Single Amino Acid Substitutions Globally Suppress the Folding Defects of Temperature-sensitive Folding Mutants of Phage P22 Coat Protein*

The amino acid sequence of a polypeptide defines both the folding pathway and the final three-dimensional structure of a protein. Eighteen amino acid substitutions have been identified in bacteriophage P22 coat protein that are defective in folding and cause their folding intermediates to be substrates for GroEL and GroES. These temperature-sensitive folding (tsf) substitutions identify amino acids that are critical for directing the folding of coat protein. Additional amino acid residues that are critical to the folding process of P22 coat protein were identified by isolating second site suppressors of the tsf coat proteins. Suppressor substitutions isolated from the phage carrying the tsf coat protein substitutions included global suppressors, which are substitutions capable of alleviating the folding defects of numerous tsf coat protein mutants. In addition, potential global and site-specific suppressors were isolated, as well as a group of same site amino acid substitutions that had a less severe phenotype than the tsf parent. The global suppressors were located at positions 163, 166, and 170 in the coat protein sequence and were 8–190 amino acid residues away from the tsf parent. Although the folding of coat proteins with tsf amino acid substitutions was improved by the global suppressor substitutions, GroEL remained necessary for folding. Therefore, we believe that the global suppressor sites identify a region that is critical to the folding of coat protein.

The primary amino acid sequence of a polypeptide encodes all the information necessary to direct the folding pathway so that the final three-dimensional structure of a protein is achieved (1). However, the mechanism by which the code directs folding is not well understood. Deciphering the code for protein folding has implications for human health, as there are a number of diseases, such as the amyloid-based diseases, cystic fibrosis, and prion-like diseases, that have been related to the misfolding and aggregation of proteins (2,3). Additionally, recombinant DNA technologies utilized to produce valuable proteins, such as insulin, often have problems with inclusion body formation when those proteins are expressed in heterologous hosts (4 -6). Understanding how the amino acid sequence controls both folding and misfolding will allow us to understand how these diseases and other processes related to misfolding occur.
The complex process of folding has been predominantly investigated in vitro by following the refolding of purified polypeptide chains from the chemically or physically denatured state. Another valuable approach to understanding the nature of protein grammar has been through the use of genetic analysis. Site-directed and randomly induced mutations have revealed residues that are important for different functions, such as stability or activity of the protein (7)(8)(9). Genetic methods have also identified residues that are specific for folding. Residues of this type have been found in various proteins including tailspike protein of bacteriophage P22 (10), bacterial luciferase (11), D-lactate dehydrogenase (12), and interleukin-1 ␤ (13).
To study the grammar encoded within the amino acid sequence, we utilize bacteriophage P22 coat protein as our model system. Eighteen amino acid substitutions at seventeen different sites in P22 coat protein have been identified that render the production of viable phage temperature-sensitive (14,15). In vivo, the temperature-sensitive folding (tsf) 1 amino acid substitutions dramatically reduce the yield of soluble coat protein because the newly synthesized coat polypeptides aggregate to form inclusion bodies prior to reaching the assembly-competent state (14,16). Furthermore, the chaperonins GroEL and GroES (17) have been shown to recognize and rescue a thermolabile folding intermediate (16,18). In vitro, the tsf amino acid substitutions do not destabilize the folded monomeric subunit because differential scanning calorimetry and denaturation by pressure or urea coat proteins with tsf substitutions have shown them to be as stable as WT coat protein (19,20). Recently, Teschke (21) demonstrated that proteins with tsf amino acid substitutions display altered secondary and tertiary structure as well as increased surface hydrophobicity, which may explain the increased propensity of the folding intermediates with substitutions to aggregate, as they do in vivo. Based on these observations, a basic model for the folding of coat proteins is suggested (Fig. 1). We envision the newly synthesized polypeptide forming an intermediate that at permissive temperature proceeds to an assembly-competent coat monomer, which, along with other bacteriophage proteins, proceeds to the mature virion. However, at the nonpermissive temperature, the tsf amino acid substitution creates an aggregationprone intermediate that forms inclusion bodies. The suppressor substitution would prevent the tsf action.
