and the duration for the alkaline exposure was 30 minutes in each round. The total

time of exposure to 0.5 M NaOH was 7.5 hours.

To investigate the structural stability of Z and different mutants thereof in alkaline

conditions, CD-spectra at two different pH, 7.6 and 13.7, were measured. The

parental Z possesses a remarkably high stability at high pH-values, almost appearing

unaffected. In addition, Z(F30A) shows lower secondary structure content in high pH

solution. The results imply that the N23T-mutation slightly stabilizes the secondary

structure of Z(F30A). Comparing the CD data for Z with the C2 scaffold, both

domains showed high thermal stability while CD measurements in GdnHCl solutions

indicated that the Z domain showed higher stability than C2 (Gülich et al. 2000). In

addition, the Z domain also showed high a-helicity at high pH, while CD

measurements on the C2 domain indicated no secondary structure at all in the same

conditions (data not shown).

0

10

20

30

40

50

60

70

80

90

100

0 100 200 300 400 5000 1 2 3 4 5 6 7 80

Capacity [%]

Time of Exposure

[h]

Z(F30A,N6A)

Z(F30A,N23T,N43E)

Z(F30A,N28T)

Z(F30A,N23T)

Z(F30A,N11S)Z(F30A,N11S,N23T)Z(F30A,N6A;N23T)Z(F30A)Z(F30A,N43T)Z(F30A,N52A)

Z(F30A,N21A)

58

Concluding remarksIn this thesis, attempts have been made to illustrate the benefits of using protein

engineering for biotechnological applications. Discussions have emphasized site-

directed mutagenesis for exploring protein-protein interactions, and improving the

performance of small protein receptor domains from bacteria.

A mutagenesis strategy was applied to investigate the binding surface of the HSA

binding domain originating from SPG. The residues contributing to the affinity where

found to be located mainly in the second helix. Additionally, electrostatic interactions

were found to be important for the association between ABD and HSA. This has also

been shown to be important for other protein-protein interactions (Clackson et al,

1998; Schreiber and Fersht 1996). The localization of the HSA-binding region at

ABD should provide new ways to enhance and improve ABD for different

applications; including the possibility to introduce new binding specificities into the

small domain, while still retaining the affinity for HSA.

In the stabilization projects, three different bacterial domains were analyzed, ABD

and the C2 domain originating from SPG and the Z domain originating from SPA. In

all cases, the most sensitive asparagines were found to be located in loops between a-

helices, or between a a-helix and a b-strand. This agrees with previous reports that

the fragility of asparagine towards modifications is dependent on different secondary

structure elements. In two cases, the stabilized variant of ABD, and C2 were

genetically fused to provide dimers to increase the column capacity during affinity

chromatography. The rational design strategies used to increase the tolerance to

alkaline pH levels, clearly show upon the benefit of understanding underlying reasons

for degradation. This strategy might also be applicable to increase the alkaline

tolerance of other proteins and expand the use of proteinaceous ligands in different

biotechnological applications.

59

Acknowledgements

I would like to thank all my former and current friends and colleagues at theDepartment of Biotechnology; not only for all their research-related input includingseminars, but also for the fantastic camaraderie we enjoyed, with “musik-kvällar”,social activities, conferences and science.

I wish to specially thank the following people:

Professor Mathias Uhlén, for accepting me as a PhD student, and for your inspirationover the years. Sophia Hober, my amazing supervisor. It has been a great pleasureworking with you. Per-Åker Nygren for valuable discussion regarding science and forproofreading of this thesis. My laboratory colleagues at EDEN - Maria and Nina. Myproject support in Kaspar and Anneli, for their well performed “exjobb”, without yourhelp I would definitely not have come this far. Thanks also to Karin, Elin and Anna atAffibody AB for valuable collaboration. To Dr Sussie Gülich for helping and guidingme through my “exjobb”.

I would also like to thank all of my friends outside the lab, especially Morgan, Petra,Kaj, Lisa, Jonas, Björn J, Björn H, Annica H, Staffan, Magnus and Annica O - for lotsof laughs.

Also the “K93 årgång” - Stoffe, Henrik, Anna B. and Stina - imagine 11 years havegone by just like that!

Thanks too to my family - Mum, Bengt, Dad, Sonja, both my brothers with families,Claes, Johan, Malin, Sunna and the juniors - Lina, Orm and Lova, for just being there.Also my “bonus” siblings with families- Petter, Anna, Eva and Mille. Good luckAnna and Mille!

My thanks here also to the “Forshaga family”- Gunno, Yvonne, Martin, Markus,Zandra and Sandra.

My beloved Caroline, for your love and encourage through the years.

My daughter Elvira, for remanding me it is a world outside the lab.

Tusen tack!

60

AbbreviationsABD albumin binding domainC2 protein G domainCD circular dichroismCH2 constant heavy chain 2CH3 constant heavy chain 3CIP cleaning in placeCm midpoint of denaturant induced transtitionC-terminal carboxy-terminalDa DaltonELISA enzyme linked immunosorbent assayEDTA ethylenediaminetetraacetic acidFab- fragment antigen bindingFc fragment crysttizableFRET Fluorescence resonance energy transferFv fragment variableGdnHCl guanidine hydrochlorideHSA human serum albuminIgG immunoglobulin Gka association rate constantkd dissociation rate constantKa affinity constantKd dissociations constantMab monoclonal antibodiesNHS N-hydroxysuccinimideNMR nuclear magnetic resonanceN-terminal amino- terminalPCR- polymerase chain reactionRU resonance unitSDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresisSPA staphylococcal protein ASPG streptococcal protein GTm midpoint of temperature-induced transitionZ engineered protein A domain

61

ReferencesAbrahmsen, L., Moks, T., Nilsson, B., Hellman, U., and Uhlen, M. 1985. Analysis of

signals for secretion in the staphylococcal protein A gene. Embo J 4: 3901-3906.

