Temperature-sensitive mutations in the phage P22 coat protein which interfere with polypeptide chain folding.

Temperature-sensitive mutations in the coat protein of phage P22 severely restrict formation of infectious particles at restrictive temperature. A set of 25 temperature-sensitive strains, which had been localized to regions of the coat gene (Casjens, S., Eppler, K., Sampson, L., Parr, R., and Wyckoff, E. (1991) Genetics 127, 637-647), define 17 sites of single amino acid substitutions by DNA sequencing. Particles assembled from the mutant proteins at permissive temperature were not thermolabile at restrictive temperature, nor defective in the infectious process. At restrictive temperature, ts mutant polypeptide chains were synthesized at near wild-type rates. These inactive chains were not degraded within the cells. The ts chains failed to interact with scaffolding proteins to form the procapsid precursor shell, and they did not polymerize with each to form aberrant shells. Rather, the mutant coat protein accumulated as insoluble aggregates, with the amorphous morphology of inclusion bodies. The results indicate that the chains fail to reach the conformation needed for subunit-subunit or subunit-scaffolding interaction. These mutations appear to be of the class of temperature-sensitive folding mutations, which destabilize an intermediate in the intracellular folding pathway.

The capsids of viruses are constructed as polymers of one or a small number of major capsid proteins (Caspar and Klug, 1962). In the mature viruses, these subunits make intimate contact with their neighbors, forming stable protective coats for their nucleic acids packaged within. In many doublestranded DNA-containing viruses, including the bacteriophages T4, X, P22, T7, and $29 (Hendrix, 1985), herpesviruses (Newcomb and Brown, 1991), and adenoviruses (Honvitz, 1991), the mature capsid is not formed directly from the polymerization of viral coat proteins. The coat subunits first polymerize with scaffolding proteins to form a precursor shell, called the procapsid, which does not contain DNA. Upon removal of the scaffolding subunits, the DNA is driven into the shell, and the procapsid lattice transforms into the mature virion particle.
Coat protein molecules which comprise icosahedral viral shells often assume several different conformations. Crystal * This work was supported by Grant GM17980 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Supported by a predoctoral fellowship from the Howard Hughes Medical Institute.
§To whom correspondence and reprint requests should be addressed 16-535, Massachusetts Institute of Technology, Cambridge, MA 02139. structures of RNA plant viruses (Harrison, 1984) and SV40 (Liddington et al., 1991) demonstrate that coat protein molecules with the same covalent structure have multiple conformations in the shell. For P22, multiple bonding interactions within the lattice of the procapsid precursor shell and a different set of multiple interactions within the lattice of the mature expanded shell have been observed (Prasad et al., 1993). The coat protein of phage P22 is not covalently modified during the folding and assembly process, so that these states are believed to be encoded in the primary amino acid sequence.
The P22 coat protein, encoded by gene 5 of phage P22, contains 430 amino acids and has a predicted molecular weight of 47,000 . In the wild-type P22 assembly pathway (Fig. I), 420 molecules of coat protein, approximately 300 molecules of scaffolding protein, and minor proteins polymerize into a procapsid (King and Casjens, 1974). Upon DNA packaging, the scaffolding molecules exit and the shell expands (Earnshaw et al., 1976). Other minor proteins are added to seal the DNA in the shell (Strauss and King, 1984).
Temperature-sensitive mutations have been isolated in genes of many organisms, including viruses (Edgar and Lielausis, 1964;Schaeffer et d., 1978), bacteria (Hubscher and Kornberg, 1980), yeast (Game, 1976) and Drosophila (Suzuki, 1970). Of those viral coat ts mutations which have been studied in detail, some have been shown to perturb the viral assembly process while others seem to prevent assembly from initiating (Ginsberg, 1979;Behm et al., 1988). This latter class of mutations is not well understood.
A group of well-characterized temperature-sensitive mutations in phage T4 lysozyme have been shown to act by destabilizing the native state of the protein (Hawkes et ul., 1984). In contrast, a group of temperature-sensitive mutants of D-lactate dehydrogenase (Truong et al., 1991), luciferase (Sugihara and Baldwin, 1988), and phage P22 tailspike protein (Goldenberg et al., 1983) act by impairing the folding of polypeptide chains carrying these mutations at restrictive temperature, but not the stability of the native folded molecule at restrictive temperature. These mutants, in which folding intermediates are destabilized, have therefore been termed temperature-sensitive folding mutants (Sturtevant et al., 1989;King et al., 1990).
We have been particularly interested in the relation of the folding of coat protein molecules to their assembly into structures in which they assume multiple conformations. For a number of multimeric proteins whose folding and subunit assembly pathways have been well characterized (Goldenberg and King, 1982;Jaenicke, 1987), the monomers are not completely folded and are not in the conformation found in the final multimer.
