Proteolytic selection for protein folding using filamentousbacteriophagesPeter Kristensen and Greg Winter

Background: Filamentous bacteriophages have been used for the selectionof folded peptide and protein ‘ligands’ by binding the phage to ‘receptor’-coated solid phase. Here, using proteolysis, we have developed a techniquefor the selection of folded and stable proteins that is independent of theirbinding activities.

Results: When a 21-residue peptide comprising a protease cleavage site wasintroduced into the flexible linker between the second and third domains of theminor coat protein p3 of filamentous bacteriophage, the phages could be cleavedby trypsin and were rendered non-infective. By contrast, phages displaying mutantbarnases at this site were resistant to proteolysis, but were cleaved and theirinfectivity was destroyed as the temperature was raised. By mixing phagesbearing two barnase mutants of differing stability, and adding protease at atemperature at which one mutant was resistant and the other was sensitive, wewere able to enrich by 1.6 × 104-fold for phages bearing the more stable barnase.

Conclusions: The approach provides a means for the selection of folded andstable proteins, and may be applicable to the selection of de novo proteins.

IntroductionAttempts have been made to design proteins that are morestable than native proteins [1–3] and also folded de novoproteins [1,4]. This has posed significant challenges. Analternative approach is the use of screening or selectiontechnologies. Enzymes with improved stability have beenisolated by screening repertoires of mutants at higher tem-peratures [2] and de novo ‘proteins’ by their ability tosurvive degrading enzymes in bacteria [5–7].

Although mass screening is powerful, the use of selectiontechnologies allow the screening of even greater numbers.In particular, the use of filamentous bacteriophage hasallowed the isolation of synthetic folded human antibodiesfrom repertoires of > 1010 members built by the assemblyof different structural elements [8–10]. After assembly, theantibody genes are cloned into filamentous bacteriophageby fusion to the bacteriophage coat protein p3 such thateach phage encapsidates a set of antibody genes and dis-plays the encoded antibody fragment on its surface. Phagesare selected from the repertoire by their binding to solid-phase antigen. As the antibody needs to be folded to bindantigen, selection for binding also selects for folding. Thisprinciple has also been used for the selection of foldedpeptides, for which binding is mediated by a discontinuousepitope [11–14].

The binding activities of a protein may be unknown, how-ever, or the ligand (such as a substrate or product analog forbinding to enzyme) may be unavailable, requiring a means

of selection that is independent of the binding properties ofthe folded protein. Proteolysis has already been used as ameans of screening for protein folding; unfolded proteinsare readily digested by proteases whereas folded proteinsare often resistant, cleavage requiring the polypeptide chainto bind and adapt to the specific stereochemistry of the pro-tease active site and therefore be flexible, accessible andcapable of local unfolding [15,16]). Here, we sought to useproteolysis as a means of selection; because filamentousbacteriophage are reported to be resistant to proteolysis(allowing their use as ‘substrate’ phage [17]), we devised ameans of selection for protease-resistant proteins displayedon phage.

The phage p3 protein has three domains (D1, D2 and D3);D1 binds to the tolA receptor (required for penetration ofthe phage DNA), D2 binds to the F-pilus, and D3 anchorsthe protein to the phage [18]. Peptides and proteins can beinserted at the domain boundaries without abolishinginfectivity [19,20], but the presence of all the domains isessential for phage infectivity [21]. Proteolytic cleavage ofa protein inserted between the domains should thereforelead to a loss of phage infectivity. Here, we constructedsuitable phage vectors and have demonstrated the principleof proteolytic selection for folding.

ResultsPhage stabilityPhage was incubated under a range of denaturing con-ditions in vitro and then restored to native conditions

Address: MRC, Centre for Protein Engineering,Hills Road, Cambridge, CB2 2QH, UK.

Correspondence: Peter Kristensen andGreg WinterE-mail: pk2@mrc-lmb.cam.ac.uk andgw@mrc-lmb.cam.ac.uk

Keywords: phage display, protein folding,proteolytic cleavage

Received: 12 May 1998Revisions requested: 02 June 1998Revisions received: 08 June 1998Accepted: 12 June 1998

Published: 09 July 1998http://biomednet.com/elecref/1359027800300321

Folding & Design 09 July 1998, 3:321–328

© Current Biology Publications ISSN 0960-9822

Research Paper 321

immediately before infection of bacteria. The incubationof phage in 10 M urea or extremes of pH (as low as pH 2;and as high as pH 12) and temperature (as high as 60°C)did not lead to a major loss of infectivity (Table 1). Thisindicates that the phage is either resistant to denaturingconditions or that if it does unfold it is able to refoldrapidly. With guanidine hydrochloride (GdnHCl), how-ever, a fivefold loss in phage infectivity was observedabove 5 M and a further fivefold loss at 8 M (Table 1).