Achari, A., Hale, S.P., Howard, A.J., Clore, G.M., Gronenborn, A.M., Hardman,K.D., and Whitlow, M. 1992. 1.67-A X-ray structure of the B2immunoglobulin-binding domain of streptococcal protein G and comparisonto the NMR structure of the B1 domain. Biochemistry 31: 10449-10457.

Akasako, A., Haruki, M., Oobatake, M., and Kanaya, S. 1995. High resistance ofEscherichia coli ribonuclease HI variant with quintuple thermostabilizingmutations to thermal denaturation, acid denaturation, and proteolyticdegradation. Biochemistry 34: 8115-8122.

Albeck, S., and Schreiber, G. 1999. Biophysical characterization of the interaction ofthe beta-lactamase TEM-1 with its protein inhibitor BLIP. Biochemistry 38:11-21.

Alfthan, K., Takkinen, K., Sizmann, D., Soderlund, H., and Teeri, T.T. 1995.Properties of a single-chain antibody containing different linker peptides.Protein Eng 8: 725-731.

Amersham Biosciences. 1997. Affinity chromatography. Principles and methods.Ankerst, J., Christensen, P., Kjellen, L., and Kronvall, G. 1974. A rountine diagnostic

test for IgA and IgM antibodies to rubella virus: absorption of IgG withStaphylococcus aureus. J Infect Dis 130: 268-273.

Arai, R., Ueda, H., Kitayama, A., Kamiya, N., and Nagamune, T. 2001. Design of thelinkers which effectively separate domains of a bifunctional fusion protein.Protein Eng 14: 529-532.

Argos, P. 1990. An investigation of oligopeptides linking domains in protein tertiarystructures and possible candidates for general gene fusion. J Mol Biol 211:943-958.

Arnold, F.H. 2001. Combinatorial and computational challenges for biocatalystdesign. Nature 409: 253-257.

Asplund, M., Ramberg, M., and Johansson, B. 2000. Development of a cleaning inplace protocol and repetitive application of Escherichia coli homogenate onSTREAMLINE Q XL. 35: 1111-1118.

Bass, S.H., Mulkerrin, M.G., and Wells, J.A. 1991. A systematic mutational analysisof hormone-binding determinants in the human growth hormone receptor.Proc Natl Acad Sci U S A 88: 4498-4502.

Björck, L., and Kronvall, G. 1984. Purification and some properties of streptococcalprotein G, a novel IgG-binding reagent. J Immunol 133: 969-974.

Blaber, M., Zhang, X.J., and Matthews, B.W. 1993. Structural basis of amino acidalpha helix propensity. Science 260: 1637-1640.

Bornscheuer, U.T., and Pohl, M. 2001. Improved biocatalysts by directed evolutionand rational protein design. Curr Opin Chem Biol 5: 137-143.

Burgoyne, R.F., Priest, M.C., Roche, K.L., and Vella, G. 1993. Systematicdevelopment and validation of sanitization protocols for a chromatographicsystem designed for biotherapeutics purification. J Pharm Biomed Anal 11:1317-1325.

62

Burnouf, T., and Radosevich, M. 2001. Affinity chromatography in the industrialpurification of plasma proteins for therapeutic use. J Biochem BiophysMethods 49: 575-586.

Canziani, G., Zhang, W., Cines, D., Rux, A., Willis, S., Cohen, G., Eisenberg, R., andChaiken, I. 1999. Exploring biomolecular recognition using opticalbiosensors. Methods 19: 253-269.

Carter, P.J., Winter, G., Wilkinson, A.J., and Fersht, A.R. 1984. The use of doublemutants to detect structural changes in the active site of the tyrosyl-tRNAsynthetase (Bacillus stearothermophilus). Cell 38: 835-840.

Cedergren, L., Andersson, R., Jansson, B., Uhlen, M., and Nilsson, B. 1993.Mutational analysis of the interaction between staphylococcal protein A andhuman IgG1. Protein Eng 6: 441-448.

Chakrabartty, A., Doig, A.J., and Baldwin, R.L. 1993. Helix capping propensities inpeptides parallel those in proteins. Proc Natl Acad Sci U S A 90: 11332-11336.

Chen, R. 2001. Enzyme engineering: rational redesign versus directed evolution.Trends Biotechnol 19: 13-14.

Clackson, T., Ultsch, M.H., Wells, J.A., and de Vos, A.M. 1998. Structural andfunctional analysis of the 1:1 growth hormone:receptor complex reveals themolecular basis for receptor affinity. J Mol Biol 277: 1111-1128.

Clackson, T., and Wells, J.A. 1994. In vitro selection from protein and peptidelibraries. Trends Biotechnol 12: 173-184.

Crasto, C.J., and Feng, J. 2001. Sequence codes for extended conformation: aneighbor-dependent sequence analysis of loops in proteins. Proteins 42: 399-413.

Cunningham, B.C., Jhurani, P., Ng, P., and Wells, J.A. 1989. Receptor and antibodyepitopes in human growth hormone identified by homolog-scanningmutagenesis. Science 243: 1330-1336.

Cunningham, B.C., and Wells, J.A. 1989. High-resolution epitope mapping of hGH-receptor interactions by alanine-scanning mutagenesis. Science 244: 1081-1085.

Cunningham, B.C., and Wells, J.A. 1993. Comparison of a structural and a functionalepitope. J Mol Biol 234: 554-563.

Cutler, P. 1996. Affinity chromatography. Methods Mol Biol 59: 157-168.Daniel, R.M., Cowan, D.A., Morgan, H.W., and Curran, M.P. 1982. A correlation

between protein thermostability and resistance to proteolysis. Biochem J 207:641-644.

Danson, M.J., and Hough, D.W. 1998. Structure, function and stability of enzymesfrom the Archaea. Trends Microbiol 6: 307-314.

de Chateau, M., and Björck, L. 1994. Protein PAB, a mosaic albumin-bindingbacterial protein representing the first contemporary example of moduleshuffling. J Biol Chem 269: 12147-12151.

Deisenhofer, J. 1981. Crystallographic refinement and atomic models of a human Fcfragment and its complex with fragment B of protein A from Staphylococcusaureus at 2.9- and 2.8-A resolution. Biochemistry 20: 2361-2370.

