Nature and distribution of sites of temperature-sensitive folding mutations in the gene for the P22 Tailspike polypeptide chain

J. Mol. Biol. (1988) 294, 607-619

Nature and Distribution of Sites of Temperature-sensitive Folding Mutations in the Gene for the P22 Tailspike Polypeptide Chain Robert Villafane and Jonathan King Department of Biology Massachusetts Institute of Technology Cambridge, MA 02139, U.S.A. (Received 5 April

1988, and in revised form

10 June 1988)

Temperature-sensitive folding (tsf) mutations in gene 9 of bacteriophage P22 interfere with the folding and association of the tailspike polypeptide chain at restrictive temperature. We report here the location and amino acid substitutions for 24 independent tsf mutants. The distribution of these and previously identified mutations is distinctly non-random; all of the 32 unambiguous sites of tsf mutations are located in the central 350 residues of the 666 residue tailspike polypeptide chain. No ts mutation has been found among the N-terminal 140 amino acids, and none among the C-terminal 170 amino acids. Since the physiological defect in these mutants is the destabilization of an early intermediate in the folding pathway, the localization of the mutants suggests that the central region of the chain is critical for formation or stabilization of this early intermediate. The majority of amino acids that served as sites for the tsf mutations were hydrophilic residues. Sixty percent of the replacements of these residues represented charge changes. This probably reflects the selection for mutant sites at the mature protein surface where the substitutions can be best tolerated without interfering with function. None of the sites of tsj” mutations were at aromatic residues, and only one proline site was found. Substitutions at these residues may cause lethal folding defects which are not recovered as tsf mutants. The local sequences at tsf sites resemble those reported for turns. Structural studies identify /I-sheet as the dominant secondary structure. These mutations may disrupt the formation of conformational features of p-sheets which are repeated, such as turns, associations between pairs of strands, or sheet/sheet packing interactions. Such a model accounts for the occurrence of tsf mutations with similar defective phenotypes at multiple positions along the chain.

of the amber sites in the gene for the Lac repressor remained fully functional when substituted with a variety of non-wild-type residues (Coulondre & Miller, 1977; Miller, 1983). Many of the recent successes in protein engineering require that the modified protein fold up correctly with altered amino acids in the active site (Borgford et al., 1987; Fersht, 1987; Howell et al., 1986). These manipulations presumably select for sites which are not critical for conformation. To determine which residues in a polypeptide chain are important in chain folding and association, we have studied tss (temperature sensitive synthesis) mutants specifically defective in these processes (Smith et al., 1980; Goldenberg & King, 1981; Smith & King, 1981). Such temperaturesensitive folding (tsf) mutants provide a means to investigate the genetic control of polypeptide chain

1. Introduction A number of lines of evidence indicate that all amino acids in a polypeptide chain do not contribute equally in determining the conformation of the polypeptide chain. Comparison of the amino from acid sequences of homologous proteins different species known to have the same threedimensional structure reveals a great deal of sequence variation (Dickerson & Geis, 1983; Rossman & Argos, 1981). Among over 200 hemoglobin sequences, the character of the residue side-chains is conserved at about 40% of the positions (Bashford et al., 1987). Many of these residues are likely to be important in determining the folding pathway. Genetic analysis reveals a similar pattern. For example, the missense proteins generated at many 607 0022~2X36/88/230607-13

$03.00/O

0 1988 Academic Press Limited

608

R. Villafane

_-

Temperature-sensltlve h

and J. King

stages

Chains associated but not fuity folded

Stable to 80 OC

Figure 1. Folding and chain association pathway for the tailspike protein (Goldenberg 6i King, 1982: Goldenberg P/ al., 1982; Haase-Pettingell & King, 1988). A series of intermediates are formed after ribosomal synthesis of the tailspikr polypeptide chain. The protrimer intermediate immediately precedes the formation of the native tailspike t,rimer. At the restrictive temperature, tef mutants fail to form the protrimer intermediate: instead the early folding intermediates form

an intracellular aggregate (Haase-Pettingell & King: 1988) folding and association (King, 1986; King & Yu, 1986). The tailspike protein is a trimer of a 666-amino acid residue polypeptide chain encoded by gene 9 of P22 (Botstein et aE., 1973; Goldenberg et al., 1982: Sauer et al., 1982). The mature protein has an endorhamnosidase activity and cleaves the Salmonella O-antigen (Iwashita & Kanegasaki, 1976; Rerget & Poteete, 1980). Laser Raman spectroscopy studies indicate that the secondary structure of the tailspike is 50 to 60% /? structure (Sargent et al., 1988). The tailspike is unusually heat-stable requiring above 80°C for denaturation temperatures (Goldenberg & King, 1981; Sargent et al., 1988). The native protein is not denatured by SDS under conditions in which almost all other phage and Ral,monella proteins are converted to SDS-polypeptide chain complexes and is resistant to protease digestion (Goldenberg et al., 1982). These treatments serve as the basis for distinguishing incompletely folded intracellular intermediates from the native trimeric tailspike. The in viva folding and chain association pathway is shown in Figure 1. After release from folded interthe ribosome, the early partially mediate converts to a form sufficiently structured for chain-chain recognition. This interacts to form t’he protrimer in which the chains are associated but not fully folded (Goldenberg & King, 1982). This species then folds further to generate the SDS/ protease/heat-resistant native tailspike. The intermediates are protease and SDS-sensitive. Since chain folding continues after chain association, there is no species corresponding to a “native” monomer (Goldenberg & King, 1982). There are no known covalent modifications in the tailspike maturation pathway. The early intermediate(s) is thermolabile: the fraction of polypeptide chains successfully proceeding to the native state decreases with increasing temperature of incubation in viva, and is less than 20% at 40°C (Goldenberg et al., 1982). In contrast the protrimer to native trimer transition is cold-sensitive, permitting the trapping of the protrimer in the cold (Goldenberg & King, 1982). An initial set of 38 tss mutants of the tailspike

