, o l . 2 8 5 ,N o . 5 7 6 0 ,p p . 7 8 - 8 i ,M a y 8 1 9 8 0 R e p r i n t efdr o mN a t u r eV @ Macmillan JournalsLtd. 1980

Mutationswhich alter the function of of the the signalsequence maltosebinding protein of Escherichiacoli Philip J. Bassford51t, AudreeV. Fowler*, lrving Zabin+, HugesBedou.1ls.$, Jon Beckwitht & MauriceHofnung'$ * Unité de Génie Génétique'Inslitut Pasteur'75015 Paris'France 02115 + Departmentof Microbiologyand Molecular Genetics,Ha.vard MedicalSchool,Boston,Massachusetts f DepartmentoI Biol;gical Chemistry,UCLA Schoolol Medicineand MolecularBiologyInstitute,Univercityof Califomia,t s Angeles,Califomia

d into theextemalpeiplasmiccompartmentof thecellby uirne of The malrosebindingproteinof Escherichiacoli is secrete an amino-terminalsignal sequence.Using DNA sequencing,we haoedeterminedthe precisenatureof mutationsin the signal sequencewhich preuentthe exponof the maltosebindingprotein,causingit to accumulatein the cytoplasmin its thechangeof a singlehydrophobicor unchatgedamino acid to a chargedamino acid within piecursor'form. In most cases, is sufrcientto block the seÛetionprocess' the signalsequence Mosr secretedproteinsin both prokaryotic and eukaryoticcells are initially synthesisedwith an amino-terminal extension of some 15 to 30 amino acidswhich is subsequentlyexcised.It has 'signal sequence' is responsible for been proposed that this initiating passageof the polypeptide through the membrane while it is still a nascentchain attachedto the ribosome'-'. The exact mechanismby which this is accomplished,and the nature of the putative receptorin the membrane,are unknown. Among the different signal sequencesthus far determined,there seems to be no signifiôantprimary sequencehomologyo.However, one feature common to all signalsequencesis a central core ot9-24 predominantly hydrophobic amino acids.In most cases,this is preceded by one or more charged residues, usually lysine and/or arginine.One approachto defining thosefeaturesof the signalsequencewhich are essentialfor initiating protein export is to generate mutations which alter the signal sequenceof a secretedprotein, resultingin the failure of the cell to export the protein from the cytoplasm. The Gram-negative bacterium Escherichia coli provides a convenientsystemfor isolatingmutants which affect secretion, as geneticmanipulationsof it are relatively easyand many of its proteins are localisedin the cell envelope,and, therefore, must be secretedthrough the cytoplasmicmembrane.Evidenceexists that signalsequencesare involved in the export of proteins to at least iwo of the bacterial envelope compartmentss. These compartments are the outer membrane and an aqueous layer located between the two membranes,called the periplasmic space. One of the proteins found in the periplasmicspaceof E. coli is the maltose binding protein (MBP), a key component of the maltose transport system6.We recently described a selection procedure which enabled us to isolate mutants of E. coli which iail to secretethe MBP into the periplasmT.In the mutants we obtained, the MBP accumulates in the cytoplasm in its higher molecular weight precursor form. From the genetic data, we suggestedthat the mutations in these strains resulted in single amino acid substitutions in the MBP signal sequence.We now report the DNA sequencecoding for the first 50 amino acids of $ To whom reprint requests should be addressed'

the wild-type MBP precursor, the identification of the signal sequencewhich was done through partial amino acid sequence determination, and the DNA sequencechangesfor severalMBP export-defectivemutants.