DeLano, W.L. 2002. Unraveling hot spots in binding interfaces: progress andchallenges. Curr Opin Struct Biol 12: 14-20.

Delbruck, H., Mueller, U., Perl, D., Schmid, F.X., and Heinemann, U. 2001. Crystalstructures of mutant forms of the Bacillus caldolyticus cold shock proteindiffering in thermal stability. J Mol Biol 313: 359-369.

63

Dennis, M.S., Zhang, M., Meng, Y.G., Kadkhodayan, M., Kirchhofer, D., Combs, D.,and Damico, L.A. 2002. Albumin binding as a general strategy for improvingthe pharmacokinetics of proteins. J Biol Chem 277: 35035-35043.

di Guan, C., Li, P., Riggs, P.D., and Inouye, H. 1988. Vectors that facilitate theexpression and purification of foreign peptides in Escherichia coli by fusion tomaltose-binding protein. Gene 67: 21-30.

Doig, A.J. 2002. Recent advances in helix-coil theory. Biophys Chem 101-102: 281-293.

Eijsink, V.G., Veltman, O.R., Aukema, W., Vriend, G., and Venema, G. 1995.Structural determinants of the stability of thermolysin-like proteinases. NatStruct Biol 2: 374-379.

Eriksson, J., Nordstrom, T., and Nyren, P. 2003. Method enabling firefly luciferase-based bioluminometric assays at elevated temperatures. Anal Biochem 314:158-161.

Falkenberg, C., Björck, L., and Åkerstrom, B. 1992. Localization of the binding sitefor streptococcal protein G on human serum albumin. Identification of a 5.5-kilodalton protein G binding albumin fragment. Biochemistry 31: 1451-1457.

Fan, Z., Larson, P.J., Bognacki, J., Raghunath, P.N., Tomaszewski, J.E., Kuo, A.,Canziani, G., Chaiken, I., Cines, D.B., and Higazi, A.A. 1998. Tissue factorregulates plasminogen binding and activation. Blood 91: 1987-1998.

Fersht, A.R. 1999. Structure and mechanism in protein science. W.H Freeman, NewYork.

Foster, M.P., Wuttke, D.S., Clemens, K.R., Jahnke, W., Radhakrishnan, I., Tennant,L., Reymond, M., Chung, J., and Wright, P.E. 1998. Chemical shift as a probeof molecular interfaces: NMR studies of DNA binding by the three amino-terminal zinc finger domains from transcription factor IIIA. J Biomol NMR12: 51-71.

Foster, P., McIntosh, RV. 1992. The principal elements of plasma productmanufacture. Wiley, Chichester, pp. p 19-45.

Frenken, L.G., Egmond, M.R., Batenburg, A.M., and Verrips, C.T. 1993.Pseudomonas glumae lipase: increased proteolytic stability by proteinengineering. Protein Eng 6: 637-642.

Frisch, C., Schreiber, G., Johnson, C.M., and Fersht, A.R. 1997. Thermodynamics ofthe interaction of barnase and barstar: changes in free energy versus changesin enthalpy on mutation. J Mol Biol 267: 696-706.

Gabdoulline, R.R., and Wade, R.C. 1999. On the protein-protein diffusionalencounter complex. J Mol Recognit 12: 226-234.

Geiger, T., and Clarke, S. 1987. Deamidation, isomerization, and racemization atasparaginyl and aspartyl residues in peptides. Succinimide-linked reactionsthat contribute to protein degradation. J Biol Chem 262: 785-794.

George, R.A., and Heringa, J. 2002. An analysis of protein domain linkers: theirclassification and role in protein folding. Protein Eng 15: 871-879.

Gershenson, A., and Arnold, F.H. 2000. Enzyme stabilization by directed evolution.Genet Eng (N Y) 22: 55-76.

Girot, P., Moroux, Y., Duteil, X.P., Nguyen, C., and Boschetti, E. 1990. Compositeaffinity sorbents and their cleaning in place. J Chromatogr 510: 213-223.

Goetsch, L., Haeuw, J.F., Champion, T., Lacheny, C., N'Guyen, T., Beck, A., andCorvaia, N. 2003. Identification of B- and T-cell epitopes of BB, a carrierprotein derived from the G protein of Streptococcus strain G148. Clin DiagnLab Immunol 10: 125-132.

64

Goward, C.R., Irons, L.I., Murphy, J.P., and Atkinson, T. 1991. The secondarystructure of protein G', a robust molecule. Biochem J 274 ( Pt 2): 503-507.

Graille, M., Stura, E.A., Corper, A.L., Sutton, B.J., Taussig, M.J., Charbonnier, J.B.,and Silverman, G.J. 2000. Crystal structure of a Staphylococcus aureus proteinA domain complexed with the Fab fragment of a human IgM antibody:structural basis for recognition of B-cell receptors and superantigen activity.Proc Natl Acad Sci U S A 97: 5399-5404.

Grimsley, G.R., Shaw, K.L., Fee, L.R., Alston, R.W., Huyghues-Despointes, B.M.,Thurlkill, R.L., Scholtz, J.M., and Pace, C.N. 1999. Increasing protein stabilityby altering long-range coulombic interactions. Protein Sci 8: 1843-1849.

Gronenborn, A.M., and Clore, G.M. 1993. Identification of the contact surface of astreptococcal protein G domain complexed with a human Fc fragment. J MolBiol 233: 331-335.

Gronenborn, A.M., Filpula, D.R., Essig, N.Z., Achari, A., Whitlow, M., Wingfield,P.T., and Clore, G.M. 1991. A novel, highly stable fold of theimmunoglobulin binding domain of streptococcal protein G. Science 253:657-661.

Gräslund, T., Lundin, G., Uhlen, M., Nygren, P.Å., and Hober, S. 2000. Chargeengineering of a protein domain to allow efficient ion-exchange recovery.Protein Eng 13: 703-709.