protein were mapped to 33 sites in gene Y (Smith et 1980). Characterization of the polypeptide chains synthesized at high temperature and the proteins formed at low temperature revealed that all but one of these sites represent defect,s in t’hr folding and chain-association pathway (Goldenberg & King, 1981; Smith & King, 1981; C. HaasePettingell, M.-P. Corr. K. Haubert, A. Marra & J. King, unpublished results). Once correctly assembled and folded at low temperature, the native forms of the tsf proteins are thermostable: their melting temperature is >8O”C (Goldenberg 8r King, .J. M. 1981; Sturtevant . personal communication). The mutant polypeptide chains released from ribosomes at restrictive temperatures (37 to 40°C) do not form native tailspikes but remain as partially folded chains (Goldenberg et al., 1982: Smith & King, 1981). They are SDS and proteasesensitive and do not react with antibody against, the native protein. If the cells are shifted to permissive temperature soon after synthesis, the chains proceed through the productive pathway and form native tailspikes. If the tsf mutant chains remain at high temperature they form an intracellular aggregate (Haase-Pettingell & King, 1988). The simplest interpretation of the effect of t.hesr amino acid substitutions is that they further destabilize the thermolabile intermediate in the wild-type pathway. The very high stability of the native protein mitigates against the recovery of the tl (thermolabile) class of ts mutations, such as those in the gene for phage T4 lysozyme (Albers et al.. 1987). Yu & King (1984) determined the amino acid substitutions of 11 tsf mutations by DNA sequencing. The results indicated a number of interesting features of the mutants, but the sample was not large enough to draw significant conclusions. To determine which amino acids direct the chain folding and association pathway for the tailspike polypeptide chain, we determined the specific nucleotide changes associated with 24 additional tsf mutations in gene 9. These data define a region of the polypeptide chain that is crucial for the assembly and folding of the tailspike, and indicate that the substrates for tsf mutations are al.,

Distribution

limited to polypeptide

a subset of local chain and protein.

sequences

in

the

2. Materials and Methods (a) Bwterial

and bacteriophage strains

Salmonella typhimurium hosts, DB7000 (Zeu A414) and DB7004 (Zeu A414 supEZ0) were derivatives of LT2 from the collection of David Botstein. The P22 mutant phage strains tslil through tsU56, and tsN42 through tsNll1 were described by Smith et al. (1980). Additional mutants (tsU75 and mutants with higher numbers) were isolated by the same procedures (M.-P. Corr, K. Haubert, A. Marra & J. King, unpublished results). All of the phage strains carried the cl-7 clear plaque allele. Some of the phage strains used for DNA sequencing also carried an amber mutation in the delayed lysis gene. The multiple mutant strains were gene 9 alleles tsU1, tsU2, tsU3, tsUl3, tsUl4, tsU56, tsN42 and tsNll1. JM109 and M13(mp18) were generously provided by J. Messing (Rutgers University) and have been described (Messing, 1983: Yanisch-Perron et al., 1985). (b) Media and chemicals Media for the propagation of Salmonella bacterial cultures and phage have been described (Fuller & King, 1981; Smith et al., 1980). Media and growth conditions for Escherichia coEi and M13, as well as ligation buffers and conditions, were described by Messing (1983). (c) Concentrated phoqe stocks Large phage lysates for the isolation of DNA were prepared and concentrated by polyethylene glycol (PEG-F) precipitation and a cesium chloride step-gradient centrifugation. A fresh bacterial culture, grown overnight at 3O”C, was diluted 1 : 100 into 250 ml of superbroth and aerated by bubbling at the permissive temperature, 30°C. Incubation of infected cells was continued under these conditions for about 3 h to a concentration of lo*, which was monitored in a Petroff-Houser counter. The cells were then infected at a multiplicity of infection of 0.05 and incubation continued with aeration until lysis or until 4 h after infection. Typically 6 large-scale infections were done in parallel. At lysis 1 ml of chloroform was added and aeration continued on ice for 15 min. To remove cell debris the phage lysate was centrifuged in a GSA rotor at 10.000 revs/min for 15 min. After centrifugation the sample was removed and plated at permissive and restrictive conditions to check the temperature-sensitive phenotype. The supernatant was made up to 7% PEG and 0.5 MNaCl by addition of solid chemicals to the supernatant (Yamamoto et al., 1970). The PEG-phage was placed at 4°C for at least 1 h but generally overnight. This step precipitated the phage as a PEG-phage complex. The PEG-phage suspension was centrifuged in a GSA rotor at 10,000 revs/min for 15 min. The pellet was resuspended in 6 ml of phage buffer (50 mM-Tris (pH 7.6), 100 mMMgCl,. 10 mM-Nacl). (d) Isolation

of phage DNA

The PEG-precipitated phage was further purified and step gradient concentrated on a cesium chloride t Abbreviations used: PEG, polyethylene kb, lo3 bases: u.v., ultraviolet light.

glycol;

609

of tsf Mutations

centrifuged at 22,000 revs/min at 4°C for 2.5 h in a SW27.1 rotor in a Beckman LS-55M Ultracentrifuge. The step gradient contained cesium chloride made up in phage buffer at densities of 1.3, 1.4, 1.5 and I.7 g/ml. The phage band was collected by puncturing the side of the tube with a 20-gauge needle. The C&l phage band (about 1 ml) was dialyzed at 4°C in 4 1 of phage buffer for 2 h, then changed. It was changed 3 times at 4 h intervals or left overnight and changed once thereafter. To obtain a yield of 200 to 400 pg phage DNA/ml, the phage were diluted to 10 A,,, units of phage/ml. Two ml of phage were mixed with an equal volume of phenol equilibrated with 0.1 M-Tris (pH 7.6) made 0.2O/, (w/v) in %hydroxyquinoline; the mix was agitated gently at room temperature for 20 min to extract the DNA. The phenoltreated phage was centrifuged at 5000 revs/min for 10 min in an SS34 rotor to separate the phases. The DNA aqueous upper phase was phenol-extract,ed as described. The residual phenol in the final upper aqueous phase was removed by dialysis in the cold (4°C) in 4 1 of DNA buffer (10 mM-Tris (pH 7.6), 7 mM-NaCl, 1 mM-EDTA). The buffer was changed after the first 2 h of dialysis and thereafter every 6 h for 2 changes of buffer. The concentration of the phage DNA was determined by an A 260 reading, assuming that an A,,, of 1 is equivalent to 50 pg DNA/ml. (e) Restriction digestions and cloning for sequencing The entire tailspike gene is located on a 7,2 kb EcoRI restriction endonuclease fragment, EcoRI-C fragment (Chisholm et al., 1980; Jackson et al., 1978; Rutila & Jackson, 1981). There are 3 BamHI restriction sites on the P22 genome. Two are found within this EcoRI-C fragment, creating a 2.2 kb BamHI fragment and 2 EcoRI-BamHI fragments (2 kb and 3 kb). The coding sequences for amino acids 1 to 505 are located on the 2 kb EcoRI-BamHI fragment, and the adjacent 2.2 kb BamHI fragment contains the remaining 506 to 666 amino acid coding sequences. Fine structure mapping studies had indicated that most of the mutations should lie in this 2 kb EcoRI-BamHI fragment. DNA extracted from mature mutant P22 phage particles were used as the source of the gene 9 fragment. Phage DNA was digested with EcoRI and BamHI to completion and ligated to mpl8RF (double-stranded, replicative form) previously cleaved with the same enzymes. Several candidate white plaques from each mutant were chosen. Single-stranded DNA was prepared from these prospective clones and analyzed by one sequencing reaction (T tracking) to identify the correct inserts. The recombinant Ml3 DNA was then sequenced with custom-made primers. (f) Primers Five 18-base gene g-specific primers, which initiate synthesis from sequences at codons 105 (TGGATCGTACTTCAATAC), 200 (CTTGATAACCCATGGTGT), 298 (ATCTTTACCTCCGCTTGG), 399 (CGGATTCATGTCAGTGTC) and 592 (TGGCTCAGCGTATACTCA), were synthesized by the Whitehead Nucleic Acid Synthesis Facility (MIT) on an Applied Biosystems DNA synthesizer. The primers were purified as described by Lo et al. (1984). These primers were constructed to have minimum homology to the rest of gene 9 and to the Ml3 sequencing vector. They were used for the Sanger enzymatic sequence determinations (Sanger & Coulson, 1975; Sanger et al., 1977).

