8 TO 5.3 KCALÆMOL) NMR SPECTRA OF THESE PROTEINSSHOWED SHARP PEAKS...

1.8 to 5.3 kcalÆmol

⁾¹

. NMR spectra of these proteinsshowed sharp peaks, suggesting folded proteins wereEffect of flexible loops and partially foldedselected. Detailed structural information is needed tointermediate on selectiondemonstrate its final success.In a more recent test for the core-directed designDepending on the position of protease cutting sites in theproposal, Chuet al. [18] have converted a partially unfoldedstructure, the existence of flexible loops and partiallystate, apocytochromeb

₅₆₂

, to a fully folded four-helixunfolded states could have a significant effect on the resultbundle protein in the absence of any cofactors. In thisof selection. If the cutting sites are in the flexible loop of thework, the authors used the method similar to that ofnative structure, they could prevent the selection of stableproteins. By examining the structures of the proteins studiedFinucane et al. [11] except that the protein A B-domainby the phage-display and proteolysis, we found thatinstead of His-tag was used to select the folded proteins.Cytochromeb

562

(cytb

562

) is a four-helix bundle proteinprotease cutting sites exist in the loop regions for bothcases (barnase and ubiquitin) in which the selection did notwith a heme holding the N- and C-terminal helices(Fig. 2A). In the absence of heme, apocytochromeb

562

produce very stable proteins (Table 1). A more seriousadopts a partially unfolded conformation with theproblem can arise from the existence of partially unfoldedC-terminal helix largely unfolded while the other threestates that have the protease digestion sites in their unfoldedhelices remain folded. To create a four-helix bundle proteinregions (Fig. 3). This is because the mutations in the foldedin the absence of heme, four residues at positions 7, 98, 102regions of the intermediate do not significantly change therelative population between the intermediate and the fullyand 106, that are expected to form a hydrophobic core andsubstitute the heme, were mutated. Residue 7 was changedfolded state. Therefore, little evolution pressure can beadded for selection of stable proteins if mutations are madeto Trp to provide a fluorescence probe for studying thein the folded region. To be able to select stable mutantsprotein’s physical properties. The other three positions wereusing phage-display and proteolysis, it is necessary thatrandomly mutated. In addition, residue 99 in the region forthe protease cutting sites be close to the mutation sites orredesign was substituted with Arg to provide a specificin the region that is exposed only upon global unfolding.cutting site for protease Arg-c. This library of mutants wasdisplayed on the surface of phage and challenged with pro-The stable region may be determined by the existence of theslowest exchanging amide protons.tease Arg-c to select stably folded proteins. The consensusstructure of this protein has been solved using the X-raycrystallography method [23]. It is found that the B-factors ofthe mutated residues are much higher than those of otherresidues, indicating that there are significant dynamicmotions in the redesigned structure, which may contributein part to the thermodynamic stability. We also examinedanother computer-designed protein G B1-domain variantby Malakauskas & Mayo [24]. This redesigned protein isalso more stable than the wild type by4 kcalÆmol

⁾¹

. In thiscase, the dynamic behavior of the redesigned protein is evenmore dramatic. Several cross peaks that correspond to theredesigned residues in the

¹

H-

¹⁵

N HSQC spectrum havevery weak intensities even though the structure of theredesigned protein has been solved using NMR [24].Examination of the mutations in the two computerredesigned proteins shows that most of the mutations arefrom polar to hydrophobic residues. Thus, the two designshave essentially reversed the earlierde novodesign practice,in which polar residues were incorporated into the designedhydrophobic core to obtain unique conformation at theexpense of protein stability [25]. Regarding this issue, itshould be noted that these redesigned proteins have 1D

¹

H-NMR spectra that look very much like those of native-likestructures. Therefore, it suggests 1D

¹

H-NMR spectrum isinsufficient for determining whether a redesigned proteinhas a more dynamic motion on a fine level and

¹

H-

¹⁵

NHSQC spectra may be a minimum requirement for char-acterizing the dynamic behavior of redesigned proteins inthe future. As proteins with heterogeneous structures anddynamic behavior in the native state are likely to be moresensitive to protease digestion than those with well-packedstructures, phage-display coupled with proteolysis may beuseful for solving this difficult problem. The backbonedynamics [26] and the 3D structure of the redesignedapocytb

₅₆₂

determined by NMR clearly show that theprotein has a uniquely folded state.Combinatorial computation versusphage-displaySignificant progress has been made using combinatorial

Fig. 3. Effect of a partially unfolded intermediate on the selection result.

computation to design proteins [1,13,22,27,28]. The advant-

If the cutting site for protease is in the unfolded region of an interme-

age of the computational methods is that they can examine

diate state, selection of stable proteins will not be achieved because these

very large numbers of mutations [27]. The limitation of the

mutants will not change the free energy difference between the inter-

current computational methods, however, is that most of the

mediate (I) and the native (N) states. U represents the unfolded state.

computer programs need to have the backbone conforma-tions completely fixed in order to make the computationefficient [22,27,29]. The fix of the backbone conformationsDesign of native-like proteinscould potentially prevent selection of alternative attractiveThe major difficulty encountered in protein design has beenstructures that are slightly different in terms of backbonethat designed proteins often have more heterogeneousconformation. Earlier work on the T4 lysozyme revealedstructures than those of typical natural proteins. Thethat over-packed core mutants typically responded by slightinitially designed structures often had the correct secondaryalteration of the main chain, preserving near-ideal rotamericstructure and topology but lacked the well-packed hydro-side chain conformations [30]. Some efforts have been madephobic core that is characteristic of most natural proteinstowards solving this problem. For example, backbone[19,20]. Iterative experimental design processes are normallyfreedom was considered in designing proteins by usingrequired to achieve the final target [3,21]. This problemalgebraic parameterization of the backbone for proteins withbecomes a more critical issue because the proteins designedsimple motifs [1] and by manipulating the relative orienta-recently by computational methods have also failed in thistions of super secondary structural elements [31]. A moreaspect. Nauliet al. [22] have redesigned the secondb-hairpingeneral method has also been explored by Desjarlais &Handel [32]. Another concern is that computationalof the protein G B1-domain and obtained a protein thatis more stable than the wild type by4 kcalÆmol

⁾¹

. Themethods generally lack the consideration of multi-bodyinteractions. Therefore, long range effects of a mutation,