Gülich, S., Uhlen, M., and Hober, S. 2000. Protein engineering of an IgG-bindingdomain allows milder elution conditions during affinity chromatography. JBiotechnol 76: 233-244.

Guss, B., Uhlen, M., Nilsson, B., Lindberg, M., Sjoquist, J., and Sjodahl, J. 1984.Region X, the cell-wall-attachment part of staphylococcal protein A. Eur JBiochem 138: 413-420.

Gårdsvoll, H., Dano, K., and Ploug, M. 1999. Mapping part of the functional epitopefor ligand binding on the receptor for urokinase-type plasminogen activator bysite-directed mutagenesis. J Biol Chem 274: 37995-38003.

Hale, G., Drumm, A., Harrison, P., and Phillips, J. 1994. Repeated cleaning of proteinA affinity column with sodium hydroxide. J Immunol Methods 171: 15-21.

Hanes, J., and Pluckthun, A. 1997. In vitro selection and evolution of functionalproteins by using ribosome display. Proc Natl Acad Sci U S A 94: 4937-4942.

Harakas, N.K. 1994. Protein purification process engineering. Biospecific affinitychromatography. Bioprocess Technol 18: 259-316.

Hensel, R., König, H. 1988. Thermoadaption of methanogenic bacteria byintracellular ion concentration. FEMS Micr. Letter 49: 75-79.

Ho, S.N., Hunt, H.D., Horton, R.M., Pullen, J.K., and Pease, L.R. 1989. Site-directedmutagenesis by overlap extension using the polymerase chain reaction. Gene77: 51-59.

Hol, W.G. 1985. The role of the alpha-helix dipole in protein function and structure.Prog Biophys Mol Biol 45: 149-195.

Holliger, P., and Bohlen, H. 1999. Engineering antibodies for the clinic. CancerMetastasis Rev 18: 411-419.

Hopp, T.H., Prickett, K.S., Price, V.L., Libby, R.T., March, C.J., Cerretti, D.P., Urdal,D.L., Conlon, P.J. 1988. A short polypeptide marker sequence useful forrecombinant protein identification and purification. Bio/Technology 6: 1204-1210.

65

Horovitz, A., Matthews, J.M., and Fersht, A.R. 1992. Alpha-helix stability inproteins. II. Factors that influence stability at an internal position. J Mol Biol227: 560-568.

Hubbard, S.J. 1998. The structural aspects of limited proteolysis of native proteins.Biochim Biophys Acta 1382: 191-206.

Hutchison, C.A., 3rd, Phillips, S., Edgell, M.H., Gillam, S., Jahnke, P., and Smith, M.1978. Mutagenesis at a specific position in a DNA sequence. J Biol Chem253: 6551-6560.

Janin, J. 1997. The kinetics of protein-protein recognition. Proteins 28: 153-161.Jansson, B., Uhlen, M., and Nygren, P.Å. 1998. All individual domains of

staphylococcal protein A show Fab binding. FEMS Immunol Med Microbiol20: 69-78.

Jendeberg, L., Nilsson, P., Larsson, A., Denker, P., Uhlen, M., Nilsson, B., andNygren, P.A. 1997. Engineering of Fc(1) and Fc(3) from humanimmunoglobulin G to analyse subclass specificity for staphylococcal proteinA. J Immunol Methods 201: 25-34.

Jendeberg, L., Persson, B., Andersson, R., Karlsson, R., Uhlen, M., and Nilsson, B.1995. Kinetic analysis of the interaction between protein A domain variantsand human Fc using plasmon resonance detection. J Mol Recognit 8: 270-278.

Johansson, M.U., Frick, I.M., Nilsson, H., Kraulis, P.J., Hober, S., Jonasson, P.,Linhult, M., Nygren, P.Å., Uhlen, M., Björck, L., et al. 2002. Structure,specificity, and mode of interaction for bacterial albumin-binding modules. JBiol Chem 277: 8114-8120.

Johnson, W.C., Jr. 1990. Protein secondary structure and circular dichroism: apractical guide. Proteins 7: 205-214.

Kelly, S.M., and Price, N.C. 2000. The use of circular dichroism in the investigationof protein structure and function. Curr Protein Pept Sci 1: 349-384.

Klemba, M., Gardner, K.H., Marino, S., Clarke, N.D., and Regan, L. 1995. Novelmetal-binding proteins by design. Nat Struct Biol 2: 368-373.

Kosky, A.A., Razzaq, U.O., Treuheit, M.J., and Brems, D.N. 1999. The effects ofalpha-helix on the stability of Asn residues: deamidation rates in peptides ofvarying helicity. Protein Sci 8: 2519-2523.

Kossiakoff, A.A. 1988. Tertiary structure is a principal determinant to proteindeamidation. Science 240: 191-194.

Kotsuka, T., Akanuma, S., Tomuro, M., Yamagishi, A., and Oshima, T. 1996. Furtherstabilization of 3-isopropylmalate dehydrogenase of an extreme thermophile,Thermus thermophilus, by a suppressor mutation method. J Bacteriol 178:723-727.

Kraulis, P.J., Jonasson, P., Nygren, P.Å., Uhlen, M., Jendeberg, L., Nilsson, B., andKördel, J. 1996. The serum albumin-binding domain of streptococcal proteinG is a three-helical bundle: a heteronuclear NMR study. FEBS Lett 378: 190-194.

Kronvall, G., and Williams, R.C., Jr. 1969. Differences in anti-protein A activityamong IgG subgroups. J Immunol 103: 828-833.

Kunkel, T.A., Bebenek, K., and McClary, J. 1991. Efficient site-directed mutagenesisusing uracil-containing DNA. Methods Enzymol 204: 125-139.

Kunkel, T.A., Roberts, J.D., and Zakour, R.A. 1987. Rapid and efficient site-specificmutagenesis without phenotypic selection. Methods Enzymol 154: 367-382.

Lakey, J.H., and Raggett, E.M. 1998. Measuring protein-protein interactions. CurrOpin Struct Biol 8: 119-123.