Phage was then incubated under native conditions with arange of proteases (trypsin, Factor Xa, IgA protease, Asp–N, chymotrypsin, Arg–C, Glu–C, thrombin, thermolysinand subtilisin) with different specificities. There was noloss in infectivity except for subtilisin, which has beenreported to cleave the p3 protein [22]. If phage was incu-bated under denaturing conditions in the presence of pro-teases such as trypsin in 3.5 M urea (or > 47°C), infectivitywas lost. This indicates that under denaturing conditionsthe unfolding of the phage coat proteins is sufficient tomake sites available for proteolysis.

A sequence (PAGLSEGSTIEGRGAHE) comprising sev-eral proteolytic sites was inserted in the flexible glycine-richregion between the D2 and D3 domains of the phage p3.Incubation of the phage (fd-K108) under native conditions

with trypsin, thermolysin or subtilisin resulted in an almostcomplete loss of infectivity (from 107 to < 10 TU/ml) andincubation with Glu–C and chymotrypsin resulted in amajor loss (from 107 to 104 TU/ml). This indicates thatthese proteases cleave the new linker. Incubation withFactor Xa, Arg–C or thrombin, however, did not lead to aloss in infectivity, despite the presence of potential cleav-age sites for these enzymes. Presumably, the presence ofthe D2 and D3 domains may block access or cleavage forthese enzymes.

Protease-cleavable helper phage and phagemidFusion of proteins to p3 should lead to a multivalentdisplay of the protein on the phage. But if the protein isfused to p3 encoded by a phagemid (such as pHEN1[23]), and the bacteria harbouring the phagemid is rescuedwith a helper phage (such as VCSM13), the fusion proteinhas to compete for incorporation into the phage with thehelper p3. This leads to so-called ‘monomeric’ phage, inwhich usually less than one copy of the fusion protein isattached to each phage particle [24].

The use of monomeric phage might be expected to bemore sensitive to proteolysis because only a single copy offusion protein need be cleaved for the phage to be ren-dered non-infectious and interactions between multimersof fusion protein would be avoided. This would alsorequire the construction of a protease-cleavable helperphage, however. We therefore introduced the protease-cleavage sequence between the D2 and D3 domains togenerate the helper phage KM13.

KM13 was shown to rescue the phagemid pHEN1. Fur-thermore, trypsin was shown to cleave a major fraction(~50%) of p3 of the rescued phage, as shown by Westernblot and detected with an anti-D3 mAb (Figure 1). Phageinfectivity was hardly altered by the cleavage; it thereforeappears that only a fraction of the p3 needs to be intact tomediate bacterial infection.

KM13 was also shown to rescue a pHEN1 phagemidencoding a single-chain antibody fragment [25]. Here,cleavage by trypsin resulted in a 50-fold loss in phageinfectivity (data not shown), consistent with indicationsthat only a small fraction of the phage express fusionprotein when rescued with helper phage [24,26].

We also constructed a protease-cleavable phagemid. Thephagemid could be rescued with KM13 or VCSM13. Asexpected, infectivity of this phagemid rescued with KM13(but not VCSM13) was destroyed by trypsin (data notshown). Later experiments showed that this phagemidvector was prone to deletions in the D2–D3 linker; bychanging the codon usage in the linker regions on eitherside of the protease-cleavable site, and shortening thelength of these linker regions, we created a more stable

322 Folding & Design Vol 3 No 5

Table 1

The stability of fd-DOG in different denaturing conditions.

Condition Infectivity (TU/ml × 1010)

Urea (60°C, 90 min)Control 0.562 M 0.644 M 0.326 M 0.328 M 0.8010 M 0.68

GdnHCl (37°C, 90 min)Control 0.722 M 0.604 M 0.705 M 0.166 M 0.137 M 0.168 M 0.03

pH (37°C, 30 min)Control 1.5pH 2.2 0.46pH 4.0 1.3pH 7.4 1.5pH 10 1.4pH 12 0.40

Temperature (30 min)Control 9.722°C 8.337°C 9.660°C 12.0

The infectivity (TU/ml × 1010) was measured (see the Materials andmethods section) and has an estimated error of about ± 6%.

vector (pK1, Figure 2). In a second vector (pK2), thesequence of the polylinker was arranged in order to placeD3 out of frame to render re-ligations within the polylinkernon-infectious (Figure 2).

Selection for foldingBarnaseBarnase is a small RNAse of 110 amino acids whosefolding has been extensively studied (for review see [27]).Barnase contains multiple sites for trypsin cleavage,although the folded protein is resistant to cleavage (data

not shown). Phage with barnase cloned between D2 andD3 should therefore be resistant to protease cleavage andcapable of selection.

Because barnase is toxic to Escherichia coli, we clonedmutant A (His102→Ala), which is catalytically inactive butstable [28,29], into the phagemid pK2. We also clonedmutant B (His102→Ala, Leu14→Ala) with lower stability;Leu14 is buried in the hydrophobic core and its mutationcreates a large cavity in the core affecting the packing ofdifferent structural elements [30]. The phages (rescuedwith KM13) were shown to bind to the inhibitor barstar byELISA, and therefore display the mutant barnases in afolded form (Figure 3).