As phage P22 is amenable to genetic manipulation, we have isolated intragenic second site suppressors of the coat protein * This work was supported by National Institutes of Health Grant GM53567 and the Patrick and Catherine Weldon Donaghue Foundation Grant 95-001. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  carrying tsf amino acid substitutions. Traditionally, a search of second site suppressors has been utilized to determine gene product interactions (22)(23)(24)(25)(26). However, in this case, a survey of second site suppressors of coat proteins with tsf amino acid substitutions could be expected to reveal interactions between residues in the affected folding intermediate, and not with proteins that interact with the tsf coat protein because the folding of coat protein into monomeric units normally proceeds without the interaction with other phage proteins (27). If all the tsf substitutions affect the same intermediate in a folding pathway, a set of suppressor amino acid substitutions that stabilize the intermediate might be expected to be repeatedly isolated from several tsf coat protein mutants. Such amino acid substitutions would be known as global suppressors (28). Intragenic global suppressors of folding mutants have been discovered in tsf mutants of P22 tailspike protein (29), a temperature-sensitive mutant of the human receptor-like proteintyrosine phosphatase LAR (30), missense mutants in ␤-lactamase (31), and destabilizing mutants of cytochrome c (32). Alternatively, if the tsf amino acid substitutions do not affect the folding in the same way, it would be unlikely that the same suppressor sites would be isolated for several tsf coat protein mutants; instead, site-specific suppressors of tsf coat protein mutants would be isolated. Such suppressors have been found in tsf mutant phage of P22 tailspike protein (29,33). In the case of P22 tailspike protein, both global and site-specific suppressors have been identified, suggesting that its folding intermediate can be stabilized in a variety of ways.
Here, we have isolated intragenic second site suppressors of coat proteins that have tsf amino acid substitutions. We found three global suppressors in close proximity to one another that are capable of alleviating the folding defects of seven tsf coat proteins that are located 8 -190 residues in the linear amino acid sequence away from its respective suppressor substitution. The majority of suppressor amino acid substitutions are still dependent on GroEL/S action. In addition, we have identified numerous same site and site-specific suppressors, as well as potential global suppressors. The global suppressors identify a region critical to proper folding of coat protein.

EXPERIMENTAL PROCEDURES
Bacteria-The bacteria used for most of the experiments were derivatives of Salmonella typhimurium LT2. The amber suppressor minus host DB7136 (leuA414-am, hisC525-am) and its amber suppressor plus derivative DB7155 (leu A414-am, his C525-am, supE20-gln) have been described previously (34). The strains DB7000 (leuA414-am) and DB7155 were transformed with pOF39, which carries the groEL/S operon behind its own promoter (35). This plasmid conferred ampicillin resistance to the cells. The Escherichia coli strains (36, 37) DW720 (WT groEL and groES) and DW716 (groEL44) were transformed with the plasmid pPR1347 (38), which encodes for the rfb gene cluster and rfc gene so that the E. coli synthesized the O antigen needed for P22. The plasmid conferred kanamycin resistance to the cells.
Bacteriophage-All phage strains used in these experiments carried the c1-7 allele, which prevents lysogeny. The P22 bacteriophage used in this study were either WT in gene 5 or carried the tsf mutations in gene 5 that lead to 18 amino acid substitutions (14) (Fig. 2).
Media-Luria broth was prepared as described by Life Technologies, Inc. and was used to support bacterial growth for plating experiments, preparation of phage stocks, and single step growth curve experiments. M9 medium was 0.6% Na 2 HPO 4 , 0.3% KH 2 PO 4 , 0.05% NaCl, 0.01% NH 4 Cl. M9/Mg 2ϩ medium was used for phage stock preparation and contained M9 with 2 mM MgCl 2 . Dilution fluid was used to dilute the phage stocks and contained 0.1% Tryptone, 0.7% NaCl, and 2 mM MgSO 4 (15) Preparation of Phage Stocks-A single fresh plaque was added to 30 ml of LB with logarithmically growing DB7136 cells and aerated by shaking at 30°C until lysis occurred or a maximum of 6 h. Two to three drops of chloroform were added to ensure complete lysis of infected cells. Samples were chilled on ice for 15 min, and the cell debris was pelleted at 10,000 rpm for 10 min in a SL-50T Sorvall rotor. Phage in supernatant were pelleted by a 90-min centrifugation at 18,000 rpm in the SL-50T rotor. The pellets were drained and resuspended overnight in 2 ml of M9/Mg 2ϩ . A second 10-min centrifugation at 10,000 rpm removed residual debris (15).