610

R. Villafane (g) Sequencing strategy

The nucleotide sequence of the first 505 amino acids is located on a 2 kb EcoRI-BarnHI fragment of P22 DNA, and the sequence for the remaining 161 amino acids is located on a 2.2 kb BarnHI fragment. These fragments were separately cloned onto M13mp18. Since we had available primers at approximately every 106 amino acid positions on the cloned gene, we used the fine structure mapping to aid in localizing the mutation. If the sequence change was not found in the interval then the adjacent interval was scanned with the next primer.

3. (a) Sequencing

Results the tsf mutations

ts Thirty-seven out of 38 of the original mutations mapped by Smith et al. (1980) were found to be defective in chain folding or subunit assembly (Goldenberg & King, 1981; Smith & King, 1981; Yu & King, 1988; C. Haase-Pettingell & J. King, unpublished results). Sixty-three additional temperature-sensitive mutants were isolated in the tailspike gene after ultraviolet mutagenesis following the protocols used by Smith et al. (1980). These were mapped against lysogens with endpoint,s in gene 9 and by two-factor crosses (M.-P. Corr, K. Haubert, A. Marra & J. King, personal communication). Reversion analysis for these mutants ranged from 10m5 to 10m7 (Table l), suggesting that most of these mutants also represented single mutations. Yu & King (1984) determined the nucleotide changes by the method of Maxam & Gilbert (1980) for a limited set of tsf mutants located in the central of a region of gene 9, due to the availability In order to convenient restriction fragment. determine the general features of gene 9 tsf mutations we determined the sites and substitutions of additional mutations. Two groups of mutants were selected: mutants originally described been by Smith et al. (1980), which had subsequently studied biochemically (Yu & King, 1988) but whose substitution had not been identified; and mutants which nucleotide sequences mapped to corresponding to the N and C termini of the chain, which included a group isolated in the second round mutagenesis described above. We were of particularly interested in mutations located in the amino and carboxyl-terminal regions of the polypeptide chain. Gene 9 comprises 1998 base-pairs (Sauer et al., 1982) so that sequencing of the entire gene for each mutant, was impracticable. The limited number of suitable restriction enzyme sites for specific endlabeling of fragments, together with the large number of mutants, prompted us to change to the Banger enzymatic synthesis method (Sanger & Coulson, 1975; Sanger et al.. 1977) with multiple primers (Itakura et aZ., 1984). This made it possible to sequence different regions of the gene from a single insert in a recombinant Ml3 sequencing vector (Messing, 1983; Yanisch-Perron et al.. 1985). DNA extracted from mature P22 phage particles

and J. King was used as the source of the gene 9 fragment. The entire tailspike gene is located on a 7.2 kb EcoRI restriction endonuclease fragment (Chisholm et nl., 1980; Rutila & Jackson, 1981). Purified phage DNA was digested with EcoRI and RumHI t)o completion and ligated mpl8RF (double-stranded, to replicative form) previously cleaved with the same enzymes. The existing tine structure maps served as a guide to indicate which DNA regions of the cloned gene 9 would be sequenced first. Tf a nucleotide change was not found with the initial primer then the adjacent primer was used. Figure 2 shows an example of a DNA sequencing gel displaying carboxy-terminal sequences of gene 9 fragment,s from a set of tsf mutants. The Sanger DNA sequence reaction generally allowed determination of the sequence of about 300 bases (equivalent in Fig. 2 to sequences coding for amino acids 400 to 500 with the universal primer). The nucleotide changes found for mutant 1)NAs were determined on t.he region of polyacrylamide gel where fragments were readily distinguished from adjacent ones. In this experiment nucleotidr changes were observed for tsU75. tsZi143 and taZJ770 and are indicated by arrows in t,he Figure. These mutants contain the most, carboxy-terminal mutations found to dat)e. Mutant tsZr170, whose nucleotide change was located at amino acid codon 434 in t)he upper part, of t,he Figure, was rerun to confirm its nucleotide change. Table 1 contains the sequence information new- t.o this report. This Table also includes the actual sequences searched in order to obtain these changes. The mutants are listed in this Table in order of increasing tsf allele number. Of 29 phage strains carrying isf mutations included in these experiments, we found nucleotide substitutions in 24 of t.he strains. The nucleotide substitut.ions in the retnaining five (tsL~13, tsZJ46, tsI.182. tsZ’727 and tsNIII) remain to be identified. Twenty of the strains displayed nucleotide substitutions. which corresponded t.o single amino acid substitutions in the polypeptide chain. Two strains, tsCJ135 and tsZT146, each contained base changes in two codons, resulting in double amino acid subst,itutions. Two strains. tsU157 and tsN4Y. displayed three substitutions each. tsL?P and ~~(~759 each had nucleotide changes in two adjacent positions of a single codon. Three mutations occurred independent,ly in two different strains; Gly > Ala at position 175 in tslJi’31 and tslJ734; He > Asn at position 180 in tsZJ135 (a double mutant) and tsZ!126; and Asp> Val a,t residue 230 in tsU56 and fsN49 (a briple mut’ant). teZ:.?4 and tsUZ59 had mutations in codon 163, tsZT75.Y contained t)wo mutations in adjacent nucleotides. One allele represents a valinr substitution and the other a glutamic a.cid subst it,u tion (Fig. 3(a)). Of the 29 phage mutant, strains listed in Table 1. the majority (22) were isolated by u.v. mutagenesis. These 22 mutant strains represent 26 nucleotide