The phages were then incubated with trypsin at a range oftemperatures (Figure 4). After incubation at 10°C, therewas a decrease in phage infectivity of 5–10-fold for bothmutants, suggesting that only a small fraction of the phagesdisplay the fusion protein. There was no further loss ininfectivity on cleavage until 30°C (for mutant B) or 37°C(for mutant A). In both cases, the major transition was atleast 10°C below that expected for the reversible thermalunfolding of the mutants.

We then mixed phages A and B in different ratios andincubated the mixture at 20°C with trypsin, a temperatureat which both mutants are stable to cleavage, or at 37°C, atwhich only phage A is stable. After ‘proteolytic selection’the phages were plated and analysed by PCR followed byrestriction digest, which distinguishes the mutants. Asshown in Table 2, after a single round of selection at 37°C,mutant A was enriched by a factor of 1.6 × 104, and aftertwo rounds by 1.3 × 106. No enrichment could be detectedat 20°C.

Research Paper Selection for folding Kristensen and Winter 323

Figure 1

Cleavage of phages with protease sites. Phages were prepared byrescue with KM13 (pHEN1, A + B), or with VCSM13 (pK1, C + D).Uncleaved (A + C) or cleaved with trypsin (B + D). 5 µl, 2.5 µl and 1 µlphages were loaded as indicated. Molecular weight markers are in kDa.

Domain 3

Full-lengthGene III

228102

71

46

28

19

14

5 2.5 1 5 2.5 1 5 2.5 1 2.5 2.5 1A B C D

Folding & Design

Figure 2

AmpR

pK14477 bp

BamHI

SignalFlag

D1 D2 D3

Polylinker

SfiI SalI ApaI KpnIpK1 AAT GCT GGC GGC GGC CCA GCC GGC CTT TCT GAG GGG TCG ACT ATA GAA GGA CGA GGG GCC CAC GAA GGA GGT GGG GTA CCC GGT TCC GAG GGT GGT TCC GGT TCC GGT GAT TTT GAT N A G G G P A G L S E G S T I E G R G A H E G G G V P G S E G G S G S G D F D

pK2 AAT GCT GGC GGC GGC CCA GCC GGC CTT TCT GAG GGG TCG ACT ATA GAA GGA CGA GGG CCC ACG AAG CAG CTG GGG TAC CGG TTC CGA GGG TGG TTC CGG TTC CGG TGA TTT TGA TTA N A G G G P A G L S E G S T I E G R G P T K Q L G Y R F R G W F R F R * F * L

Folding & Design

The phagemid vectors pK1 and pK2. These vectors contain a protease-cleavable sequence between D2 and D3 of the phage p3 protein. In pK1,D2 + D3 are in frame; in pK2, D3 is out of frame.

VillinThe 35 amino acid subdomain of the headpiece domain ofthe f-actin-bundling protein villin [31] is much smaller thanbarnase, but is stable to temperature and to proteolysis; fur-thermore, its stability does not rely on disulfide bonds orbinding ligands [32]. The villin subdomain, which containsseveral potential trypsin cleavage sites, was cloned betweenthe D2 and D3 domains of the phage, and incubated withtrypsin at different temperatures (Figure 4). The profile forloss of infectivity was not as sharp as with barnase, with themajor transition below 35°C, considerably below the ther-mal unfolding of villin (70°C; [31,32]). The phage display-ing villin were mixed with pK1 and incubated with trypsin.After a single round of proteolytic selection, the fusionphage were enriched 8.7 × 103-fold (Table 3).

DiscussionOur results show that the infectivity of the phage is rela-tively resistant to temperature, pH, urea and GdnHCl, andto several proteases, but if a flexible linker comprising a pro-tease-cleavage site is inserted between domains D2 and D3of the phage coat protein p3, the phage becomes sensitiveto cleavage. By contrast, if the protease-cleavage sites com-prise a folded protein domain, such as barnase or villin, thephage is resistant to cleavage. This allows proteolytic selec-tion for protein folding with enrichment factors of > 104-foldfor a single round of selection. Selection was evident forbarnase, an average sized [33] domain of 110 amino acids,and for villin, a small domain of 35 amino acids.

Although the approach seems general, there are several lim-itations. The phage has to be capable of resisting the diges-tion conditions; in future it may be possible to select phageswith improved stability to act as carriers for proteins to beselected. The protein of interest has to be exported throughthe bacterial membrane into the periplasm and also has tocontain a cleavage site for the chosen protease (thus requir-ing the use of different proteases or a cocktail of proteases).Furthermore, the protein has to fall apart on nicking. Insome cases the cleaved portions are expected to remain non-covalently attached, but are likely to have a lower stabilitythan the native structure [34–36] and should be amenableto being teased apart with denaturants (see below).

Discrimination between structures of different stabilitiescan be accomplished by increasing the stringency of prote-olytic selection. Thus, with increases in temperature, bothbarnase and villin became susceptible to cleavage, reflectingprotein unfolding. The main impact of protease cleavage,however, was at a temperature lower than the unfoldingtransition as measured by CD [37]. This may reflect the factthat the unfolding transition is a fully reversible process,whereas cleavage by proteases (of unfolded structure) is a

324 Folding & Design Vol 3 No 5

Figure 3

Binding of phage–barnase to barstar. Phages displaying differentfusion proteins are incubated with biotinylated barstar captured on astreptavidin-coated plate and are detected by ELISA.