Isolation of Revertants of tsf Mutant Coat Proteins-To obtain independent revertants, plaques were isolated at the nonpermissive temperature for each phage carrying coat protein with a tsf amino acid substitution by plating approximately 10 6 phage on each plate. Each plaque is an independent isolation of the tsf phage (33,39). Ten to 20 revertant plaques from each plate for each tsf allele were picked and purified three times by plating a single plaque sequentially. Phage stocks were prepared from one purified plaque at 30°C.
PCR Amplification of Specific Alleles (PASA) -To screen the tsf revertants, PASA analysis (40 -42) was utilized to discriminate between WT revertants and pseudorevertants. Pseudorevertants are mutants that retain the tsf amino acid substitution but have an additional substitution elsewhere. Phage stocks of concentrations 10 10 to 10 11 phage/ml were filter sterilized (Nalgene syringe filter, 0.2 m syringe filter). Each 100-l polymerase chain reaction (PCR) vessel contained 1 l of sterile filtered phage stock (10 7 to 10 8 phage), 2.5 units of Amplitaq DNA polymerase (Perkin-Elmer), 200 M each nucleotide (Perkin-Elmer, Gene Amp dNTPs), 1ϫ PCR Buffer II, 1 mM MgCl 2 , 50 pmol each of reverse primer and specific PASA primer (Table I). PCR vessels were exposed for 4 min at 94°C, 25 cycles of 30 s at 94°C, followed by 30 s at annealing temperature (varying 70 -80°C), extension for 1 min at 72°C, and finally 7 min at 72°C, and then chilled to 4°C in a Perkin-Elmer Gene Amp PCR system 2400. The optimal conditions for the various PASA-specific primers were determined by changing the annealing temperature. A reverse primer was synthesized to be complimentary to bases 6480 -6459 (43), and another primer was synthesized to be complementary to either the WT nucleotide sequence or to the tsf nucleotide sequence (Table I). The nucleotide numbering is based on the P22 DNA sequence in GenBank TM (accession number M59749). For primer descriptions, the first number represents the 5Ј-end of the primer and the last number represents the 3Ј-end of the primer. An agarose gel (1%) was used to determine whether amplified product was present.
PCR Amplification and Purification-PCR was utilized to amplify gene 5, which encodes coat protein, from bacteriophage stocks. One l of filtered phage stock (10 7 -10 8 phage) was amplified with final concentration of 2.5 units of high efficiency polymerase (Pwo DNA polymerase FIG. 1. Model of the folding pathway of P22 coat protein carrying tsf amino acid substitutions. A newly synthesized polypeptide forms an intermediate that at permissive temperature proceeds to an assembly-competent coat monomer that is necessary to form a capsid. At a nonpermissive temperature, the tsf amino acid substitution results in an intermediate that is aggregation-prone and forms inclusion bodies (14,16). A suppressor amino acid substitution prevents the tsf action.
from Roche Molecular Biochemicals); 200 M each ATP, CTP, GTP, and TTP (deoxynucleoside triphosphate set, PCR grade, Roche Molecular Biochemicals); 1ϫ PCR buffer with 0.2 mM MgSO 4 ; and 50 pmol of each forward (14) and reverse primer ( Table I). The PCR vessels were exposed to the same procedure as described above except that the annealing temperature was 55°C. The amplified PCR product was purified using a high pure PCR purification kit (Roche Molecular Biochemicals). An agarose gel (1%) was used to determine quantity and purity of PCR-amplified product.