We describe here the nature of the intracellular defect associated with a set of temperature-sensitive mutations of the P22 coat protein. The results indicate that the ts mutations prevent the polypeptide from assembling into the procapsid at restrictive temperature. The assembly-competent conformation of coat protein is not stably formed in these infections. The coat protein molecules accumulate in the misfolded inclusion body state, which is consistent with a defect in protein folding (Mitraki and King, 1989).
Bacteriophage-The strains carrying mutations in the coat protein gene are described in Table I. Most of these temperature-sensitive mutations were isolated by mutagenizing phage and screening phage for temperature-sensitive growth. A second group was isolated by screening apparent revertants of cold-sensitive mutations in gene 1 for temperature sensitivity (Jarvik and Botstein, 1975). A third group was isolated by screening apparent revertants of a gene 1 amber mutation growing on cells which inserted an amino acid not yielding functional gene 1 protein for temperature sensitivity (Jarvik and Botstein, 1975). Mutations were mapped to the coat protein gene (gene 5) by complementation and recombination. Other mutations carried by phage used in these experiments are cl-7, which prevents lysogeny, gene 13 amHlO1, which delays lysis, and gene 3 amN6, which prevents DNA packaging (King et al., 1973).
Phage Crosses-Exponentially growing DB7155 cells in broth at a concentration of 2 X 108/ml were infected with phage strains carrying desired alleles, each at a multiplicity of infection of 5. Progeny phage were screened for desired alleles. All mutants discussed in this work were crossed into a cl-7 background.
DNA Sequencing-Eppler et al. (1991) have sequenced the P22 DNA packing genes, which include gene 5 (encoding the coat protein) and surrounding DNA. Following their notation, gene 5 extends from nucleotide 4804 to 6093. Casjens et al. (1991) have mapped most of the coat protein mutations discussed here by deletion mapping to intervals of gene 5, which they call EE, FF, GG, HH, 11, and JJ. We exploited this information in the design of oligonucleotide primers and choice of gene 5 regions to be sequenced. In short, symmetric PCR' was performed on phage DNA, generating double-stranded DNA spanning regions of interest. An asymmetric PCR reaction The abbreviation used is: PCR, polymerase chain reaction.  *Where more than one amino acid substitution is present, the one marked with an asterisk is believed to be responsible for the ts phenotype.
Wild-type amino acid which is substituted is shown in bold.
A.  (plus wild-type phage) were sedimented in a Microfuge and washed. Portions of the resuspended pellet and supernatant for both the high temperature and low temperature labeling were electrophoresed through SDS gels. The amount of labeled coat protein appearing in gel lanes was quantitated with a PhosphorImager. A, percent of coat protein pelleted for each phage lysate labeled at low temperature. B, percent of coat protein pelleted for each phage lysate labeled at high temperature.
utilizing one primer was then performed to generate single-stranded DNA for sequencing.
In more detail, 10O-pl PCR reaction vessels for symmetric PCR contained 5 pl of phage stock, 10 pl of 10 X PCR buffer (500 mM KC1, 100 mM Tris-HC1, pH 8.3, 15 mM MgZ'ClZ), 0.2 mM concentration each of ATP, CTP, GTP, TTP (Perkin-Elmer Cetus Instruments), 0.5 p1 of Amplitaq polymerase (Perkin-Elmer), and 100 pmol of each primer (all oligonucleotide primers were obtained from the MIT macromolecular synthesis facility). Reaction mixtures were typically exposed to 2 min at 94 "C, 30 cycles of 1 min at 94 "C, 2 min at 60 "C, 2 min at 72 "C, and 4 min at 72 "C in a Perkin-Elmer Cetus DNA Thermal Cycler 480. One primer was synthesized to be complementary to nucleotides 6193-6169 (primer 1): the other primer used was the same as nucleotides 5199-5223 (primer 2) or 4675-4699 (primer 3). For primer descriptions, the first number represents the 5' end of the primer; the last number represents the 3' end of the primer.
Typically, 15 p1 of these reactions (now containing several pg/pl DNA of expected size) were electrophoresed through a 1% agarose gel. The band containing the amplified DNA was cut from the gel, precipitated with Geneclean (BiolOl), and resuspended in 10 pl of 10 mM Tris, 0.1 mM EDTA, pH 8.0 (TE buffer). A PCR reaction designed to generate single-stranded DNA was then performed using this double-stranded DNA and one of the aforementioned three primers under conditions as above. This reaction mixture was precipitated with Geneclean and resuspended in 15 pl of TE buffer. Typically, 7 pl of this DNA were sequenced with Sequenase (U. S. Biochemical Corp.) using 5-10 ng of primer, with choice of primers described next.
Primer 3 was used for sequencing regions EE and FF (singlestranded DNA generated with primer 1). Primer 4 (corresponding to nucleotides 5000-5016) was used to sequence region GG (singlestranded DNA generated with primer 1). Primer 2 was used to sequence HH (single-stranded DNA generated with primer 1). Mutations mapped to region I1 were sequenced both with primer 5 (corresponding to nucleotides 5460-5476), with the single-stranded DNA having been generated by primer 1, and primer 6 (complementary to nucleotides 5908-5892), with the single-stranded DNA having been generated by primer 2 or 3. Primer 1 was used to sequence regions JJ and KK (single-stranded DNA generated with primer 2 or 3 ) .