Partial determinationof the MBP signalsequenceusing ^ malU-lacZ hybrid protein Becausethe MBP precursor does not accumulate to appreciable amounts in wild-type cells, material for determining the MBP signal sequencewas obtained from another source. We had previously isolated E. coli strains in which the malE gene encoding the MBP is genetically flsed to the lacZ gene enThe resultant hybrid gene coding the enzyme p-galactosidasen. codes for a hybrid protein which has at its amino-terminus an amino-terminal portion of the malE gene product and, at its carboxy-terminus, enzymatically active p-galactosidase. One such strain, PB4-81, synthesisesa hybrid protein which is cytoplasmic and which seems, from the genetic and biochemical data, to includeonly a very small, amino-terminalportion of the malE gene product. Consequently,we suspectedthat this protein containedonly a portion of the MBP signalsequence.Thus, this protein shouldprovide material for a partial signalsequence determination. The determination of the PB4-81 hybrid protein sequence was made possible by techniquesdeveloped for the sequence analysisof a hybrid protein between B-galactosidaseand the /ac repressorn.A partial signal sequencefor alkaline phosphatase had been determined previously in this waytt'. The procedure used to generate hybrid proteins must result in the removal of at least the first 17 amino acids of wild-type p-galactosidase,thus eliminating the first methionine residue. If the amino acid sequence which replaces the normal p-galactosidaseaminoterminus does not contain any methionine (other than the initiating methionine), only one new cyanogen bromide fragment (CNBr2*) should result from treatment of the hybrid protein. This proved to be the case with the hybrid p-galactosidasefrom strain PB4-81. Sequenceanalysisof the CNBr2* indicated that the first 14 residues are not derived from pgalactosidase(Table 1). These residues are derived from the

Tabfe 1

Sequencer analysis of the malE-lacZ

hybrid protein from strain PB4-81

Amino acid identified Residue no.* 1 Met 2 Lys 3 Ile 4 Lys 5 Thr 6 Gly 1 Ala 8 Arg 9 Ile l0 Leu l1 Ala 12 Leu 13 Ser 14 Ala 15 Ser

Intact proteint

CNB12*t

Amino acid identified

CNBr2* T4$

+ T 1 -j-

+ -t

+ + + + + + +

+ f

+ -l-

+

+ + + + + +

+ + T

+

Residueno.' * 16 Pro * 1 7G l y * 1 8V a l *19Thr *20 Gln *21 Leu *22 Asn *23 Arg *24Leu *25Ala *26 Ala *27 His *28 Pro *29Pro *30 Phe 31Ala

Intact proteinT

CNBr2**

+ + + +

+ + + + + + +

+

CNBr2* T4$

1

+ + + T

+

All sequencing lr]asdoneusing0.1 M Quadrolin a Beckman890Csequencer. The peptidesweresequenced usinga duâl washprogram . The intâcl cârboxymelhylaled proteinsas scqu€nced wilh a programsimilarto that ofHunkapiller and Hood . AII other procedur€s $ere asdescribedpreviously , exc€plthal recentlyidentification wasalsocaried out with HPLC. + Indicates that the amino acid indicated c,asdeiecred. A blânk indicates either ân ambigrous result or that it wasnoi possibleto dete€t the amino acid by this technique. * Amino acidsidentifiedwith an asteriskcorrespondro residues19-34 of nativeE crii 8-galactosidase. TThehybridp-galactosidâs€waspurifiedbythesameprocedureasfor{he/r.repressor-lt-galactosidasehybrid.SirainPB+81wasg.ownina150-lN€wBrunswickfermentor qo q0 purifed grceroland prolein al30"Cin medium63containingI was 4o-foldw'th a €ldof36q0basedon û-g 0 2 ûalrose.Thehybrid 9oqopure on SDS g€1s22 and migrated with a mobility closeto lhat of wild-typ€ f-gâlactosidase. protein The hybrid was carboxynethylated and digested with cyanosen as The CNBr2* peptidewâspurifiedusinsCM-celluloseând t bromide describedelsewhere'zr. SP-Sephadex ion-exchange chromatogrâphyin 8 M ureaand gelfiltrationon S€phadexG-50 in 30olûaceticacid'?r. The râdioirnrnunoassay usedto monitorpurificalionon CNBr2* has been describedpreviouslye. The yield of CNBf2* from purified hybrid proteir was 25ol.. The elution volum€ on SephadexG-50 sussesteda size of about 90 residues.Anino âcidcompositiondata indicatedthat the PB4-81 CNBr2* containslysinewhereaswild-tpe CNBr2 has no lysine. $ To confirm lh€ presenæ of serine at residue 14 of CNBr2*, the peptide was €leavedù'ith trrpsin and lhe resulting peplides purified. Amino âcid anâlysisof CNBi2* (corresponding to re8idues9-23 of the inlâcl protein)irdicated two residuesof serineând three residuesofleucine. Detailsofthe proteiûând peptidepurificationsând (4.v.F., I.2., P.J.B. Jr and J.B. in preparation). sequence analysislrill be presented elsewhere