DNA Sequence Analysis-The P22 packaging genes have been sequenced, including gene 5, which encodes coat protein (43). To determine the sequence of the tsf pseudorevertants, Sequenase, Version 2.0, DNA sequencing kit (Amersham Pharmacia Biotech) was used following the manufacturer's instructions with the following modifications. Nine l of purified PCR product were denatured in the presence of 1 l of sequencing primer (5-50 pmol/l) ( Table I) for 5 min at 100°C. The samples were then cooled on ice for 5 min. Dithiothritol, diluted labeling mix (1:5), T7 Sequenase reaction buffer, [␣-35 S]dATP (NEN Life Science Products), and diluted T7 Sequenase were added to the ice-cold annealed DNA mixture. The reactants were mixed and incubated for 5 min at room temperature. Sequencing reactions were terminated by incubation at 45°C for 5 min with the dideoxy nucleotides. The sequencing primer was selected based on the region of gene 5 to be sequenced (Table I). The reactions were heated to 80°C prior to loading (4 l) on a 6% acrylamide/bisacrylamide gel containing 7 M urea and 1ϫ glycerol-tolerant gel buffer. GTG buffer was prepared at 20ϫ and contained 1.78 M Tris, 0.57 M taurine, and 0.01 M EDTA Na 2 ⅐2H 2 0, pH 8.9. The sequencing gel was dried and exposed to film (Biomax-MR Kodak) and developed the following day.
Efficiency of Plating-Phage were either WT in coat protein, carried coat protein with tsf amino acid substitutions, or coat protein with tsf and suppressor amino acid substitutions (tsf/su). The phage were plated at various temperatures from 16 to 41°C on S. typhimurium DB7136 or DB7155 with and without the plasmid pOF39 or on E. coli DW720 and DW716 strains with the plasmid pPR1347 (38). Plating efficiencies were calculated by dividing the titer of the phage grown at each temperature by the titer produced when phage were grown at 22°C on each bacterial strain (14).
Single-step Growth Curves-Salmonella DB7136 host cells were infected at the nonpermissive temperature with WT phage, phage carrying coat protein with tsf amino acid substitutions, or the double tsf/su amino acid substitutions in coat protein, at a multiplicity of infection of 10. At times after infection, samples were lysed with chloroform and plated at permissive temperature (15). The number of phage produced per cell was calculated.

RESULTS
Previously, a group of amino acid substitutions at multiple sites in P22 coat protein have been characterized that cause a tsf phenotype (14). These amino acid substitutions identify residues important for proper folding of coat protein. Suppressors of 18 tsf coat protein mutants have been isolated in order to determine the positions of additional amino acids crucial to stabilizing the folding intermediate(s) affected by the tsf amino acid substitutions. Here, we report the isolation of global suppressors of folding mutants in phage P22 coat protein.
Isolation of tsf Revertants-Over 260 independent revertants of 18 different phage carrying coat protein with a tsf amino acid substitution (Fig. 2) were isolated on S. typhimurium with and without increased GroEL/S levels at the nonpermissive temperature of the tsf parent phage. The reversion frequencies, which ranged from 10 Ϫ5 to 10 Ϫ7 , were consistent with single nucleotide substitutions (data not shown). To determine whether the revertants were wild-type revertants or pseudorevertants, an analysis known as PASA, described in detail under "Experimental Procedures," was used to discriminate between the two (40 -42). To confirm the results of PASA analysis, gene 5 of P22 was amplified, and the region corresponding to the original tsf nucleotide substitution was sequenced. Those suppressor mutants that retained the original tsf nucleotide substitution were further sequenced to identify the site of the suppressor nucleotide substitution.
Identification of Second Site Suppressors-The location and identity of second site suppressors was determined by DNA sequence analysis. Intragenic suppressors were isolated from phage carrying the tsf coat protein substitutions A108V, D174N, D174G, S223F, S262F, G282D, T294I, D302G, P310A, G403D, F353L, Y411H, and P418S. While second site suppressors were isolated from the majority of these tsf coat proteins, four of the tsf mutant coat proteins only reverted to the WT amino acid. These mutants were W48Q, P238S, V297A, and V300A. The suppressors of the 14 remaining tsf mutant coat proteins were grouped into five categories: same site substitutions, site-specific suppressors, global suppressors, potential global suppressors, and extragenic suppressors. Intracodon nucleotide substitutions resulting in the change of the tsf substitution to another amino acid were found for D174N, D174G, S262F, D302G, and G403D and have been categorized as same site substitutions (Table II). In addition to these same site suppressors, there were a few suppressors that were found with only one tsf coat protein parent and at a low frequency, which we have categorized as site-specific suppressors (Table  III). We also isolated one potential extragenic suppressor of G232D.