Distribution

tsU56 n

C’GT

rsu75 Ac-‘r,T

isu/z7

of tsf Mutations

hut45

tsUl60

611

f.slJ/70

ACGTACGTACGTACGT

4

--

492

,493

Figure 2. Autoradiogram of products of DNA sequencing reactions displayed on a 6% polyacrylamide gel for DNA fragment, s cloned from 6 different tsf phage. The lanes are labeled according to the particular dideoxynucleotide used to inhibit tb le DNA synthesis reaction (A, C, G or T). The fragments synthesized are from the non-coding strand of the tailspike gene. The arrows depicted in the Figure indicate the resultant new fragments found for the mutants. The location (3f the altered codon is indicated to the right. For example the sequencing reaction for tsZJ75 mutant DNA which now ends in A (adenosine nucleotide) instead of T. In the coding strand. this reaction produced a new fragment at codon position 492. As a consequence. isoleucine is replaced by leucine. illustrate: 3 an A + T transversion

R. Villafane

612

and J. King

Table 1 Newly

sequenced gene 9 tsf mutab

Codon change

Amino acid change

Codon number

Sequences searched

ACC > GC(’

Thr > Ala

307

GAT > AGT

Asp > Ser

238

ts1:3 tx 1,:13

GAA > GTA

Glu > Val

309

ts1:14

GAG > AAG

Glu > Lys

405

ts1:34 tsT746

AAG > GAG

Lys > Glu

163

GAA > AAA GAT > GTT ATA > TTA

Glu > Lys Asp > Val Ile > La

196 230 492

GTA > GCA

Val > Ala

471

TTA > T(:A ATT > AAT

Leu > Ser Ile > Asn

224 180

GGA GGA GTT AAT TCT CTT GAT CGA GGA GAT AAG CCA TCT ACA CGC

> > > > > > > > > > > > > > >

GCA GCA GCT AAT TTT TTT GTT CAA GAA GTT GTG CTA TTT AAA TGC

Gly Gly Val Tie Ser Leu Asp Arg Gly Asp Lys Pro Ser Thr Arg

> > > > > > > > > > > > > > >

Ala Ala Ala Asn Phe Phe Val Gln Glu Val Val Leu Phe Lys Cys

175 175 141 180 493 144 149 148 155 216 163 250 227 199 ‘434

250-390 405-500 95-195 199-293 183-390 70-279 300-378 39&470 70-279 300-378 390470 70-193 70-279 300-378 343-500 91-293 95-500 390-500 l-293 70- 195 343-500 191-279 l-195 70-500 71-279 71-279 71-279

AGC GAT AGA GAA

> > > >

AA( AAT AAA AA$

Ser Asp Arg Glu

> > > >

Asn Asn Lys Lys

299 230 285 344

9 tsj allele

l.s1’157

t
tsSll1 t These reversion values were obtained from Smith

70x10

i

I 3x lWh 7.5x 10 q 1.2x lo-’

1.9x IO h

390-500 71-195 95-293

95-195 163-500 l-383 7 l-293 163-248 390-500 250-390 163-395

9.3 x IO- i 3.6 x lWh 1+x lo- 5 1.3 x lo- h 64 x 10 -f*

250-390

1.6 x 10 4

et al. (1980)

changes including double and triple mutants. Most of the base substitutions in the ultraviolet-induced mutants are transitions (14j24). (tsU2 and tsU159 have been omitted since they contain multiple nucleotide substitutions in a codon.) Of these transitions 9114 are of the GC -P AT type. There are ten transversions. The AT + TA transversion type accounts for most (7/10) of these transversions. The AT + CG transversion was not found. The isolation of these mutants required the production of a functional polypeptide chain at permissive temperatures. This selects against the recovery of addition and deletion mutations. A striking feature of the results was the absence of substitutions in the region from amino acids 1 to 140 at the N-terminal region of the protein, and from residues 500 to 666 at the C-terminal end of

the protein. All of the sites identified here, and previously by Yu & King (1984), are located in the central 350 residues of the tailspike polypeptide chain. (b) Nature of the amino acids that are sites for tsf mutations The third column of Table I shows the amino acid substitutions derived from the DNA sequence changes newly determined for this report. For strains carrying multiple mutations in gene 9, the identification of amino acid substitutions with the tsf phenotype is ambiguous. For the 20 unambiguous assignments, three of the sites are glutamic acid (substituted by 2 lysine and 1 valine) three are serine (substitut,ed by 2 phenylalanine and 1

:

‘, 600 .

/*

600

\,400

i 0

\

1

400 /@I

, TRP 202

GLN

u/4

I GLU 405

LYS

I

I I

I I

I

I I

I 1

I

100

U//6

u/33

620 I

580 I I

420 6 I

II

I

200

.

/1

,

nrr

.t

I I ARG434

I THR 368 UI 70 CYSt-

ILE ,I

H301

SER ASP 227 230

U2

I LYS 163

VAL

u/59

I

H304

GLU

u34

40 I

H303

I,

!

U/I U/60

U38

l

I

-GLY 435

-tGLU

640 I

560 1 I

440 I I

360 I

.

UP4

660 I8

540 I

460 I I

340 I

LEU 260 1 1 ILE 258

!40 I

60 !

300

400

9./

I

!

ALA

lJ85

I /

.

I GLY 323

ASP

H302

500

I ‘i

I I ( VAL 471

1 -SER 333

*ASN

N48

VAL 270

GLY f

RE-RH-RAF

666 4

ALA334

VAL-

l~lllil~ II ;I

.

I

I I

I /

1

ILE SER ARG SER LEU 240 I I I I I THR ASP GLY PRO 235 238 244 250

H300

u53

.,

l/56

I ,

-u/34

PHE VAL

U/62+5

GLY -FLY 177 175

A?G *ALA

SER 380 I I ARG 382

220 I

ILE180

ASN,

20 I

520 I

480 I !

320 1

280 I

120 II

80 I

u/9

I

U/-l/I6

-ALA

I 600

I I

ILE492

I

LEU+

u75

-SER 493

-PHE

u/43

I1 , LEU - GLU -THR 311 309 307

H”l’s * ::L

ARG 285

t

LYS

II

I -\

\ I

i 666

500,: /

500 -\

309: 1

\

\ !