Gel purifying the double-stranded DNA generated in the first reaction was essential for obtaining consistently readable sequencing gels.
Puke-Chase Experiments-A 1:lOO dilution of an overnight culture of DB 7136 cells was grown at 30 "C in minimal media to a concentration of about 10M/ml. Cells were pelleted, resuspended at a concentration of 4 X 10M/ml, and put on ice. 0.2 ml of phage at a titer of 2.8 X 109/ml was placed at 39 "C for 90 min, at which time 0.2 ml of cells were added. 12 min after infection, a portion of the culture was placed at 28 "C. The low temperature portion of the infection was labeled at 60 min with 2 pCi/ml "C-amino acids (Du Pont-New England Nuclear), chased with casamino acids at a final concentration of 2% at 64 min, and iced at 94 min. The high temperature portion of the infection was labeled with 2 pCi/ml of 14C-amino acids at 45 min after infection, chased with casamino acids at 49 min after infection, and put on ice 74 min after infection.
PelletlSupernatant Separation-Samples labeled in the above pulse-chase experiment were frozen at -20 "C, thawed, then frozen again in dry ice/ethanol, and thawed. 5O-pl samples were centrifuged for 3 min in a Microfuge. The supernatants were removed and the pellets were washed in 50 pl of M9/Mg2+ and recentrifuged. These supernatants were removed and combined with the corresponding supernatants from the first centrifugation. The pellet was resuspended in 100 ~1 of M9/M$+. Samples were mixed 1:2 with 3 X SDS sample buffer and electrophoresed through SDS gels. Bands were quantified by exposing gels to PhosphorImager screens and using the ImageQuant software (Molecular Dynamics).
Sucrose Gradients-5 ml of exponentially growing cells in minimal media at a concentration of 4 X 108/ml were added to 5 ml of phage at 39 "C, at an multiplicity of infection of 7. This infected high temperature culture was labeled 45 min after infection with "Camino acids at 2 pCi/ml and chased 49 min after infection with casamino acids at a concentration of 2%, immediately after which half the culture was shifted to 28 "C. Cultures were put on ice 79 min after infection. Cells were concentrated OX, frozen in dry ice/ethanol, thawed, and then stored at -20 "C. After thawing at room temperature, lysate samples were layered onto 5 to 20% (w/v) linear gradients of sucrose in M9/MgZ', on top of 60% sucrose shelves. Gradients were centrifuged for 30 min in an SW50.1 rotor at 4 "C at 30,000 rpm. Approximately 18 fractions per tube were collected, which were subjected to scintillation counting and SDS-gel electrophoresis.
Thin Section Electron Microscopy-10 mI of exponentially growing DB7136 cells in LB at a concentration of 4 X 10M/ml were added to 10 ml of phage, at a mutiplicity of infection of 7, at either 38 "C or 28 "C. The 28 "C infections were placed on ice 2 h after infection; the 38 "C infections were placed on ice 90 min after infection. Fixation, embedding, and sectioning of phage-infected cells was performed essentially as described in Lenk et al. (1975). Electron microscopy was performed with a JEOL 1200 operating at 80 kV.

RESULTS
Twenty-five temperature-sensitive phage strains carrying mutations in the coat protein gene were analyzed. These strains were able to form plaques at 30 "C and were restricted for growth at 39 "C. Twenty-two strains gave very low titers at 39 "C, in the range of reversion frequencies seen for single point mutations ( Table 11). Three of the strains (tsRU221, tsRH58H, and tsl0) formed tiny plaques at 39 "C. The frequency of large plaques formed at 39 "C by tsRU221, ts-RH58H, and tsl0 was also in the range expected for reversion of a single point mutation.
Sites and Amino Acid Substitutions of the Temperaturesensitive Coat Protein Mutations-These mutations had been mapped to regions of the coat protein gene by Casjens and co-workers . To determine the nucleotide changes in the coat protein gene which caused the temperature Radioactivity is plotted as a function of sedimentation through the gradient. The bottom of the gradient is at figure left, corresponding to fraction 1. sensitivity, we sequenced these regions of the coat protein gene, or in some cases the whole gene, using a PCR-based protocol described under "Experimental Procedures." Wildtype DNA was sequenced in parallel. For one strain of the set, tsI0, we saw no nucleotide changes by DNA sequencing in the region of the coat protein gene to which the mutation had been mapped . Sequencing the whole coat protein gene revealed a single nucleotide change outside this region, whose deduced amino acid alteration we infer is responsible for the temperature sensitivity. Twenty-one of the strains had single nucleotide replacements corresponding to single amino acid substitutions; three strains had two nucleotide replacements corresponding to double amino acid substitutions; one strain had a triple nucleotide replacement corresponding to a triple amino acid substitution.
