1 THE STRUCTURE AND l<UNCTION OF YEAST KI TOXIN

HONG ZHU DEPARTMENT OF BIOLOGY McGILL UNIVERSITY MONTREAL,OUEBEC SEPTEMBER 1990

A thesis submitted to the Faculty of Graduate Studies and Research in

partial fulfillment of the requirements for the degree of Ph.D

(C)HONG ZHU, 1990

l

Abstract

The carboxyl-termmal sequences of the ex and ~ subunits of the sc3creted

yeast K1 toxin have been determined by protein sequencing and ammo acid

analysis of peptide fragments generated fram the purified tOXIn. It revealed that the

CI. and f3 subunlts consist of amino acid residues 45-147 and 236-316 fram the

prepratoxln, respectively The preprotoxin configuration can be represented as:

preprapeptlde-ArgPro-a-ArgArg-y-LysArg-f3. This structure for the preprotoxin

defines a specifie pracessing pathway ln yeast involving a dibasic endoprotease,

encoded by the KEX2 gene and a carboxypeptidase B like enzyme which is

prabably encoded by the KEX1 gene.

By using the patch-clamp technique, it is shown both in vivo wlth sensitive

yeast spheroplasts and in vitro wlth asolclctin liposomes that the toxin forms ion

permeable channels. The toxin induced ion channels are voltage independent

with a unit conductance of 118pS, often appearing in pairs and prefer monovalent

cations.

K1 toxin has a rTluch wider killing spectrum at the spheroplast level than at

the who le cell level as demonstrated by the fact that the toxin kills spheroplasts

from the genera Candida, Kluyveromyces, and Schwanniomyces, whose cells are

toxin insensltlve A toxin binding study shows that the wall receptor can define

toxin speeificity Hnd is necessary but not sufficient for toxin action on Intact cePs.

Usmg various mutagenesis techniques, a set of mutations throughout

regions encoding the CI. and f3 slJbunits that allow secretion of mutant toxins were

generated. By analyzmg the phenotypes of these mutant toxin , the ion channel

formlng domain is assigned exclusively to the hydrophobie ex subunit and the eell

wail reeeptor blnding domam is localized to bc,th the a and f3 subunits.

1

.,

Il

RÉSUMÉ

La séquence de l'extrémité carhoxy-terminak d~" sous-\Inité~ lX

et B de la toxine KI chez le,.; levures a été détcrminé~ par "L;qllL'n~'agL'

de protéines et par l'analyse des acides aminés pro\'cnant dl'

fragments peptidiques générés à partir de la tO\.llle plJlIIIL;L'. CCl'I a

permis de démontrer que les sous-ulllté" (X ct g COITl'\pondcnt

respectivement aux ré~ldus d'acide" aminés 45-147 ct 2.~6-.~16, a

partir de la préprotoxlllc.

La configuration de la préproto\.ine peut donc êtlL' 1 L'\)rL; "C Il téL'

comme su it: prépropeptlde-ArgPro-ex.- ArgA r g -y-Ly~t\ rg IL Cet te

structure définit un enchaînement de maturation SpéClflquL' l'ho les

levures impliqu,lIlt une endoprotéase dihaslque codée par Il' g01ll' 1\ F '\ :!

ainsi qu'une enzyme sImilaire ~l la carhox:., ,H.:ptida~c B plllhahklllL'llt

synthétisée par le gène K E XI.

En se servant de la technique de "patch clamp", Il l':-.t lk;1ll0Iltll- 111

vivo en utilisant des sphéropla~tes de IcYure~ sen"lblc~ Ü la IOXlnl' KI.

de même que in vltro gTâce à des liposomes 1'I'IU'l d'a'lolcctilll' que la

toxine forme des canaux perméables aux ions. Ces canaux ,ont

indépendants au voltage avec une unité de conductance de 1 1 Xp~~ et

apparaissent souvent en paires, préfèrant les catIons monovalent,

La toxine K 1 a un spectre d'activité beaucoup plm y,lste au

niveau des sphérop1astcs qu'à celui de la cellule entière. En effet, la

toxine tue les sphéroplastes du genre Candlda, K I/(vve rOn!VCl' sand

SchwanniomYLes mais n'a aucun effet au niveau de leurs cellules

entières. Une étude de couplage de la toxine démontre que le récepteur

membranaire aide à défInir la spéCIficité de la toxine et C'It néce\\aire

mais insuffissant pour l'action de la toxine sur le" cellule~ intacte,.

Utilisant diverses techniques de mutagénèse, un groupe dc

mutations a été généré dans la région a et 13 cie la protéIne, permettant

la sécrétion de toxines mutantes. En analysant le phénotype de ce~

toxines, le domaine de la protéine nécessaire à la formatIon du canal

d'ions est localisé uniquement à la sous-unité hydrophohlque fi tandl\

que la région "contact" du récepteur membranatre e\t locall\ée aux

sous-uni tés ex. et B .

traduit par Anne-Marle Sdlcu

l Table of Contents

Ahstract

Re'\ume

Table of Contents

Acknowlcdgemcnts

Preface

Abbreviations

Lbt of Figures and Tables

Chapter Literature Review

1.1 Introduction

1.2 Genetic basis of the killer toxin in

1.2.1

1.2.2

S.cerevisiae

The MI-dsRNA genome

The LI genome

1.2.3 Maintenance and expressIOn of the

dsRNA genome

1.3

1.4

104.1

104.2

104.3

1.5

Processing and secretion of toxin

Toxin action

Cell wall receptor for the KI toxin

Toxin action at the plasma membrane

Functional domain assignment of KI toxin

Immunity

1.6 Toxins from Prokaryotic and Eukaryotk

1.6.1

1.6.2

organisms

Colicins

A-B type toxins

111

11

111

VI

Vlll

ix

x

1

1

2

2

3

4

8

8

1 0

1 2

1 3

1 6

1 6

20

1

..

1.6.3

1.6.4

Chapter 2

2.1

2.2

2.3

2.4

Chapter 3

3.1

3.2

3.3

3.4

Chapter 4

4.1

4.2

4.3

Other toxins l'rom ycasts

Rationale

Determination of the carboxyl tcrminl of

the u and p subunit" of ycast KI klllcr

toxin: requiremcnt of a carboxypcptidasc

B-Iike activity for maturation

Introd uction

Materials and methods

Results

Discussion

Yeast KI killer toxin forms ion-channcls

in sensitive yeast spheroplasts and in

liposomes .

Introduction

Materials and methods

Resul ts

Discussion

The KI killer toxin of Saccharon!.v('es

cerevisiae kills spheroplasts of many

yeast specle~

Introduction

Materials and methods

Results and Discussion .

1 \'

25

25

26

2X

~5

40

40

41

44

5 1

54

54

54

55

~

... Chapter 5

5.1

5.2

5.3

5.4

Summary

References

Mutational analysis of functional

donains of KI killer toxin l'rom

Saccharomyces cerevisiae

Introduction

Materials and methods

Results

Discussion

Contributions to original knowledgt.

Appendix Complete nucleo.~de sequence of the Ml

dsRNA preprotoxin gene

v

62

62

63

69

75

82

87

105

106

1

\' 1

Acknowledgements

1 wish to express my deep gratitude to Dr Howard Bussey for hls gUidances.

encouragement, patience and help thraugtlout the course of my tl éllnl!lg 1 éllso

thank Dr Gregory Brown and Dr Ronald Poole, who have ail served on Ply

supervisory commlttee, for thelr tlme and advlce

1 am very grateful to Dr Michael W Clark for hls constant 3ncouragemenl.

support and Immense help at the last stage of thls work, especléllly for bis pntl81lCe

and tima in readlng thls thesls.

1 like to thank my collaborators, Alexander W Bell III Blotechnology

Researcb Instltute, Montreal and Bons Martlf'lac, AndrzeJ KIJbalskl, Xln-Ll3ng Zhou

in Wisconsin for their excellent !echnical help Thelr !mpreSSlve skills and

knowledge or protein chemlstry and electrophyslology have made the

collaborations frultful. Above ail, 1 have truly enJoyed worklng wlth them and

learning fram them. Dr. Ching KLJng' s generous support and valuable dIScus5Ion~;

throughout the collaboration III Wisconsin IS speclally acknowledged

While 1 was in Wisconsin, many people III Dr Ching Kung' s Lao and ln Dr.

Michael Culbertson's Lab have provlded me wlth thelr fnendshlp and generous

help both III and out of the Lab. Yoshlra Salml, Robin Preston and Karla Peuluvar

should be especlally mentioned 1 wlsh in the future, 1 Will have more

collaborations like this, fun, enJoyable and successful

Dunng my stay ln ~!:)Iogy Department, McGl1i University for the past few

years, my dear fnends and colleagues have always been my unendl:lg supports

on ail matters. They generausiy provlded me wlth unlimlted sources of spiritual

food and scientiflc Information Includlng tE:chnlcal assistance, Ideas and

diSCUSSions at the various stage of thls work. It IS beyond my abillty to express how

much 1 have appreciated ail thelr helps and 1 thank ail of them, partlcularly,

Margraet Ahmad, Charles Boone, Antony Cooper, Dorota Czerucha, Hong-Mel

l ..

VII

Gao, Zhl- l'un Gong, Kathryn Hill, Petra Kuhl, Marc Laroche, Ananna Lee, Terry

Roerner, Carol Saavedra, Anne-Mane Sdlcu, Josephine Wagner, Cun-Le Wu,

Xlao-Mlng Yang, Mmg-Da Zhang and Plilg Zhao From th&m 1 have learned and

shared S0 much

1 must also thank Karen Ketchum, Vahe Saraflan for the diScussions about

electr'Johyslologlcal work, Linda Hougan for help wlth the hydroxylamme

mutagenesls, Thierry Veinet for provldmg plasmlds, Bruce William for mterests ln

thls work and Claire Bonflls fo'" bHlng my tenniS partner

A special appreclatlon goes ta Linda Anderson for her excellent assistance

ln almost ail my dealmg wlth McGlI1 University throughout many years whlch has

made my stay in Monlreal much easler and also for 11er kindness and

understandmg. 1 also hke to thank Kim Bartlett and Eduardo Turcott RIos for thelr

fnendshlp

1 would hke to express my heartfelt appreciatlon to Xlu-Ming Yang for her

special fnendshlp and helpfulness, kindness, cheerfulness and support over man y

years. Mrs. Youko Karino and Dr. Takeshl Kanno ments a sp9cial mention for their

constant encourgement, support and fnendship.

To my parents, slster, bruther and Toshihisa Asakura for their unfailing

moral support, encouragement, understanding and love, 1 wish to express my

deep appreclatlon and love

My fmal thanks go to Nankai University and Canadian International

Development Agency for providlng me thls chanco to study in McGill Ulliversity

where 1 have learned and understood ~o much about both in sCience and IIfe.

1 would IIke to dedlcate thls thesis to my motherland-China, and 1 sincerely

Wish It to be strong and prosperous.

•

\' III

Preface

This thesis IS assembled ln accordance wlth the regulatlons of tlle Faculty of

Graduate Studles and Research of McGlI1 University It IS composed of an abstract,

a literature revlew ( chapter 1), followed by four chapters ( 2, 3, 4.5) ln a form

sUltable for publication ln order to avold redundancy. the introductions for Cl1apter

2, 3, 4, and 5 have been revised Ali clted literature has been combmed and

placed at the end of the thesls

Chapter 2 has been published as a paper by Zhu, H ,H Bussey, 0 Y

Thomas, J. Gagnon and A W Bell 1987. In J BIOl Chem 262 10728-10732 Ali

of the results presented in thls chapter are the work of the author

Chapter 3 has been published as a paper by Martlnac, B ,H Zhu, A

Kubalskl, X. L Zhou, M. Culbertson, H Sussey, and C. Kung 1990, ln Proe. Nat!.

Aead. Sei. (USA) 87: 6228-6232. The results reported ln thls chapter represent a

collaborative effort. Patch-clamp of the yeast spheroplasts and asolectln liposomo

vesicles was done by B. Martinac, A. Kubalski, and X. L. Zhou. Punflcatlon of

native K1 tOXIn, deSign of the expenments includlng effiCient incorporation of K1

toxin into yeast spheroplasts and artlflcial liposomes and preparation of such

materials were carried out by the author.

Chapter 4 has been published as a paper by Zhu and Bussey, 1989, ln

Appl. Environ. Mlcrobiol. 55 2105-2107. Ali of the results presented ln this chapter

are the work of the author.

Chapter 5 has ceen submltted as a paper to Mol Cell BIOl by Zhu and

Sussey. It is in press. Ali of the results presented ln thl5 chapter are the work of the

author, except that the Initial hydroxylamm9 mutagenesl5 of the killer plasmld was

done by L. Hougan.

AOH

ADP

ATP

bp

cONA

dsRNA

ER

HEPES

kd

MES

NAD

NMR

ORF

pS

ScV

SOS-PAGE

Tris

VLP

YEPO

Abbreviatïons

alcohol dehydrogenase

adenosme dlphosphate

adenoslne tnphosphate

base pair

comp!ementary deoxyribonuclelc acid

double-stranded nbonuclelc aCld

endoplasmlc retlculum

N-2-hydroxyethyl plperazme N-2'-ethanesulfonic acid

kilodalton

2-(N-morpholino) ethanesulfonlc acid

nlcotinamlde adenme dinucleotide

nuclear magnetlc resonance

open readmg frame

plcoslemen

Saccharomyces cerevlsiae virus

IX

sodium dodecyl sulfate polyacrylamide gel electrophoresis

tris ( hydroxymethyl) amine methane

Virus-like partiele

Yeast extract peptone dextrose

•

1

Chapter 2

Fig. 1

Fig. 2

Fig. 3

Table 1

Chapter 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Chapter 4

Fig. 8

List of Figures and Tables

Purification of KI toxin

HPLC purification of the COOIl-tcrminal

fragments .

Structure and proccsslT1g of the ycast KI

killer preprotoxin

Amino acid composition of purificd KI

killer toxin and the COOH-tcrminal

peptides from the a and 13 subunits

Channel activity of excised patch l'rom

\

30

32

yeast spheroplasts exposed to killer toxin 46

Channel activity detected l'rom a~olcctin

liposome~ containing toxin

Current-voltage (i-Vp) plot~ for lipos()Jl1c~

in asymmetric K and K/Na solutions

Opening probabilities of the toxin-induccd

channels

Effects of killer toxins on ycast

40

50

Fig. 9

Table 2

Chapter 5

Fig.IO

Fig.ll

Table 3

spheroplas ts

Binding of killer toxin by sensitive,

insensitive, and resistant yeast cells

Sensitivities of yeast spheroplasts to KI

and K2 killer strains

Localization and phenotypes of

mutations in the coding regions of the

ex and ~ subunits.

Pustulan-Sepharose column

chromatography of killer toxin

Mutations in the tox in precursor gene

and a phenotype summary

Xl

57

61

58

72

76

7 1

1 Chapter 1 Literature revicw

1.1 Introduction

The killer phenomenon in the yeast Saccharomyces cl'rl'\'Ïsilll' \Vas

first discovered by Makower and Bevan in 1963. ft was found lhat

certain yeast strains secreted a protein, which was IClhal 10 sensitive

yeast cells. The strains producing this protein tox in, howcver, wcrc

immune to the specifie toxin they produced. Utilizing the power of

yeast genetics and DNA recombination technology, extensive

investigation of this killer system has generated data upon a vancty of

cellular processes, such as, virology, protein maturation, secretion and

protein-cell surface interaction.

Makower's killer yeast system contains two specles of dsRNA

genome: a 1.9 and 4.7 kb called Ml and LI dsRNA (Bevan and Sorners,

1969; Bevan et al., 1973; Mitchell et al., 1973; Sweeney ct al., 1976;

Vodkin and Fiuk, 1973). Both Ml and LI dsRNA are present in the œil

cytoplasm and encapsidated in virus-like particles called ScV-M 1 and

ScY-LI, respectively (Herring and Bevan, 1974; Buck et al., 1973;

Hopper et al., 1977; Bostian et al., 1980a, b). The LI dsRNA genomc

provides encapsidation, replication and maintenance funetions for bolh

viruses. The Ml dsRNA on the other hand provides the genetic

information that produces the KI killer toxin.

In addition to the two cytoplasmically inherited dsRNA gcnorncs,

the killer system is also dependent on numerous complex inlcracl ion~

with gene products provided by the host genome. Thcre arc al Ica~t :n

MAK genes (maintenance of killer) and 7 SKI genes (supcrkillcr )

controlling the maintenance and copy number of the dsRNA pla .... mi(\I"

(

2

While the maturation of the primary translation product of Ml

(preprotoxlü) and secretion of the toxin are regulated by a number of

host SEC (secretion ) and KEX (killer expression) genes. In addition,

many host-encoded products, such as REX (resistance expression), and

KR E (ki ller resistance), are involved in the process of immunity, and

sensitivity to toxin.

Characterization of the secreted KI toxin has been quite thorough.