I

300 I SER.\\ ‘, 299

ASN

N42

100,’ /

100 I.

Figure 3. t.vf mutations in gene .9. The region of the map containing the mutations is emphasized by a thicker line. (a) Locations and amino acid substitutions for all tsf mutants whose sequence changes have been found to date. The diagram is calibrated in amino acid position co-ordinates, with 666 amino acids in the complete tailspike chain (Sauer et al., 1982). For example, in the tailspike mutant tsiJl4 at amino acid position 405, a glutamic acid was substituted by a lysine residue. (b) A linear map in which the amino acid locations of the tsf mutants are represented by filled lollipops.

ib)

(a)

199 196

RIHl29/IA

‘1,200 .

\

i

:

u55

LYS LYS

lJ166

I

0

614

R. Villafane

and J. King

Table 2

asparagine); two are isoleucine ( > asparagine and leucine), two are threonine (>alanine and lysine): two aspartate (>serine and valine); and one site each is arginine ( > cysteine), glycine ( > alanine), ( > serine), proline leucine ( > serine), valine (>alanine) and lysine (>valine and glutamic acid). The lysine site at amino acid residue position 163 yielded two different mutant amino acids because a double nucleotide one of the mutants contained substitution. These substitutions are predominantly hydrophilic in nature (lSj20, 80%). This is much higher than the fraction of hydrophilic residues in the central region of the chain (173/350 = 49%). The complete set of tsf mutanm whose DNA has been sequenced comprises 41 single mutant strains which represent 32 unique sites (Table 4). 24/32 sites (75%) of amino acid substitutions were hydrophilic residues (5 glycine. 4 serine, 4 threonine, 4 glutamir acid. 3 arginine, 2 aspartic acid, 1 lysine and 1 proline). Two-thirds of the amino acid substitutions at these 24 hydrophilic sites represented changes in net charge. None of the tsf mutations occurred at sites of tyrosine, phenylalanine or tryptophan. In this central region of the polypeptide chain 38 codons of the 350 (11 Oh) are tyrosine, tryptophan and phenylalanine (Table 2). Only eight sites of mutation were aliphatic residues (1 alanine, 3 isoleucine. 2 leucinr and 2 valine). There are 127 aliphatic/aromatic amino acids in the central region of the chains, so that the underrepresentation of hydrophobic and aromatic residues is significant (Table 2). Only one proline codon (WA at codon 250) was a site of mutation even though the protein contains 28 proline codons. The residue replacements in the mutants were

fjistribution

of amino acid rrsidues and t.sf mutations

S-terminal IL150 A. Hydrophilic

(‘entral I 5 I -5cH)

(“-terminal 501468

Sumher of tsf mutations

rrsidu~s

I!) II .i 11 !j

0 0 0 0 0

diverse; many of the substitutions replaced a hydrophilic residue with a hydrophobic enc. However. in no case was an alipha.tic amino acid substituted by a charged amino a,cid.

Table 3 Mutable

codon distribution Ibtribution

Amino acid Ala AF A’F ASI'

Glu GIU Gly Gly Qly Ill? Ill? Leu IRU

Lys PM SW Ser

Thr Thr Thr YLLI Yal

(‘odon

Allele

(‘odor] no.

GCC AGA

t&.1 ts1’19

CG(’

ts1:53;

GA+ GAA GAG GGA GGC GGG ATT ATA TTA (‘TT AAG CCA TCT AGC ACA A(‘( ACT GTC GTA

tsl:56: kl’5i.

334 285 382: 434 230; 238 196: 309; 344

of radons

l-149

150-,500

3 0 0

6 2

ts1:14

405

I

.i I4 IO 3

tsC131; ts1’134; tsr:9; tsH304 tsH302 tslr3X tsl:116 tsU24. tuf:ic5 km6

175. 177. ’ 224 323’ 435

0 3 I

9 9 13

lsIT7

31 1

L&55;

ts1:1io

ts&

t&c:2 ;.d:x

ts1;34: L9UlsI tsVl1: kG160; tsH303 tsl:5; tsC’162: tsT’l43 lsN-12: tsN4R tslJ166 tS1; tsl:IX; tsH300 tsH301 tsRAF; tsRE; tsRH ts I’&5

I-2 3

:iOl-ti66 2 2

.i 7 4 I 2 2 7

180

x

II

Ii

258: 492 224

2 I

II) x

4 IO

2

:1

2

163

2

Ii

-,

250 227: 493

3 4

7

.)

li

.-5

299; 333

I

7

4

199

2

4

;i

307. 236 368‘

2 3

9 !)

r, 2

270

2

3

,I

171

*A

6

:s

Distribution

(c) The non-rmdm distribution of tsf mutations is not a reJection of mutagenic spec@city The majority of the 41 sequenced single codon change mutants (31/41) were isolated after U.V. mutagenesis (Tables 1 and 4). The allele designations of these mutants are prefixed by a U. Of the remaining 10 tsf mutants, three were isolated as suppressors of a cold-sensitive gene 1 mutation (lcsHl37): tsRAF, tsRE and tsRA (Jarvik & 1975). The five other mutants were Botstein, mutants (tsH300, derived as hydroxylamine tsH301, tsH302, tsH303 and tsH304). Two of the after nitrosoguanidine were isolated mutants mutagenesis (tsN4.2 and tsN48).

of tsf Mutations

We considered the possibility that the concentration of temperature-sensitive sites in the central region of the chain was due to effects at the level of mutagenesis. A possible explanation for the absence of tsf mutations in the two terminal regions of the gene (codons 1 to 140 and 500 to 666) is that the codons which were the sites of mutation are not regions. Table 3, present in these outlying organized alphabetically by amino acid, shows that with the exception of glycine and arginine codons all target codons are present in all three genetic regions. For example, only one of the six possible codons for serine was a site of mutation (TCT) in the generation of tsf mutants, but this codon is present four times in the first region, six times in

Table 4 Sites of tsf mutations in the tailspike gene Codon IlO.