Eighteen different patterns of substitutions were found in the 25 strains sequenced. Five nucleotide substitutions were found in more than one (Table 111). Three nucleotide substitutions were duplicated in strains with similar names (tsNl3 and ts13.1, tsN26 and ts26.1, ts5 and ts5.1) and probably represent stocks derived originally from the same mutant whose names have been altered. The two strains whose deduced amino acid substitution is Asp'74 + Asn, three of the strains whose substitution is SePZ3 + Phe, and the two strains whose substitution is Pro418 -Ser are probably independent isolates.
The reversion frequency data suggested that the temperature sensitivity resulted from single nucleotide substitutions. In the four cases where multiple substitutions were seen in regions to which mutations had been mapped, we attempted to determine the substitution primarily responsible for the temperature-sensitive phenotype. Apparent revertant plaques of tsN53, tsRH137B, tsRH58E, and tsRH137D were picked from restrictive plates and stocks prepared under permissive conditions. These phage did not exhibit temperature sensitivity a t 39 "C. We sequenced the region of DNA to which these mutations had been mapped for several apparent revertants of each of these four mutant strains. In each case, we found at least one revertant in which one of the original substitutions had reverted to wild-type. These substitutions, which are marked with an asterisk in Table  111, are presumably those responsible for the ts phenotype. The other secondary substitutions appear silent under the conditions tested. For simplicity in the subsequent discussion, we refer to these multiple change mutants by their primary substitution only.
These coat protein ts mutations, which have been gathered from a variety of screens and selections (Table I), are diverse and distributed throughout the coat protein gene, although they appear concentrated in the C-terminal half of the protein. In total, the 25 mutants sequenced have deduced amino acid changes at 17 different sites in the coat protein gene (counting only the primary substitution in the four cases with multiple substitutions). Both hydrophilic and hydrophobic sites are represented. Six of the 18 different amino acid substitutions seen (not counting secondary changes) represent changes in charge + Asn, Asp'74 + Gly, GlyZ3' -Asp, G1yZ8' + Asp, Asp3'' "-* Gly, and Gly403 + Asp).
The Nature of the Temperature-sensitive Defect-That these coat protein mutants propagate poorly at high temperature indicates that a function of the coat protein important for phage growth is impaired at restrictive temperature. Major stages of coat protein function which have been identified are folding, procapsid assembly, DNA packaging, transformation into stable phage particles, and infectivity. We have at-  Fig. 4 were electrophoresed through 10% SDS gels, without preboiling, and were subject to autoradiography. A , wild-type coat protein at 37 "C. B, V300A at 28 "C after late shift-down. C, V300A at 37 "C. The vertical arrow marks the position of procapsids, about s~~. " , .
tempted to determine the principal in uiuo step in P22 growth which has been impaired by each mutation. For the five cases in which the same substitutions were found in more than one mutant, one was chosen for analysis.
The product of phage gene 3 is necessary for DNA packaging into the procapsid particle ( Fig. 1). Sixteen of the eighteen ts mutants carrying different nucleotide substitutions were crossed into a gene 3 amber background in order to locate the stage of impaired coat protein function with respect to the DNA packaging event. Functional procapsid particles containing the coat protein accumulate in infections with phage carrying the gene 3 amber allele (King et al., 1973). A gene 13 amber allele, which delays premature cell lysis, was also introduced into these phage strains by genetic crossing.
Are Phage Particles Formed a t Low Temperature Functional a t High Temperature?-It was possible that input phage particles themselves were themolabile. Heating phage stocks a t 39 "C for 90 min, followed by titering a t 30 "C, did not result in a reduction in the number of plaques formed (not shown). This result indicated that the phage particles were not thermolabile a t restrictive temperature.
T o measure protein synthesis, cells were infected with ts mutants of interest and exposed to radioactive amino acids to follow newly synthesized proteins. The phage were preincubated a t 39 "C, and the infection was performed a t 39 "C. Twelve min after infection, half the culture was shifted to 28 "C. The high and low temperature cultures were pulsed with I4C-amino acids, chased, and incubated further for 25 and 30 min, respectively. Samples of the lysates were electrophoresed through SDS gels (Fig. 2).
It was possible that the ability of some ts mutants to infect cells was impaired a t restrictive temperature. Such mutants would be defective in their ability to induce late phage protein synthesis in this protocol, since we had preheated the input phage and performed the infection at restrictive temperature. The pattern of protein synthesis in uninfected cells is shown in Fig. 2, lane 19. The expected pattern of phage-specific late protein synthesis is clearly evident in all of the mutantinfected cells (King et al., 1973). This result indicates that phage particles containing mutant coat protein were not defective in stability or infectivity a t restrictive temperature.