Mature, secreted KI toxin encoded by Ml dsRNA is composed of two

disulfide bond Iinked, nonglycosylated subunits of 11.4 and 9.0 kd,

designated as (l and ~, separately (Bostian et al., 1980a; 1984 ).

The construction and expression of the cDNA copy of the KI killer

toxin in yeast (LoUe et al., 1984; Hanes et al., 1986) has allowed

detailed molecular analysis of KI toxin possible.

1.2 Genetic basis of the killer toxin in S.cerevisiae

1.2.1 The Ml-ds RNA genome

The 1.8 kb Ml-ds RNA genome encodes the preprotoxin gene,

which confers both the killing and immunity phenotypes to host yeast

strains (Bevan et aL, 1973; Mitchell et al., 1973; Vodkin and Fink, 1974;

Sweeney et al., 1976; Fink and Styles, 1972; Wickner, 1974). Sequence

analysis of Ml-dsRNA revealed (Skipper et aL, 1984; Bostian et al.,

1984; Thiele et al., 1982; 1984) that it consists of approximately 1.0 kb

and 0.6 kb segments separated hy an adenine-uracil (AU) rich

denaturation "bubble" sequence, which is about 200 bp long (Skipper,

1983; Fried and Fink, 1978). The preprotoxin sequence is within the 1.0

kb segment and starts from bases 14-964 on the left-hand end (5'

primer end) of Ml plus strand, followed by the AU -rich denaturation

..

3

"bubble" and 0.6 kb segment. The denaturation "bubblc" and the 0.6 kb

region does not contain any significant open rcading frames (Thiclc ct

al., 1982b; 1984a; Hanning et al., 1986).

From studying naturally occurring dclction mutants of the MI

dsRNA, it was found that the signaIs for transcription. packaging. and

replication were probably within the left hand end of Ml dsRNA from

bases 1· 230 and the rightmost approximately 500 bp (Fried and Fink.

1978; Thiele et al., 1984 a, b). These regions probably aet as recognitioll

sequences for the regulatory gene products provided by the host

genome.

1.2.2 The LI genome

The LI dsRNA genome exists in most Saccharomyces strains,

regardless of whether the cells express the killer phenotype or not

(Bevan et al.,1973; Vodkin et al., 1974). However, the existencc of the

ScV -Ml partide is dependent on Ll-dsRNA. In vitro translation of

denatured LI dsRNA in a wheat germ system produced the major coat

protein (80 kd) of the virus like particles in which both it and M dsRNA

are encapsidated (Hopper et al.,1977; Bostian ct al., 1980b). A complete

cDNA clone of the LI dsRNA has been obtained and scqucncc(1 (lcho

and Wickner, 1989). ft was found there are two open reading frames III

the sequence. ORFI encodes a 680 amino acid residue protein with a

molecular weight of 76 kd. close to the prcvious estimates for the major

coat protein. ORFI and ORF2 overlap by 130 bp and ORF2 i~ in the -1

reading frame with respect to ORFI. It is predicted that ORrl and ORF2

together encode a 180kDa fusion protein, probably by a -1

translational frame shift, a scheme commonly used by many

..

4

retroviruses to produœ fusion proteins (Jacks et al., 1988). This 180 kd

fusion protein has i.nmunological cross-reactivity with the major coat

protcin (lcho and Wlckner, 1989).

Other families of L ds RNA, denoted LB, and Le, have been found

in most Saccharomyces strains (Sommer and Wickner,1982b). It is not

clcar whether these L-B e dsRNAs have any functional relation ta the

killer phenomena, since sorne KI killer strains lack L-BC entirely

(Sommer and Wickner,1982b). It IS known that Ml does not require

either LB or Le for its maintenance or replication.

1.2.3 Maintenance and expreSSlOn of he dsRNA genome

The VLPs encapsidating LI and Ml dsRNA are found to contain

an RNA polymerase activity that can synthesize a full-length plus

strand message of each dsRNA (BeTTing and Bevan, 1977; Welsh, et aL,

1980; Welsh and Leibowitz, 1980; 1982; Bruenn et al., 1980). These

full-Iength plus strands (ssRNAs) are released from the VLPs and have

message activity as weIl as being a template for the synthesis of the

minus strand in replication (Bostian et al., 1983a; Sc1afani and Fangman,

1984; Williams and Leibowitz, 1987). It is not c1ear what genes encode

the VLP RNA polymerase(s). The Ll-dsRNA and a number of

chromosomal genes are the possible candidates.

A large number of chromosomal genes, called M AK and SKI have

been found to be involved in the replication and maintenance of the

dsRNA genome. Among them, only 4 MAK genes, 3,8,31 and 32 are

required for maintenance of LI (Sommer and Wickner, 1982; Wlckner

and Toh-e, 1982~, ail of them are needed for maintenance of M

(Wickner and Leibowitz. 1976b; Wickner, 1978; Guerry-Kopecko and

l Wickner. 1980). Several MA K genes have been isolated. '\eqlll'Ilcl'd. and

their functions determined. MA K 8 (also called TC M 1) encode ..

ribosomal protein L3 (Wickner el al.. 1(82). MAKI (l'OP!) l'nL'odes a

DNA topoisomerase 1 (Th rash et al.. 1(84). IloweVl'r. the me(:hani"illl of

their effects to the replication and the maintenance of M l-dsR N A 1"

totally unknown.

Another set of genes involved in maintenance of both M and L

dsRNA are the SKI genes. Most of them have ncgative contlOls ovel the

dsRNA copy number (Toh-e et al.. 1978). In some cases. ""/..1- mutatIons

suppress many of the mak- mutations (Toh-e and Wickncr. 1 ()80) as

weil as conferring Ml maintenance and replicatlon independcnt of

[HOKj, the non-Mendelian gene which supplies thc hclpcl fUllction

needed by Ml for replication in a wild-type strain (Wickncr and Toh-c.

1982; Ridley et al., 1984). Also ski- mutations cause cold sensitivity loI'

cell growth (at SOC ) when an M replicon is present (Ridley ct al.. 1(84),

due to elevated copy number of M dsRNA.

The interaction between host gene products and dsRNA( s) arc

very important for control1ing the host cell from over-replication of

killer genomes. However, the mechanism of these complcx intcractions

remains to be further explored.

1.3 Processing and secretion of toxin

The primary in vitro translation product of dcnaturcd M l-d"iRNA

IS a 32 kd protein immuno-reactivc with antibodic~ raiscd againl.,t

secreted KI killer toxin (Bostian et al., 1980a). Whcn tramlatloll occlIrc.,

in the presence of increasing quantities of dog pancrca, rncrnhranc

vesicles, a mixture of two protein~, 30.4 kd and 42 kd arc round al.,

6

produets of the primary translation produet (Bostian et al., 1983b). The

42 kd protein was sen'\itive to endo H treatment consi~tent with its

having undcrgone core glycosylation while the 30.4 kd protein was

proce~sed :0 be the signal clcaved product of the 32 kd species.

Analysis of intracellular immunoreactive protein species in pulse­

labcled cells indicated that the in vivo toxin precursor IS &lso a

glycoprotein of approximately 42 kd molecular weigh> (Bostian et al,

1983b; Bussey et al, 1982). Therefore, it was cIear that the secreted

toxin was processed from a precursor. With the determination of the

DNA sequence for the entire KI toxin coding region, it became cIear that

the preprotoxin consists of a leader peptide followed by ex. and p subunits which are separated by a glycosylated y peptide (Eostian et al,

1984 ).

The 0 domain of the toxin precursor ha~ typical ER signal

sequence characteristics, including a hydrophobic membrane spanning

peptide sequence between iesidues 12 and 27 (Bostian et al, 1984) and

a putative leader endopeptidase c1eavage site at residue 26. When the ù

domain plus 8 amino acids of the a. subunit was fused to bacterial

cellulase, this leader directed secretion of the cellulase protein to the

extrace\lular medium (Skipper et al., 1985). These results strongly

suggest a signal-like function for the KI leader.

Early genetic analysis of secretion-defective mutant strains of

yeast harboring the Ml virion demonstrated that secretion of toxin is

depcndent on these gene products and showed that the killer toxin

precursor is processed along the genetically defined yeast secretory

pathw41Y (Bussey et al, 1983). In addition, proteolytic processing of the

toxin prccursor is also dependent on at least two other genes, K EX 1 and

1

7

KEX2 which are also required for the maturation of thL' mating

pheromone, a factor (Wickner and Leibowi li, 1976a; Dmocho",!'\h.a \.'t al,

1987). Mutant strain<; of kex/ and f...c'(:! are unable to "L'L'Jell' active

toxin, but retain immunity (Wicknl'r and Leibowiu, 1(76). A fmlhl'r

clarification of the procl'ssing l'vents illvolved III prep[oto\1I1

maturation came from the isolation and DNA ~eqllenL'l' analy"\', 01 the

KEX2 gene (Julius et al, 1984). K EX2 is now known to cncOlk an

endoprotease which is h calcium dependent, neutral serine protcase

with homology to subtilisin (Julius f-~t al, 1984; Mlzuno et al, \<)87; 1988;

Fuller et al.. 19R8). The K EX2 product cleaves specifically hetwecll

paired basic residues Iike Lys-Arg or Arg-Arg. There are Ihree

potential dibasic K EX2 cleavage sites within thc y peptide of

preprotoxin. Cleavage at the C-terminus of the y peptide, reIcascs the li

subunit (Bostian et al , 1984). It is not clcar whcther thc othcr Iwo

K EX2 sites in the y peptide are cIeaved ln VlVO by the K h'X '2 product. An

additional sequence surrounding the N -terminus of a "uhull Il, Pro t\ rg

GluAla, may be also a substrate f('r the K EX2 cndoprotcase, ~incc 1 Il

vitro assay of a partially purified K E X'2 en.lyml' demonstrated Ihat \lIch

site (ProArg) in synthetic peptides (an be cleaved (Mi/.Ltno cl al, 1<)Xl)

The KEX 1 gene has recently been cJoned, scquenced and charactefl/cd

(Dmochowska et al., 1987). It i~ a saine cm boxypl'pwla<;e wlth

homology to the yeast vacuolar protcase, carhoxypeptlda ... c Y

(Dmochowska et al, 1987; Cooper and Bu\~cy, 19H9). The exact cleavage

site within the preprotoxin for K EX / product wa<; rccently revealcd hy

the determination of the C-termini of the secreted toxin (o.,ee chapler 2).

(

8

1.4 Toxin action

The KI to)<.in IS active and stable within a narrow pH range from

4.2-4.7, whlch is close to its pl value; higher pH causes inactivation of

the toxin. The toxin is also heat labile, but this instability can be

partially overcome by addition of 15% glycerol, therefore, the optimum

as say mg condition for KI toxin is al ways carried out at pH 4.7,

temperature 18-22oC (Bussey, 1981).

By combining genetics, physiology. and biochemistry, it has been

demonstrated that the to"in action on sensitive cells involves a series of

surface reactions including; binding to a cell wall receptor. and

disruptmg the normal state of electrochemical ion gradients across the

plasma membrane with subsequent cell death.

1.4.1 Cell wall receptor for the KI toxin

The killer toxin acts first by binding to a receptor on the yeast

cell wall. Evidence for this initially came from studying the toxin­

resistant mutants, kre 1 and kre 2, These mutants are resistant to toxin

and have il reduced level of toxin binding to celIs relative to wild type

(Russey et al, 1973b; AI-Aidroos and Bussey, 1978). Using purified 35S­

labclled toxin binding to sensitive cells it was shown that this binding is

energy independent, and rapid, being complete within three minutes at

200C, pH 4.7 (Bussey et al., 1979). It was also determined that there are

1.1 x 1 () 7 receptors/per sensitive cell with an association constant of

2.9 xl 0- 6 M, such receptors are missing from kre 1 mutant cells.

Later, it was l'ound that zymolyase which contaills an endo-

(1 ~ 3 )-P-D-glucanase, solub;ti'led cell wall extracts l'rom sensitive cells

which could competitively inhibit toxin action, whereas, such

..

solubilized wall extracts from the kre 1 mutant could not (llutrhins and

Bussey, 1983). Binding of toxin to a trypsin-rcsist:'nt, periodatr­

sensitive ccli wall component implicated a glucan in sensitive œlls.

Further biochemical studies by polysaccharide competition, ccII wall

fractionation, and enzymatic digestion identificd (146)-~-D-glucan as

the toxin cell wall receptor. In addition, this (l ~ 6 )-~-D-glucan Irvcl in

the kre 1 mutant strain was reduced approximately 50% comparcd with

a sensitive wild type strain. Furthermore, the active toxin will

selectivdy bind to a Sepharose column that has a (146)-I3-D-glucan

analog such as pustulan attached to il. This toxin hinding to pustulan

occurs only .:1 acidic pH (pH 4.7). When the pH is raiscd to 7.6, the toxin

is eluted. Affinity purification of the active KI killer toxin is hascd on

such a proper!y (Hutchins and Bussey, 1983).

Recently, the c1oning, sequencing and rnolecular analyzing of the

KR E genes (Boone et al., 1990; Meaden et aL, 1990) has provided insighl

into the synthesis and assembly of the (1-4 6 )-~-D-glucan. The K N l~' J

gene encodes a 32 kd serine, threonine-rkh secrèted protein neccssary

for the addition of 1-6-~-linked outer chains to a mixed 1 inkcd core I~­

glucan structure (Boone et al., 1990). Null mutants of KR El are toxin

resistant and have a reduced level of (1-46 )-P-D-glucan. Analy~ls of the

residual (1~6)-P-D-glucan in the krel mutant by linkage as~ay, C l1

NMR spectroscopy, and molecular sizing indicates that the mutant

(1-4 6 )-~-D-gi ucan is of iower molecular weight with tewcr 1-6-p -1 in kcd

residues than that from the wi Id type structure. ThIS altercd ... tructurc

of (1-4 6 )-~-D-glucan in the kre 1 mutant expl'.lIlcd why, dc'>pite the

presence of (1-4 6)-~-D- glucan, it is still toxin re~i~tant. Thc kre 1

mutant lacks the effective functional cell wall rcccptor for the KI toxin

(

(

10

(Boone et al., 1990). The K RES gene product is a large, hydrophilic,

secretory glycoprotein which contains the COOH-terminal endoplasmic

reticulum retcntion signal, HDEL, implicating a possible localization of

the KRE5 gene product in the ER. Since nuit mutants of KRE5 are

completely toxin resistant and have no-detectable levels of alkali­

insoluble (1 ~ 6)-I3-D-glucan in the cell wall, it was suggested that the

KRE5 gene product functions in the early step of the assembly of the

(1~6)-I3-D-glucan (Meaden et a1., 1990).

1.4.2 Toxin action at the plasma membrane

The ccII wall receptor is necessary for toxin action, but it does not

appear to be the only component in the killing process. Several lines of

evidence suggest that a second step, probably on the plasma

membrane, is invol ved. ft was noticed that in a sensitive strain, (S 14a),

although the abundant (1.1xl07 per ccli) wall recet'tor appeared

necessary for tox in action, as few as 2.8x 1 ()4 toxin molecules were

necessary ta kill a sensitive ccII of S 14a (Bussey et al., 1979). Thus, it

appears as if sorne other component(s) is limiting the lethal process.

When resistant mutant cells kre 1 and 2, were converted ioto

spheroplasts. they became sensitive ta toxin, suggesting the existence of

the second step (AI-Aildroos and Bussey, 1978).

The third piece of evidence came from a mutant (kre3) which

bound toxin normally to the wall receptor but was toxin resistant (AI­

Aidroos and Bussey. 1978). The KRE3 gene product may fuaction as the

second receptor for killer toxin.

In addition, p~ysiological studies of toxin action suggested that

the second step (or the final target) is at the plasma membrane level.

•

t t

Initially, it was noticed that sensitive cells in the exponcntlat pha~l' of

growth were most sensitive to the toxin. Howcver, only 5WÏt' morlalily

could be obtained after 40 mInutes from the addition of 10\111 mea'\lIled

by ability to fmm (;ulufti~s on agar-contaHling medium (dl..' la Pella.

1980). MaÀimum killing was attained only al' ter two to thrce hour'\ of

toxin exposure depending on the strain (Bussey, t 972; Skipper and

Bussey, 1977). The toxin trea(ed cells shrank in volume, hut dIt! not

lyse (Bussey, i 974). Measurements of metaholtc and lllacIOlllolecular

biosynthesis after toxin addition showed that after a lag perlOt!. DNJ\

and protein synthesis were shut off (Bussey and Sherman, 1(73) and

coincided Nith plasma membrane damage measured r.y loss of K + ion~

and ATP (Skipper ana Bussey, 1977). Later, more Immediate cffccts

after toxin addition were addresscd by the de la Pena group (19XO,

1981). It was found that amino acid uptake by glucose l'cd ccll~, as weil

as, proton pumping to the medium were inhibited shortly aftcr toxin

addition. Such inhibition coincided with that of the uptaj..c of pota'\siul1l

ions which are thought to be accumulated by yea'\t cells in ordcr to

neutralize the membrane poter.tial created becausc of the extru~i()n of

protons (de La Pena, 1981). In addition, it was found that the tOXIfl

acidified the cell contents and reduced the proton graJlcnt acrm,..,

plasma membrane resulting in K+ efflux. Thercforc, il wa~ ... uggc~tcd

that the toxin III sorne way perturbed an energl7eJ mcmhrane "tate,

but whether it acted to inhibit sorne componcnt of the proton l'lump Pf

more directly by forming an ion channel wal.l nol clear.