The majority of second site suppressors that were not categorized as site-specific or same site substitutions were found in the amino-terminal region of coat protein. The suppressor amino acid substitutions D163G, T166I, and F170L were independently isolated from seven tsf coat protein mutants, A108V, D174N, D174G, S223F, G282D, T294I, and F353L, and ranged from 8 to 190 in the linear amino acid sequence from the original tsf amino acid substitution (Table IV). These types of suppressors are described as a global suppressors as they were able to suppress the phenotype of multiple tsf parents. In addition, a group of potential global suppressors has been isolated. These suppressors have been categorized as such because the number of tsf parents from which the suppressors had been isolated was fewer than that of the global suppressors. A more extensive search may have found these suppressors to be global suppressors (Table V). In addition, two of the site-specific suppressors, D135H and A136D, could be global suppressors because the folding defect of the tsf parents of these particular suppressor substitutions was also alleviated by the global suppressors, and they were found next to one another in sequence (Table III). Gordon and King (15) also performed a limited search for intragenic second site suppressors to confirm that the tsf amino acid substitutions affected the folding process. In that search, a number of true revertants were isolated, as well as S223F/F170L and P310A/A263S (15). In addition, the silent mutations that Gordon and King (14) had identified in various tsf phage strains were also found in our study.
Effect of Suppressor Substitutions on Phage Biogenesis-To determine the limits of the effectiveness of the suppressor amino acid substitutions to mitigate the folding defects, efficiency of plating experiments were performed with tsf and tsf/su phages (Fig. 3). The presence of a suppressor amino acid substitution alleviated the defect due to the tsf amino acid substitution at the temperature used for selection; however, Each of the phage carrying a tsf coat protein was used in the independent isolation of intragenic suppressors. The complete sequence of gene 5 has been determined by Eppler et al. (43). The tsf sites were identified by Gordon and King (14). The amino acid preceding the position number is the WT amino acid. The amino acid after the position number is the tsf amino acid substitution. some were more effective at restoring growth at higher temperatures than others. For instance, of the site-specific and potential global suppressors of the tsf coat protein parent P310A, P310A/A337G increased the minimum restrictive temperature from 33 to 41°C, which is higher than P310A/D85V, P310A/A263S, or P310A/D303G, which could not suppress the folding defect at temperatures greater than 37°C (Fig. 3). In comparison, the global suppressors of S223F and F353L completely restored the phenotype to that of WT phages, which can grow at temperatures up to 41°C. The relative plating efficiencies were also determined for other tsf coat protein parents and their suppressors. Overall, the global suppressors restored the phenotype of the tsf coat protein mutants to that of WT phage (data not shown). Additionally, we discovered that the tsf coat a Number in parentheses is the number of times the same site substitution was independently isolated from the tsf parent phage.
b The nucleotide preceding the position number is the WT nucleotide. The nucleotide after the position number is the intracodon nucleotide substitution.
a Number in parentheses is the number of times the nucleotide substitution was independently isolated from the tsf parent phage.
b The nucleotide preceding the position number is the WT nucleotide. The nucleotide after the position number is the suppressor nucleotide substitution.
a Number in parentheses is the number of times the suppressor substitution was independently isolated from the tsf parent phage.
b The nucleotide preceding the position number is the WT nucleotide. The nucleotide after the position number is the suppressor nucleotide substitution.
a Number in parentheses is the number of times the suppressor substitution was independently isolated from the tsf parent phage.
b The nucleotide preceding the position number is the WT nucleotide. The nucleotide after the position number is the suppressor nucleotide substitution. protein parent, G282D, and the suppressor substitution, N414T, isolated from Y411H and P418S resulted in a coldsensitive phenotype (data not shown). A substitution of serine at position 414 has previously been identified as a cold-sensitive mutant (14,21,44).