175 175 177 244 323 435 250 250 250 199 235 368 307 307 227 227 299 333 493 285 382 434 163 163 230 230 238 196 309 344 405 334 224 311 180 258 492 270 270 270 471

Amino acid substitution Gly > Gly > Gly > Gly > Gly > Gly > Pro > Pro > Pro > Thr > Thr > Thr > Thr > Thr > Ser > Ser > Ser > Ser > Ser > Arg > Arg > Arg > Lys > Lys > Asp > Asp > Asp > Gl” > Glu > Glu > au > Ala > Leu 1 IRU > Ile > Ile > He > Vai > Val > Val > Val >

Ala Ala Arg Arg Asp Glu Ser Leu Leu Lys Ile Ile Ala Ala Phe Phe Asn Asn Phe Lys Ser Cys Val Glu Val Val Ser Lys Val Lys Lys Val Ser His Asn Leu Le” Gly Gly Gly Ala

615

Local sequence Ala Lys Ala Lys Phc Ile Val Lys Asn Tyr Leu Leu Glu Thr Glu . Thr Glu Thr Met Glu Gly Tyr Thr Trp Asp. Giy Asp Gly Thr . Leu Thr Leu Ala Asn Gly SetGln Ile Gly Phe Asn Leu Asn . Leu Asp Phe Asp. Phe Glu . Ser Glu Ser Pro Thr Gly Val Ile Ile 4sn t Gly Asp . Met Ser ‘Val Val . Val Thr Phe Asp Gly Lys Gly Asn . Gin Glu ‘. Cys Glu Cys Glu Cys Thr . His

. Phe Phe Gly Phe t Val Val Leu . Leu Leu Ser Gin Gin t Ile Ile Lys Lys Asn Val Tyr . Leu Qln Leu Gly Gly Lys Lys Val Phe Thr Cly Asn ‘Ser Ala Glu . Asn Gin Ile Ile Ile lie . Glu

Ile Ile Asp Pro Ile Arg Leu Leu Leu Thr Pro Gly Ile Ile Gin Gln Pro Ser Ile Phe Phe Val Gly Gly Thr Thr Ser Met Phe Phe Pro Ser Thr Asn Leu Asn Tyr Gly GIy Gly Ser

G&. Asp 1Gly Asn Leu m.Asp.Gly .Asn.Leu G1y. Asn Leu Ile Phe @J. Ile Glu Thr Leu m.Gly .Arg.Thr .Ser C&.Ala .Leu.Gly .Val ProPro.Asn.Ala .Lys m.Pro .Asn.Ala .Lys m.Pro .Asn.Ala .Lys Thr. thr . Pro Trp Val m. Val . Ser Asp Tyr J&. Val Gly Ser Thr m.Phe.Glu.Asn.Leu J’&.Phe,Glu.Asn.Leu &r Lys Thr ‘4s~ Gly & Lys Thr Asp Gly & Gly Gly Lys . Asp &g Ala Gin . Phe . Leu & Cly . Ala Cys Arg &. Gly . Cys . His t Phe &. Asp . Ser . Val Val As Gly Ala Leu Gly &. Val Leu . Thr Ile & Val Leu Thr Ile &. Gly . Tyr . Glu . Pro As. Gly Tyr Glu Pro &.Tyr.Val .Lys.Phe Glu Ser Thr . Thr Thr GluAsn.Leu.Ser .Gly aArg Asp. Gly . Gly Glu.Leu.Asp.Arg.Pro & Gin Phe t Leu Arg m. Lys Gln Ser . Lys &,Ser .Gly .Asp.Trp & Phe Thr Lys Leu & ‘Thr ‘Ser “I’hr ‘Leu & .Ser .Gly ‘Ala .(lys &I . Glu Val . His t Arg )‘aJ.Glu .Val .His .Arg W . Glu Val His Arg Val Phe Thr Asn Ile

8. Villafane and J. Kiny

616

the second region, which contains all the tsj mutations, and five times in the third region. The codons outside the central region are just as likely to be mutagenic targets as codons in the central region. Since mutagenesis is often context-dependent (Miller, 1983, 1985; Skopek & Hutchinson, 1982) the same codons could display different mutational sensitivities in different’ regions of the gene. For mutants isolated after U.V. mutagenesis, if the DNA strand is chosen such that. the base to be substituted in the mutant. is a pyrimidine, the tailspike mutations are located preferentially adjacent to other pyrimidines (not shown). This finding is consistent with the known mode of action of ultraviolet light on DNA (Haseltine, 1983; Miller, 1985) and in agreement with what Miller and his collaborators have observed in the lacl gene of Escherichia coli (Coulondre & Miller, 1977; Miller. 1983, 1985). From the results cited it is clear that U.V. mutagenesis should have been able to target any codon for mutation. Thus. the mutagenesis procedure does not limit the number of mutable sites. The same mutable codons are found in all three regions. U.V. irradiation has also been used to isolate over 100 amber mutants of gene 9. These nonsense mutations have been genetically mapped to all three regions of the tailspike gene (Fane & King, 1987). Taken together, these results indicate that the absence of tsf mutations in the parts of the gene corresponding to the protein termini is not due to the method of mutagenesis. (tsUl35 and The two double mutant strains tsU146) had reversion frequencies associated with single mutations. One of the changes noted for ts11135 is Ilel80Asn (isoleucine substituted by asparagine at codon number 180) found in the single mutant tsUl16. This suggests that the second substitution at residue 141 is silent. All three of the substitutions in tsN49 correspond to ts mutations. These multiple mutant strains are under further study. Although we have found many of these mutations by using the primer that is closest to the genetically determined location we have not formally proven that these mutations are the cause of the temperature-sensitive phenotype. However, the general correspondence of the nucleotide substitutions with the genetic map, and the lack of nearby substitutions in the majority of the strains, indicate that the majority of these amino acid subst,itutions represent the actual ts mutation. This is also supported by the absence of silent nucleotide substitutions. Further characterization of revertants currently underway should resolve the ambiguous cases of the multiple nucleotide substitutions.

4. Discussion For enteric bacteria such as S. typhimurium and E. coli normal growth and protein maturation

occurs across a range of temperat,ure greater than 20°C. At the higher end of the normal bacterial growth t,emperature, phage I’22 maturation is inhibited, due to the thermolability of partiall? folded intermediates in the tailspike maturation pathway (Qoldenberg et al.. 1982). The tsf mutations curtail phage growth by further lowering the yield of properly folded and associated t,ailspike chains (Smith & King, 1981). The most likely locus of action of the tsf mutations is the amino a,cid int.eractions maintaining the conformat.ion of critical folding intermediates. (a) tsf substitutions are located central region of th,e polypeptide

in the chain

The nucleotide changes of all 4 1 straius carrying are localized to t,hch trnt~ral region of the protein corresponding to amino acids 140 to 493. No amino acid substitution has been found in nucleotide regions corresponding to amino acid residues 1 to 140 or 494 to 666. The prevalence of tsf mutations in the central region nf t’he polypept*ide chain suggests that interactions among residues in this region are responsible for t,he format,irm or stability of the early thermolabile thlding inter-mediate in the t,ailspike maturation pathway. The ot,her regions of t,he chain probably achieve their conform&ion after t)he t,hermolabilr steps. A large set of nonsense mut,ations have been isolat,ed and mapped in gene 9 (Fane 8r King, 1987). Many of these mapped in the X-terminal region. establishing that, there is no barrier to mutagenesis in this region. The missense proteins generated at these sites with different host’ tRNA suppressor strains were biologically a,ctive. This result, together with the lack of t$ mutations iti this region indicates that t,he N terminus is not cxritic:al for t,he folding or association of the polypeptidtl chains. The N terminus probably takes its conformation late in the pathwa,y, and IS determined by inberactions with the rest, of the struc*t,urcb. Analysis of gene 9 amber mutant amHIOl4, located at, amino acid codon position 479 (J. Schwarz & P. Berget,. personal communication) in the C-terminal region of the tailspike, has shown that this site is involved in the catalytic function of the tailspike protein (Berget & Poteetr. 1980). A set of amber sites in this region generatIed let,hal phenotypes when propagated on various host tRNA suppressor strains (Fane $ King. 1987). The C-terminal part of the protein ma!; have the catalytic site for cleavage of the 0-antqen. It also may be critical for chain/chain as&:iation. or attachment of the mature tailspike to phage heads.