Work on another killer toxin from yeast Pic/lia kluyvl'!ï wh 1 c h

has similar physiological properties on "cnl.lItivc S ('ereVI.\llle ccl'" a.., the

KI toxin (Middelbeek et al., 1980) demonl.ltrated thal thl.., loxin i..,

1 2

able to form ion channels in vitro In a black-lipid ùilayer system

(Kagan, 1983). The~e channels showed the existence of two conductance

~tates of 140 and 220 pS. They were relatively non-selective for

common physiological catiolls and anions, and were weakly voltage­

dcpcndent. Kagan suggcsted ttltl t these channels were probably the

basi" of kllling by Pu:hia kluyveri toxin. Howe'. er, direct evidence for

the formation of ion channels by the KI to,c in from S. cerevisiae had not

becn obtailled.

The sccreteJ toxin exists as a disulfide bond linked a-p di mer

with pl of 4.34 which is soluble in aqueous buffers. Studies of the

active toxin protein in a native gel system indicated that the basic (l- P

dimer is capable of f0rming multimers (up to octamers can be seen ),

though whether these are necessary for toxin action is unknown

(Bussey et aL, 1988).

1.4.3 Functional domain assignment of KI toxin

Primary structural analysis of KI toxin has revealed that the

secreted toxin contains ex. and 13 subunits linked by disulfide bonds. The

(l subunit con tains two highly hydrophobie regions (from residues 72-

91 and 112-123). Each of these regioils have a high potential for being

membrane-spanning and are separated by a hydrophilic region

(Bostian et al., 1984). It was postulated that the region From isoleucine

72 to threonine 123 may form the transmembrane cation channel, with

the central carboxylic acid residues acting to promote the transport of

protons across the membran~ (Bos tian et al., 1984). The p subunit,

howcvcr, lacks uny obvious membrane-spanning regions, its most

hydrophobie segment has a hydrophobicity of only 0.9, inconsistent

13

with it being an integral membrane protein. The killer to\in may

resemble the abrin and ricin class of toxins in which reL'eptot hllldlllg

and toxic domains reside on separate. disul fide-bonded polYPL'pt "Il'

chains (Olsnes, and phil. 1973). If SO, then ~ should ha\'e all',llly l'nt the

(l-t6)-()-D-glucan wall receptor and the (1 SUbUlllt ~hould IIltl'l.ll·t \Vith

the membrane. Experimental evidencc ta support the abovl'

sp.;~ulations are very limited. Mutations in the toxin gcne oltell !cd to

failure to secrete signific;ant levels of toxin, making analysis or tOXIll

phenotypes difficult (Boone et al., 19R6; Sturl.;y et .:1., 19X6). j\

mutation, PTXI-234, which changcd the first two I.lmillo aCltb of the I}

subunit from TyrVal to AspPro, led to a mutant toxin which 'oult! not

kili who le ceUs but which killed spheroplasts (Sturlcy ct al, 1 ()X(). This

result is consistent with a rol~ of () in binding to a ccII wall rl'CL'plor

(Sturley et al., 1986).

1.5 Immunity

Genetic evidence c1early demonstrates that M I-dsRN A is the

determinant of both toxin production and immunity. First, CUI ing ot the

Ml-dsRNA results in the loss of both tOXtn production and tox III

immunity (Fink and Styles, 1972; Mitchell et al, 1973. Wickner. IlJ74).

Second, neutral mutants of MI-dsRN A (phenotypically K -1{ +). are

defective in active toxin production, but thcy rctain imrnunity (Solllcr

and Bevan, 1969; Bussey et al., 1982). Third, ~uicidal mutanh of Ml

dsRNA (K+R -) are defective in conferring tmmunity hut ai,' ahle to

produce active toxin (Bussey et al., 1982). The IInmunity i', al<.,o VIIIOI\

specifie, for example, the KI immunity ~y ... tem i~ not immune to the K2

produced toxin system and vice versa.

(

14

The cDNA clone of the toxin precursor gene expressed from the

ADH1 promoter in yeast can confer immunity, indicating that the

immunity is caused in sorne way by this precursor molecule or a

component of it (Lolle et al., 1984 ). From examination of the primary

structure of the toxin precursor, it was speculated that the interstitial y

glycoslyated peptide was a possible candidate for the immunity

component (Bostian et al, 1984). To test this and alternative models of

immunity action, mutations were generated throughout the precursor

gene to localize the region coding for the immunity domain (Boone et al,

1986; Sturley et al, 1986). Such mutation analysis localized the

immunity domain to the a. subunit extending through the C-terminal

half of the (l into the N-terminal part of the y glycopeptide. Hence, the

idea that the y glycosylated peptide alone was the immunity component

was ruled out.

The fact that neutral mutants in which the precursor is

synthesized but no toxin is secreted, retained immunity (Boone et al,

1986, Bussey et al, 1982, 1983) suggested that the toxin precursor

itsclf cou Id confer immunity, or at tcast there is no requirement for

precursor processing for immunity action. How the immunity is actually

conferred to the host cells to prevent its own toxin killing remains a

mystery. Several models have been proposed based on very limited

evidencc. One suggests that the immunity determinant precursor acts

as a competitive inhibitor of toxin action by binding to a receptor

(K R E3) via the same site as the a. peptide of the mature toxin (Boone et

aL, 1986). In growing cells, the precursor/receptor immunity complex

forms during the progtession along the secretory pathway following

synthesis of these molecules. so that the receptor would be occupied by

, '1

. .

1 5

a precursor prior to active toxin processing late in the pathway. In this

way, no unoccupied receptors would reach the plasma membrane to he

available to interact with mature toxin in the growth medium (Boone et

al., 1986). The key of this model is a receptor common to both the

precursor and mature toxin. One candidate is the KR E3 gcnc producl. Il

was noticed earlier that toxin bound to the kre3 mutant ccII wall

receptor as weU as wild type, but faHed to kill kre3 spheroplasts (AI­

Aidroos and Bussey, 1978).

An alternative model was proposed by Sturley et al (1986). It

was based on the observation that a p22 product could be detected III

the ceU extracts when the toxm was overexpressed. The molccular

weight of the P22 species corresponded in size to a fragment that

resuIts when KEX2 cleaved the preprotoxin at residue 188 in the y

region, instead of at residue 234. Under this circumstance, P22 may no

longer be a substrate for cleavage between a and y. According to

Sturley et al (1986), this P22 product which contains a leader, <X sub­

unit and a small fraction of 'Y peptide could be the immunity com­

ponent competitively binding to the membrane receptor with secreted

toxin. These authors also suggested that the REX J gene product may he

the candidate for an enzyme cleaving protoxin at residue 188, or clse­

where in 'Y producing the P22 peptide, since rex J mutants express im-

munit y poorly (Wickner and Leibowitz, 1976; Sturley el aL, 1986). Thi'\

model, in fact, can be further tested by investigating the effeet of k ex

and rex mutations on the formation of P22, by further charactcrization

of the structure and location of this component and by dctermining the

effects on immunity of mutations of the LysArg sequence al 188. But '\0

far, no compelling experimental resuIts have becn obtaincd.

1

1.6 Toxins from Prokaryotic and Eukaryotic organisms

1.6.1 Colicins

16

Colicins are secreted toxic proteins produced by Escherichia coli

and c10sely related bacteria (Reeves, 1965). They are usually single

large polypeptides (40-60 kd) encoded by plasmids that also eJlcode

the immunity components. This immunity component provides the

toxin producing strain with resistance to their own toxins.

According to their mode of action, colicins can be divided into two

classes: 1. Colicins that kilt the cells by enzymatic cleavage of DNA or

16s ribosomal RN A; representatives of this family of proteins are

colicins E2 and E3. 2. Colicins that kill the cells by de-energizing the

cytoplasmic membrane. Colicins El, A, B, la, lb and K are in this class

(Konisky et al, 1982; Pugsley ,1984a, b).

Here, the characteristics of the second class of colicins; Eland A,

will be discussed, since their killing mechanism at the physiologically

level resemhles to that of the Kl toxin from S.cerevisiae.

It has been demonstrated that colicins Eland A kill sensitive E.

coli cells in three steps: 1) binding to a specific receptor located on the

outer membrane; 2) translocation across the outer membrane; 3)

Binding to and de-energizing the cytoplasmic membrane , probably by

forming ion channels causing cell death. Combining various approaches

including mutational analysis, hybrid protein constructions and

proteolytic c1eavage of the colicin polypeptides (Baty et al, 1988;

Cramer et al, 1983; Lazdunski et al, 1988), it has been shown that the

three defined steps in colicin action are associated with three distinct

domains on a single polypeptide chain. The ion channel formation

l

1 7

domain is localized at the C-terminal end of the molecule. The oull'I

membrane receptor binding domain is in the central region. l'hl' N­

terminal end of the molecule is involved in the interaction \\'ith the

translocation machinery.

The outer membrane receptor for colicill El and A has heen

identified to be encoded by the Btll B locus. This locus also fUllctions 111

the uptake of vitamin B 12 (Cavard et al, 1981; Cramer ct al.. 1983;

Sabet and Schnaitman, 1971). Following the binding to the outer

membrane receptor, colicin Eland A are translocated through the outer

membrane to the cell membrane, the~e events arc required hy tlw

tolQRAB gene cluster pathway. The gene cluster (tolQR AB) codes for

four proteins involved in the entry of colicin El, A and single-stranded

DNA of infecting filamentous bacteriophages into E. col! (SUIl and

Webster, 1986,1987). It was noticed among colicin El, A and th~ ssDNA

bacterophages that they ail contain a glycine and prol ine-rich rcgion at

the amino terminus. These similarities in primary structure suggest

that this region may be involved in interacting with some products of

the tolQRAB gene cluster (Pugsley, 1987, Ohno-Iwashita and ImahoTl,

1982).

From studying the in vitro colicins-planar 1 ipid bi layer and "pitt

vesicle systems (Cramer et al., 1983; Davidson et al., 1984a, h; 1 985;

Frenette et al., 1989, Pattus et al., 1983; Xu et aL, 1988; Bullock ct al.,

1983; Raymond et al., 1985 ), it has becn demomtratcd that the colicin ...

bind to negatively-charged lipids and acidic pHs catalyze thi~ hilltllng

process. Following binding to the membrane surface, i n~erti()n of the

protein into the hydrophobie core of the bilaycr will takc place fOrll1lng

an ion-channel.

1 8

Colicin Eland A can form voltage-dependent ion channels with a

single channel conductance of 15-20 pS in 1 M NaCI at neutral pH and

poor selcctivity between anion and cations. Vesicle permeation studies

as weil as selectivity measurements in planar lipid bilayers revealed

that molecules 1 ike sucrose and NAD are able to penetrate membranes

containing colicin produced ion-channels. The colicin ion channel must

thus have a channel size diameter of 8-10 A (Raymond et al., 1985;

Slatin, 1988) .

Proteolytic digestion of the colicin molecules can release a peptide

of approximately 20 kd from the C-terminus of colicins. This peptide

itself can form ion channels in a planar lipid bilayer (Davidson et al.,

1984b; Martinez et al., 1983 Cleveland et al., 1983; Bullock et al., 1983).

ft is interesting to note that this peptide is soluble in aqueous solution

while also being able to insert into the membrane forming ion channels.

This characteristic is consistent with its primary structure, which is

found to contain several (6) membrane spanning amphipathic helices

(Pattus et al., 1985). Detailed mutational analysis of colicin A

demonstrated that only three helices can span the lipid bilayer, white

the others belong to the mouth of the pore ( Baty et al., 1987). This

amphipatic structure must undergo a large conformational change from

soluble phase to lipid bilayer, as weB as when the ion channel opens

and closes (Raymond et al., 1986: Statin, 1988).

The data for determination of the molecularity of the colicin

channel are still conflicting. Rased on conductance measurements, the

diameter of the pores is about 8A (Raymond et al., 1985) and requîres

6 a-helices to form such a pore. Il has been suggested that the pore

may be formed by a di mer or a trimer of colicins (Pattus et al., 1985).

1

1 9

There is experimental evidence however. against the ahove proposaI.

Both in vivo and in vitro evidence suggest a molecularity of 1 for colicin

El and A (Cramer and Philips. 1970; Schcin ct al.. 1l)7~; Bruggl'Ill;\Il and

Kayalar, 1986).

Although there lS plenty of in vitro evidcncc to suggcst thal

colicin El and A are able to fonn ion channels. rcal in \'1\'0 cvidl'ncl' IS

still missing . Physiologic • .tlly, analysis of intoxicatcd cells by collcin El

showed that the inner membrane potential gradient hall collapscd

coincident with a very large K+ eff! ux (Feingold. 1970; Gould and

Cramer, 1977). This effeet cou Id be the direct cause of ion channel

formation by colicin El and A.

It is a common trait that toxin produeing cells in nature aIl have a

special mechanism to protect themselves from being Idllccl by Iheir

own toxins. Colicin producing strains have a similar il11l11unity

phenomenon to that observed from yeast ki 11er system, in tcrms of

specificity. For example, a colicin El proclucing strain is only immune 10

colicin El, but sensitive to other eolicins which have identical modes of

action (Bishop et al., 1985). Sorne of the immunity protcins have heen

identified, for ex ample, the immunity protein to colicin la and rh

(Weaver et al., 1981; Mankovich et al., 1986), to colicin El «(Joldman et

al., 1985 ). and to colicin A (Lloubes et al., 1984; Geli ct al., 19X6). They

are cytoplasmie membrane spanning protcins. So the immunity i~

probably operated at the cytoplasmic membrane level. Since the C()()II­

terminal domains of sorne of the eolicins are identificd to interact wlth

the immunity protein (Bishop et al.. 1985; Mankovich ct al., 1 ()X6), It 1\

suggested that the immunity IS obtained through forming colicll1\­

immunity protdn complexs. Thcse complexc\) then prcvcnt the colicin\

1

20

from forming functional channels. However, no direct evidence show

the existence of a coIicin-immunity protein complex either in vitro or in

vivo.

1.6.2 A-B type toxins

In nature, there is a distinct group of protein toxins produced

from bactcria to plants, I\uch as diphtheria toxin from Corynebllcteriunl

diphthertae and ricin, abrin from the seeds of Ricinus commwzÎs and

Ahrus precatorills respectively. These toxins usually consist of two

disulfide bond linked dissimilar peptide chains (A and B) carrying

different functions. Since sorne of these toxins are the cause of human

diseases, many studies have been carried out and a body of knowledge

has been aeeumulated on their structure and funetion.

Diphtheria toxin is one of the best charaeterized of this group of

toxins. Il consists of a longer B chain and a shorter A chain, linked by a

disulfide bond. Il kills sensitive cells by (1). binding to a cell surface

reeeptor which is probably a glycoprotein (Praia et al., 1979); (2).

translocating the A chain through the cell surface and reaching ilS

target; (3). catalyzing the ADP-ribosylation of elongation factor 2 (EF-2),

and shutting off protein synthesis. Molecular analysis of diphtheria

tox in has localized the receptor binding and translocation domains to

the B chain (Ittelson and Gill, 1973; Zanen et al., 1976), and enzymatic

activity is within the A chain (Chung and Collier, 1977; Oppenheimer

and Bodley. 1981).

When diphtheria toxin kills cells, reduction of the disulfide bond

is required (Collier and Kandel, 1971; Gill and Dinius, 1971). However,

purified A chain itself can not kill sensitive cells. AIthough it is

~ 1

enzymieally active, the A chain can not hind to the reccptors and thus

can not enter the cytoplasm of the target cclI. lIow this to .... in IS

translocated From the cell surface to its cytoplasmic target IS still not

known. There lS evidence that a considerahle amounts of li Iphthcria

toxin can be internalized by receptor-mediated endocyll)sis (Lcppla l'I

al., 1980), but it is not knawn whether this mechanism IS nel'essary 1'01

the intoxication process. It is reported that low pH can facilitatc entry

of surface-bound diphtheria toxin directly through the plasma

membrane (Sandvig and Olsnes, 1980). There is also evidencc which

shows that the amino terminus of the B chain (Kagan et al., 19X 1) and

whole diphtheria toxin (Donovan et al., 1981) l'orm channds in IIpid

bilayers. Therefore, it IS proposed that thesc channels provide the

pathway for the A chain ta across membranes.

Structurally and mechanistically, ricin and abrin are similar 10

diphtheria toxin (Van Heyningen, 1982).