amino-terminus of the protein to which B-galactosidasehas been fused, in this case, the malE gene product. Two factors suggestedthat these 14 residues,along with the amino-terminal methionine from the hybrid protein, correspondedto the first 15 amino acids of the MBP precursor. First, this l5-residue sequence, with three basic residues among the first eight followed by several predominantly hydrophobic amino acids, is similar to other signal sequences.Second,none of this sequence was encountered when the protein sequence of the aminoterminal 25 amino acids of the mature MBP was determined (data not shown, see Fig. 1 legend).

DNA sequenceof the wild-type malU signal sequenceregion Cleavageof plasmid pHC5, which carries the malB operon (see Fig. 2\, with the restriction endonucleaseEcoRI, generates a fragment o11,227 base pairs, comprised of 850 base pairs from the malB region of the E. coli chromosome and 377 base pairs derived from the vector (plasmid pBR322) (Fig. 2). This EcoRI,EcoRI hybrid fragment includes the promoter-proximal portion of the malE gene which codes for the MBP signal sequence region. Beginning with this fragment, a region of 600 nucleotides extending from the EcoRI cleavage site in the malK (ref.. 11) gene in the direction of the malE gene has been sequenced on both strands,using the dideoxy/nick-translation technique of Maat and Smitht2. This region includes the entire regulatory interval between the two malB operons and will be described elsewhere(H.B., in preparation). We have taken the.EcoRI cleavagesite in malK on the strand encoding the MBP as the origin for numbering purposes. Starting from this point and examining the sequence towards the malE gene, the first ATG initiation triplet which is not soon followed in phase by a nonsense triplet is at nucleotides 457459. Just before this point, at nucleotides 443-447, there is a 5-nucleotide sequencecomplementary to the 3'-OH end of the 165 rRNA (Shine and Dalgarno sequencet'). This sequence suggeststhe existenceof a translation start which is used in uiuo. In fact, nucleotides 457-498 correspond to the first 14 amino acids of the MBP precursor, as determined from the protein sequence of the PB4-81 malE-lacZ hybrid protein described

above. Nucleotides 535-600 correspond to the frrst 22 amino acids of the mature MBP. Consequently, we deduce that the complete MBP signal sequence,that part of the MBP precursor which is excised to yield the mature protein, must be 26 amino acids long and encoded by nucleotides 457-534 (summarisedin Fig. 1 legend). Note that the serine at residue 15 of the PB4-81 hybrid protein (Table 1) is not derived from either the malE or the lacZ gene; neither can it arise from a simple fusion of the malE and lacZ geneswithin a codon. This serine residue may have been generated during the in uiuo gene fusion technique which used bacteriophage Mu to obtain the malE-lacZ hybrid gene*.

Sequencingthe DNA of the malB region of MBP export-defective mutants From the DNA sequenceof the regulatory interval between the two malB operons, we identified a cleavage site for endonucleaseHinf.I at nucleotide 406, and a cleavage site for AuaII at nucleotide 445. Restriction analysis revealed that the enzymes HinfI and AuaII cleaved the EcoRI-EcoRI hybrid fragment of plasmid pHC5 at only one site each, this being the one identified from the sequence (Fig. 2). Simultaneous digestion of the hybrid fragment with Hir?f I and AoaII yields two large fragments, as well as one small 39-base pair fragment corresponding to nucleotides 407-445 on the numbered strand. To determine the sequence of the putative malE signal sequence mutations we had previously isolated, we used the technique developed by Sanger and co-workersrawith the 39fragment serving as a primer. The base pair Hinfl-AvolI advantagesof this technique were as follows. (1) The genetic data on these mutations indicated that they were located early in the malE gene'. (2) These mutations were isolated directly on a transducing phage carrying the region of interest. Thus, phage DNA for each of the mutants could be used as template for the sequencing procedure after strand separation. (3) The AUaII end of the Hinf[-AuaIl fragment is located l, L nucleotides from the beginning of the malE gene. This allowed us to read approximately 150 nucleotides into the malE gene in a single experiment. Using this procedure, we have verified the wild-type DNA

rl8-l ) fl0-ll

(r4-r)