0

0.1

0.2

0.3

0.4

0.5

OD405

Barnasemutant A

Barnasemutant B

Villin No phageFolding & Design

Figure 4

Temperature denaturation of phage fusion proteins. Phagemids wererescued with KM13. Infectivity is shown after incubation and cleavagewith trypsin at given temperatures. Fusion is with the villin subdomain,barnase mutant A, barnase mutant B and chloramphenicol-resistantpHEN1.

103

104

105

106

107

108

109

1010

1011

1012

0 10 20 30 40 50 60Temperature (ºC)

Infe

ctiv

ity (T

U/m

l)

Folding & Design

Villin subdomainBarnase mutant ABarnase mutant BpHEN1 (ChlorR)

kinetic and irreversible process, pulling over the equilib-rium from folded to unfolded (and cleaved) structure. Thisis consistent with the CD unfolding transition seen for villin[32] for which at temperatures as low as 35°C there is evi-dence of unfolding; this is the same point at which villinstarts to become susceptible to protease attack.

We anticipate several areas for the application of prote-olytic selection for protein folding because it should allowmuch larger numbers of proteins to be processed than withscreening. Proteolytic selection may, for example, allowthe isolation of mutant proteins with improved stability [1];for example, from combinatorial libraries of mutants inwhich residues at several sites are varied simultaneously[38,39] or from random mutants or by recombination [2,40].Proteolytic selection may also facilitate the isolation of pro-teins de novo [5–7,41]; although such proteins may wellretain elements of secondary structure, they often lack thestable tertiary interactions characteristic of the folding ofnative proteins, suggesting the presence of molten glob-ules [3]. It may be possible to improve packing and stabletertiary interactions by using several rounds of mutationand increasingly stringent proteolytic selection, much likethe affinity maturation of antibodies.

We also anticipate other uses for proteolytic selection. Forphage display repertoires many of the phage do notdisplay the protein [24], but are infective by virtue of the

p3 contributed by the helper phage. As these phages con-tribute to the background of non-specific binding toantigen-coated solid phase, the use of protease-cleavablehelper phage should destroy their contribution to the back-ground and thereby improve selection efficiencies. Thisrequires the fusion of the protein to intact p3, however,either at the N terminus or between domain boundaries.Furthermore, it may be possible to use proteolytic selec-tion for the identification of interacting protein elements.Thus, if two such elements linked by a protease-cleavablelinker were cloned between the D2 and D3 domains fordisplay on phage, the only infectious phages after proteoly-sis should be those in which the D2 and D3 domains areheld together by non-covalent interactions between theinteracting protein elements. This should provide an alter-native to other phage selection strategies for molecularinteractions based on the association of the domains ofp3 [42–44].

Materials and methodsMaterialsAll restriction enzymes and T4 ligase were obtained from New EnglandBiolabs. Taq DNA polymerase were obtained from HT Biotechnology.Pfu DNA polymerase was obtained from Stratagene. Ultrapure dNTPwas from Pharmacia. Proteases and the protease inhibitor Pefabloc werefrom Boehringer Mannheim, except chymotrypsin and trypsin TPCK-treated, which were obtained from Sigma. All other chemicals wereobtained from Sigma.

Phage preparationEscherichia coli TG1 [45] was used for cloning and propagation ofphage. TG1 harbouring fd-DOG [46] or derivatives was grown over-night in 2 × TY containing 15 µg/ml tetracycline. Phagemids wererescued using KM13 or VCSM13 as described [25]. Phage particleswere prepared by two PEG precipitations [47].

Vector constructionThe phage vector fd-DOG [46] was used as the parent vector for theconstruction of the protease cleavable fd-K108. Unique restriction sites(SfiI and KpnI) were introduced into the glycine-rich spacer regionbetween D2 and D3 using the Sculptor in vitro mutagenesis system(Amersham) and the oligonucleotide pklinker (Table 4). Further restrictionsites (ApaI and SalI) and sequence encoding a protease-cleavage sitewere cloned between the SfiI and KpnI sites using the oligonucleotidespolyXafor and polyXaback to create the vector fd-K108.

Research Paper Selection for folding Kristensen and Winter 325

Table 2

Selection of barnase mutants.

Round Ratio of phage A : phage B

1:1* 1:102 1:104 1:106 1:108

Round 1 Phage A 16 (14†) 24 20 0 ndPhage B 8 (10†) 0 4 24 ndEnrichment – – 1.6 × 104 – nd

Round 2 Phage A nd nd nd 24 0Phage B nd nd nd 12 36Enrichment nd nd nd 1.3 × 106 –

Mixtures of barnase mutants (A + B) in ratios from 1:1 to 1:10–8 wereselected by proteolysis at 37°C; 24 (in round 1) or 36 (in round 2)phage clones were analysed and the numbers of each mutant were

noted. *Selection at 20°C, at which temperature both mutants areexpected to be stable. †Before selection. nd, not detected.