Effect of GroEL and GroES on Phage Biogenesis-To understand the effect of the chaperonins GroEL and GroES on the biogenesis phage with tsf and tsf/su coat proteins, a series of plating experiments were performed. The phage were plated on cells that overexpress the chaperonin complex, GroEL/S by approximately 9-fold (18). The minimum restrictive temperature of the phages with tsf and tsf/su coat protein was increased by approximately 2°C, unless the suppressor already completely restored the tsf phenotype to that of WT phage (Fig. 3). This result was consistent with previous experiments (16,18) and suggested that the folding intermediates of coat protein with tsf amino acid substitutions are recognized by GroEL/S.
To determine whether functional GroEL is required for the folding of tsf and tsf/su coat protein mutants, phage carrying these substitutions were plated on a strain of cells that carry a point mutation in GroEL that causes a substitution from glutamic acid to glycine at position 191 (GroEL44) (37). Cells that contain this substitution in GroEL are dysfunctional, because a number of different bacteriophage can not grow on this strain (37). Our plating experiments revealed that the minimum re-strictive growth temperature decreased for both the tsf and tsf/su coat protein mutants when plated on the GroEL44 strain, although the suppressors still maintained a higher minimum restrictive temperature than the tsf parent. This result indicates that functional GroEL was important for proper folding of tsf coat proteins, even with the suppressor amino acid substitution present. Interestingly, P310A/D85V maintained the same minimum growth restrictive temperature on the GroEL44 strain as when plated on the normal cell line. From this experiment, we might conclude that P310A/D85V does not require functional GroEL to fold productively. However, increased levels of GroEL/S increased the minimum restrictive temperature of P310A/D85V. Therefore, for this suppressor functional GroEL improves folding efficiency, perhaps because the suppressor amino acid substitution did not completely correct the folding defect, but GroEL may not be as essential for productive folding at P310A/D85V minimum restrictive temperature. Nevertheless, tsf/su coat protein mutants that we examined still required GroEL for proper folding.
Effect of Second Site Suppressor Mutations in Liquid Culture-To examine the effect of the suppressor amino acid substitutions on the kinetics and yield of phage production, single step growth curves at the restrictive growth temperature of the tsf parent were performed. The single step growth curve shows the genesis of viable phage as a function of time after infection. The phage with the coat protein mutants S223F and P310A were selected for two reasons. First, the suppressors of each mutant represent the main types of suppressors: global, potential global, and site-specific suppressors. Second, the tsf coat protein mutants S223F and P310A are temperature-sensitive at 37 and 33°C, respectively; therefore, the rate-limiting step in phage morphogenesis would not be the production of tailspike protein, as it is at high temperatures (29). These curves, then, indirectly reflect the ability of each coat protein to productively fold and assemble into a mature virion.
The phage with the tsf coat protein mutant P310A is restricted for growth at 33°C, whereas the second site suppressors of the parent mutant increase the minimum restrictive temperature, as seen by efficiency of plating experiments (Fig.  3). The growth curve of P310A and its tsf suppressors at 33°C showed differences in both the rate of production and yield of phage (Fig. 4). The coat protein mutant P310A had a burst of ϳ20 phage/cell, in contrast to the wild-type phage strain, which had a burst of ϳ600 phage/cell at 33°C. The tsf/su phage, P310A/D85V, yielded a burst of ϳ60 phage/cell, P310A/A263S a burst of ϳ160 phage/cell, P310A/D303G a burst of ϳ70 phage/ cell, and P310A/A337G a burst of ϳ800 phage/cell; the range was therefore from 10% to approximately 130% of the WT burst.