The distinct domain lacalization of this group of toxins, 111 which

the A chain carries the toxicity and the B chain is responsiblc for

binding ta the cell surface receptor and for translocation of the A chai Il

entry the cell, has led to the development of synthctic imlllunol()xin~.

Immunotoxins (hybrid toxins ) usually consist of the potcncy of the

protein toxin A chain linked with specifie ligands (such as monoclonal

antibodies ) to help direct the hybrid toxin to the <;pcciflc targel ccll~.

Much progress has been achieved in using thcsc immunotoxill"i in the

treatment of certain diseases though none are complctcly sati~·Jactory

(Neville, 1986).

1.6.3 Other toxins From yeasts

(

,

22

The discovery of the killer phenomenon In S .cerevisiae b y

Makower and Bevan (1963) led subsequently to the search for other

killing actlvitics among different yeast species. Since then, killer

strains which produce different toxie activities have been found in the

ycast gcnera Candida, Cryptococcus, Debaryomyces, H ansenula,

Kluyveromyces, Plchia and Torulopsis (Philliskirk and Young, 1975;

Stumm ct al, 1977). By examination of the cross-killing reactivities

hetween strains, at least Il distinct killer activities have been detected

and the y probably represent Il biochemically different toxins (Young

and Yagiu, 1978: Rogers and Bevan, 1978) .

The biochemical properties known for a few purified killer toxins

show that they are either proteins or glycoproteins, such as the K2, and

KT28 toxins from S. cerevisiae (Pfeiffer and Radier, 1984; 1982) and

toxins from Pichia kluyveri (Middelbeek et al., 1979), Hansenula mrakii

(Yamamoto, et al., 1986) and Kluyveromyces lactis. Apan from the

toxin from H. mrakii which is thermostable and pH stable, other toxins

are very unstable and usually have a very narrow pH range from pH

4.0-5.0 for killing activity.

The genetic basis of most of these toxins are not known. So far,

the identified genes which encode yeast killer toxins are aIl

cytoplasmically inherited. K2, KT28 toxins from S. cerevisiae and toxins

(KPl, KP4, KP6) from Ustilago mtJydis, a fungal pathogen of maize, are

encoded by dsRN As encapisidated in virus like particles, a system

resembling that of KI toxin from S.cerevisiae. Toxin from K. lactis is

dependent on two linear DNA plasmids kt and k2 (Gunge et al., 1981).

It is now known the kl DNA plasmid codes for the toxin subunits (Niwa

ct al.. 1981; Wesolowski et al., 1982; Stark and Boyd, 1986).

1

•

23

Recently. the cioning and sequencing of sorne of thcsc hl\ins. ha"

generated more information on their structural organi7ati on

The secreted K2 toxin consists of two polypeptides \\'ith 1ll0lCl'lIIal

weights of 21 kd and 9 kd separatcly. Thcse two polypeptidL'~ app\..'al

to be glycosylated and do not appear to he associated hy dlsul! Hk

bonds. The primary nucleic acid sequence show~ no identlly wllh thal

coding for the sequence of the KI toxin, and this non-idcntity cxtcnds

to the protein level (Whiteway, M. unpuhlished rcsults).

The KP6 toxin is one of the secreted toxins from a fllngal

pathogen of maize, Ustilago maydis. It 1S encoded hy a dsRNA gcnol1lc

called P6M2. KP6 toxin consists of two small subunits (a and 13 )

processed from a preprotoxin in a similar way to thal of the KI toxin

from S.cerevisiae. At the primary structural level, no sigmlïcant

homology of KP6 toxin to other known toxins has been round (Tao ct al..

1990).

The killer toxin from K. lactis consists of three polypeptides (u, 13

and y). It is found th3.t the toxin activity is associated wi th the Huec

subunits since exposure of the toxin to ~-mercaptoethanol Icads to

dissociation the three subunits and destroys toxin activity (Stark and

Boyd, 1986).

In keeping with the scarce information on the under~tandHlg of

the structures of these fungal toxins, only a little is known abOUt lheir

mechanism of killing. It has been suggested that the K2 toxin actIon al

the cell wall and physiological levels are very similar to that 01 the KI

toxin (D. Rogers Ph.D Thesis, University of London, ItJ76). The

glycoprotein toxin l'rom Pichia klyverit i~ '\hown to bc ahlc to lorm Ion

channels in a lipid bilayer system (Kagan, 1983), Implicating channel

24

formation in the killing mechanism for this toxin. The KT28 toxin from

S.cerevisiae is found primarily to bind to the mannoprotein of the cell

walJ of sensitive yeasts (Schmitt and RadIer, 1987). Mutants mnn l,

mnn2 and mnnS which cause large defects in the mannoprotein, were

resistant to KT28 toxin (Schmitt and Radier, 1988). This toxin was

reported to inhibit DNA synthesis. Wh ether this is the primary site of

action of thÎ1i toxin remains unknown (Schmitt et al, 1989). The mode of

action by toxins from K. lactis and U.maydis are still not known.

1.6.4 Rationale

In this thesis, 1 report studies on the yeast KI toxin: its structure,

processing, mode of action and structure-function relationship, and

describe results obtained using biochemical, electrophysical and

molecular genetic approaches.

1

,

1

1.

l

25

Chapter 2 Determination of the carho\.yl termini of the ex and ~~

subunits of yeast KI killer to\.in

Requiremcnt of a carboxypeptldasl' B-like activity for

maturation

2.1 Introduction

The mature, secreted KI toxin consists of ex and ~ subunits

processed from a precursor. which is crmposed of a signal peptide

foltowed by the a subunit. a glycosylatcd y peptide. and the I~ ~lIhllnlt

(Bastian et al., 1984). This structure has been assigned on the ba"'l~ of

DNA sequence analysis of the preprotoxin gene and NlI2-termlllal

sequence analysis of the purified a and ~ subunlto; and of the

synthesized in vitro preprotoxin (Bostian et al.. 1984).

Maturation of yeast a factor and killer toxtn display a high degree

of similarity (Tipper and Bostian, 1984). Both require several common

activities including the product of the KEX2 genc (Juliu<.; et al., 19H4).

The product of K EX2 gene has been idcntified as a dibask

endopeptidase that initially cleaves following pairs of bao;ic amlno acid ...

(Julius et al., 1984) and such sites are apparently operational 111 both u

f!lctor (Julius et al.,1984) and killer toxin maturatIOn (Bo<.;tJan ct al,

1984). A dipeptidyl aminopeptidase activity, providcd by the product

of the STE /3 gene (Julius et al., 1983), and a carboxypcpttdaw B-like

activity have been implicated in a factor maturation (Fuller el al.,

1985). Mutant analysis indicatcs that the product of K I~'X J genc play ... a

role in toxin maturation (Wickner and Leibowitz, 1976' Bu ...... ey cl al.,

1983); however, its site of action has not hecJ1 Idcntlflcd. 'l'wo pO'>'>lhle

sites for KEX / processing of toxin exist: the propcptidc-u Junction and

(

(

J

26

the a-y junction. We h~lve determined the COOH termini of the Cl and 13

subunits of toxin in order to furtht>r analyze the functional domains and

processing cvents. The COOH-terminal sequences of the two subunits

are reported here, and the sequence at the COOH terminus of the a

subunit implicates a process; r.g pathway which is more similN to a

factor maturation than r: èviously realized.

2.2 Materials and Methods

Purification of killer toxin

S. cerevisiae KI killer strain T158c/S14a was used for production

of killer toxin. Cells were cultured as indicated by Palfree and Bussey

(1979). The concentrated cell-free medium prepared by the method of

Palfree and Bussey (1979), in 50 mM acetate, 15% glycerol buffer, was

loaded onto and eluted from a Sepharose CL-6B gel filtration column

(l.6x75 cm) equilibrated with 50 mM sodium acetate buffer.

Fractions at elution volume 56-66 ml which contained the bulk of

the 280-nm absorbing material, were pooled into Spectra Por dialysis

tubing (6000-8000 molecular weight cutoff) and coated with Aquacide

II to remove water. The concentrated material was redissolved in 1 %

sodium dodecyl sulfate (SOS), 0.1 M Tris, pH 6.8, heated at 1000C for 3

min, and then passed through a second Sepharose CL-6B column (1.6x

75 cm) equilibrated with 0.1 % SOS, 0.1 M Tris buffer. This heat

treatment inactivates the toxin. Fractions (2.0 ml) were collected. The

purity of the killer toxin-rich fractions was estimated by sodium

dodecy! sulfate-polyacrylamide gel electrophoresis (SO~-PAGE) and the

matcrial eluted between 106 and 112 ml was pooled on this basis.

•

&

- -------------.-

27

Other methods

Reduction, alkylation, succinylation, and cyanogen bromide (CNBr)

cleavage were performed as described by Christie and Gagnon (1982).

The only modification to their methods was that the concentration of

protein in these reaction was much lowcr (20 ug/ml). R::duced.

alkylated, and succinylated killer toxin was subjected t,) mild acid

hydrûlysis in 70% formic acid at 370 C for 72 h (Landon, 1977).

Amino acid analysis

Samples of HPLC-purified peptides or dialyzed whole protein

were dried by speed vacuum centrifugation in c1eaned acid hydrolysis

tubes. New acid hydrolysis tubes were c1eaned by overnight pyrolysis

at 5500 C in a muffle furnace. Acid hydrolysis openttions were

performed using a Pico-Tag Work Station (Waters, Division of Millipore

) employing the following methods. Constdnt boiling HCI (Pierce

Chemical CO. ) containing phenol (1 % v/v) was added to the bottom of

the hydrolysis container; th en the container was purged three times

with nitrogen as prescribed by Waters (Waters, Division of Millipore).

After the third purging, the container, at a reduced pressure, wa~

sealed and placed in the heating block, set at 1500 C, for the de~ircd

time. Values for cysteine, methionine, and tryptophan were derived

from the DNA sequence (Bostian et al., 1984; Skipper et al., 1(84).

Analyses were performed on a Beckman System 6300 Iligh

Performance Analyzer/ Model 7000 Data Station according to the

general procedures of Spackman et al. (1985).

28

Sequence analysis

Automated Edman degradation of peptides and reduced and

alkylated toxin were conducted on a Model 410A Gas Phase Sequencer

with an on-line Model 120A Phenylthiohydantoin Analyzer (Applied

Biosystems Inc., ABI) using the general protocol of Hewick et al. (1981).

Samples wele appli~d to precycled filters containing 1.5mg of

polybrene plus 0.1 mg of NaCI (Biobrene Plus). Standard precycling of

the polybrene and sequencing of samples employed the 03 RPRE and

03RPTH programs (ABI), respectively.

2.3 Results

Purification of toxjn

Toxin was purified using a two-step gel filtration procedure. The

first gel filtration, under non-denaturing conditions, gave relatively

impure toxin which was further purified by a second gel filtration

column under denaturing conditions. Purified inactive toxin was

identified by SDS-PAGE (Fig.1) and confirmed to be authentic toxin by

NH2-terminal sequence analysis. Sequence analysis of purified toxin

(data not shown) confirmed the presence of two polypeptide chains of

approximately equimolar concentrations. The two sequences agreed

with the predicted gene sequence and the previously determined NH2

termini of the secreted toxin subunits (Bostian et al., 1984). Purity of

the toxin was estimated from the SOS-PAGE (Fig.l), and the ammo acid

composition (Table 1) is consistent with that predicted from the DNA

sequence afler assignment of the COOH-terminal residues (see belo\\) of

1 the a. and J3 subunits. The recovery of purified inactive toxin was

approximately Img per 19 lIlers of cell growth medium.

29

Purification and determination of the COOH-terminal s~quenccs of the (X

and J3 subunits of killer toxin

The initial strategy for COOH-terminal sequence detcrmination of

the a. and J3 subunits was to isolate short peptide fragments, including

the COOH termini. These fragments would then be puri ficd by

reversed-phase chromatography and characterized by sequence and

amino acid analysis.

Based on the protein sequence predicted from the nuclcic acid

sequence, the CNBr-induced cleavage reaction at methionine residues

(Gross and Witkop, 1962) was employed to specifically cieavc the r~

subunit into peptide fragments which would meet the above criteria. In

the process of obtaining the purified COOH-terminal CNBr fragment of

the J3 subunit the CNBr digest was first fractionated (data not showll) hy

gel filtration on a Superose 12 (Pharmacia P-L Biochcmicals) column

employing a fast protein liquid chromatography system, (Pharmacia p­

L Biochemicals). The column was developed in 70% formie acid at a t low

rate of 0.5 ml/min, monitored continuously at 280 nm, and fract i()n~

were collected manually. Several of the pools were subjcctcd to

sequence analysis in order to identify which contained the anticipatcd

COOH-terminal CNBr peptide starting with the sequence Lys-Phc-Ilc

(Lys-296 of preprotoxin, Fig.3). Several sequences, corre~pondlng to

peptide bind cleavage at methionine residues 238, 249, 255, and 195 of

preprotoxin, and the NH2-terminal sequence of the r3 ~uhunJl could he

ictentified from the data for one pool. The~e re'iult~ wcrc con~i\tcnt

30

Fig. 1. Purification of KI toxin.

SOS-PAGE on an 8-18% polyacrylamide gradient gel according to

the method described by Laemmli (1970). Lane a, purified toxin

obtained by gel filtration on Sepharose CL-6B under denaturing

conditions. Lane b, concentrate from the cell-free medium.

1

f ~,

1

a b •

• '

J;;" ~, . "' .. "' . •

'.

« 3 1

with the COOH terminus of the ~ subunit extending at least to Gly-315

of preprotoxin. Further purification of the components in this pool (Fig.

2) rcsulted in the isolation of a peptide with the amino acid composition

(Tablel) predicted by the DNA sequence corresponding to the CNBr

peptide from Lys-296 to His-316 of preprotoxin (Fig. 3). Failure to

detect the COOH-terrninal histidine residue of rhis peptide during

sequcnce analysis was probably a result of a combination of the

following reasons: 1) low recovery of charged residues, particularly

histidine, from the Ga!' Phase Sequencer (ABI User Bulletin 12, dated

Aug. 15, 1985); 2) potential washout of short peptides during sequence

analysis (Klapper et al., 1978; Tarr et al., 1978); 3) the low quantity

(about 50 pmoI) of the COOH-terminal peptide initially coupled.

The initial amino acid composition data (Table 1) of the purified

toxin, after taking into consideration the length of the ~ subunit (see

above ), suggests that the a subunit extended further than Trp-130 (as

previously proposed by Bostian et al.(1984) of preprotoxin.

Consequently, the Me2S0/HCl CNBr cleavage reaction (Huang et al.,

1983), which has a high degree of specificity to cleave peptide bonds at

the COOH-terminal position of tryptophan residues, was employed. The

cleavage mixture generated from KI toxin by treatment with CNBr In

MC2S0/HCl was subjected to sequence analysis. The data revealed a

sequence consistent with clcavage at Trp-130 of pTeprotoxin~ the

fragment rrobably extending 6 or 7 residues beyond thc Asp-Pro

sequence located at positions 140-141 of preprotoxin. Based on the

above result, reduced, alkylated, and succinylated K 1 toxin was

subjected to mild acid hydrolysis (Landon, 1977) with the expectation

32

Fig. 2. HPLC purification of the COOH-terminal fragments.

A, isolation of the ~ subunit COOH-terminal peptide. The 13 suhllnit

COOH-terminal peptide-rich pool obtained by gel filtration on SlIperose

12 (see Results) was lyophilized, redissolved in 10% acetic acid. and

applied to a uBondapak C18 co1umn (4.6 x 250mm). The peptides were

eluted at a flow rate of 1 ml/min using a Varian Vista 550{) HPLC. The

peak (arrow) with retention time 27 min was shown by amino acid

analysis (Table 1) to correspond to the COOH-terminal peptide of the ~

subunit.

B, isolation of the a subunit COOH-teminal peptide. The reaction

mixture after mild acid hydrolysis (see Materials and Methods and

Results) of KI toxin was applied to a Hypersil OOS (C18) column (100 x

2.1mm). The peptides were eluted at a flow rate of 200ul/min using a

Hewlett-Packard 1090 Liquid Chromatograph. The peak (arrow) with

retention time 9.0 min was shown by sequence (data not shown) and

ami no acid analysis (Table 1) to correspond to the COOH-terminal

peptide of the a subunit. In both cases an acetonitrile gradient

(1 %/min) in the presence of trifluoroacetic acid was used. Solvent A:

0.1 % trifluoroacetic acid/ wateT. Solvent B: 0.1 % trifluoroacetic

acid/acetoni trile. The peptides were collected man u ail y.

1

•

1 02 A

01

~ 0 c

-01 l 8 ~ ~ 0 ~ ~ n ~ ~

Retention Tlme(mln)

B

0.4

0.3

8 0.2 N ~ 0.1 1

0

-0.1

10 20 30 50 60 Retention time (min)

1

33

of cleaving the Asp-Pro peptide bond (sec below). A peptide \Vas

isolated from the mild acid c\eavage mixture (Fig. 2) which bascd 011

sequence analysis (data not shown) and amino acid analy~is (Tahle 1)

corresponds to the COOH-terminal peptide of the a suhullit. This peptide

was generated by scission of the Val-136 Ser-IJ7 peptide bond of

preprotoxin and is Il residues in length, extcnding to Ala-147.