(16-I)

, ,t Jô-

I

|

)

Fig. I The DNA sequence of the first 144 nucleotides of the malE gen€, corresponding to the first 48 amino acids of the MBP precursor. CCC CAA AAC AGGACG Only one strand, that with the same polarity as the molE message,is shown. Those amino acids underlined represent residues which were also A T G A A A A T A A A A A C A G G T G C A C G C A T C C T C G C A T T A T C C G C A T T A A C G A C G A T C A i J T T T ? C C G C C T C G Gdetermined CTCTC from protein sequencing. The sequence beginning at residue 26 was determined on the mature MBP protein (A.V.F. and Me tL s I L e L SJ'NTGL A LaA r. I L eL e u A L a L e u Se r A L a L e u T h r T h r M e t 1 4e t P h e S e r A L a Se r A L a L e u LZ., unpublished). On the basis of protein 'lC 15 sequencing, the evidence for glutamic acid at ?O 25 residue 3 of the mature MBP is tentative only. The arrow after residue 26 indicates the position at which the precursor is cleaved to yield the mature MBP. The five codon changes corGCCAAAATCGAAGAAGGTAAACTCGTAATCTCGATTAACCCCGAIAAAGGCTA?AI.CGGTCTCGCTCAA responding to the different signal sequence mutations are indicated above the position in the sequence where they occur, along with the ALaLlsILeGLuGLL,lLULl.tsLeuVaLiLeTrpfLeAsnGLltA.sr;LltsGL!.tTltrAsnGLllLeuALaGLu predicted changes for the amino acid sequence. Glu

I rittt

|

30

L!s

I

35

I

ArgArg

11

40

sequencefor the first 144 nucleotides of the malE gene which had previously been obtained by the dideoxy/nick-translation technique (Figs 1, 3).

Phenotype of MBP export-defective mutants The mutations thought to result in MBP signal sequencealterations were initially obtained in a malE-lacZ hybrid, gene carried on the À transducing phage, À,p72-47 (ref . 7). In this case, the hybrid gene encodes a hybrid protein that includes a substantialportion of the MBP, including the entire MBp signal sequences. The mutations, which weré located in the malE portion of the hybrid gene,were subsequentlyrecombined into a wild-type malE gene on the bacterial chromosome. The resulting recombinants are Mal when scored on maltose tetrazolium agar, indicating that their ability to utilise maltose as a carbon source has been impaired. When they are induced for the expression of the malEFG operon, they accumulate the MBP precursor in the cytoplasm. However, these strains are not absolutely defective, for they grow slowly on maltose minimum medium and export a small amount of native MBP to the periplasm (see ref. 7). Furthermore, the mutants vary in their ability to utilise maltose, indicating that they may each affect the secretion of MBP to different extents (Table 2).

DNA sequencealterationsin MBP export-defectivemutants Using the same sequencing procedure described above, we determined the DNA sequenceof the promoter-proximal 144 nucleotides of the malE gene for nine spontaneous, independently isolated MBP export-defective mutants. The DNA used as template was obtained from the À transducingphage on which the mutations were initially isolated (derivatives of À p72-47). Each of the nine mutants sequencedexhibited an alteration in the region of the malE gene corresponding to the MBP signal sequence. Five different alterations were recognised (Fig. 7). The transition CTC (Leu) to CCC (Pro), resulting in a change in residue 10 of the MBP precursor, occurred in one mutant. The transversions GCA (Ala) to GAA (Glu) at position 14, ACG (Thr) to AAG (Lys) at position 16, and ATG (Met) to AAG (Arg) at positions 18 and 19 were encountered 3, 1,2 and2 times, respectively.