Is Mutant Coat Protein Present at High Levels within Cells?-As can also be seen in Fig. 2B, the levels of the ts mutant coat proteins at restrictive temperature are for the most part similar in infections with the temperature-sensitive mutants and phage-containing wild-type coat protein. This suggests that impaired synthesis of the mutant coat protein is not the source of the temperature sensitivity. In addition, it appears that newly synthesized ts coat protein was not significantly degraded during the chase period.
Does the Mutant Coat Protein Assemble into Procapsids a t Restrictive Temperature?-Translation of scaffolding protein is regulated in phage-infected cells, such that the rate of scaffolding synthesis is inversely proportional to levels of soluble scaffolding protein . In gene 3 amber infections, scaffolding protein is sequestered in procapsid particles and does not participate in repression of further synthesis . Therefore, a high rate of scaffolding synthesis in coat protein tslgene 3 amber infections would suggest that the coat protein can still associate with scaffolding protein in procapsids. A low rate of scaffolding synthesis in these mutant infections would suggest that scaffolding protein cannot stably associate with coat protein.
In comparison to rates of scaffolding synthesis seen in the infection with wild-type coat protein a t 39 "C (Fig. 2B, lane  1 ), scaffolding synthesis is extremely low in infections with the temperature-sensitive coat protein mutants in the gene 3 amber background (Fig. 2B, lanes 3-18). The rate of scaffolding synthesis is comparable to that seen in the coat protein umber infection (Fig. 2B, lane 20), which cannot form procapsids. A possible explanation, which we confirm below, is that the levels of procapsid formation are reduced and that the mutant coat protein is not interacting with scaffolding protein. Fig. 2A shows that for some mutants levels of scaffolding protein synthesis are also reduced a t 28 "C, suggesting that the mutations are impairing procapsid production even at the lower temperature.
T o determine whether the mutant coat protein was in an essentially soluble or insoluble state, we performed a pellet/ supernatant separation on lysates prepared with the 18 different mutants at a high and low temperature. In this procedure, procapsids and phage particles remained in the super-natant. The supernatant and resuspended pellets were electrophoresed through SDS gels. The distribution of coat protein between pellet and supernatant at 28 "C is shown in Fig. 3A and at 39 "C is shown in Fig. 3B.
The wild-type coat protein, when assembled into procapsids in the 3am/13am background, or into phage particles in the 13am background, was not significantly pelleted at both temperatures examined. Coat protein amber fragments, which are unable to form native subunits or shell structures were recovered in the pellet, representing an aggregated state. At restrictive temperature, the coat protein from all of the ts mutant infected cells was found primarily in the low speed pellet (Fig. 3 B ) . At low temperature, some of the mutant proteins were also recovered in the supernatant, presumably assembled into procapsids. The remaining mutants showed a significant decrease in soluble protein even at permissive temperature (Fig. 3A).
The lysate made with the N-terminal scaffolding amber mutation accumulates wild-type coat protein. Some of these molecules polymerize into aberrant shell-like species (Earnshaw and King, 1978). The percentage of ts coat protein which was pelleted was greater than the percentage of wild-type coat Wild-type coat protein pelleted in the scaffolding amber background. This suggests that the defect exhibited by the temperature-sensitive coat protein mutants is not simply loss of the ability to interact with scaffolding protein. It appears that at restrictive temperatures the coat protein mutations give rise to conformations leading to insoluble aggregates (as seen by the low speed pelleting) in place of those necessary for coat/scaffolding or coat/coat interaction. T o analyze levels of procapsid formation with increased precision, we fractionated lysates of phage-infected cells on 5-20% sucrose gradients. Lysates of eight of the ts mutants were analyzed, all in a gene 3 amber background. Cells infected at high temperature were exposed to a pulse of radioactivity, after which half of the culture was shifted to lower temperature (late shift-down). Lysates were centrifuged through sucrose gradients such that 240 S procapsids should run near the center of the gradient. The gradients were fractionated and radioactivity per fraction was quantified by scintillation counting. Results are presented in Fig. 4.
As expected, the wild-type coat protein control yields a peak of procapsids both at high temperature and at low temperature after shift-down (Fig. 4A). The coat protein amber mutant infection yielded no detectable procapsid peak. The levels of procapsids in the ts mutant infections at restrictive temperature were very low for all 8 mutants studied. These polypeptide chains are apparently unable to assemble into stable procapsids.
Procapsid yields for the ts mutants after the late shift-down increased significantly for six of the mutants. VaP" "-f Ala and Tyr411 + His showed only slight increases. In this sucrose gradient experiment, procapsid yields after the late shift-down reflect both the degree of transient reversibility of newly synthesized coat protein chains and the efficiency of procapsid formation at low temperature.
The gradient fractions were electrophoresed through SDS gels to examine protein distribution. All of the sucrose gradient procapsid peaks shown displayed both coat and scaffolding proteins. In the restrictive temperature lysates, the ts coat protein was recovered primarily at the bottom of the gradients (Fig. 5).