66

Lehmann, M., Kostrewa, D., Wyss, M., Brugger, R., D'Arcy, A., Pasamontes, L., andvan Loon, A.P. 2000. From DNA sequence to improved functionality: usingprotein sequence comparisons to rapidly design a thermostable consensusphytase. Protein Eng 13: 49-57.

Lehmann, M., Loch, C., Middendorf, A., Studer, D., Lassen, S.F., Pasamontes, L.,van Loon, A.P., and Wyss, M. 2002. The consensus concept forthermostability engineering of proteins: further proof of concept. Protein Eng15: 403-411.

Lehmann, M., and Wyss, M. 2001. Engineering proteins for thermostability: the useof sequence alignments versus rational design and directed evolution. CurrOpin Biotechnol 12: 371-375.

Leszczynski, J.F., and Rose, G.D. 1986. Loops in globular proteins: a novel categoryof secondary structure. Science 234: 849-855.

Lian, L.Y., Barsukov, I.L., Derrick, J.P., and Roberts, G.C. 1994. Mapping theinteractions between streptococcal protein G and the Fab fragment of IgG insolution. Nat Struct Biol 1: 355-357.

Libon, C., Corvaia, N., Haeuw, J.F., Nguyen, T.N., Ståhl, S., Bonnefoy, J.Y., andAndreoni, C. 1999. The serum albumin-binding region of streptococcalprotein G (BB) potentiates the immunogenicity of the G130-230 RSV-Aprotein. Vaccine 17: 406-414.

Ljungquist, C., Jansson, B., Moks, T., and Uhlen, M. 1989. Thiol-directedimmobilization of recombinant IgG-binding receptors. Eur J Biochem 186:557-561.

Lura, R., and Schirch, V. 1988. Role of peptide conformation in the rate andmechanism of deamidation of asparaginyl residues. Biochemistry 27: 7671-7677.

Lyu, P.C., Liff, M.I., Marky, L.A., and Kallenbach, N.R. 1990. Side chaincontributions to the stability of alpha-helical structure in peptides. Science250: 669-673.

Ma, B., Wolfson, H.J., and Nussinov, R. 2001. Protein functional epitopes: hot spots,dynamics and combinatorial libraries. Curr Opin Struct Biol 11: 364-369.

MacLennan, J. 1995. Engineering microprotein ligands for large-scale affinitypurification. Biotechnology (N Y) 13: 1180-1183.

Main, E.R., Xiong, Y., Cocco, M.J., D'Andrea, L., and Regan, L. 2003. Design ofStable alpha-Helical Arrays from an Idealized TPR Motif. Structure (Camb)11: 497-508.

Makrides, S.C., Nygren, P.Å., Andrews, B., Ford, P.J., Evans, K.S., Hayman, E.G.,Adari, H., Uhlen, M., and Toth, C.A. 1996. Extended in vivo half-life ofhuman soluble complement receptor type 1 fused to a serum albumin-bindingreceptor. J Pharmacol Exp Ther 277: 534-542.

Mansfeld, J., Vriend, G., Dijkstra, B.W., Veltman, O.R., Van den Burg, B., Venema,G., Ulbrich-Hofmann, R., and Eijsink, V.G. 1997. Extreme stabilization of athermolysin-like protease by an engineered disulfide bond. J Biol Chem 272:11152-11156.

Markert, Y., Koditz, J., Mansfeld, J., Arnold, U., and Ulbrich-Hofmann, R. 2001.Increased proteolytic resistance of ribonuclease A by protein engineering.Protein Eng 14: 791-796.

Martin, A., Sieber, V., and Schmid, F.X. 2001. In-vitro selection of highly stabilizedprotein variants with optimized surface. J Mol Biol 309: 717-726.

67

Martins, L.O., Huber, R., Huber, H., Stetter, K.O., Costa, M. S. DA., Santos, H.,.1997. Organic Solutes in Hyperthermophilic Archaea. Appl. Env. Micr. 63:896-902.

McCafferty, J., and Glover, D.R. 2000. Engineering therapeutic proteins. Curr OpinStruct Biol 10: 417-420.

McLendon, G., and Radany, E. 1978. Is protein turnover thermodynamicallycontrolled? J Biol Chem 253: 6335-6337.

Minor, D.L., Jr., and Kim, P.S. 1994a. Context is a major determinant of beta-sheetpropensity. Nature 371: 264-267.

Minor, D.L., Jr., and Kim, P.S. 1994b. Measurement of the beta-sheet-formingpropensities of amino acids. Nature 367: 660-663.

Miyazaki, K., and Arnold, F.H. 1999. Exploring nonnatural evolutionary pathways bysaturation mutagenesis: rapid improvement of protein function. J Mol Evol 49:716-720.

Moks, T., Abrahmsen, L., Nilsson, B., Hellman, U., Sjöquist, J., and Uhlen, M. 1986.Staphylococcal protein A consists of five IgG-binding domains. Eur JBiochem 156: 637-643.

Moks, T.A., L., Österlöf, B., Josephson, S., Östling, M., Enfors, SO., Persson, I.,Nilsson, B., Uhlén. M. 1987. Large-scale affinity purification of humaninsulin-like growth factor I from culture medium of Escerichia coli.Bio/Technology 5: 379-382.

Munoz, V., and Serrano, L. 1994. Elucidating the folding problem of helical peptidesusing empirical parameters. Nat Struct Biol 1: 399-409.

Myers, J.K., Pace, C.N., and Scholtz, J.M. 1997. Helix propensities are identical inproteins and peptides. Biochemistry 36: 10923-10929.

Myszka, D.G. 1997. Kinetic analysis of macromolecular interactions using surfaceplasmon resonance biosensors. Curr Opin Biotechnol 8: 50-57.

Nicholson, H., Becktel, W.J., and Matthews, B.W. 1988. Enhanced proteinthermostability from designed mutations that interact with alpha-helix dipoles.Nature 336: 651-656.

Nilsson, B., Moks, T., Jansson, B., Abrahmsen, L., Elmblad, A., Holmgren, E.,Henrichson, C., Jones, T.A., and Uhlen, M. 1987. A synthetic IgG-bindingdomain based on staphylococcal protein A. Protein Eng 1: 107-113.