The suppressors of S223F alleviated its folding defects as demonstrated by efficiency of plating experiments (Fig. 3). The levels of phage produced by phage with the tsf coat protein mutant S223F were less than 1 phage/cell at 37°C, compared with WT phage, which produced ϳ530 phage/cell (Fig. 4). The suppressors improved the yield of phage produced significantly when compared with the tsf parent. With the global suppressors S223F/D163G, S223F/T166I, and S223F/F170L and potential global suppressor S223F/D135H, the amount of phage produced per cell was greater than levels produced by the sitespecific and potential global suppressors of P310A. The exception was P310A/A337G, which improved the yield and rate of kinetics of phage production to levels greater than that of WT phage. Thus, the suppressors were better able to alleviate the folding defect of S223F than P310A, which might be expected because the folding defect caused by P310A is more severe than that caused by S223F. Of the global suppressors of S223F, S223F/T166I restored the phage growth curve to that of WT phenotype. DISCUSSION Amino acid residues, as defined by both their position and chemistry, have functionally different roles in the structure of a protein, including its activity, stability and folding. While the function of one amino acid does not exclude the possibility of another function for that amino acid, only a portion of the amino acids of a polypeptide are necessary for directing the conformation of the chain through its folding intermediates to its final folded state (10 -13). Isolating mutants that interfere with polypeptide folding is one method of identifying the residues that are critical for defining the conformation of the folding intermediates (45). Another way to identify residues important to the folding process is to isolate suppressors of folding mutants (28 -32, 39). These suppressors may also be useful in analyzing the steps that lead to nonproductive folding pathways, such as inclusion body formation, because these now correct the folding defect of the original mutants hosts (5, 6, 46 -51).
Analysis of the Location of the tsf Mutations-The nucleotide substitutions resulting in a temperature-sensitive phenotype in gene 5 of P22 were isolated by chemical and UV mutagenesis, as revertants of mutant bacteriophage gene product 1, or as revertants of a gene 5 amber mutant (14,22,52,53). Using a variety of methods, the amino acid substitutions in coat protein of P22 were discovered to be important to the folding process (14 -16, 18 -21, 54 -56). Although the tsf amino acid substitutions are found throughout the entire sequence of coat protein, the majority of the tsf substitutions are located in the carboxylterminal portion of the protein. The bias in the location of the tsf amino acid substitutions could be due to incomplete saturation of the coat protein gene during the mutagenesis that produced these phages. Alternatively, the amino terminus of coat protein may not give rise to tsf amino acid substitutions either because the substitutions lead to a lethal phenotype or because they are silent mutations, neither of which would be identified in a screen for conditional-lethal mutants. These possibilities are currently being investigated. 2 At present, the locations of the tsf amino acid substitutions in the three-dimensional structure of coat protein are unknown, as the structure has not been solved. Using the Kyte-Doolittle hydropathy calculation, Gordon and King (15) were unable to discover any pattern in the tsf coat protein amino acid substitutions. However, based on current experiments in our lab, we are able to conclude that the phages carrying coat protein with a tsf amino acid substitution, D174G/N, S223F and G232D are located on the surface of procapsids. 3 To help us analyze the effects of the tsf and tsf/su amino acid substitutions, the Discrimination of Secondary structure Class prediction model of King and Sternberg (51) was utilized to generate a secondary structure prediction of P22 coat protein (Fig. 5). This program was chosen because the fractions of predicted secondary structure of P22 coat protein closely resemble data attained by circular dichroism and Raman spectroscopies (54,57,58). In addition, the proteins that were used to generate the data base of protein structures contain percentages of secondary structure similar to that of P22 coat protein.
Using this model, the tsf amino acid substitutions are predicted to be in helix, sheet, and irregular structures. However, 89% of the tsf amino acid substitutions are predicted to be in either ␤-sheets or irregularly structured regions. The high frequency of tsf amino acid substitutions isolated in these regions may be due to the nature of these structural elements. The ␤-sheets require interactions that are distant in the primary sequence to stabilize the ␤-strands, and irregularly structured regions are necessary for turns in the structure, which allow interactions of other structural elements (59,60).
The Sites of the Global Suppressors-Based on DNA sequence analysis, we identified the nucleotide substitutions of the tsf pseudorevertants. Three amino acid substitutions at positions 163, 166, and 170 of coat protein were able to suppress folding defects resulting from a number of different tsf amino acid substitutions. The repeated isolation of the three global suppressors, located in close proximity to one another, strongly supports the notion that these amino acid substitutions have an important role in coat protein folding.