Table 3

Selection of villin.

pK1 : villin–phage

Sample 1:1 1:10–2 1:10–4 1:10–6

pK1 0 (16*) 0 7 24Villin 24 (8*) 24 17 0Enrichment – – 8.7 × 103 –

Mixtures of villin–phage and pK1 rescued with KM13 in ratios from 1:1to 1:10–6 were selected by proteolysis at 10°C; 24 phage clones wereanalysed and the number of each was noted. *Before selection.

The protease-cleavable helper phage KM13 was prepared fromfd-K108 by transplanting into the helper phage VCSM13 a BamHI–ClaIfragment generated by PCR and primers fdPCRBack and LIBSEQfor.

A protease-cleavable phagemid vector was derived from fd-K108 asabove except using pCANTAB 3 (Pharmacia). A FLAG-tag was intro-duced at the N terminus of D1 by cloning a NotI–SfiI fragment gener-ated by PCR and primers Flagprimer and LSPAback. To circumventdeletions resulting from repeated sequence in the D2–D3 linker, thecodon usage of the polylinker region was changed in two steps: usinga BamHI–SfiI fragment generated by PCR and primers RECGLYforand LIBSEQfor, screening recombinants by PCR and the primersLSPAfor and LSPAback; and using a KpnI–ClaI fragment generated byPCR and the primers RECGLYback and LIBSEQback, screeningrecombinants using LSPAfor and LSPAback. The entire p3 gene wassequenced using PCR cycle sequencing with fluorescent dideoxychain terminators (Applied Biosystem) according to [48]. The ‘out offrame’ vector pK2 (Figure 2) was derived from pK1 by site-directedmutagenesis using the oligo delCKpn and the Sculptor Amersham kit.

Cloning of barnase and villinThe vectors encoding the single barnase mutants, His102→Ala andLeu14→Ala [28,49], were used as templates for PCR amplificationwith primers Barnasefor and BarnaseH102Aback and Pfu polymerase.The PCR products (encoding the single mutant His102→Ala and thedouble mutant His102→Ala, Leu14→Ala) were digested using therestriction enzymes SfiI and KpnI, and ligated into vector pK2 to givethe phagemids pK2BA and pK2BB, respectively, and the barnasegenes sequenced using PCR cycle sequencing.

The 35 amino acid thermostable fragment of the headpiece of thef-actin-binding protein villin [32] was amplified from chicken bursacDNA using PCR primers villinfor and villinback with Pfu polymerase.The PCR products were cloned as above to give the phagemid pK2V.

Resistance of phages to denaturants, pH and proteasesFor resistance to denaturants, 10 M urea in PBS (25 mM NaH2PO4,125 mM NaCl pH 7.0) or 8 M GdnHCl and 50 mM Tris–HCl pH 7.4,1 mM CaCl2 (buffer A) was added to 10 µl phage stocks (108–1010 TU)to give a volume of 1 ml and the conditions specified in Table 1. Thephage were incubated for 1–2 h, then a 100 µl aliquot was added to 1 mlTG1 (OD600 ~0.5) and serial dilutions were plated on TYE plates with15 µg/ml tetracycline. For resistance of phages to extremes ofpH (2–12), Tris–glycine or Tris–HCl buffers (0.1 M glycine or 0.1 M Trisrespectively) were added to 10 µl phage stocks, and to neutralise each100 µl aliquot we added 50 µl 1 M Tris–HCl pH 7.4 before infection. Forresistance to temperature, buffer A was added to 10 µl phage stocks togive a volume of 1 ml and incubated at a given temperature (20–60°C)for 1 h. 100 µl aliquots were added to TG1 and plated as above. For

resistance to proteases, 100 mM NaCl, 50 mM Tris–HCl, 1 mM CaCl2pH 7.4 (Factor Xa 100 ng/ml or trypsin, chymotrypsin, thrombin, ther-molysin and subtilisin, all 100 µg/ml) or 50 mM Tris–HCl, 1 mM EDTApH 7.4 (IgA Protease 10 ng/ml) or 50 mM NH4CO3 pH 8.0 (Arg–C100 µg/ml, Glu–C 100 µg/ml) or 25 mM NaH2PO4, 125 mM NaClpH 7.0 (Asp–N 40 ng/ml) was added to 10 µl phage stocks (fd-DOGand fd-K108) to give a volume of 100 µl and a final concentration ofprotease as indicated. Digestions were incubated for 15 min at roomtemperature, samples (100 µl) were then infected into TG1 as above.

For resistance to proteases in the presence of denaturants, sampleswere prepared as above for urea and temperature denaturation. To90 µl aliquots, 10 µl trypsin (1 mg/ml) was added; after 5 min at roomtemperature 4 µl Pefabloc (100 mM) was added and the samples wereinfected into TG1 as above.