tsf mutations

(b) The sites of tsf substikutions preferentially hydrophilic

are

Residues which were sites of thtb tsJ’ mutations were predominantly hydrophilic, if glycinr is included in this category. Of the substitutions at. t.hese sites, more than 60!/; represented changes in net charge on the protein. We have, included the

Distribution

lysine-for-arginine replacement in this group, since the mutant protein displays an altered isoelectric point and electrophoretic mobility (Yu & King, 1988). The proteins carrying these charge changes have altered electrophoretic mobilities (Goldenberg et al., 1982). Yu & King (1988) studied the electrophoretic properties of purified native forms of a set of 15 tsf mutants. The results indicate that the substitutions were occurring at the surface of the native form of the protein and directly altering the electrophoretic properties. Hydrophilic residues are found more frequently at’ the protein surface than in the solvent-inaccessible interior (Miller et al., 1987; Bashford et al., 1987). The surface location would explain the enrichment for hydrophilic residues. Yu & King (1988) proposed that the surface location was a result of the initial selection for tsf mutants. This requires that the substitutions, which disrupt or destabilize folding intermediates at high temperature, be accommodated in the nat’ive structure at permissive temperature. The location on the native protein likely to be most tolerant of substitutions is of course the protein surface. Thus, t’he mutants identify amino acid sequences which (1) influence the conformation of folding intermediates. and (2) end up on the surface of the native protein. As suggested above, we suspect that’ substitutions of interior residues by bulky or caharged amino acids would prevent the formation of the native t’ailspike at all temperatures. The preferential surface location of tsf sites can account for the higher frequency of hydrophilic residues among the mutant sites, and can explain bow an a,rginine can be accommodated at the site of a glycine wit,hout disrupting the structure or function of the native (low temperature) form of the protein. An alternative model would be that the substitutions, not necessarily themselves charged, disrupted the local conformation unmasking other charged groups. This model predicts the existence of neutral substitutions which cause alterations in electrophoretic mobility or isoelectric point. The substitutions in six of the 15 proteins studied by Yu & King had not been identified. We found the substitutions for five of these: tsU2, tsUl4, taI:34, tsl’55 and tsX49. In all cases the changes result in an alteration of net charges. We have not found a mutant’ whose protein has indicated a mobility difference from wild-type protein and whose sequence has not revealed a charge change. The mutant amino acid replacements are distributed more equitably between hydrophilic (19) and hydrophobic (12) residues. This is not since there is no selection against surprising hydrophobic residues occurring at’ the protein surface (Miller et al., 1987). (c) 7’h,e central region probably rlssential

aromatic

residues

are

at all temperatures

The absence of’ aromatic occurrence of only one proline

residues and the residue as a mutant

of tsf Mutations

617

site were unexpected. The absence of aromatic residues as sit,es of tsf mutations indicates either that these sites are not critical or that they are critical but independent of temperature; that is, substitution causes a lethal mutation, which is not recovered as a temperature-sensitive mutation. Fane & King (1987) found that at more than ten sites in the tailspike chain, substitution of the amino acid residue caused lethal mutations. These would not be recovered in screening for tsf mutations. A subset of these lethal sites are tyrosine and tryptophan residues (B. Fane, It. Villafane & J. King, unpublished results). P. Berg& & J. Schwarz (personal communication) have also isolated absolute lethal sites in the tailspike gene. In their study of aromatic-aromatic interactions in proteins, Rurley & Petsko (1985) found that the side-chains of aromatic amino acids such as tryptophan, tyrosine and phenylalanine are very often found as aromatic pairs within proteins. Such aromatic interactions generally involved networks of at’ least three aromatic pairs. Disruptions of these energetically favorable interactions in a predominantly P-sheet-structured protein might be lethal. Tt seems likely that the low frequency of tsf mutations at proline sites also represents the critical role of such residues over the entire temperature range of phage growth. (d) Local conformations

at the sitrs

of mutation

Raman spectroscopic analysis has shown that the protein consists of 50 to 600,b of P-pleated sheet secondary structure (Sargent et al., 1988). Given the predominance of fi structure and the elongated morphology of the tailspike, the /?-sheet is likely to be of the cross-j family (Geddes et al., 1968) with short strands and many turns. Adenovirus penton fibers (Green et al., 1983) and phage T4 tailfibers (Earnshaw et al., 1979), both structures known to be largely /? structure, contain their fl structure as cross-j3 strands. Turns in polypeptide chains are concentrated at the protein surface (Kuntz, 1972; Richardson, 1981). In a cross-/l structure surface turns are numerous. Inspection of the sequences flanking tsf sites given in Table 4 reveals that many of them resemble sequences found at high frequency in fi turns (Chou & Fasman, 1977, 1979; Sibanda & Thornton, 1985). This correspondence between the local amino acid sequences and protein turn sequences strongly suggests that many of the tsf sites are associated with turns in t)hr polypeptide If the tsf substit’utions disrupt, these chain. we would expect the occurrence of processes. mutations with similar phenotypes at many sites through the region. The requirements for making various turns of a cross-p struct’ure must be somewha,t similar, despite the lack of repeated sequences. One appealing tailspike structural model is t,hat the central region contains much of that, /l structure

618

tl.