What Sort of Aggregates Are Formed by the Mutant Coat
Proteinz-The pellet/supernatant separation and sucrose gradient experiments established that mutant coat protein was forming large aggregates. Thin sections of infected cells were examined by electron microscopy to determine the form of these aggregates in situ. Six of the coat protein mutants were examined. Sample micrographs in Fig. 6 are of the infection with wild-type coat protein, the scaffolding amber mutant, and the coat ts mutants Asp'74 + A s n and Th?g4 -+ Ile, all in the gene 3 amber background, at a low and high temperature. Procapsid particles are evident in the wild-type infections (Fig. 6 A ) . In the scaffolding amber mutant infections, which are defective in assembly, we saw predominantly aberrant particles. These are spirals and shells of various sizes which appear deformed and lack the inner scaffolding core (Fig. 6 B 1. In contrast, amorphous inclusion bodies are present in the ts mutant infections at 38 "C (Fig. 6, C and D). This suggests that the ts amino acid substitutions interfere with the ability of the coat protein to fold into a stable conformation, rather than causing incorrect polymerization.
Even at 28 "C some inclusion bodies are present in the ts mutants, as well as procapsids. In these electron microscopy experiments, the 28 "C infections were maintained a t 28 "C throughout the experiment; there was no shift-down. This establishes that these coat protein ts mutants also exhibit some defect in coat protein maturation at low temperature. Results from this experiment are tabulated in Table IV.
Formation of Inclusion Bodies-To determine if ts mutant chains forming inclusion bodies were capable of re-entering the productive pathway at permissive temperature, a temperature shift-down experiment was performed. The mutant Phe353 + Leu, which formed inclusion bodies at restrictive temperature but gave a large yield of procapsids at permissive temperature, was used for this experiment. T o examine the i n vivo reversibility of inclusion body formation, we infected cells with Phe353 + Leu and labeled a t 39 "C, and then transferred samples as a function of time to 28 "C. Samples were lysed by freezing and thawing and fractionated into low speed pellets and supernatants. As shown in Fig. 7 A , labeled coat protein which remained at 39 "C for 30 min before being transferred to 28 "C was largely pelleted, indicating that inclusion body formation was irreversible. Newly synthesized coat protein appeared increasingly in the supernatant as the time of incubation at 39 "C was reduced, presumably assembled into procapsids. This suggests that some early step on the pathway leading to inclusion body formation was reversible or that at early times after labeling the newly synthesized coat protein had not yet attained the conformation leading to inclusion body formation.
The inclusion bodies formed at high temperature could be forming as a result of destabilization of procapsid particles. TO examine the possibility that the inclusion bodies were derived from procapsids dissociating at restrictive temperature, we infected cells with Phe353 + Leu and labeled at 28 "C and then transferred samples as a function of time to 39 "C. Samples were lysed and fractionated into low speed pellets and supernatants.
As shown in Fig. 7 B , coat protein in samples incubated at low temperature, and then shifted up to 39 "C at 12 or more minutes after labeling a t 28 "C, remained largely in the supernatant, suggesting that the ts mutant procapsids were not unstable. While it is possible that procapsids formed at low temperature could have fallen apart at high temperature without irreversible coat protein aggregation, this result indicates that the inclusion bodies do not arise from this process. Coat protein derived from samples transferred to 39 "C at 4 min after labeling was largely pelleted, suggesting that some FIG. 7. Reversibility of mutant phenotype. Exponentially growing cells were infected with F353L, in a 3aml 13am/cI-7 background. After labeling and chasing at indicated times and temperatures, infected cells were lysed by freezing and thawing. A pellet/supernatant separation was then performed. Samples were electrophoresed through 7.5% SDS gels and exposed to films, which are shown. A, infection, labeling, and chase were at 39 'C. At times indicated, portions of the samples were shifted to 28 "C and incubated further for 30 min. Lane labeled none was put on ice 33 min after labeling at 39 "C and was never shifted down. B, infection labeling and chase were at 28 "C. A t times indicated, portions of the samples were shifted to 39 "C and incubated further for 30 min. Lane labeled none was put on ice 33 min after labeling a t 28 "C and was never shifted up. ' 4' 8' 12' 16' 20' 24' 28' 32' 33' 8/12' 16'20'24'28'32' $ 4' 8' 12' 16'20'24'28'32' 3' 4' 8' 12' 16' 20' 24' 28' 32'33' Tailspike timer -_ _ _ l _ Non-natlve tallspike +

Coat protein -+
Time of shift-up early step on the procapsid pathway was reversible or that the coat protein had not yet passed through the conformation leading to inclusion body formation. Coat protein which was labeled after shift-up and maintained a t high temperature aggregated, while coat protein which was labeled after shift-down and maintained at low temperature remained in the supernatant (not shown). In these temperature shift experiments, the degree to which the coat protein was pelleted depended on the time of incubation a t high temperature, but not the temperature at which the labeled protein was synthesized or the prior temperature of incubation. These results establish that it is the temperature a t which coat folding and assembly occur which is of primary importance in determining the phenotype observed.