Western blotPhages (pHEN1 rescued using KM13 and pK1 rescued usingVCSM13) were subjected to SDS-PAGE [50] before or after cleavageby trypsin (50 ng/ml). After semi-dry transfer to PVDF membranes thefilter was processed essentially as described [25]. The primary anti-body, monoclonal anti-gIII (MoBiTec), was added in a 1:5000 dilutionfollowed by anti-mouse HRP-conjugated antibody (Sigma) in a dilutionof 1:50,000. Finally, the filter was developed using the luminol-basedChemiluminescence Western Blotting kit (Boehringer Mannheim).

ELISAPhage displaying barnase mutants were analysed for binding to theRNase inhibitor barstar as described [47]. 10 pmol biotinylated barstar(P. Kirkham, D. Neri and G.W., unpublished observations) was mixedwith approximately 1010 phage displaying barnase mutant A, or barnasemutant B, villin or buffer A. Phage binding barstar was captured onStreptavidin coated plates (Boehringer Mannheim) and developed usingHRP conjugated anti-M13 antibody (Pharmacia) and 2,2′-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) (Sigma). Absorbance readings weretaken at 405 nm.

Temperature denaturationAt each temperature ~1010 phage displaying the barnase mutants orvillin (ampicillin resistant) were mixed with a cleavable control fd-K108(tetracycline resistant), and a non-cleavable control phagemid, a chloro-amphenicol-resistant derivative of pHEN1 (P. Wang, unpublished obser-vations), rescued with KM13 in a total volume of 90 µl of buffer A. Afterequilibration for 20–30 min at the temperature indicated, 10 µl trypsin(5 µg/ml) was added and the incubation continued for 2 min. Trypsinwas neutralised by adding 4 µl 100 mM Pefabloc. Infection and serialdilution was performed in TG1 as above and aliquots were plated onTYE plates containing 100 µg/ml ampicillin + 1% glucose, 30 µg/mlchloroamphenicol + 1% glucose or 15 µg/ml tetracycline.

326 Folding & Design Vol 3 No 5

Table 4

Primer sequences.

pklinker 5’ GGCACCCTCAGAACGGTACCCCACCCTCAGAGGCCGGCTGGGCCGCCACCCTCAGAG 3’polyXafor 5’ GGTGGCGGCCCAGCCGGCCTTTCTGAGGGGTCGACTATAGAAGGACGAGGGCCCAGCGAAGGAGGTGGGGTACCCCCTTCTGAGGGTGG 3’polyXaback 5’ CCACCCTCAGAAGGGGGTACCCCACCTCCTTCGCTGGGCCCTCGTCCTTCTATAGTCGACCCCTCAGAAAGGCCGGCTGGGCCGCCACC 3’fdPCRBack 5’ GCGATGGTTGTTGTCATTGTCGGC 3’LIBSEQfor 5’ AAAAGAAACGCAAAGACACCACGG 3’LIBSEQback 5’ CCTCCTGAGTACGGTGATACACC 3’LSPAfor 5’ GTAAATTCAGAGACTGCGCTTTCC 3’LSPAback 5’ ATTTTCGGTCATAGCCCCCTTATTAG 3’Flagprimer 5’ CAAACGGGCGGCCGCAGACTACAAGGATGACGACGACAAGGAAACTGTTGAAAGTTGTTTAGCAA 3’RECGLYfor 5’ CCCCTCAGAAAGGCCGGCTGGGCCGCCGCCAGCATTGACAGGAGGTTCAGG 3’RECGLYback 5’ GAAGGAGGTGGGGTACCCGGTTCCGAGGGTGGTTCCGGTTCCGGTGATTTTG 3’delCKpn 5’ CCCTCGGAACCGGTACCCCAGCTGCTTCGTGGGCCC 3’Barnasefor 5’ CTGGCGGCGGCCCAGCCGGCCCTGCACAGGTTATCAACACGTTTGAC 3’BarnaseH102Aback 5’ CTCGGAACCGGTACCTCTGATTTTTGTAAAGGTCTGATAAGCG 3’villinfor 5’ GGCGGCCCAGCCGGCCTTTCTCTCTCTGACGAGGACTTCAAGGC 3’villinback 5’ CCTCGGAACCGGTACCGAAGAGTCCTTTCTCCTTCTTGAGG 3’

Selection experiments10 µl of serial dilutions of the barnase mutant phage A was mixed with10 µl of the non-diluted barnase mutant phage B in 70 µl buffer A. After30 min incubation at 20°C or 37°C, 10 µl trypsin (5 µg/ml) was added.Following 2 min of digestion, 4 µl Pefabloc (100 mM) was added. Thephage were infected into TG1 as above. A second round of selectionwere performed by scraping bacteria in 3 ml 2 × TY, 50 µl inoculated into50 ml 2 × TY/ampicillin/glutamate and the phagemid rescued and phageprepared as above. Clones were analysed by PCR using the primersLSPAfor and LSPAback followed by restriction digestion using DdeI.

Selections between pK2V and pK1 phage particles were performed asabove, except that the selection was performed at 10°C. Clones wereanalysed by PCR using the primers LSPAfor and LSPAback.

AcknowledgementsThe authors would like to thank Nick Foster for providing the barnasemutants used in this study. P.K. has been supported in part by a grant fromthe Danish Natural Science Research Council.