We believe that the global suppressors are located at or near the surface of the coat protein. This hypothesis is based on a number of observations. First, as discussed above, the amino acid at position 174 has previously been determined to be located at the surface of the coat protein. 3 Second, the suppressor amino acid substitutions at positions 163 and 166 themselves are nonconservative in nature. Nonconservative substitutions are often localized in regions of proteins that can tolerate such substitutions, such as loops and other flexible conformational motifs (61). Whereas the suppressor substitution of leucine from phenylalanine at position 170 is a relatively conservative substitution, substitutions of aromatic residues with leucine have been shown to have a significant impact on the distribution of folding intermediates and the stability of the folding intermediates and native-like structure (62,63). In addition, Van der Schueren et al. (64) recently demonstrated that a single site substitution of leucine to phenylalanine prevented aggregation of the truncated choramphenicol acetyltransferase, thereby improving the folding. Intriguingly, the Discrimination of Secondary structure Class analysis of P22 coat protein placed the global suppressor substitutions in a predicted irregularly structured region (Fig. 5). Based on a model suggested by Tuma et al. (65) in which the coat protein is proposed to consist of two domains joined by a hinge region, and on our biochemical data, one feasible hypothesis is that the suppressor amino acid substitutions are located in or very near to this hinge region. The hinge region has been suggested to be critical to the domain movement of coat protein that occurs during expansion of the capsid lattice. Proper formation of this hinge region may also be important during folding.
Models of Suppression-The global suppressor amino acid substitutions localize to a predicted irregularly structured re-gion (Fig. 5). Because the tsf amino acid substitutions are far apart in sequence, we believe they are unlikely to each interact with the global suppressors in a site-specific fashion. Consequently, it is more probable that the global suppressor amino acid substitutions stabilize an aggregation-prone folding intermediate, change the kinetics of folding or unfolding in a way that compensates for the effect of the tsf amino acid substitutions, retard the rate of the off-pathway aggregation reaction, or destabilize incorrect conformations (48). Conversely, the site-specific suppressor amino acid substitutions may suppress the folding defect through direct side chain interactions.
One conundrum of the global suppressor amino acid substitutions is why these amino acids are not found in the WT coat protein sequence because they appear to improve folding. They may not have evolved for reasons other than folding, such as a change in the stability of coat protein or a decrease in the efficiency of other interactions necessary to permit proper assembly and maturation of the virion. It is also possible that individually the suppressor substitutions may have a deleterious effect on folding. A similar mutational analysis was performed with P22 tailspike protein (29,39,66). The search for suppressors of tailspike tsf mutants realized two global suppressors, V331A and A334V (29,33). The global suppressor amino acid substitutions have been shown to improve folding efficiency in vitro and in vivo. Based on biochemical studies and the solved structure of truncated P22 tailspike protein (67,68), it has been determined that the two suppressor substitutions operate through different modes of action (69). The suppressor, V331A, alleviates the tsf folding defects by stabilizing the completely folded protein, and based on its location in the tailspike structure, the substitution functions by removing a steric hindrance. The suppressor, A334V, alleviates the folding defects by accelerating unfolding at high temperature, which decreases stability of the tailspike protein by improving hydrophobic stacking in an early folding intermediate but destabilizes the native protein. These global suppressor substitutions have not evolved naturally, as they are located in the active site region of tailspike protein, where their respective amino acid substitutions improve folding but interfere with function.
Here we have isolated three global suppressors, D163G, T166I, and F170L, which are capable of alleviating the folding defects of multiple tsf coat protein parents. While the suppressors improve the folding of the tsf parents, the tsf/su mutant coat proteins are still dependent on GroEL/S. Future investi- FIG. 5. Secondary structure prediction of P22 coat protein. Discrimination of protein Secondary structure Class analysis is based on the study of King and Sternberg (51). The protein data set comprises 126 globular proteins that have 32% ␣-helix conformation, 22% ␤-strand conformation, and 47% coil formation, including turns. Discrimination of protein Secondary structure Class analysis evaluates every amino acid residue in a window of 17 to predict secondary structure. The irregularly structured regions are represented by thick solid lines, ␣ helices are represented by a barrel-like structure, and the ␤-strand is represented by an arrow. Placed above the predicted secondary structure are stars, which represent the sites of the suppressor substitutions, and open circles, which identify the sites of tsf amino acid substitutions. Solid lines are drawn between a tsf parent and the global suppressor isolated from it. gations will probe the mechanism by which the global suppressor substitutions alleviate the folding defects of the tsf parents.