The sequence results for this II-residue peptide contïrmed the

presence of the aspartyl-prolyl pcptide bond which surprisingly \Vas

not the major acid cleavage site under the conditions employcd.

However, the sequence analysis data indicated that hydrolysis of this

aspartyl-prolyl peptide bond did occur during the automated Edman

degradations, consistent with the sensitivity of this bond (Brandt et al.,

1982; Piszkiewicz et al., 1970). The facts that a wide range 01 miltl acid

hydrolysis conditions have been employed for aspartylprolyl pcpt ide

bond cleavage (Landon, 1977) and that the inclusion of dcnaturants

during mild acid hydrolysis can increases the yicld of products

(Hermodson, 1982) are indications of the varied susceptibi lit y of

aspartyl-prolyl peptide bonds and probably reflections of stahle

secondary structures. Acid-catalyzed cleavage of valyl-seryl peptide

bonds have been documented (lwai and Ando, 1967); howcvcr.

different conditions (6N HCI, 210C) were used. These faets may

implicate a highly constrained structure for KI toxin under the mi kl

acid hydrolysis conditions employed. Furthcr to thc uncxpcctcd

insensitivity of the KI toxin aspartyl-prolyl peptide hond tn rnlld acid

hydrolysis, succinylation appeared to be a prerequi~ite lor qh ... crvcd

valyl-seryl peptide bond cleavage, possibly furthcr implicating

secondary structure constraints.

:1

':"able :: Amine aCld Canp:lSH:-cn cf pur1.fied Kl luller and :,.'1e C-:er:-.::-.al peptlœs frcm t:".e a a::d ~ si...bur..::sa .

:: ... - .:.:~ç, .!.:;::.à Kl K1.':'':'e~ :.:x..:... ,,":b C-:e~~al ?ept:des

Ca Ce As? J (13) (1) ( 3)

Asx 23.0 2.1 4.6 Asn :-J (12) (1) (3) nu- T 10.3c (8) 1.0 (1) 0.9 ( 1) Ser S 16.5c (15) 1.0 (2) 1.8 ( 1) Glu E (11) 3.9 (3)

Glx 18.0 1.4 Gln Q (5) (1) Pro P 4.9 (4) 1.0 (l) Gly G 19.2 (18) 1.7 (1) 3.5 (2) Ala A lS.7 (15) 2.1 (2) 1.6 (1) Val V 12.1d (13) 1.0 (1) Met M Se (5) Ile l 8.2<i (10) 1.4 (2) leu L 10.4 (11) 0.6 -Tyr y 7.0c (8) Phe F 5.5 (6) 1. 5 (2) His li 3.5 (4) 0.9 (1) Lys K 8.7 (10) 1.0 (1) Arq R 3.1 (3) -Cys C 6e (6) O.gh(l) Trp W ge (9)

fot:)lecular Weight 20 65at Total Res1dues (186) (11) (21)

a Values in parentheses were detennineà fran the tNA sequence. b Average ot analyses tor 2-, 3- and 4-h hydrolysates except where

ott.erwise now. e Determ1ned Dy extrap:;lation :0 zero-time ot hydrolys1s. d Value à:)tained tran 4-h hydrolysate. e Value à:)tained fran the !NA sequence. f Mil calculated tran the ~ sequence, assuning rrature prote1n coot.ains 3

èl.sd ~ide bo.~ës .. C] ca :..~d c~ ccr:-espc::d to 'C..":e ':-te!:':"!1.:!..~al pe;:t1.œs !:-an tr.e a ar.d ~

s-..:.o"::-.lts, res;:ect~vely. Sl.."l;:'e ar.alYS1S of a 2h hyèrolysate. h :e:e~.! .. .rled as =~bcxy.':'et..":yl cyste:::e.

Il 'i

.

1

1

35

Assignment of the COOH tennini of the ex and ~ subunits of KI

toxin allows absolute determination of the amino acid composition as

predicted by the nucleotide sequence (Bostian et al. 19H-+~ Sk.ippl'f (' (

al., 1984). The amino acid composition of purificd toxin rrable 1) is in

good agreement with the DNA predicted composition whcrc the (X and I~

subunits correspond to residues Glu-45 to Ala-147 and Tyr-234 tn lIis-

316 of the precursor, respectively. The a and ~ suhunits are prl'dictl'd

to be 103 (l1,122g/mol) and 83 (9543g/mol) amino acid rcsidues.

respectively. Mature toxin, containing three disulfide bonds (at Icast

one intersubunit), has a calculated molecular weight of 20.658.

2.4 Discussion

The KI killer toxin of S. cerevisiae lS a secreted heterodimeric

protein encoded by a cytoplasmically inherited Ml doublc-stranded

RNA viral genome which confers both the killer and immunity

phenotypes on yeast cells. The toxin/immunity gene encodes a

precursor protein containing an amino-terminal leader sequence

followed by the ex subunit of the secreted killer toxin, an interstitial

glycosylated y peptide, and the COOH-terminal ~ subunit of the toxin

(Bos tian et al.. 1984). Site-directed mutagenesis (Boone ct al., 1 <JX6) has

defined the immunity domain largely as the peptide segmcnt l'rom Val-

86 to Ala-147 of the precursor. Results in this papcr, aSl\ignmcnt of the

COOH-terminus of the ex subunit, indicate that the above dclïncd

immunity domain resides completely in tre ex subunit. Proce,~ing of the

precursor, however, is not a prerequisite for immunity, al\ evidenced hy

the fact that sorne mutated toxin genes, which gencratc propcrly

glycosylated precursors but fail to secrete or ~ecrete greatly reduccd

(

36

levels of toxin (Boone et al., 1986; Bussey et al., 1982, 1983; Lolle et al.,

1984; Wickner, 1974), retain full immunity. This is also true for the

kexl and kex2 processing mutants (Bussey et al., 1983; Wickner, 1974)

where synthesis of the precursor is not affected.

Maturation of preprotoxin requires several processing events,

including c1eavage by the signal peptiJase and the K EX 1 and K EX2 gene

products. The signal peptidase is believed to cleave between Ala-26

and Leu-27 of the precursor (Fig. 3, LoBe and Bussey, 1986), 18

residues upstream from the NH2 terminus of the ex. subunit of mature

toxin (Bostian et al., 1984). The processing enzyme involved in cleavage

at the junction of the propeptide and the ex. subunit has not been

identified, and this site has been proposed as a possible target for the

product of the KEXI gene (Bussey et al., 1983). Another potential KEX 1

processing site, at the a-y junction, was proposed, based on

chymotrypsin-like activity being involved in toxin maturation (Bussey

et al., 1983), to be the peptide bond between Trp-130 and Gly-I31 of

the precursor (Bostian et al., 1984). The involvement of a dibasic

endoprotease in toxin processing was realized upon identification of the

y-~ junction at the peptide bond following the pair of basic residues,

Lys-232 Arg-233 (Bostian et al., 1984). Consistent with mutant studies

(Bussey et al., 1983; Leibowitz and Wickner, 1976), the product of K EX2

gene, which has been identified as a dibasic endoprotease that cleaves

the peptide bond following pairs of basic residues (J uilius et al., 1984),

appears to fullfill this requirement in toxin maturation. There are

three other potential KEX2 sites in the toxin precursor: Arg-99 Lys-100,

Arg-148 Arg-149 and Ly~-187 Arg-188. Evidence presented here

suggests that the pair of basic residues at positions 148-149 is also

37

used as a processing site and that these rcsidues mark the a- j 1 \: ,JO

in the precursor. The requirement for a carboxypeptidase B-like

activity in toxin maturation thus becomes apparent. This

carboxypeptidase B-like activity may be provided by the product of the

KEXI gene.

The similarity between toxin maturation and ex. factor maturation

has been recognized (Tipper and Bostian, 1984). Processing of Cl factor

from either two genes , MF al or MF a2 (Kurjan and lIerskowitz, 19R2:

Singh et al., 1983) which contain 4 and 2 repeats of the mature

pheromone, respectively, apparently involves a dibasic endoprotease. a

dipeptidyl aminopeptidase, and , for aIl but the COOH-terminal copies of

the a factor in the precursors, a carboxypeptidase B-like activity. The

strict requirement of the di basic endoprotease and the dipeptidyl

aminopeptidase in ex. factor maturation has aided in the identification of

the genes responsible for these activities (Julius et al., 1983. 19R4);

however, the gene for the carboxypeptidase B-like activity in toxin

maturation suggests that the processing pathw.lY for KI toxin and a

factor are more similar than previously anticipated. The possibility that

the product of the KEXI gene is involved in a factor maturation has now

been tested, and its involvement has been confirmed. The gencral

scheme of processing events at pairs of internai basic residucs is

c1eavage of the proximal COOH- terminal peptide bond followcd hy

removal of the two basic residues by a carboxypeptidase B-like

protease ( Fuller et al.,1985; Steiner et al., 1980). If K EX 1 providc'i the

carboxypeptidase B-like activity required in a factor and toxin

processing, then the KEX 1 processing events must occur after the KfX2

lX

Fig. 3. Structure and processing of the yeast Ki killer preprotoxin.

The sequence of the COOH-terminal fragments (Ca and C~) of the ex

and f3 subunits of KI toxin, the location of these peptides within the

mature Cf. and r~ subunits, the location of the ex and ~ subunits within

prt;protoxi n and putati ve proteolytic processing sites wi thin

preprotoxin arc shown. LoBe and Bussey (1986) have proposed that the

signal peptide is removed by signal peptidase cleavage of the peptide

bonJ between alanine 26 and leucine 27 (AL). The enzyme that cleaves

the peptide bond (R44-E45) at the junction of the propeptide-ex subunit

has not been identified. Both o.--y and y-~ junctions are f1agged by a pair

of b~lSic residues (R 148-R 149 and K232-R233) which are potential

processillg sites for the K EX2 -encoded dibasic endoproteinase.

Subsequent to K EX2 processing and prior to maturation of the ex

subunit, the pair of basic residues are removed. The product of the

K EX 1 gene, purportedly a carboxypeptidase B-like enzyme, rnay be the

enzyme that removes the basic residues during toxin maturation. the

location of W (tryptophan) 130, the previously speculated COOH

terminus of the ex peptide (Bostian et al., 1984) is also shown. The single

Ictter ami no acid abbreviations (see Table 1) are used.

, ..

" , '-

- •

~ ~ ~ ~ ~ ~ ~ . ~ . ~ . Preprotoxin

" AL 'RE AR' J H l ,~ a ~ t ~ i so".,l ? f KEX2 \11 ~X2" 1

Peptidase KEXl (2) KEXl (2)

E Y~.l4 "1' Mature a-subunit Q Mature j3-subunit ~

(l

Mature toxin 1 1 i3

J

39

events, in contradiction to the provisional ordering of these mutants by

the precursor stability studies of Bussey et al. (1983).

The dipeptidyl aminopeptidase, the product of the STE 13 gene

involved in a factor processing, cleavage NH2-terminal Glu-Ala ( and

Asp-Ala-repeats (Julius et al., 1983). The first two residues of the

mature a subl1nit of toxin are Glu-Ala (Bostian et al., 1984). These

residues may serve a role during maturation; however, removal of this

dipepttde is probably hindered by the nature of the peptide bond

which wou Id be cleaved. The third residue in the Cl subunit is a proline

residue, and peptide bonds involving proline tend to be resistant to

many proteases.

1

> ..

40

Chapter 3 Yeast KI killer toxin forms ion-chanlll'is in

sensitive yeast spheroplasts and in Iipnsollll.'s

3.1 Introduction

Previous genetic and biochemical studies of KI killcr toxin action

on sensitive yeast cells revealed that it involves a set of speciflc ccll­

surface interactions. These include binding to a ccII wall-rcceptor.

which has been identified to c0ntain (l---76)-~-D-glucan (liutchins and

Bussey, 1983; Boone et al, 1990). Following wall binding, encrgy

dependent processes result in lethal physiological changes. Ion leakage

at the plasma membrane appears to be a primary cause for the

lethality. In rnetabolically active cells, a rapid inhibition of net proton

pumping from the cells, along with an inhibition of potassium and

amino acid uptake upon exposure to the killer toxin ha~ bccn rcportcd

(de La Pena et al, 1980,). Furthermore, intoxicated celIs showed a

reduced proton gradient across the membrane of yeast ccII with

acidification of ceU contents and potassium efflux (de La Pena et al,

1981). At later stages of killer toxin treatment, potassium ions, ATP and

smalt metabolites were lost from cells (Skipper and Bussey, 1(77).

Based on such data, it was suggested that the killer toxin perturhed an

energized membrane state. Wh ether it inhibited sorne component of Ihe

proton pump, or acted more directly by forming a protein channel,

remained unclear. The amphipathic character of killer toxin prolelfl

(Bostian et al, 1984) is consistent with the idca that thi" loxin dircctly

perturbs the cell membrane. Kagan (1983) showed that a killer loxin

from Pichia kluyveri, with physiological propertie~ similar to the KI

41

killcr toxin of S.cerevisiae, could form ion channels in vitro in a black­

lipid bilaycr system.

ln tiiis chapter, we report results of patch clamp experiments

which show that a fraction enriched in KI toxin forms ion-channels in

vivo in sensitive yeast spheroplasts and in vitro when the KI toxin 15

incorporatcd into artificial liposomes.

3.2 Materials and methods

Strains

The following strains of Saccharomyces cerevisiae were used: 1.

TI58C/S 14a MA Ta/MATa. his4C-864/HIS4 ade2-5/ADE2 (ATCC26427),

a KI killer strain containing the Ml virus. 2. T158C/SI4a. H.C., a

sensitive non-killer stTain derived from T158C/S 14a by heat curing of

the MI virus (Wickner, 1974; Bussey, et al., 1988). These strains were

used as sources for killer toxin preparations, and toxin fTee controis. 3.

AH22 MATa leu2.3,2.112 his 4.519 canI, a sensitive yeast strain, was

used as a toxin sensitive strain in studies designed to detect toxin­

induced channel activity by patch clamp examination of its

spheroplasts. AH22 was transformed with a killer expression plasmid

pL308 (Lolle et al., 1984), such transformants secreted KI killer toxin.

In addition AH22 was transformed with a control plasmid pL361 (Lolle

et al, 19R4), Vv'hich lacked the killer gene. 4. H4a MATa his3 ura3 was

transformed with a killer toxin mutation plasmid ZH20, in which Ile129

in the C-terminus of the a. subunit was substituted by Arg129.

Concentrated extracellular protein preparations from the se

1

J

transformants were used as sources of killer toxin or as toxin-free

controls.

Yeast concentrated culture filtrates and partially purificd toxin

preparations

Yeast cells were grown at 18()C in liquid millimal medium (pli -1..7)

containing 1 x Halvorson salt (Halvorson. 1958). 2C'/tl glucose and 1 WX,

glycerol. After the cultures entered the stationary phase of growth. the

cells were pelleted by centrifugation. The medium was removed and

concentrated from 200x -IOOOx by ultrafiltration using a PM 10

membrane (Amicon).

Active wild type killer toxin (from strain TI58C/S 14a ) was

partially purified by affinity chromatrography using pustulan. a wall

receptor (1 ~ 6)-Il-D-glucan analog coupled to Scpharosc as descrihcd

(Hutchins and Bussey,1983). Sorne 70% of the protcin 111 such

preparations was KI killer toxin, based on Coornassie Brillant blue

staining, following SDS polyacrylamide gel electrophoresis.

Toxin activity assay

Toxin activity was tested by a standard killing zone assay.

Sensitive yeast cells (AH22) were seeded into a YEPD agar plate. An

aliquot of concentrated toxin culture filtrate was then spottcd on top of

the plate. The inhibition of cell growth was shown by formation of a

clear zone around the spotted culture fiItrate (Bus'icy. 1981).

Yeast spheroplast preparation

43

Killcr sensitive strain, AH22, was grown at 300C in pH 4.7 YEPD

to the carly stage of logarithmic growth, and the affinity purified toxin

was added at approximately 5ug/ml of cell culture, which was allowed

to grow for another 15-20 min. The yeast cells were then subjeeted to

Yeast Iytie enzyme (73,400 U/g from IeN 8ioehemicals ) digestion at 80

ug/pcr ml in the solution of 1 M sorbitol, 0.1 mM Na2EDTA , 5 mM

HEPES (pH 7.2) for 20 min at 300 C (Gustin et al, ] 988). The yeast

spheroplasls were patch-c1amped immediately. The pipette and bath

solutions contained CsCI or NaC) instead of KC), to suppress the activity

of the K+ channel reported in yeast spheroplasts (Gustin, et al., 1986).