References1. Sauer, R.T. (1996). Protein folding from a combinatorial perspective.

Fold. Des. 1, R27-R30.2. Zhao, H., Giver, L., Shao, Z., Affholter, J.A. & Arnold, F.H. (1998).

Molecular evolution by staggered extension process (StEP) in vitrorecombination. Nature Biotechnol. 16, 258-261.

3. Munson, M., et al., & Regan, L. (1996). What makes a protein aprotein? Hydrophobic core designs that specify stability and structuralproperties. Protein Sci. 5, 1584-1593.

4. Dahiyat, B.I., Sarisky, C.A. & Mayo, S.L. (1997). De novo proteindesign: towards fully automated sequence selection. J. Mol. Biol. 273,789-796.

5. Kamtekar, S., Schiffer, J.M., Xiong, H., Babik, J.M. & Hecht, M.H.(1993). Protein design by binary patterning of polar and nonpolaramino acids. Science 262, 1680-1685.

6. Davidson, A.R. & Sauer, R.T. (1994). Folded proteins occur frequentlyin libraries of random amino acid sequences. Proc. Natl Acad. Sci.USA 91, 2146-2150.

7. Davidson, A.R., Lumb, K.J. & Sauer, R.T. (1995). Cooperatively foldedproteins in random sequence libraries. Nat. Struct. Biol. 2, 856-864.

8. Hoogenboom, H.R. & Winter, G. (1992). By-passing immunisation.Human antibodies from synthetic repertoires of germline VH genesegments rearranged in vitro. J. Mol. Biol. 227, 381-388.

9. Winter, G., Griffiths, A.D., Hawkins, R.E. & Hoogenboom, H.R. (1994).Making antibodies by phage display technology. Annu. Rev. Immunol.12, 433-455.

10. Griffiths, A.D., et al., & Winter, G. (1994). Isolation of high affinityhuman antibodies directly from large synthetic repertoires. EMBO J.13, 3245-3260.

11. Braisted, A.C. & Wells, J.A. (1996). Minimizing a binding domain fromprotein A. Proc. Natl Acad. Sci. USA 93, 5688-5692.

12. Riddle, D.S., et al., & Baker, D. (1997). Functional rapidly foldingproteins from simplified amino acid sequences. Nat. Struct. Biol. 4,805-809.

13. O’Neil, K.T., Hoess, R.H., Raleigh, D.P. & DeGrado, W.F. (1995).Thermodynamic genetics of the folding of the B1 immunoglobulin-binding domain from streptococcal protein G. Proteins 21, 11-21.

14. Gu, H., et al., & Baker, D. (1995). A phage display system for studyingthe sequence determinants of protein folding. Protein Sci. 4,1108-1117.

15. Hubbard, S.J., Eisenmenger, F. & Thornton, J.M. (1994). Modelingstudies of the change in conformation required for cleavage of limitedproteolytic sites. Protein Sci. 3, 757-768.

16. Fontana, A., Polverino de Laureto, P., De Filippis, V., Scaramella, E. &Zambonin, M. (1997). Probing the partly folded states of proteins bylimited proteolysis. Fold. Des. 2, R17-R26.

17. Matthews, D.J. & Wells, J.A. (1993). Substrate phage: selection ofprotease substrates by monovalent phage display. Science 260,1113-1117.

18. Riechmann, L. & Holliger, P. (1997). The C-terminal domain of TolA isthe coreceptor for filamentous phage infection of E. coli. Cell 90,351-360.

19. Smith, G.P. (1985). Filamentous fusion phage: novel expressionvectors that display cloned antigens on the virion surface. Science228, 1315-1317.

20. Krebber, C., et al., & Plückthun, A. (1997). Selectively-infective phage(SIP): a mechanistic dissection of a novel in vivo selection forprotein-ligand interactions. J. Mol. Biol. 268, 607-618.

21. Stengele, I., Bross, P., Garces, X., Giray, J. & Rasched, I. (1990).Dissection of functional domains in phage fd adsorption protein.Discrimination between attachment and penetration. J. Mol. Biol. 212,143-149.

22. Gray, C.W., Brown, R.S. & Marvin, D.A. (1981). Adsorption complexof Filamentous fd virus. J. Mol. Biol. 146, 621-627.

23. Hoogenboom, H.R., et al., & Winter, G. (1991). Multi-subunit proteinson the surface of filamentous phage: methodologies for displayingantibody (Fab) heavy and light chains. Nucleic Acids Res. 19,4133-4137.

24. Bass, S., Greene, R. & Wells, J.A. (1990). Hormone phage: anenrichment method for variant proteins with altered binding properties.Proteins 8, 309-314.

25. Nissim, A., et al., & Winter, G. (1994). Antibody fragments from a“single pot” phage display library as immunochemical reagents.EMBO J. 13, 692-698.

26. Marzari, R., Sblattero, D., Righi, M. & Bradbury, A. (1997). Extendingfilamentous phage host range by the grafting of a heterologousreceptor binding domain. Gene 185, 27-33.