Discussion We have presented here the DNA sequenceof the MBP signal sequenceand of mutant signal sequenceswhich result in defective export of the maltose binding protein. The MBP signal sequenceis similar to signal sequenceswhich have been determined for other exported proteins in both prokaryotic and eukaryotic systems.Of the 26 amino acids comprising the MBP signal sequence, almost two-thirds are hydrophobic residues. Furthermore, all three charged residuespresent (two lysinesand an arginine) are clustered in the first eight residues. Following

45

the arginine at residue 8, only the neutral amino acidsserine and threonine interrupt the hydrophobic nature of the MBP signal. The amino acid just before the cleavage site in the MBP precursor is alanine. This is consistent with other signal sequences,where the residue here usually has a side chain with no more than one carbon moietva.

ntalG

mulF

mulF

pmul

malK

lcntB

I

I

Sall

1 EcoRl

asntI iIBamHl)

1 EcoRl

'fhe Fig. 2 malB region of E. coli includes the malE gene and four other genes whose products are also components of the maltose transport system23. These are organised into two operons orientated in opposite directions and which diverge from the promotor region designated here as Pmal (ref .24). The 1,22'l-basepair .EcoRI-.EcoRI hybrid fragment shown above is obtained by EcoRI digestion of plasmid pHC6. This plasmid carries a bacterial fragment of 1,750 base pairs which extends between cleavage sites in the malB region for restriction endonucleases Sa/l (in malK) and BelII (in malE)2s . This fragment was inserted between cleavage sites for Sa/I and BamHl in pBR322'z6.Details on the construction and characterisation of plasmid pHC6 will be presented elsewhere (J. M. Clement, D. Perrin and J. Hedgpeth, in preparation). Sites for cleavage of the hybrid fragment by Auall and HinTI are also indicated (see text).

The isolation of mutants with well defined alterations in the signal sequenceof a secretedprotein and in an outer membrane protein (seeaccompanyingpaperl5) which result in the failure of the cell to secretethese proteins, provides the first genetic proof for the role of the signal sequencein the initial steps of protein secretion. Other workers on bacterial systems have suggested that the basic residues early in the signal sequenceare responsible for the initial interaction of the precursor with the negatively charged inner surface of the cytoplasmic membranel6'17. Subsequently,according to this model, the hydrophobic portion of the signal beginsto penetrate the membrane as the first step in the translocation process. Four of the five mutations we have characterised introduce a single charged residue into the hydrophobic core of the MBP signal sequence. Three of these mutations were found at least twice in independently isolated mutants. This strongly supports the notion that the hydrophobicity of the signal sequencehas a key role in initiating protein export through the membrane. In the remaining case,

c

a

C

T A G

T

c

A G

T A G

t+: Ac Fig. 3 Composite figure comparing autoradiograms of s e q u e n c i n gg e l s l o r t h e w i l d typc malE gene and the malE signal scquencemutations. C)nly the relevant parts of the gels are shown. a, The wildtype sequence. Points along the sequence where mutations were encountered are indicated on the left. The enlarged letter indicatesthe exacl nucleotide that is altered in the mutants. F/ Represent the five classesof signal sequence mutations which we obtained in this mutations analysis. The shown in e and f correspond to those present in E. coli s t r a i n sP B l l ( ) l a n d P 8 1 1 0 2 . respectively, as described in rel. 7 .

6c

6

A

TC

ccc e

d

AG6

Growth properties of malE signal sequence mutants

Wild-type l0-l l4-I lo-l

1 8 -I 1 9 -I

Wr"'çtr

ffi

w+sw

*wgP

oGc

oA6

such revertants.