To maintain killer toxin activity the bath solution was always adjusted

to pH 4.7 by MES and the pipette solution adjusted to pH 7.2 by HEPES.

Prl!paration of liposome vesicles incorporating toxin

Li posome vesicles were prepared by using Asolectin (from

Sigma) according to Delcour et al (1989) with the following

modifications. In order to maintain killer toxin activity, aIl the buffers

were adjusted to pH 4.7 hy MES (pH 4.7). The KI killer toxin was

incorpomted into the liPlds at a toxin to lipid ratio of 1: 1 000 in the

solution of IOOmM KCl plus 1 x Halvorson saIts (Halvorson, 1958)

containing 0.03 M (NH4)2S04, 0.05M K2HP04, O.05M C4H60 4, CaCL2

2. 7mM. MgS04 4.1 mM by using either partially purified toxin or toxin­

containing concentrated culture fi Itrates from strain TI58C/S 14a,or

AH22 transformed with pL308. Toxin incorporation was tested by

spotting the redissolved lipid-protein mixture onto an agar plate

seedcd with sensitive yeast cells, and scoring the formation of a killing

zone. Concentrated culture fiItrates without killer toxin came from

44

either strain T158C/S 14aHC or from AH22 transformcd with pL361.

(the expression plasmid lacking the to' in gene). and wcrc llSt'd as

controls for incorporation with lipids under the conditions dcscrihed

above.

Patch-clamp techniques

Conventional patch-clamp techniques were used (llami Il ct al..

1981). Ali recordings from liposomes were obtaincd from inside-out

membrane patches excised from toxin incorporatcd Asolectin

liposomes. The recordings from spheroplasts exposed to the partially

purified toxin were obtained from inside-out exciscd patehes or fr'Hll

whole spheroplasts. Data were stored in an FM tape recorder (Gould

6500), digitized off-line at a rate of 0.2 msee, and analyzed on

computer (lndee) with a program developed by Dr. Yoshiro Saimi.

Amplitude histograms and 1-V relations were plotted using this

program.

3.3 Results

Spheroplasts from sensitive cells previously treated with partially

purified killer toxin yielded patehes that often contained il ehar­

acteristic channel aetivity (Fig.4B) (9 out of 14 patches). In contrast. out

of 14 patehes excised from spheroplasts from cells incubated with

concentrates of secreted yeast proteins from an isogenic strain lacking

the toxin gene, none was found to have the activity of such a channel

(Fig. 4A). This activity was also never encountered ln ovcr son pate he,

from u:ltreated yeast spheroplasts examined for this ~tudy, or for othcr

investigations of ion channels native to the yeast pla,ma mcmhranc

----- ----- ---------------------

c

(

45

(Gustin et al., 1986, 1988; Zhou, unpublished observations). The CUTTent

through this conductance f1icked rapidly among three states with

kinetics largely beyond the rcsolution of the present recording system

(5KHz), hence the spiky appearance of the records. This conductance

ohen occurs in a pairwise manner. Currents through this conductance

appear lo dwell mostly at a one-unit level but frequently visit a two­

unit level, or a zero-current level (marked 01, 02, C, in Fig. 4B). These

Icvels were detcrmined by the rarer events when longer dwell times

provide the flat-top or f1at-bottom records as shown in the lower trace

of Fig. 48. The kinetics of the conductance are not markedly affected

by voltage (compare Fig. 48 left and right). The current-voltage relation

(1-V plot) from two experiments shows the unit conductance to be 114

pS in magnitude (Fig. 4C). In these two cases, no chang,~ in the open

probability of the se channels was observed in the range of voltage

plotted. We found both spheroplast patehes and whole-spheroplasts

from toxin-treated cells to be unstable and easily broken, especially at

high voltage (>60 mV). We, therefore, concentrated on using the toxin­

incorporated liposomes to characterize the toxin-related channels.

The channels found in toxin-incorporated liposomes had the same

fingerprint as those in spheroplasts from toxin-treated cells (Fig. SB).

With the toxin-to-lipid ratio of our incorporation experiments (see

material and methods). we encountered this channel about 25 times out

of 80 patehes. No sueh channel activities were observed in liposomes

illeorporated with the non-toxin-containing control filtrates (Fig. 5A).

As in sphcroplasts. the conductance flickered rapidly among the three

states. which could he resolved at higher time resolution (Fig. 5D, lower

trace). An amplitude histogram of such activities over 30 seconds

-

-

46

Fig. 4. A. Absence of killer-toxin channel activity 10 excised patehes and

whole spheroplasts containing toxin-free extracellular protcins from

T158C/S14a HC.

B. Channel activity of excised patch from yeast spheroplasts exposed to

partially purified killer toxin from TI58C/S 14a. Channel activity was

measured at ±60mV in symmetric Cs. The trace at +60mY is expandcd

below at a higher time resolution. C and 01 and 02 denote closed and

unit-one-and two -unit-open levels of the channel, respecti vcly.

C. Average single-conductance levels from two experiments were 1 14.4

and 220.7 pS. NPo [opening probability of the total unknown numbcr of

channels (N) in the patch] for the se two traces were 2.1 and 2.0. This

indicates that the channel is not voltage sensitive. 1, current; Vp•

voltage.

.. ,

+60mV

A. ..... UI " .......... , o ............ , cf.

B.

~ 0-

"'L • 4Im •• c:

•

-60mV

li "Tt * .............. ' ,.,...,.

c. +20 I(pA) Q/

,

q,.."'-

+12 ...-

", -c;6 Q p,/lY'

-80 -40 +4 /'!).~'15.

Vp(mV) ~,...'" - +40 +80 A..-~::/ -4 Ar"""- ;1"

,,- V ...-~ -12 Q ",/

...-'" ...- -20 ...-

-c -0 1

-0 2

1

,.

47

shows that the conductance most often appeared at its unit-open (01).

state (Fig.50). A separate distribution peaking al the two-unit ll(K'1l (02)

state is also evident. The closed statc (C) was apparently rarely vi ... ited.

as shown by the laek of points at the zero-current Icw!. Fig.5C plots the

unit currents and the two-unit eurrcnts at different voltage (bctwecn

80 ailJ +80m V) l'rom two separatc cxpcrimcnts). Linear regressioll

shows the unit conductance to be 118 pS and the two-unit conductance

to be 222 pS.

We altered the pipette solution and the bath solutions to

investigate the ion specificity of this conductance. With aSYllllllctric KCI

solutions, the reversai potential of the unit current is near the

calculated equilibrium potential for K+, indicating little or no CI­

permeation (Fig.6A). In Na+, K+ bi-ionie conditions. the currcnt revcrscd

near the origin, indicating no discrimination betwcen the se Iwo ions

(Fig. 68).

The possible voltage sensitivity of this conductance was

examined. No obvious changes in channel opening or closing with

membrane voltages were encountered in ail the patches examined (Fig.

48 left, right, Fig. SB left, right). In three separate patehes for which

quantitative analyses were made, the opening probabilities of the

conductances showed no systematic variations with membrane voltage

from -80 mV to 80 mV (Fig. 7).

We also examined a mutant toxin from a strain harboring a

mutation in the a. subunit coding region of the toxin gene, Z1120. This

mutant toxin, which has il specifie activity sorne 10% of the wild type

4X

Fig. 5. A. Activity of patches from liposomes containing extracellular

protein extracts of yeast strain T158C/S14aHC lacking killer toxin.

B. Traces of channel aetivity from reconstituted killer toxin recorded 10

syrnmetrie 150mM K. This channel activity eould be seen in one out of

three or four patehes (+60 and -60mV).

C. Current-voltage (I-Vp ) plots from two experiments. As can be seen In

H. there were two Icvels of conductance. In these two experiments

average single conductance was 118.4pS (diamonds and triangles) and

average double conductance was 221.7 pS (squares and inverted

triangles ).

D. Amplitude histogram for +30mV, trace shown below (reconstituted

system, symmetric 150mM K).

..

1

1

+60mV -60mV

A. __ .................. ____ ..................... .-.._,--.... ----............ __ , .... ~ ......... ""'''' ..... _ ..... ~ ... , '.~ ...

-0

B. 2 -0

1

-c Cl

:L .,. 2 .. c:

C. D. ·20 I(pA) ;"

0 0 / 9/

.12 Sl- U)

~/ -~;"

~IOOO'

+4 9 cu 100 60 -/. 0

-~

Vp(mV) ~ ~ 500· 60 100 cu -4 .0

e-Q-~""" \ .. ', ~~ /.AT' -12

3~ ...... -20 ......

E ::J c

~ ~L

82msec:

r- ~

:3 6

-c -0

1

-02

(pA)

49

Fig.6. A. Current-voltage (i-Vp ) plot for asymmetric K solution (150/50

mM K pipet/bath resconstituted system). The obtained reversai

potcntial Erev =-25.4mY and the expected Ek, equilibrium potential of

K + is -27. 7mY (-25.9m Y, al' ter activity-coefficient correction). The

expected Eci is +27.7mY. Similar results were obtained in three other

expcriments.

B. Currcnt-voItage (i-Yp ) plot for K/Na (l50mM KCl in the

pipet/l50mM NaCI in the bath reconstituted system; the same

experimcnt as in A). E rcv in K/Na is -2.4 mY near 0 mV expected In

K/K. There is no significant difference In conductance between K/K

(110.1 pS) and Na (104.5 pS).

l

j {

A.

B.

-80

-80 -40 /'/ /

/'

. 150K/50K ,,(pA) /1$ // 150K/150K

~..<) //

<> / III / /

/ ~ / /M'

/ .. /

/' //' Vp(mV)

-5

i(pA)

5

/' /

/'

IY /

,/ yA

40

80

150K/150Na /'

/ A/ /

A/ /'

Vp(mV)

80

.,I/'A ~ -5

,J!a/

/ /

/

~ ",'

50

Fig. 7. Data points from three experiments indicate that the channel is

not voltage sensitive. X, Average NPo [opening probability of the total

unknown number of channels (N) in the patch] value was between 2

and 3, which indicates that more th an one pair of channels wa~ )resent

in the patch. In spite of the large scatter however, there was no evident

voltage dependence of the channel. V p, voltage.

1

• 0

0

0 <J <>

0 <J 0

0

0-0 C\J

Z

x:

o

....-

~ -Cl.

> 0 00

0 ~

o q-

1

o 00 ,

(

5 1

on yea"it ccliii and ~pheroplasts, was likely altered in sorne liposomes

incorporalcd with this ZH20 mutant toxin, we found no channel

activitic"i of the character described in this chapter. In 8 cases we saw

aclivitics with dlrferent characteristics (data not shown), consistent

with the dctection of a mutant channel.

3.4 Discussion

We have discovered the aetivities of a type of ion channel

associated with a fraction highly enriched for the KI toxin from killer

strains of S cerevlsicll'. The channel is cation nonspecific and capable of

passing K +. consistent with the observed K+ bleeding during the process

of cell intoxication (Skipper and Bussey, 1977). Such properties befit a

destructive device that has evolved to kill sensitive cells bv draining

them electrically and encrgetically. These conductance aetivities were

not secn in untreated spheroplasts or liposomes, but how confident can

wc bc that they are killer toxin specifie? To demonstrate biochemically

that the toxin causes the conductance is inherently difficult, for n~

matter how pure the protein taxin one ean argue that some very mInor

contaminant may be responsible for the eurrents detected by the

cX'fcmcly sensitive biophysical deviee-the patch clamp. Here while the

tcxin is the principal component of the partially purified extracellular

protein prcparations, other proteins are clearly present in minor

amounts. The best cvidence for specificity of toxin action is genetic. We

are able to effcctively delcte tne toxin gene in two independent ways,

by dclction of the killer virion. or by deletion of the toxin gene from an

c'\.pn:ssion plasmid. Whcn such deletions are made in otherwise

isogcnic ycast backgrounds, wc find that preparations of extracellular

,

1

!

proteins lacking only the toxin protein do not form the oh~er\'ed

conductances in spheroplasts or liposomes. In addition 10 thl'Sl'

negative controls. we have examined the conductance aCllvtlll'S of a

mutant toxin which when incorporated into lipo~oll1l'S shows an alterl'd

activity. These controls are consistent with the cOlHluct,lIlce activltil'''

being killer toxin depcndent. and strongly suggcst that thc"l' challnd~

are the basis of killing by KI toxin.

By using planar lipid bilayers. Kagan (19XJ) showcd that

extraeellular extracts containing a toxin from the yeast. PU'IIlCl /../uYl'l'I i.

form an ln vitro channel with broadly similar teatures to that secn wlth

KI toxin, but the channels of the two toxins appear ln dll l'cl

considerably in detailed characteristics. Thc P. k/Ii\'\'l'Il toxin challnel

exists in two conductance states. of 140 and 220 pS. a\ compared to the

118 and 222 pS conductances seen with KI toxin. The\e twin

conductances hint at complexity, but as Kagan has pointcd out. wht:thcr

these represent two distinct molecular specie" or IWO conductance

states of the same channel is unclear. Using the prc\ent techniquc wc

can not determine the structure of the channel. although prc\umahly

the hydrophobie Cl subunit, identified by site-directed 1l111tagcllc\il., al.,

being necessary to kill spheroolasts, i" involvcd. One may cOIlJccturt: Oll

structural grollnds that a single monomer of the (J. "uhull1t ni the tOXIIl.

with its two transmembrane domallls. is in,>uffll.:icnt ln Ilne a

waterfilled pore. In this context, the known oligornerii'al1on nf tht: tOXtn

up to at least an octamer (Bussey, et al., 19XH), may a..,I.,lIme lunctional

signifieance. TI.,:, P. kluyven toxin channel ..,howed a m~,rJ..cd voltage

dependence above 40 mY and helow -40 mY and nnly a weak

selectivity for common physiological ion~ in thc I.,e411cnc(; K+ > Na+ > Ca f- +-

53

> CI-. With KI toxin the conductance appeared to be voltage

independenl from -80 mY to +80mV and to show a preference for

monovalent cations. The kinetics of opening and closing of the 1;).

kluyveri conductance were considerably slower than that seen with the

KI loxin, though this may be merel)' because the two toxins were

examined by very different techniques. An additional common feature

of lhese loxins is that both function at acidic rH in vivo, although the

basis for lhis pH dependence is unc\ear. Both here and in the Kagan

sludy (19L~) toxin conductances were measured at pH 4.6, though

detailed sludy of the pH dependence of these channels has not been

made. Wilh the KI toxin, liposomes incorporation of toxin was done at

pH 4.6, but following incorporation toxin conductance appeared to be

independent of a pH gradient from pH 4.6-7.2.

Our work demonstrates for the first time killer toxin induced

conductances in vivo in sensitive yeast spheroplasts, and implicates

such channels in the killing process. The fact that such conductance can

be demonslratcd in vitro in artificial Asolectin liposomes does not

necessarily imply that there are no additional components involved in

vivo in toxin action at the plasma membrane.

Biologically there is additional complexity In the process of toxin

immunity in toxin producing cells. Immunity occurs at the plasma

membrane level, and is conferred by the product of the toxin precursor

gene, wherc mutations map to the reglOn eneoding the a subunit of the

toxin (Soone et al. 1986: Chapter 2). To effeet immunity the precursor

protein must in sorne way interfere with toxin channel formation, but

how this is achieved remains to be determined.

T

Chapter 4 The KI killer toxin of Sacc/lClr011l\'(,cs ('('rl'rislcll'

kills spheroplasts of man y yeast spl'l"Ies

4.1 Introduction

KI and K2 killer toxins have a narrow host range; they are

reported to kilt only S. cerevÎswc and rorU[OpSIS ~/llhratcl (BUSSl')' and

Skipper, 1976; Young and Yagiu 1(78). Studks 01 KI to:\in al'tioll

suggest that binding to the cell wall rcceptor, and kllllllg 01

spheroplasts which involves interactions with the plasma memhl aile

and causes ion-Ieakage, are independent steps (Bussey. 1 ()81). l'hl'

killing spectra of the KI and K2 toxins at the spheropla~t \evcl have not

been examined. In this chapter, we report lhal spheroplash 1 rom

several yeast genera are sensitive to KI and K2 kil 1er toxins.

4.2 Materials and methods

The strains used in this study are Iisted 111 table 2.

Spheroplast preparation and its regeneration

Spheroplasts were made hy digestion of yealit ccliii wlth 40 ug 01

zymolyase-60,OOO per mJ~i::Sussey and Meaden, 1(85) al ::WoC froll1 1 h

to 3h, depending upon the ycast strain ulied. Spheropla~t.., wcrc

regenerated by the method of lIinnen ct al. (1978) in YEPD agar (1 g of

yeast extract, 1 g of peptone, and 2g of gluco ... c per 1 OOml) cOlltailHng

1.2M sorbitol at pH4.7. Survival of ccII.., follow1I1g the ... phcropla ... t

procedure was not greater than 1 ()-5, mca .... urcd hy platlllg on YEPD agar

without sorbitol as osmotlc liupporl.