27. Fersht, A.R. (1993). Protein folding and stability: the pathway offolding of barnase. FEBS Lett. 325, 5-16.

28. Mossakowska, D.E., Nyberg, K. & Fersht, A.R. (1989). Kineticcharacterisation of the recombinant ribonuclease from bacillusamyloliquefaciens (Barnase) and investigation of key residues incatalysis by site-directed mutagenesis. Biochemistry 28, 3843-3850.

29. Meiering, E.M., Serrano, L. & Fersht, A.R. (1992). Effect of active siteresidues in barnase on activity and stability. J. Mol. Biol. 225,585-589.

30. Serrano, L., Kellis, J.T., Cann, P., Matouschek, A. & Fersht, A.R.(1992). The folding of an enzyme. II. Substructure of barnase and thecontribution of different interactions to protein stability. J. Mol. Biol.224, 783-804.

31. McKnight, C.J., Matsudaira, P.T. & Kim, P.S. (1997). NMR structure ofthe 35-residue villin headpiece subdomain. Nat. Struct. Biol. 4,180-184.

32. McKnight, C.J., Doering, D.S., Matsudaira, P.T. & Kim, P.S. (1996). Athermostable 35-residue subdomain within villin headpiece. J. Mol.Biol. 260, 126-134.

33. Xu, D. & Nussinov, R. (1997). Favorable domain size in proteins. Fold.Des. 3, 11-17.

34. Kippen, A.D. & Fersht, A.R. (1995). Analysis of the mechanism ofassembly of cleaved barnase from two peptide fragments and itsrelevance to the folding pathway of uncleaved barnase. Biochemistry34, 1464-1468.

35. Gay, G.d.P. & Fersht, A.R. (1994). Generation of a family of proteinfragments for structure-folding studies. 1. folding complementation oftwo fragments of chymotrypsin inhibitor-2 formed by cleavage at itsunique methionine residue. Biochemistry 33, 7957-7963.

36. Wu, L.C., Grandori, R. & Carey, J. (1994). Autonomous subdomains inprotein folding. Protein Sci. 3, 369-371.

37. Kwon, W.S., Silva, N.A.D. & Kellis, J.T. (1996). Relationships betweenthermal stability, degradation rate and expression yield of barnasevariants in the periplasm of Escherichia coli. Protein Eng. 9, 1197-1202.

38. Axe, D.D., Foster, N.W. & Fersht, A.R. (1996). Active barnase variantswith completely random hydrophobic cores. Proc. Natl Acad. Sci.USA 93, 5590-5594.

39. Waldburger, C.D., Schildbach, J.F. & Sauer, R.T. (1995). Are buriedsalt bridges important for protein stability and conformationalspecificity? Nat. Struct. Biol. 2, 122-128.

40. Patten, P.A., Howard, R.J. & Stemmer, W.P.C. (1997). Applications ofDNA shuffling to pharmaceuticals and vaccines. Curr. Opin.Biotechnol. 8, 724-733.

41. Roy, S., et al., & Hecht, M.H. (1997). A protein designed by binarypatterning of polar and nonpolar amino acids displays native-likeproperties. J. Am. Chem. Soc. 119, 5302-5306.

42. Gramatikoff, K., Georgiev, O. & Schaffner, W. (1994). Directinteraction rescue, a novel filamentous phage technique to studyprotein-protein interactions. Nucleic Acids Res. 22, 5761-5762.

43. Duenas, M. & Borrebaeck, C.A.K. (1994). Clonal selection andamplification of phage displayed antibodies by linking antigenrecognition and phage replication. Bio/Technology 12, 999-1002.

44. Krebber, C., Spada, S., Desplancq, D. & Plückthun, A. (1995).Co-selection of cognate antibody-antigen pairs by selectively-infectivephages. FEBS Lett. 377, 227-231.

Research Paper Selection for folding Kristensen and Winter 327

45. Gibson, T.J. (1984). Studies on the Epstein-Barr Virus Genome. PhDThesis, University of Cambridge, Cambridge, UK.

46. Clackson, T., Hoogenboom, H.R., Griffiths, A.D. & Winter, G. (1991).Making antibody fragments using phage display libraries. Nature 352,624-628.

47. McCafferty, J., Griffiths, A.D., Winter, G. & Chiswell, D.J. (1990).Phage antibodies: filamentous phage displaying antibody variabledomains. Nature 348, 552-554.

48. Fisch, I., et al., & Winter, G. (1996). A strategy of exon shuffling formaking large peptide repertoires displayed on filamentousbacteriophage. Proc. Natl Acad. Sci. USA 93, 7761-7766.

49. Matouschek, A., Kellis, J.T., Serrano, L. & Fersht, A.R. (1989). Mappingthe transition state and pathway of protein folding by proteinengineering. Nature 340, 122-126.

50. Laemmli, U.K. (1970). Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature 227, 680-685.

328 Folding & Design Vol 3 No 5

Because Folding & Design operates a ‘Continuous PublicationSystem’ for Research Papers, this paper has been publishedon the internet before being printed. The paper can beaccessed from http://biomednet.com/cbiology/fad — forfurther information, see the explanation on the contents pages.