Mutant no

u "

N\*,"Sq.w*-.s*sS6f

the substitutionof proline for leucineat residue10 representsan exchangeof hydrophobic residues.Proline does have the property oî interrupting a-helices in polypeptide chainsr8.so it is conceivable that secondary structure of the signal sequence plays a part in the initial stepsof secretion. The changesobservedhave difteringel]ectson MBP secretion (Table 2). The two mutationswith the strongesteffectsare 18- 1 and 19- 1, which changea hydrophobic to a chargedamino acid. The three remaining mutations, which do not cause such a dramatic reduction in growth on maltose,involve a hydrophobic to hydrophobic amino acid change( 10- 1) and an unchargedto a c h a r g e da m i n o a c i d c h a n g e( 1 4 - 1 a n d 1 6 - 1 ) . T h e s e r e s u l t sa r e also consistentwith the notion that the overall hydrophobic nature of this region is the most important feature for the secretionprocess. All of the mutations we have obtained in the MBP signal sequenceare point mutationswhich readily revertT.We hope to obtain additional information on the important structural featuresof the MBP signalsequencethrough a sequenceanalysisof

Table 2

f

*W

G r o w t h o n m a l t o s em i n i m a l m e d i u m *-f++ -T +1"

+-+-* I

This researchhas been supported by grants from the Centre National de la Recherche Scientifique (LA 04271 and ACC 4218), the Délégation Générale à la Recherche Scientifiqueet Technique (ACC 79.7.0664),the North Atlantic Treaty C)rganization(grant 1297), the Institut National de la Santé et de la RechercheMédicale (Groupe Recombinaisonet E x p r e s s i o nG é n é t i q u e U . 1 6 3 ) t o M . H . , t h e N S F ( 7 9 - 1 - 9 9 1 4 ) a n d t h e U S P H S( A I 0 4 1 8 ) t o I . Z . a n d t h e N S F ( P C M - 1 6 - 2 1 9 5 5 ) t o J . B . P . J . B . J r w a s a p o s t d o c t o r a lf e l l o w o f t h e H e l e n H a y Whitney Foundation. H.B. is the recipient of a specialstipend from the Institut Français du Pétrole. We thank Jean-Marie Clément and Joe Hedgpethfor the gift of the pHC5 plasmid,and W. Boos for the mature MBP. Received l7 December 1979: accepted22 February l9tl1) L l l l o b e l . G . & D o b h e r s t e i n .B . - I . C e l l B i o l . 6 7 ' l i - 3 5 8 5 1 1 1 9 7 5 ) 2 . l n c r u y e .H . & B e c k w i t h . J . P r o c . n a t n .A c a d . S c i . U . - SA 1 1 , l 1 1 o - I ! 4 l t l 9 1 7 l 3 . R a n à a l l .l _ . l _ . .H a r c l y 'S, . i . S . & J . s e f s s o n ,L . - C . P r o c .n a t n . A c a d . S c l . L t . S . A . 7 5 , 3 6 1 3 6 5 rl978t. 4 . A u s t e n . B . M . F E B . SL c r r . 1 0 3 , 3 0 U 3 1 3 { 1 9 7 9 ) 5. Silhavy.T. J., Bassford, P. J. & Beckwith, J. R. in Bacterial outer Membrane: Bios|nthesis, A s . s t ' r n b l t ' u nFcul n t t t r s n s( e d . I n o u l e , \ ' l ' ) 2 0 1 - 2 - s '(1\ \ ' i l e y ' N c u Y o r k ' 1 9 8 0 ) (l9l'1)' 6 . K e l l ! - r m a n nO . . & S z r n c l c n r a nS. . É r r . . / . B i r x h e n t ' 4 7 . l l 9 l ' 1 9 7 . B a s s f o r t l .P . & U e c k w i t h ,J . \ a t u r t ' 2 7 7 ' - 5 l l t - 5 4 1 ( 1 9 7 q ) . -f. l 9 l l (1979) 1 3 9 ' B t t c t . J. & tleckwirh' J. J lJ. Bassford, P. Jr., Silhavy, 9 . u r a f c . A . . t. , [ - o * l e r , A . v . , z a b i n , 1 . , K a n i a , . 1 .& N ' l Ù l l e r - H i l lt,s . P r o L . n u t n . A t u d . S < i . I / . . S . A . 7 5 . . + 8 2 . 1 - . 1 t i 2 I7I 9 7 8 ) . 1 0 . S a r t h - vA , . . F o w l e r . A . , Z - a b i n ,I . & B e c k w i t h ' J J . B a c t l 3 9 ' 9 3 2 - 9 3 9 { 1 9 7 9 ) . e n .G e n e t . 1 1 4 , 2 6 1 - ) 6 7 \ 1 9 ' 7 9 t . l l . R a i b a u d .o . . c l e m e n r , J . M . & H o f n u n g , M . l v l o l e c g (1974)' 12. Maat .l.& Smith, A. J. H. Nucleic.{cidsRes.5,45-37-4545 (1974)' l 3 . S h i n e .J . & D a l g a r n o , L . P r o c .n a m A c a t l S c i U - S A 7 l ' 1 3 4 2 - 1 3 4 6 1 . 1 . S a n g e r . F . . N i c k l e n . S . & C - o u l s e n ,A . R . P r o c . n a t n . A c a d . s c i . t / . S . A . 1 4 , 5 4 6 7 - 6 4 6 7 r 1 9 7 7t . li5 1 5 . E m r . S . D . , H e d g p e r h ,J . , C l é r n e n t ,J . - N l . ' S i l h a v y ,T . - 1 .& H o f n u n g , N ' l ' N ( t / ' r e 2 E 5 ' 8 2