Boiling any of the labeled samples shown in Fig. 2 before loading them on gels did not change the amount or mobility 4' 8' 12'16'20'24'28%?' 2 4' 8' 12'16'2d 24'28'32' su pe r n atont of coat protein detected, indicating that these inclusion bodies could be solubilized in 2% SDS without boiling (not shown). The P22 coat protein contains 1 cysteine residue. T o determine whether or not the inclusion body aggregates contained intermolecular disulfide bonds, samples treated with or without the reducing agent P-mercaptoethanol were electrophoresed through SDS gels. Dimeric coat protein molecules or other aggregated species were not detected in the nonreducing gels (not shown).

DISCUSSION
The coat polypeptide chain passes through a number of conformations en route to the mature virion. These include the nascent polypeptide chain, the assembly-competent monomeric conformation, the subunit polymerized with scaffold-ing subunits in the procapsid lattice, and the subunit forming the mature lattice in the capsid (King et al., 1973;Prevelige et al., 1988;Prasad et al., 1993). The temperature-sensitive mutations we have studied in this work form inclusion bodies a t restrictive temperature which are derived from a conformation of the coat protein attained prior to shell formation. At restrictive temperature, these mutant coat protein chains do not fold into stable assembly-competent coat protein monomers.
The virus particles formed at low temperature are composed of coat proteins carrying the ts mutant substitutions. Such phage were not thermolabile or defective in the infection process at restrictive temperature. The source of the temperature-sensitive defect is in a step leading to virion formation, not disruption of the function of completed virion particles.
If the temperature sensitivity resulted from a defect in scaffolding release, DNA packaging, or the shell transformation reactions, we would expect procapsids to accumulate in cells blocked in DNA packaging. For the 16 mutants crossed into the gene 3 amber background, procapsids did not accumulate a t restrictive temperature. This indicated that the mutant defects were acting before the DNA packaging event and were preventing the formation of stable procapsids.
As seen in the pulse-chase experiments, at restrictive temperature, the ts coat protein molecules were generally synthesized at a high rate and not substantially degraded. The thin section electron microscopy experiments, in which low levels of procapsids were seen, followed the fates of total accumulated protein, not just labeled protein. These microscopy experiments indicated that the aggregation phenomena seen by the pulse-chase experiments were not restricted to chains labeled during these times.
The thin sections of infected cells revealed that these aggregates were not structured polymers of coat subunits, but amorphous inclusion bodies, for the 6 mutants examined. The formation of inclusion bodies generally reflects a defect in protein folding (Mitraki and King, 1989). In cells infected with phage carrying a wild-type coat protein gene and an amber mutant in the scaffolding gene, at late times coat subunits are recovered as aberrant spiral structures, as well as variously sized empty shells ( Fig. 6; King et al., 1973;Earnshaw and King, 1978). The assembly of coat protein into these aberrant particles is slow (Casjens and King, 1974), which indicates that unassembled wild-type coat protein is stable. If the mutants were specifically defective in their ability to be recognized by scaffolding protein, or if the mutant coat protein chains could fold into a stable assembly-competent monomeric conformation but were defective in some aspect of the assembly process, we would have expected aberrant shell structures or monomeric coat protein to accumulate.
More detailed experiments with Phe353 + Leu showed explicitly that procapsids formed a t low temperature did not form inclusion bodies when shifted to high temperature. The Phe353 -+ Leu inclusion bodies must have been derived from a conformation of coat protein present before or up to that necessary for scaffolding interaction. Once the inclusion body had formed at high temperature, the chains could not be recovered by shift-down, indicating that the inclusion bodies represented a kinetically trapped form of the mutant polypeptide chain. For several minutes after chain synthesis at high temperature or low temperature, these mutant chains retained the capacity to form either procapsids or aggregates, depending upon the subsequent temperature of incubation.
Our data indicate that the species destabilized by these mutants is a coat protein conformation occurring up to the assembly-competent monomeric conformation. Salmonella host cells which overexpress the GroEL/S chaperonin proteins substantially alleviate the coat protein temperaturesensitive defects.' GroEL has been shown to interact with folding intermediates, thereby preventing their aggregation, but not with native proteins (Viitanen et al., 1992;Zahn and Pluckthun, 1992;van der Vies et al., 1992;Martin et al., 1991). This suggests that the coat protein species destabilized by the ts mutations is a partially folded intermediate in the folding pathway. The temperature-sensitive folding phenotype has been documented for the P22 tailspike protein, for which ts mutations act by destablizing an intermediate in the folding pathway (Goldenberg et al., 1983;Haase-Pettingell and King, 1988;King et al., 1990).
These coat protein ts mutants appear to exhibit varying degrees of a similar phenotype. While they all lead to aggregation, differences in the extent of this aggregation at 28 "C were seen. The various amino acid substitutions may destabilize different conformations of coat protein or the same conformation but to different extents. Even though the original screens used to generate these mutants were for temperature sensitivity, many of the mutants also exhibited defects in growth at the permissive temperature. It seems that the destabilized conformation of coat protein is also important for folding at the permissive temperature.