Nilsson, J., Ståhl, S., Lundeberg, J., Uhlen, M., and Nygren, P.Å. 1997. Affinityfusion strategies for detection, purification, and immobilization ofrecombinant proteins. Protein Expr Purif 11: 1-16.

Nord, K., Gunneriusson, E., Ringdahl, J., Ståhl, S., Uhlen, M., and Nygren, P.Å.1997. Binding proteins selected from combinatorial libraries of an alpha-helical bacterial receptor domain. Nat Biotechnol 15: 772-777.

Nord, K., Nilsson, J., Nilsson, B., Uhlen, M., and Nygren, P.Å. 1995. A combinatoriallibrary of an alpha-helical bacterial receptor domain. Protein Eng 8: 601-608.

Northrup, S.H., and Erickson, H.P. 1992. Kinetics of protein-protein associationexplained by Brownian dynamics computer simulation. Proc Natl Acad Sci US A 89: 3338-3342.

Nygren, P.A., Ljungquist, C., Tromborg, H., Nustad, K., and Uhlen, M. 1990.Species-dependent binding of serum albumins to the streptococcal receptorprotein G. Eur J Biochem 193: 143-148.

Nygren, P.Å., Eliasson, M., Abrahmsen, L., Uhlen, M., and Palmcrantz, E. 1988.Analysis and use of the serum albumin binding domains of streptococcalprotein G. J Mol Recognit 1: 69-74.

68

O'Neil, K.T., and DeGrado, W.F. 1990. A thermodynamic scale for the helix-formingtendencies of the commonly occurring amino acids. Science 250: 646-651.

Olson, C.A., Spek, E.J., Shi, Z., Vologodskii, A., and Kallenbach, N.R. 2001.Cooperative helix stabilization by complex Arg-Glu salt bridges. Proteins 44:123-132.

Olsson, A., Eliasson, M., Guss, B., Nilsson, B., Hellman, U., Lindberg, M., andUhlen, M. 1987. Structure and evolution of the repetitive gene encodingstreptococcal protein G. Eur J Biochem 168: 319-324.

Pace, C.N., Scholtz, J.M. 1997a. Measuring the conformational stability of a protein.In Protein Structure. (ed. T.E. Creighton). Oxford University Press, Oxford.

Pace, N.R. 1997b. A molecular view of microbial diversity and the biosphere. Science276: 734-740.

Parsell, D.A., and Sauer, R.T. 1989. The structural stability of a protein is animportant determinant of its proteolytic susceptibility in Escherichia coli. JBiol Chem 264: 7590-7595.

Pauling, L., Corey, RB. 1960. The nature of the chemical bond. Cornell UniversityPress.

Pearce, K.H., Jr., Cunningham, B.C., Fuh, G., Teeri, T., and Wells, J.A. 1999. Growthhormone binding affinity for its receptor surpasses the requirements forcellular activity. Biochemistry 38: 81-89.

Perl, D., Mueller, U., Heinemann, U., and Schmid, F.X. 2000. Two exposed aminoacid residues confer thermostability on a cold shock protein. Nat Struct Biol 7:380-383.

Peters, T., Jr. 1985. Serum albumin. Adv Protein Chem 37: 161-245.Philipps, B., Hennecke, J., and Glockshuber, R. 2003. FRET-based in vivo screening

for protein folding and increased protein stability. J Mol Biol 327: 239-249.Piehler, J., Roisman, L.C., and Schreiber, G. 2000. New structural and functional

aspects of the type I interferon-receptor interaction revealed by comprehensivemutational analysis of the binding interface. J Biol Chem 275: 40425-40433.

Porath, J., Carlsson, J., Olsson, I., and Belfrage, G. 1975. Metal chelate affinitychromatography, a new approach to protein fractionation. Nature 258: 598-599.

Potter, K.N., Li, Y., and Capra, J.D. 1996. Staphylococcal protein A simultaneouslyinteracts with framework region 1, complementarity-determining region 2, andframework region 3 on human VH3-encoded Igs. J Immunol 157: 2982-2988.

Power, U.F., Plotnicky-Gilquin, H., Huss, T., Robert, A., Trudel, M., Stahl, S., Uhlen,M., Nguyen, T.N., and Binz, H. 1997. Induction of protective immunity inrodents by vaccination with a prokaryotically expressed recombinant fusionprotein containing a respiratory syncytial virus G protein fragment. Virology230: 155-166.

Price, N., Johnson, CM. 1990. Proteolytic Enzymes. A Practical Approach. IRL Press,Oxford.

Prusiner, S.B. 1998. Prions. Proc Natl Acad Sci U S A 95: 13363-13383.Quirk, A.V., Geisow, M.J., Woodrow, J.R., Burton, S.J., Wood, P.C., Sutton, A.D.,

Johnson, R.A., and Dodsworth, N. 1989. Production of recombinant humanserum albumin from Saccharomyces cerevisiae. Biotechnol Appl Biochem 11:273-287.

Raman, C.S., Jemmerson, R., Nall, B.T., and Allen, M.J. 1992. Diffusion-limitedrates for monoclonal antibody binding to cytochrome c. Biochemistry 31:10370-10379.

69

Ramos, A., Raven, NDH., Sharp, RJ., Bartolucci, S., Rossi, M., Cannio, R., Lebbink,J., OOst, J., Vos, W., Santos, H. 1997. Stabilization of Enzymes againstThermal Stress and Freeze-Drying by Mannosylglycerate. Appl. Env. Micr.63: 4020-4025.

Ribas De Pouplana, L., Atrian, S., Gonzalex-Duarte, R., Fothergill-Gilmore, L.A.,Kelly, S.M., and Price, N.C. 1991. Structural properties of long- and short-chain alcohol dehydrogenases. Contribution of NAD+ to stability. Biochem J276 ( Pt 2): 433-438.

Richardson, J.S., and Richardson, D.C. 1988. Amino acid preferences for specificlocations at the ends of alpha helices. Science 240: 1648-1652.