1

55

Ki IJing aSliay

The ~ensitivity of yeast cells and spheroplasts to the killer toxin

wa~ determined by killing zone as say in a plate test (Bussey, 1981). An

agar plate sccdcd with a tester strain (eilher whole cells or

lipheropla~ts) was patched with killer cells, or spotted with 10 ul of

concentratcd killcr yeast culture filtrates. The inhibition of cell or

spheroplast growth was shown by formation of a clear zone around

the killer cells or culture filtrate (sec Fig. 8). Controls showed that

spotting of 10 ul of buffer or water in place of killer filtrates caused no

zone of growth inhibition.

Cell wall receptor binding assay

The bi nding of the killer toxin to cell wall receptors was

pcrformed by the toxin activity removal method of AI-Aidroos and

Bussey (1978). The isolation of yeast alkali insoluble glucan was as

described hy Hutchins and Bussey (1983).

4.3 Results and discussion

Cells and spheroplasts of nme different yeast strains (Table 2)

were tested for sensitivity to KI and K2 toxins. A KI killer strain,

TI58C/SI4a, and a K2 M471 strain were used for killing activity. A

non-killcr strain. TI58C/S 14a HC, deri ved from TI58C/S 14a by heat

curing of the KI dsRNA was uscd to determine killing effects

spccifically related to the KI toxin. Apart from S. cerevisiae, cells from

ail the strains were insensitive to the toxins. However, spheroplasts

from eut/ils. llnd K lactis were found to be sensitive to both toxins,

1

]

56

white C. albicans and S. al/uvius spheroplasts were sensitive to KI to\.in

(see Table 2), although they are ail less sensitive than wild type S

cerevisiae spheroplasts to KI and K2 toxin as indicalcd hy the smalkl

killing zone formation. If mraki, C. Iminl'nsis and P J..IU\TCr

spheroplasts were insensitive to both toxins. The rcsults indicatc that

both killer toxins have a far wider spheroplast killing spectlulll than

that found at the whole cell level.

Previous work with S. cerevisiae cells indicatcd a rcquirelllcnt for

a (1 ~ 6 )-P-D-glucan cell wall receptor for toxin dcpendent kliling tn

occur (Hutchins and Bussey, 1983). Mutants with defects in the KR HI

gene of S. cerevisiae have altered wall (1 ~ 6 )-P- D-glucan, l'ail to hllld to

the toxins, and are toxin resistant (Hutchins and Bussey, 19XJ; Bonne

et al., 1990). One reason for toxin insensitivity in non-Saccllll rom "Cl'.\'

yeasts could be lack of a su ch (1 ~ 6 )-P-D-glucan receptor. 1'0 test this

possibility, C. albicans and K. laclis cells were assaycd for toxin hinding

ability. A sensitive S. cerevisiae strain,463-1 C, and an isogenic reccptor

defective resistant mutant kre l, 463-1 B, wcre used for comparat Ive

binding studies (see Fig. 9), in which toxins cou Id be scen to hind with

far less avidity to the resistant mutant than to the sensitive parent

cells. K. lactis cells showed reduced binding of toxin cOlllparcd with that

of the sensitive S cerevisiae strain, 463-1 C, and binding ~imdar to lhal

of the receptor defective resistant mutant, 463-1 B. Ttw, rcsult ~uggC ... l~

that K. lactis lacks an effective wall receptor, and 'iuch laek may he

sufficient to prevent toxin action.

Results with il toxin from Il.mrakti are con'ii ... tcnt with thi'i

explanation. The fi mrakkil toxin kill'i spheropla\t<; (Fig. X) hut not

who le cells of K.lactis. This toxin kill<; S cerevistac cclh and ahl)

57

Fig. 8. Effects of ki 11er toxins on yeast spheroplasts.

YEPD agar plates with 1.2l\t1 sorbitol at pH 4.7 were seeded with

sphcroplasts of K lactis (A) and C. alhicans ( B). The seeded plates were

patched with 10ul of the indicaled killer or nonkiller (NK) cells or

culture filtrates and incubated at lS0 C for 2 to 3 days. See Table 2 for

strains uscd.

f

•

NK Kl

A

K2 H.mrakii

NK Kl

B

K2 Kl

.....

Table 2. Sen5itivities of yeast SJilerq)lasts ta ta anl 1(2 killer strains

strains usecI f« killing activity

Straim tart:ed for: SJilerq)last _itivity

.-. .. _.-· .... ~l· \.. •• -1 ..... _-:::

~.j • '-._-J6.-"_~

9"""'isiae 56

œmvisi " ~C 1324

camlda albicans A'1œ 10261

candida utilis NCYC 707

candida buinensis Anr 58432

ltiMJiienIla arakii NCYC 500

KlYWeJa!YP!S ladis 2105-10

PMia kluyveri A'Jœ 24241

sctNannialVOeS alluvius A'Jœ 26075

Arec: AEriœrt ~ Ollture Q)Uectim, U.S.A.

Tl58c/Sl4a (ta 'lUtin)

+

+

+

+

+

..

NC'YC: National Cbllectiœ of Yeut Olltun!s, Enqlard 86: Sensitive tester stnin, vild type 2105-0: KiUer strain frc. N. CUqe

*71 (1C2 'n:Jxin)

+

+

+

+

Tl58c/S14a.H.C. (NK)

Tl58c/S14a.H.C.: Sensitive oon-killer &train derival frc. TlSSe/Sl4a by hast aarinJ of the KIIrKl ~ +: Spt1e.rq)lasts sensitive to killer strains -: Spt1e.rq)lasts insensitive ta kiUer &trains

....

-

l

:; ()

appears to use a (l~6)-P-D-glucan wall rcceptnr: Wl' havl' round Ihal S

cerevisiae kre 1 mutants are resistent to il. Lack of suc h a glucan

receptor on K lactiç cells cou Id cxplain the in ... en-.illvity of Ihl' ... e l'l'Ils 10

the H mrakf..i toxin.

C. alhicans, in contrast, showcd a toxtn-hIlH.ling capacity ... imilar 10

that of the sensitive S.cerevisiae strain yet rcmained to\in lIl ... l'IlSilive.

To exclude the trivial possibility that a protcasc or other factor

inactivated the toxin, thus leading to apparent toxin hindll1g trl our

assay, we isolated alkali-insoluble cell wall glucan from C al/J/colls and

repeated the toxin-binding assay, with rcsults similar to those

obtained with whole cells (data 110t shown). In additIon. wc showcd

that this toxin binding was revcrsible with pU; toxin activlty wa ...

removed by the glucan at pH 4.7 but could be dlssociated trom the

glucan at pH 7.6, as had been found for toxin hinding 10 S ('(' 1"(' \'1.\'llU'

glucan (Hutchins and Bussey, 1(83). The C. alhlclIns l'l'II wall" I-.nown

to contain a (1~6)-[3-D-glucan (Gopal et al.. 1(84). and hllldlllg of the

toxin is likely to be this glucan. Our findings show thal tor (' a/hl( lll/.\,

the KI killer toxin binds to a cell wall receptor and can kil!

spheroplasts, yet it fails to kill yeast cclls. Clcarly, mcrc hindtng III thL'

toxin to the cell wall is not sufficient here to cflcct action 01 the tOXIII al

the plasma membrane. It is possible that thclc arc addltional

unidentified wall components requircd for killcr toxln acllon that arc

missing in the wall of C allJlcans Alternativcly thcn: Illay he \Ollle

structural differences in the wall of C aIlJlcan.\ which prevent

accessibility of the glucan-receptor bound toxil1 to the pla\ll1a

membrane.

«

l

60

Our results have sorne bearing on the possibility of a plasma

membrane receptor necessary for toxin action (AI-Aidroos and Bussey,

J 978; Tipper and Bostian, 1984). If the toxin was nonspecific and free

to insert into any Iipid bi layer as it has been shown in chapter 3, one

would expect that the spheroplasts of ail yeasts would be tOXIn

sen~itive. Thi~ i~ not the case, as spheroplasts of three of 9 yeast strains

lcslcd herc were unaffected by toxin action. These insensitive

sphcroplastl, could inactivate the toxin or may be lacking in sorne

ncccssary rcceptor at the plasma membrane. If there is a plasma

membrane receptor, it is not species specifie, being functionally

conserved ,\fllOng at least four yeast genera.

This work demonstrates that the yeast cell ',val1 plays an

important role in determining toxin action and can establish the

apparent spccificity of these toxins, A cell wall glucan receptor is

necessary for the KI and K2 toxins to function on cells, but binding to

this receptor may not be sufficient. Such findings focus attention on our

lack of knowledge of the structure of the yeast ~ell wall and how it

functions ln permitting protein passage. The observation that toxin

resistance is conferred at the ce]] wall level in yeast cells with sensitive

sphcroplasts suggcsts that by modifying wall structure, sensitive cells

could be constructed. For example, expression of (1 ~6 )-P-D-glucan in K.

[lU'lIS may makc these cells toxin sensitive. Alternatively the idea of

constructing mutant or hybrid toxins that can recognize effective

rcccptors on the walIs of insensitive yeast cells while retaining the

ability to lill "phcroplasts could result in the construction of new toxins

for thcse ycast genera.

6 1

Fig. 9. Binding of killer toxin by sensitive, inscnsitive. and rcsistant

yeast cells. Yeast cells at a range of concentrations wcrc incuhatcd in

1 ml of medium with a fixed amount of killer toxin for 30 min at 4°('.

The cells were pelleted by centrifugation. Toxin activity rcmaining in

the supernatant was assayed by a zone test (AI-Aidroos and Busscy.

1978), and bound toxin W=lS caculated by difference. Each Error har

represents 1 standard deviat;on. 463-1C, Sensitive S.cerevisiae strain;

463-1 B, isogenic wall receptor-defective kre 1 mutant.

1

r

•

62

Chapter 5 Mut .. taonal analysis of functional domains of KI

killer toxin l'rom Sacclwromyct!s ('l'I('risÏlu'

5.1 Introduction

Previous genetic and biochemical studies dcmonstmtcd that loxin

action required at least two steps, involving a wall rcccpior and aClion

at the plasma membrane. The toxin initially binds 10 a ccli wall

receptor which contains a (1 ~ 6 )-~-D-glucan (A I-Aidroos and Bus~cy.

1978; Hutchins and Bussey, 1983). Assembly of this glucan reccplor

requires a set of nuclear K RE genes (Boone et al. 1990; Mcadcn l'( al.

1990; Bussey et al, 1990). Mutants of KR Ji gencs are rcsistant 10 KI

toxin, but when kre mutant cells are converted to sphcroplasts. they

are sensitive to killer toxin, suggesting the existence of a second step.

Physiologieal ~tudies of KI toxin action suggestcd that the toxin

perturbed an energised plasma membrane state causing ion \eakage

and subsequent cell 1eath (de la Pena et al, 1981). In chapter 3. we

have shown that by using patch clamp technique the toxin forms

voltage independent cation channels in sensitive spheroplasls and ln

artificial liposomes. Such channels are Iikely the basis of \oxin action.

A possible functional aS'iignment of domains of the KIki 11er loxin

has been suggested based on the primary structure of the IOXIn

subunits (Bostian et al, 1984). The a subunit contaJrl~ two hlghly

hydrophobie regions (residues 72-91 and 112-127) separaled hy a

short, hydrophilic region. This secondary structure "iugge"ited that the (J.

subunit may be responsible for ion channel formation. In contra\t, the r~

subunit IS very hydrop:lilic and lacks potentia! membrane \panrllllg

regions. Hence, by analogy to the abrin and ricin cla"i"i of toxi/l\ (Ol\/lc\

63

and Phil. 1(73). thi~ subunit ha~ been proposed (Bostian et al. 1984) to

bind to the (1 ~ 6)-B-D-glucan cell wall receptor. Experimental support

for thesc idca~ i\) limited. Published mutation~ in the taxin gene often

led to failure to secrete significant levels of toxin making analysis of

phcnotypc~ difficult (Boone et al, 1986; Sturley et al, 1(86). One

mutation (PTX 1-234) has becn described in the ~ subunit that led to a

toxin which faib to kill cclls but kills spheropla~ts. This rcsult is

consistent wlth a role of J3 in bindiJ1 ta a cell wall receptor (Stur!ey et

al, 1(86).

Ilere. we report results of an extensive mutational analysis of

both toxin subunits. We show that while the a subunit is necessary for

channel formation, bath a and J3 subunits appear to be required for

gluc.m hinding.

5.2 Matcrials and methods

Ycasts, bacteria strains and medium

S.ccrevislClc strain S6 (KIL-O) wild type was used as a sensitive

tester strain for toxin activity. T158C/S14a MATa/MATa./zis 4C-

864/IIIS4 acll'2-5/ADE2 (KIL-KI) (ATCC26427), containing the Ml

VITUS and TI SXC/S 14a HC, a sensitive (KIL-O) non-killer strain derived

t'rolll T158C/S14a by heat curing of the Ml virus (Bussey et al., 1(88)

wcre lIscd 10 prepare toxin and non-toxin containing culture fi Itrates.

AII22 MATalcu2-3, 2-/12 his4-519 canl (KIL-O), H4a MATahis3 lIra3

(KIL-O) strams which lack M I-ds RNA were used as recipients for

transformation with the pYT760 (Lolle et al, 1(84), and pYTIOOU

(Vernet ct al, 1(87) based expression vectors respectively.

•

Transformation of S. cere~'isiae was by the lithium chloridc

procedure of Ito et al (1983). Growth medium was the minimal medium

described previously (Lolle et al, 19X4), supplcrncntcd with

appropriate amino aGids or uracil.

The E.co/i strain MCI061 (Casadaban and Cohen. 19RO) wa~ USl'

for routine growth and maintenance of plasmids, whilc 1:' col, UT5XO

with the helper phage M 13K07 was used for production of singlc­

stranded DNA (ssDNA) as sequencing templates (Vernet ct al.. 19X7).

An E. coli ung mutant strain CJ236 was used to produce uracil­

containing ssDNA templates for efficient site-directcd tnutagencsis

(Kunkel, 1985). A Il media for routine growth and maintenance of 1:' ('oli

strains were standard.

Plasmids

The yeast KI killer toxin expressIOn vectors uscd wcrc pL30X

(Boone et al., 1986) and pVTlOOU -KT (Vernet et al.. 19X7) rcspectively.

in which the cDNA fragment of preprotoxin was placcd behind the

ADHl promoter in plasmids pYT760 (Lollc et al., 19H4) and pVTIOOU

(Vernet et al., 1987). Plasmid pL308 was subjectcd to hydroxylarninc

mutagenesis and transformed into either E coll MC 1 061 or yca"it "itrain

AH22. Mutations were confirmed by sequencing (Sanger et al.. 1(77)

the preprotoxin gene which was subcloned into the IImdlll, Bamlf)

sites of Bluescript vector M13 (Stratagene, San Diego). The pla"imid

pVTIOOU-KT is an E.coli, yeast shuttle vector which contain<, the 1-1

origin of replication allowing production of single <,.randed DN A. S!te­

directed and random site-directcd mutagenesi~ wcrc carricd out hy

u~ing this vector.

65

lIydroxylamine mutagenesis and screening mutants which secrete

inacli vc toxin

Ilydroxylamine mutagenesis was carried out according to Schauer

cl al (1985). The CsCI purified killer expression vector pL308 was

hcated al 650C for 40 min in 004 M hydroxylamme, 0.05 M potassium

pho~phate pH 6.0. The DNA was precipitated by ethanol, resuspended

in TE hutl'er and transformed into E coli Mc1061 for amplification, and

plasmid DNA was isolatcd from E coli as described by Holmes and

Quiglcy (1981) and transformed into yeast strain AH22. Hydroxylamine

treatcd DNA was also directly transformed into yeast strain AH22. The

initial scrcening of mutations which caused secretion of inactive toxin

was carncd out hy rt~pltcating yeast transformants onto plates seeded

with sensitive cells S6 and looking for the colonies incapable of

forll1l1lg a killing zone. Non-killer colonies were screened for those

secrctlllg inactIve toxin using a western dot-blot assay. Plasmids from

such yea~t mutants were isolated according to Davb et al (1980), and

transformcd to E COll MCI061 to purify the plasmids. Fmally, the

puril'ied plasmids from E col! were transformed to yeast strain AH22 to

confirm that the phenotype was plasmid dcpendent. Mutations H21,

IIEII, HE3 and HE6 were generated in this way.