Villafane and J. King

and forms the core of the protein. An intermediate in the achievement of this conformation would then be thermolabile. The tsf mutations would act by further destabilizing this intermediate. Yu & King (1984) noted sequence similarities surrounding sites of threonine mutations in the tailspike chain, and suggested that these sequences were directing the formation of similar local conformations. Inspection of sequences flanking other tsf sites shown in Table 4 also reveals similarities. For example, this can be seen in comparing the residues N-terminal to the arginine sites (tsUl9, tsUS3, tsUl70) and the glycines following. These local sequences may be directing simila,r conformational features in the folding of a-pleated sheet, intermediates. (e) Comparison

with T4 lysozyme

ts sites

Albers et al. (1987) have described the threedimensional locations of a set of phage T4 ts mutations. The sites of these mutations were quite different from those described here; t,hey were at sites of limited solvent accessibility and limited mobility. The lysozyme ts mutations are thought to act by destabilizing the native state of lysozyme. Albers et al. (1987) proposed that these sites were making the largest contribution t,o the stability of the native protein. Residues at 11 of 18 ts sites were hydrophobic. In contrast to lysozyme, the tailspike tsf mutations do not destabilize the native st’ate, but) destabilize intermediates. The results are therefore compatible; tsf sites which are not important in stabilizing t,he tailspike prot.ein are very different from the ts sites which stabilize the nat’ive state of lysozyme. The ronformational contribution of the gene 9 sites in the folding intermediates may in fact be different from their role in the native state. Crystals of the tailspike protein diffracting to high resolution have been obtained (T. Albers, I,. LSreaver. M.-H. Yu and J. King, unpublished resu1t.s) but, the structure has not yet been solved. We would like to express our gratitude and Dasa Lipovsec Sali, participants in program. for contributions made during this work. This work was supported PCM8402546 and NIH grant CMl7980.

to Richard Seet the MIT CROP various parts of by NSF grant

References .4lbers, T.. Dao-pin, S.. Nyr, ,J. A., Muchmore, 1). C‘. & Matthews. B. U’. (1987). Biochemistry. 26, 3754-~ 3758. Bashford. T>.. Chothia, (1. & Lesk. A. M. (1987). J. ilfd. Rid. 196, 199-216. Jserget. P. B. & Poteete. A. R. (1980). J. I’irol. 34, 234~ 243. f
Hioche&ry,

26.

7246-

7250.

Botstein, I).. Waddell, (‘. H. 8r King. *I. (1973). .J. Mol. Hid. 80, 669-695. Rurlry. S. K. & Petsko. G. A. (198.5). LScirnce, 229. %3-C%.

Chisholm, It. I,.. ?Jackson. K. .I.,
C”umrnings Publisher, Menlo Park. CA. Earnshaw. W. C., Goldberg, E. R. & C”rowther~. It. .\. (1979). .I. Mol. Bid. 132, 101-131. Fan?. B. 8r King. ,J. (1987). Cenatics, 117. 157 171. 26. X03 I -8037. Fersht. A. R. (I 987). Biochemistry, Fuller. M. 8r King. J. (1981). Virology, 112, .529-547. (ieddes. h. ?J.. Parker, K. I)., Atkins. J-Z. I). ‘1‘. & Rrighton. E. (1968). -1. Mol. Hiol. 32. 343S3.58. (:oldenbrrg, I). & King. J. (1981). .I. #ol. Riol. 145. 633 651. ($oldenberg, 1). & King, ,I. (1982). ~‘roc:. %t. .-lc~/. Ci.. 1X.A. 79, 3403- 3407. (:oJdenberp. I).. Berget. I’. & King. .I. (1982). ,/ Kiol. Chem. 257. 7864-7871. (ireen. N. M.. Wrigley. ?i. G., R.ussell. D-. (‘. Martin. S. R. & McLachlan. A. J). (3983). EMHO .J. 2. 1357 1765. . < (:rii%n-Shea. R. (1977). Ph.1). dissertat,ion. Massa&usrt.t+ Instit.ute of Technology. Haase-Pettingeli. (‘. & King. .I. (1988). ,/. Kiol. (‘ltur/l. 263. 4977-4983. Haselt,ine. ‘cc’. (1983). (‘ell. 33. 1%Ii. Howell. E. E.. Villafranca. ,J. E., Warren, M. S.. Oatlry. S. ,I. & Kraut, J. (1986). Science, 231, 1123-l 1%. Itakura, K.. Rossi, ,J. ,J. &, \Vallac*e, R. 13. (1984). .In,tcr. Ksa.

Biochrm.

53, 32%

356.

lwashita,, S. &. Kanegasaki, S. (1976). Eur. ./. Ijiochenc. 65, 87-!f4. .JaLc*kson. E. ,Ij.. Miller. H. I. & Adams. RI. I,. (197Xj. J. No/. Riol. 118, 347 -363. .Jarvik, J. bz Uotstein. 1). (1975). I’roc,. *Vat. AC&. Sri., 1..S’.A 72. 4738--2741. King, ,J. (19%). Bio,Techno/ogy, 4. 97~M:S. King, *J. & Ytr. M.-H. (1986). Methods Enzyncol. 131. ?‘W dfifi. Kuntz. I. I). (1976). .J. A~Lw. Ph~m. Sot. 94, 4009-4012. IAJ. Ii.. ,Jonrs. S. S., Hackett, S. R. & Khorana, H. (:. (1984). Proc. A’&. Acad. Nci.. I’.S.d. 81, 2285 -d2X9. Maxam. A. & (Gilbert. N’. (1980). M&hods Enqmol. 65. 499-550. I. <*t 101. IO 93. Messing. .I. (1983). Methods Enqmol. Miller. .I. H. (1983). Anwx. Rec. Genef. 17. 215 4:!X. Miller. ,I. H. (1985). .J. Rio/. Biol. 182. 45 -6X. >liller. S.. ,Janin. .J.. Lwk. A. 11. A (‘hothia. C’ (1987). .J. ~llol. Hiol. 196. 641 &56. J~i&ardsot~. .I, S. (1981 ). A&rcrn. l’rofri~ ( ‘k(l)/!. 34. 16s Xi!). Itutila. .J. E. B
Distribution J. M. (1985). Nature Sibanda, B. L. & Thornton, (London), 316, 170-174. Skopek, T. R. & Hutchinson, F. (1982). J. Mol. Biol. 159, 19-33. Smith, D. H. & King. J. (1981). J. Mol. Biol. 145, 653676. Smit,h, D. H., Berg& P. & King, J. (1980). Genetics, 96, 331-352. B.. Yamamot,o, K. R., Alberts. B. M.. Benzinger,

of tsf Mutations

619

Lawhorne. L. & Treiber, C:. (1970). Virology, 40, 734-744. Yanisch-Perron, C., Viera, J. & Messing, .J. (1985). Gene, 33, 103-119. Yu, M.-H. & King, J. (1984). Proc. Nat. Acad. Sci., 7J.S.A. 81, 6584-6588. Yu, M.-H. & King, ,J. (1988). J. Biol. Chem. 263, 14241431.

Edited by A. Fersht