Richman, D.D., Cleveland, P.H., Oxman, M.N., and Johnson, K.M. 1982. Thebinding of staphylococcal protein A by the sera of different animal species. JImmunol 128: 2300-2305.

Robinson, A.B. 1974. Evolution and the distribution of glutaminyl and asparaginylresidues in proteins. Proc Natl Acad Sci U S A 71: 885-888.

Robinson, A.B., McKerrow, J.H., and Cary, P. 1970. Controlled deamidation ofpeptides and proteins: an experimental hazard and a possible biological timer.Proc Natl Acad Sci U S A 66: 753-757.

Robinson, A.B., and Robinson, L.R. 1991. Distribution of glutamine and asparagineresidues and their near neighbors in peptides and proteins. Proc Natl Acad SciU S A 88: 8880-8884.

Robinson, A.B., and Rudd, C.J. 1974. Deamidation of glutaminyl and asparaginylresidues in peptides and proteins. Curr Top Cell Regul 8: 247-295.

Robinson, C.R., and Sauer, R.T. 1998. Optimizing the stability of single-chainproteins by linker length and composition mutagenesis. Proc Natl Acad Sci US A 95: 5929-5934.

Robinson, N.E., and Robinson, A.B. 2001. Prediction of protein deamidation ratesfrom primary and three-dimensional structure. Proc Natl Acad Sci U S A 98:4367-4372.

Robinson, S.M., and Lubke, K. 1978. The stability of A1,B29-adipoylinsulin tochemical and enzymatic degradation. Hoppe Seylers Z Physiol Chem 359:225-229.

Rux, A.H., Willis, S.H., Nicola, A.V., Hou, W., Peng, C., Lou, H., Cohen, G.H., andEisenberg, R.J. 1998. Functional region IV of glycoprotein D from herpessimplex virus modulates glycoprotein binding to the herpesvirus entrymediator. J Virol 72: 7091-7098.

Sauer-Eriksson, A.E., Kleywegt, G.J., Uhlen, M., and Jones, T.A. 1995. Crystalstructure of the C2 fragment of streptococcal protein G in complex with the Fcdomain of human IgG. Structure 3: 265-278.

Schmid, F.X. 1997. Optical spectroscopy to characterize protein conformation andconformational changes. In Protein Structure. (ed. T.E. Creighton). OxfordUniversity Press, Oxford.

Schneewind, O., Fowler, A., and Faull, K.F. 1995. Structure of the cell wall anchor ofsurface proteins in Staphylococcus aureus. Science 268: 103-106.

Schreiber, G. 2002. Kinetic studies of protein-protein interactions. Curr Opin StructBiol 12: 41-47.

Schreiber, G., and Fersht, A.R. 1995. Energetics of protein-protein interactions:analysis of the barnase-barstar interface by single mutations and doublemutant cycles. J Mol Biol 248: 478-486.

70

Schreiber, G., and Fersht, A.R. 1996. Rapid, electrostatically assisted association ofproteins. Nat Struct Biol 3: 427-431.

Selzer, T., Albeck, S., and Schreiber, G. 2000. Rational design of faster associatingand tighter binding protein complexes. Nat Struct Biol 7: 537-541.

Selzer, T., and Schreiber, G. 1999. Predicting the rate enhancement of proteincomplex formation from the electrostatic energy of interaction. J Mol Biol287: 409-419.

Selzer, T., and Schreiber, G. 2001. New insights into the mechanism of protein-protein association. Proteins 45: 190-198.

Serrano, L., Day, A.G., and Fersht, A.R. 1993. Step-wise mutation of barnase tobinase. A procedure for engineering increased stability of proteins and anexperimental analysis of the evolution of protein stability. J Mol Biol 233:305-312.

Sieber, V., Pluckthun, A., and Schmid, F.X. 1998. Selecting proteins with improvedstability by a phage-based method. Nat Biotechnol 16: 955-960.

Sjöbring, U. 1992. Isolation and molecular characterization of a novel albumin-binding protein from group G streptococci. Infect Immun 60: 3601-3608.

Sjölander, A., Nygren, P.Å., Ståhl, S., Berzins, K., Uhlen, M., Perlmann, P., andAndersson, R. 1997. The serum albumin-binding region of streptococcalprotein G: a bacterial fusion partner with carrier-related properties. J ImmunolMethods 201: 115-123.

Smith, C.K., and Regan, L. 1995. Guidelines for protein design: the energetics of betasheet side chain interactions. Science 270: 980-982.

Smith, C.K., Withka, J.M., and Regan, L. 1994. A thermodynamic scale for the beta-sheet forming tendencies of the amino acids. Biochemistry 33: 5510-5517.

Smith, D.B., and Johnson, K.S. 1988. Single-step purification of polypeptidesexpressed in Escherichia coli as fusions with glutathione S-transferase. Gene67: 31-40.

Smith, G.P. 1985. Filamentous fusion phage: novel expression vectors that displaycloned antigens on the virion surface. Science 228: 1315-1317.

Spitzfaden, C., Weber, H.P., Braun, W., Kallen, J., Wider, G., Widmer, H.,Walkinshaw, M.D., and Wuthrich, K. 1992. Cyclosporin A-cyclophilincomplex formation. A model based on X-ray and NMR data. FEBS Lett 300:291-300.

Starovasnik, M.A., Skelton, N.J., O'Connell, M.P., Kelley, R.F., Reilly, D., andFairbrother, W.J. 1996. Solution structure of the E-domain of staphylococcalprotein A. Biochemistry 35: 15558-15569.

Stemmer, W.P. 1994. Rapid evolution of a protein in vitro by DNA shuffling. Nature370: 389-391.

Stephenson, R.C., and Clarke, S. 1989. Succinimide formation from aspartyl andasparaginyl peptides as a model for the spontaneous degradation of proteins. JBiol Chem 264: 6164-6170.

Stone, S.R., and Hermans, J.M. 1995. Inhibitory mechanism of serpins. Interaction ofthrombin with antithrombin and protease nexin 1. Biochemistry 34: 5164-5172.