A set of temperature-sensitive mutations has been studied in T4 lysozyme. In this case, the mutations occurred preferentially in solvent-inaccessible sites as defined by the crystallographic structure (Alber et al., 1987). Some of these mutations have been shown to destabilize the native structure (Hawkes et al., 1984). The temperature-sensitive folding mutations in the P22 tailspike are believed to occur preferentially at surface sites in the native protein (Yu and King, 1988 Villafane and. As of yet, a high resolution structure of phage P22 or the coat protein alone, in which the location in the folded protein of residues involved in causing these ts mutations could be visualized, is unavailable.
It is clear that the location of an amino acid in the coat protein molecule, not just the amino acid substitution itself, is important for determining whether or not a substitution leads to temperature sensitivity. For example, the substitution Ala" -+ Val is temperature-sensitive, as are the substitutions Valzg7 -+ Ala and V a P + Ala. The substitutions Asp174 + Gly and Asp3' ' -+ Gly are temperature-sensitive, as are the substitutions GlyZ3' -+ Asp, G1y282 + Asp, and Gly403 -+ Asp. Thr258 -+ Ile appears to be a silent mutation even though ThrZg4 -Ile is temperature-sensitive.
The 25 mutant strains examined here were derived from several independent screens. The observation that this diverse set of amino acid substitutions all interfere with chain folding indicates that this is the major mechanism of temperature sensitivity for the coat protein. We believe this reflects the lability of the putative folding intermediate, in comparison with the stability of its own native polymerized state.
Three of the ts mutations analyzed here (tsRH137D, ts-RH137C, and tsRH137D) were isolated originally as second site suppressors of cold sensitive mutations in gene I , which encodes the portal protein (Jarvik and Botstein, 1975). These mutations cluster around amino acid 300 and may identify a site in the functional coat protein which interacts with the portal protein during assembly.
As can be seen in Fig. 2, the ts coat protein mutations perturb the accumulation of some other phage proteins. First, as discussed above, the scaffolding synthesis is reduced in these infections at restrictive temperature. Second, the minor proteins gp16 (Fig. 3) and gp20 (not shown) are barely de-' C. L. Gordon and J. King, manuscript in preparation. tectable in the phage mutant lysates at restrictive temperature, a result not seen in the infection with phage carrying the coat protein amber mutation. Third, the folding of the P22 tailspike protein into trimers is impaired in the coat protein mutant infections. Native P22 tailspike protein forms a trimer resistant to SDS denaturation (without heating), while partially folded tailspike or tailspike aggregated into inclusion bodies migrates as monomers on SDS gels (Haase-Pettingell and . As is shown in Fig. 3, the fraction of tailspike protein which forms SDS-resistant trimers is reduced by the presence of the coat protein mutations. In general, temperature-sensitive mutants which have been isolated in coat proteins of other viruses have not resulted in the formation of thermolabile virions. Some temperaturesensitive and absolute lethal mutants isolated in other viral coat proteins are defective in the assembly process, while other mutations prevent assembly from initiating. Two temperature-sensitive mutants in the T4 coat protein gene 23 and seventeen temperature-sensitive mutants in the T4 coat protein gene 24 were isolated by Edgar and Lielausis (1964)  . For three of these mutants, the stability of phage particles produced a t permissive temperature was examined, and they were found to show no substantial difference in their rate of heat inactivation as compared to wildtype phage. These authors then concluded that the temperature-sensitive step was acting before completion of the mature particle.
Temperature-sensitive mutants in the major coat protein of SV40 have been isolated and described (Behm et al., 1988). At nonpermissive temperature, some mutants form partially assembled virion particles, while others are deficient in the initiation of particle assembly. These defects were interpreted by assuming that each amino acid substitution caused a local perturbation in the native protein structure (Behm et al., 1988). Ginsberg (1979) and co-workers have studied temperaturesensitive mutants in the adenovirus hexon protein, which is the major coat protein of the virus. Wild-type hexons form stable hexameric structures. One set of mutants yielded hexon protein which failed to assemble into hexamers. A second set could assemble into hexamers but was deficient in transport into the nucleus. A third set formed hexamers which were transported into the nucleus but which were deficient in their ability to assemble into capsids (Ginsberg, 1979).
Katsura has reported the isolation of a set of missense mutants in phage X gene E, which is the major coat protein of X (Katsura, 1980). The mutants he isolated were absolute lethals which were propagated as prophage. Of 71 such mutants studied, Katsura reported that 41 produced no headrelated structures, 16 producedpolyheads (tubular structures), 8 produced petit X particles, and 7 produced structures of normal size (Katsura, 1980). The mutants which produced no head-related structures were not characterized further.
Thus, in SV40, adenovirus, and X, mutations have been identified which prevent shell assembly from initiating. The studies reported in this work may provide an explanation for such mutations.