1

The mutants examined here are given numbers according to the amino acid p o s i t i o n i n t h e p r o t e i n . M u t a n t s l 8 - l a n d 1 9 - l c o r r e s p o n dt o m u t a n t s 1 1 0 1 a n d i 102, r".p..tively, as described in ref. 7. The growth phenotype was determined by observing colony size on maltose minimal agar at 37 "C after 1 day. The hierarchy of growth rates corresponded exactly to the gradient of colours seen on maltose ietraiolium agar. For example, the wild-type parent forms white colonies, whereas mutant 18-1 formed the darkest red colonies on this agar. Mutant 16-1 forms a pink colony on the same agar, but apparently secretesenough MBP to the periplasm to allow near normal growth on maltose minimal media'

16. 17. Itl. 19. 20. 21. 22. 23. 24. 25. 26.

{ I9l{0). (1978). D i R i e n z o , . l . M . , ) . l a k a m u r a ,K . & I n o u y e , M . A . R e u . B i o c h e m . 4 7 , , 1 1 1 1 - 5 3 2 H a l e g o u a ,S . & I n o u y e , M . J . m o l e c B i o i . 1 3 0 . 3 9 - 6 1 ( 1 9 7 9 ) . C h o u . P . Y . & F a s m a n .G . D A . R e t . B i o c h e m .1 7 , 2 5 1 - 2 7 6 t 1 9 7 8 ] r ' (1975). B r a u e r . A . W . , M a r g o l i s , M . N . & H a b e r , E . B i o c h e m i s t r , I-t l . 3 0 2 9 - 3 ( ) 3 - 5 (197U)' H u n k a p i l l e r , M . W . & H o o d . 1 , . E . B i o c h e m i s t r v1 1 , 2 1 2 4 - 2 1 4 4 F o w l e r . A . V . & Z a b i n , I . l . h i o l . C h e m . 2 5 3 ' - s 4 9 9 - 5 5 0 4( 1 9 7 8 ) ' l-aemmli, [J. K. .Varlre 227,6tt0-68-s (1970). S i l h a v y ,T . l . e r a l . M t t l e c .g e n . C e n e t l 7 1 , 2 4 9 - 2 5 9 ( 1 9 7 9 ) H o f n u n g , M . G e n e t i c s7 6 , 1 6 9 - 1 8 4 ( 1 9 7 4 ) . (197U)' M a r c h a l , C . , G r e e n b l a t t ,J . & H o f n u n g , M . I . B a c t . 1 3 6 ' I 1 0 9 - 1 I 1 9 S u t c l i f i e . J . G . N u c / e l c A c i d s R e . s .5 , 2 ' 7 2 1 - 2 ' 7 2 8 \ 1 9 ' 7 8 ) .

p r i n t e d i n G r c a t B r i t a i n b y M a c m i l l a n P r o d u c t i o n L t d . . B a s i n g s t o k e .H a m p s h i r c