Ståhl, S., and Nygren, P.Å. 1997. The use of gene fusions to protein A and protein Gin immunology and biotechnology. Pathol Biol (Paris) 45: 66-76.

Ståhl, S.H., S. Nilsson, J. Uhlén, M. Nygren, P.-Å. 1999. Genetic approaches tofacilitate purification. In Isolation and purification of protein of proteins. (ed.B.H.-K. Mattiasson, R). Marcel Dekker, New York.

71

Takeda, S., Kamiya, N., Arai, R., and Nagamune, T. 2001. Design of an artificiallight-harvesting unit by protein engineering: cytochrome b(562)-greenfluorescent Protein chimera. Biochem Biophys Res Commun 289: 299-304.

Tashiro, M., Tejero, R., Zimmerman, D.E., Celda, B., Nilsson, B., and Montelione,G.T. 1997. High-resolution solution NMR structure of the Z domain ofstaphylococcal protein A. J Mol Biol 272: 573-590.

Tawfik, D.S., and Griffiths, A.D. 1998. Man-made cell-like compartments formolecular evolution. Nat Biotechnol 16: 652-656.

Tomizawa, H., Yamada, H., Hashimoto, Y., and Imoto, T. 1995. Stabilization oflysozyme against irreversible inactivation by alterations of the Asp-Glysequences. Protein Eng 8: 1023-1028.

Tyler-Cross, R., and Schirch, V. 1991. Effects of amino acid sequence, buffers, andionic strength on the rate and mechanism of deamidation of asparagineresidues in small peptides. J Biol Chem 266: 22549-22556.

Unneberg, P., Merelo, J.J., Chacon, P., and Moran, F. 2001. SOMCD: method forevaluating protein secondary structure from UV circular dichroism spectra.Proteins 42: 460-470.

van den Burg, B., and Eijsink, V.G. 2002. Selection of mutations for increased proteinstability. Curr Opin Biotechnol 13: 333-337.

Van den Burg, B., Vriend, G., Veltman, O.R., Venema, G., and Eijsink, V.G. 1998.Engineering an enzyme to resist boiling. Proc Natl Acad Sci U S A 95: 2056-2060.

Vaughan, C.K., Buckle, A.M., and Fersht, A.R. 1999. Structural response to mutationat a protein-protein interface. J Mol Biol 286: 1487-1506.

Vijayakumar, M., Wong, K.Y., Schreiber, G., Fersht, A.R., Szabo, A., and Zhou,H.X. 1998. Electrostatic enhancement of diffusion-controlled protein-proteinassociation: comparison of theory and experiment on barnase and barstar. JMol Biol 278: 1015-1024.

Vriend, G., Berendsen, H.J., van den Burg, B., Venema, G., and Eijsink, V.G. 1998.Early steps in the unfolding of thermolysin-like proteases. J Biol Chem 273:35074-35077.

Wang, Q., Buckle, A.M., Foster, N.W., Johnson, C.M., and Fersht, A.R. 1999. Designof highly stable functional GroEL minichaperones. Protein Sci 8: 2186-2193.

Wang, Y., Shen, B.J., and Sebald, W. 1997. A mixed-charge pair in humaninterleukin 4 dominates high-affinity interaction with the receptor alpha chain.Proc Natl Acad Sci U S A 94: 1657-1662.

Watson, J.D, and Crick, F.H. 1953. Molecular structure of nucleic acids: a structurefor deoxyribose nucleic acid. Nature 171: 737-738.

Wenk, M., Jaenicke, R., and Mayr, E.M. 1998. Kinetic stabilisation of a modularprotein by domain interactions. FEBS Lett 438: 127-130.

Wider, G., and Wüthrich, K. 1999. NMR spectroscopy of large molecules andmultimolecular assemblies in solution. Curr Opin Struct Biol 9: 594-601.

Wikström, M., Drakenberg, T., Forsen, S., Sjöbring, U., and Björck, L. 1994. Three-dimensional solution structure of an immunoglobulin light chain-bindingdomain of protein L. Comparison with the IgG-binding domains of protein G.Biochemistry 33: 14011-14017.

Wikström, M., Sjöbring, U., Drakenberg, T., Forsen, S., and Björck, L. 1995.Mapping of the immunoglobulin light chain-binding site of protein L. J MolBiol 250: 128-133.

72

Williams, R.W., Chang, A., Juretic, D., and Loughran, S. 1987. Secondary structurepredictions and medium range interactions. Biochim Biophys Acta 916: 200-204.

Willis, S.H., Rux, A.H., Peng, C., Whitbeck, J.C., Nicola, A.V., Lou, H., Hou, W.,Salvador, L., Eisenberg, R.J., and Cohen, G.H. 1998. Examination of thekinetics of herpes simplex virus glycoprotein D binding to the herpesvirusentry mediator, using surface plasmon resonance. J Virol 72: 5937-5947.

Winter, G., Fersht, A.R., Wilkinson, A.J., Zoller, M., and Smith, M. 1982.Redesigning enzyme structure by site-directed mutagenesis: tyrosyl tRNAsynthetase and ATP binding. Nature 299: 756-758.

Wojcik, J., Mornon, J.P., and Chomilier, J. 1999. New efficient statistical sequence-dependent structure prediction of short to medium-sized protein loops basedon an exhaustive loop classification. J Mol Biol 289: 1469-1490.

Wright, H.T. 1991. Sequence and structure determinants of the nonenzymaticdeamidation of asparagine and glutamine residues in proteins. Protein Eng 4:283-294.

Xie, M., Aube, J., Borchardt, R.T., Morton, M., Topp, E.M., Vander Velde, D., andSchowen, R.L. 2000. Reactivity toward deamidation of asparagine residues inbeta-turn structures. J Pept Res 56: 165-171.

Åkerström, B., Brodin, T., Reis, K., and Björck, L. 1985. Protein G: a powerful toolfor binding and detection of monoclonal and polyclonal antibodies. J Immunol135: 2589-2592.