01 igon lIC Icotide -di rected mutagnesi s

Oligonucleotide-directed mutagenesis was carried out as

descrihcd hy Zollt'r and Smith (1982) using uracil-containing templates

made l'rom pYTI()()U-KT and Bluescript M13-KT according to the

methods of Kunk.el ()985). The mutagenic olîgonucleotides were 5'

GATCAACCCGGGATI3' ( pL7), S' GGCAAGAAGC'ITACCTGGGGTY (1.1123). 5'

CATCATCIT AGGTGGGGTGGT3' (ZI-I20), S' TCGC AGTC AAGGCCTAATC ;CI('3'

(ZH21), 5'AACCTGTATAAGCTTGCAGAA3' ( ZII2 .. n. 5'ATTGAGGAACCGATfGATAAC3' (ZIl3). S'CCCCTCClAGGGCiT (pL 13),

5'CTTTGAAGGCC1TGACACAGGCCA3' (17B). and 5'

GGCTGTGACTAGTCGCACTAG3' (ZIII). Ali the mutations wcrc conflllllcd

by sequencing the region surrounding the location of the dcsilt'd

mutation. ln the case of ZH23, ZH3, ZH 1 the mutatcd killer tOXIIl !!.l'IlC

was isolated frorn the replicative form of Bluescript M 13-KT as il

HindIII-BarnHI fragment and cloned into a pYT760 cxprc"sion vector.

Site-directed random mutagenesis

This mutagenesis was carried out according to Kaldcron et al

(1984). An deoxyoligonucleotide covering the second hydrophobIe

reglOn was synthesized with redundancies in the second po~ition wlthill

specified triplet codons. ft was as follows: 5' GCT lTA C1'* A CITe AGe

AT*T TTT GT* A GCA GT*T ACA T*CC GGe-3' ( * marking po~Jti()n" wherl'

10% of the oligonucleotides synthesized have bases othcr than wild

type). The above oligonucleotide was annealed to pYTI()()U-KT ... ingle­

stranded template synthesized 111 ung E coli strain CJ236. primer

extended and strain Mc 1 061 transformed with thc rcactlon mlxturc.

The total population of tramforrnants (around 3000-50(0) wa" poolcd.

Plasmids were isolated and transformed into yeast strain 114a. M utanl,;

which secreted inactive toxin were screcncd (R4-3S, R4-32) a ...

described in the hydroxylamine mutagencsis section ahovc.

67

Mutant toxin activity estimation

Isolatcd mutants were grown at 180 C in liquid minimal medium

containing 1 x lIalvcrsion \ salt, 2% gl ucose and 10% glycerol. After the

culturCI., cntcrcd the ~tationary phase of growth, the cells were pelleted

by centrifugation. The medium was removed and concentrated from

200-500 timl'~ ty ultrafiltration usmg a PM 10 membrane (Arnicon).

The activity of each mutant toxin was assessed by spotting 25 ul of

concentrated medium onto an agar plate seeded with 1 x 106 cells/ml of

strain S6. The size of the killing zone was measured and killing activity

calculatcd as a percentage of the pL308 wild type. Values for wild type

pL308 killing activity were from a calibration curve obtained From S6

plates from a dilution series of IOOOx concentrated medium from an

idcntically grown pL308 transformant.

Estimation of secreted toxin from mutants

The amount of toxin secretion from mutants was determined

mainly by pulse-chase experiments (occasionally supplemented by

loading concentrated medium directly on SDS-polyacrylamide gel). Each

mutant culture was grown to mid-log phase in liquid minimal medium,

labcllcd with I~SSI-methionine for approximately 12 min at 300 C, and

chase for 40 min with non-radioactive methionine. Growth medium was

obtaincd by centrifugation, and concentrated in ultrafiltration cones

(CF-25, Amicon Corp). Concentrated medium was subjected to SDS­

polyaccry lamide gel electrophoresis and radioactive proteins detected

by fluorography. Band intensities of the toxin proteins were estimated

-----------------------.

Western Blot assay

Antitoxin antiserum raised in rabbit was purified uSlIlg a eN Br

activated Sepharose (Pharmacia) affinity column coupled with to,in

free concentrated medi um protei ns from T15HC/S l..ta IIC (a~~ordi ng to

the manufacturer's instructions). Quick detectlOn of sc~n.·ted tO\1Il l'rom

the mutants was done using a dot-blot technique. Mutant ycast l'oloniL':-'

grown at 300 C on selective agar medium wcre overbyed with a

nitrocellulose membrane and incubated overnight at 300C. when the

nitrocellulose was removed and allowed to air dry. The presence of the

toxin on the nitrocellulose was detected immunologkally using a fI' 1111 t y

purified antitoxin antiserum, and the ABC Vcctastam Kil (Vretor

Laboratories, Burlingame, CA), used according to the manufaelurer's

instructions.

Immunity assay

The immunity of each mutants was determined by sccding 2%

agarose complete medium plates pH 4.7 with 20 ul of stationary phase

cultures of yeast transformants harboring the variou~ pl asmids and

then spotting with 10 ul, a ~tationary phase concentrated culture

filtrate of killer strain T158c/S 14a. Plates were incubatcd overnight at

I8 0e, and for a further 24h at 300C sensitive non-immune strain" gave

a zone around the killer spot, immune strain did not.

Sensitivity of spheroplasts to mutant toxins

Yeast spheroplasts from strain S6 were made and rcgencratcd a.,

described in chapter 4. Mutant cells which containcd variou., mutatcd

69

Yea\t ~pheroplastli from strain S6 were made and regenerated as

dcscribed in chapter 4. Mutant cells which contained various mutated

toxin plasrnidli werc patched onto agar plates seeded with spheroplasts.

The inhibition of spheroplast growth was shown by formation of a clear

zonc around the patched yeast cells.

CcII wall receptor binding assay

Concentrated culture filtrates from mutants were passed through

a 10 by t.O cm pustulan-Sepharose 6B column as described previously

(liutchins and Bussey, 1983). The fractions were collected and proteins

wcrc precipitated by acetone at -200 C overnight. The pellet was

collected by centrifugation at 10,000 RPM, and dissolved in SDS-sample

buffer containing J:'-mercaptoethanol for SDS-polyacrylamide gel

elcctrophoresis (13%), usin6 mini-proteinlI Dual Slab Cell System (Bio­

Rad Laboratories, Richmond. CA). The toxin band was visualized by

silvcr staining (Wary et al.,1981). In arder to exclude the possibility

that hinding to the column of toxins from mutant filtrates was an

artifact due to contamination by wild type toxin, a thorough wàsh by

0.1 M pH 7.6 Tris buffer containing 15% glycerol was given after a wild

type extract was passed though the column, prior to application of

another sample. In sorne cases, a new pustulan column was used for

illdividual mutants.

5.3 Rcsults

Our strategy to determine the toxin functional domains was to

systematically construct mutations throughout the a. and ~ encoding

regions. It was hoped that such mutations would lead to only local

70

bind to the cell wall receptor and to kill spheroplasts. the rl!spectivl!

domains would be determined.

Estimation of levels of secreted toxin from mutants and thcir killing

acti vit y on sensitive cells

Our set of mutations were obtained usmg a range of tcchntqucs

inc1uding site-directed mutagenesis (pL 7, pL318. Z1I20. Z1l21. pL 13.

ZH3, ZH24 and 178); random hydroxylamine mutagencsis (1121. IIEII.

HE3 and HE6); and site-directed random mutagencsis (R4-35. R4-32).

An overall summary of the location, modification and phenotypes of the

mutations is shown in Table 3 and Fig. 10.

Initial tests by western dot-blot assay indicated that ail the above

mutations secreted detectable toxin. Further, more accuratc.

quantitation was done by radio labelling the toxin from transformcd

yeast cells harbouring mutations in the toxin gene. 35S- mc thionine

labelled, secreted toxins were analyzed by SDS-polyacrylamidc gel

electrophoresis, and visualized by fluorography. The intensity of each

mutant toxin band was determined by densitometry and compared 10

that of the wild type toxin. The killing activity of each mutant tox in wa"

arbitrarily determined by plate as say (see Materials and Mcthods). and

compared to that of the wild type.

These resuIts are shown in Table 3. Many mutant toxins (pL 7.

ZH23, H21, HE 11, pL318, ZH22, R4-32, R4-35, HE3. IIE6, and ZIII) have

greatly reduced levels of activity «0.05%), others (ZH20 and Z1I21)

•

TABLE 3. Mutations in tha toxin precursor gene and a phenotype summary

Mutations Protein modification Phenotypes Toxin Toxin activlty secretion on whole cells

Cell wall Sheroplast IlIIIUll ty (X) (X) receptor killing bincling

pl308, Plasmids containing the • • • 100 100 pVT100UKT killer toxin gene pYT760, Plasmids containing no pVT100U killer toxin gene 0 0 pL7 Insert NPGL following 152 + + 20 <0.05 ZH23 Change 571-172 to K71-L72 + 20 <0.05 H21 Change G83 to D83 • 66 <0.01 pL31a* Insert LE following V85 • 30 <0.01 HE11 Change C92 ta Y92 65 <0.02 pL336* Change 0101 ta Rl01 • • 30 <0.05 ZH22 Change L114 to K114 59 <0.01 R4-35 Change l115 ta F115 41 <0_01 R4-32 Change S124 ta P124 21 <0.01 ZH20 Change 1129 to ~129 t** t t 80 10 ZH21 Change D140 to R140v t** t t aD 10 HE3 Change P237 ta 5237 • + 68 <0_1 HE6 Change Y256 to M256 + + 65 <0_2 ZH24 Change G264 to l264 .** • • 75 75 pl13 Insert PLEG following S287 .** • • 40 35 ZH3 Change 5305 to P305 .** • • 30 30 17B Change C312 to l312 .** + + 57 50 ZH1 Delete T314-G315-H316 + + 31 <0.02

Mutations which do not affect cell wall receptor binding, spheroplast killing and illllUlity are inclicated as (+), otherwise (-). Mutations which cause partlJl toxin phenotypes are indicated as (t).

*These mutations were made by Baone etaI (1986), but lacked detailed analysis of their toxin phenotypes. **The cell wall receptor bincling abllity of these toxins was determined only by plate assay.

,.

72

Fig. 10. Localization and phenotypes of mutations in the coding regions

of the Cl and ~ subunits.

The diagram at the top shows the structure of the preprotoxin and it"

hydropathic profile. Below are the mutations and their corrcsponding

effects on the killing of cells and spheroplasts. and on glucan hinding

and immunity. The circled G's represent sites of attachment of N-linkcd

glycosyl residues.

1

umml!::., 'Pllilil

Illlllllllt 1111111 1., ,.'""

''':!~Iitlll , ,1 "

'''''1 "'Ill 1111 .... '

,lIii ......... , 1111111111

':""

"~!!IIIUI ., '::dmn

11111 l'

, lliiiil:/II u ,

''''UlIIII''IIIIIHIl~mlllllli ... .. nIlUl!!Ilu!~""",

.... "I~!IIII~!!~mllllll 111111111 .....

, ... IIIII!;::: .... ·

'""'111l

• ;1111!li iiiilllllllilli 111111

........ _ .......................................................................... . , 1 1 1 1 1 1

1 1 1 1 0 1- :- 1 otQO<IdoIlMH otIll(dolIlAH

II: lai CI

~

C

M o -o .. A-• ..

Do.

:::,] +

+- ZH24

+-tE6J •

+- tt=.3

+- ZH21 ] +1 4- ZH20

4- R435 • 4- R432]

+- ZH22

4- pL336 ~ HEll • 011= PL318]

Hé.) +- ZH23

4- pL?

.. c o -• -::1 2

.! 'i u

c o

+

+1

+

+

al .5

c~ o .-.Q

>­~c > • - u - ::1 u_ oCa

+

+

+

+1

+

.. -.. c • o Ci. ~~ .; . .- oC Ua. oC ..

+

+

+

+1

+

+

>­!: c ::1 E E -

l

73

have intermediate levels about 10%. In some cases, rcduced to\.in

activity corresponds with a reduced alllount of secreted ttl\.ill (ZII2·1.

pLI3, ZH3 and 17B), and rhcse mutant to:-..ins are prohahly Ilot

functionally defective.

Mutations defective in spheroplast killing, and putative IOn channel

formation are confined to the a subunit

Transformed yeast cells containing mutations III the toxin gene, 01

concentrated culture filtrates from such mutant strains were tested for

their ability to kil 1 yeast spheroplasts; the resliIts are slIllllllaril.ed in

Table 3 and Fig.ID. Mutants with defects in either of the hydrophoblc

regions of the a subunit (ZH23, H21, pL318 in the t'irst hydrophobie

region, ZH22, R4-32, R4-35 in the second hydrophobie rcgiofl) totally

abolished the ability of the toxin to kill spheroplasts. In addition, a

linker insertion mutation pL 7, located in the region coding for the

hydrophilic N-terminus of the a subumt, was also inactive 111 killing

spheroplasts. Two mutations in the region encoding the hydrophilic C­

terminus of the a subunit produced toxins partially defective towards

whole celIs and sphtroplasts. AlI of the mutations in the B S li h li nit

coding region retained the ability to kill sphcroplasts.

Mutations affecting immunity

The immunity phenotype of ycast tran"iformanh containing

mutations in the toxin precursor was tC"ited. In the fi. ~uhunlt, many

mutant toxins 10st both thc immunity and "iphcropla~t killlng

phenotypes. The N-terminal mutation pL7 rctained Immunlty, whde

ZH20, ?nd ZH21 retained partial immunity. AIl of the mutatIon", in the I~

74

1986), mapping immunity to the a subunit region of the toxiu

precursor.

Mutations in both the a and ~ subunits of the toxin lead to dcfectivc

toxin binding to the cell wall receptor

The finding that inactive toxins secreted from mutants failcd 1O

kili sensitive ceIls, yet retained the ability to kill spheroplasts

suggested that the toxins may be defective in binding to the wall

receptor. Mutat.ions at the N- and C ·termini of the P subunit (HE3, IIE6

and ZH1) gave such a phenotype (See Table 3 and Fig. 10). A large

group of mutations in the a subunit (See Fig. 10) led to toxins that were

inactive toward both yeast cells and spheroplasts. 'fo directly

determine if these mutant toxins retained cell wall receptor binding, an

in vitro celI wall receptor binding assay was used (Hutchin and Bussey,

1983). Active native toxin can bind in a reversible, pH-dependent

manner to the (1--7 6)-~-D-glucan polymer, pustulan, which serves as a

cell wall receptor analog. Pustulan can be coupled to Sepharose, and

toxin binding to a column of this material can be estimated (see

Materials and Methods) essentially, toxin binds 10 the column at pH 4.7,

and this bound toxin can be eluted at pH 7.6 Mutant toxins defcctivc ln

ceU wall receptor binding should fail to bind to such column. The

extracellular concentrates of mutants from the a subunit pL7, ZH21,

H21, pL318, HEll, ZH22, R4-35 R4-32 were subjected to thi" affinity

chromatography binding assay. Among these mutations, it was round

HEll, and ZH22 (see FiglO) :ts weil as R4-35,R4-32 (data not shown),

produced toxins unable to bind to a pustulan Sepharose column (sec

1

75

produced toxins un able to bind to a pustulan Sepharose column (see

Fig. ]] bottom panel), whereas, other tested mutant toxins pL 7, pL318,

1-121 and ZH23 were found to bind (Fig. 1 1 upper panel).

A toxin band eluting in the pH 4.7 fraction of the wild type and

the tested mutants can be seen in the A lanes of Fig.ll, and represents

an inactive toxin fraction which has been seen previously (Hutchins and

Bussey, 1983), and which is irrelevant to the analysis.

Toxins from mutations localized in the ~ subunit (HE3, HE6 and

ZHl) were also subjected to the pustulan column chromatography,

none bound (Fig. 1 1 bottom panel).

5.4 Discussion

We have identified functional domains of the KI toxin by analysis

of mutations in the toxin encoding regions of the a and /3 subunits that

allowed toxin secretion. These mutations are summarized in Table 3

and Fig. 10.

We found that mutations affecting cell wall receptor binding are

localized to regions encoding both subunits as represented by

mutations HE Il, ZH22, R4-32, R4-35, pL336, HE3, HE6, and ZHI. The

inability of these matant toxins to interact with the ~-glucan receptor

was determined by both spheroplast killing and binding to a (1-7 6 )-~­

D-glucan column. We reasoned that mutants that secreted a toxin

inactive toward cells, but which retained the ability to kill spheroplasts

were Iikely affected in a cell wall receptor binding domain. Two

mutations (HE3 and HE6) at the N-terminus of the ~ subunit and one

small deletion mutation (ZH 1) at the C-terminus of the /3 subunit

76

Fig. Il. Pustulan-Sepharose column chromatography of killcr toxin.

Concentrated culture-medium from wild type killer-plasmid

transformed cells (H4a-KT) and from mutated killer plasmid

transformed cells were passed through a pustulan -Sepharosc 6B

column, washed with 0.1 M sodium acetate containing 1 )C'io glyccrol at

pH 4.7 and bound toxin eluted from the column with 0.1 M Tris

containing 15% glycerol at pH 7.6. Fractions were collected, precipitatcd

and subjected to SDS-PAGE followed by silver staining (sec methocls

and materials). Lane A and lane B indicate the fractions collected al plI

4.7 and pH 7.6 respectively. The upper panel shows mutant toxins

which bind to the column as indicated by a disulfide bond reduced 9 kd

toxin band eluted with a.IM Tris at pH 7.6, lane B. Bottom panel shows

mutant toxins which do not bind to the column, and lack the 9kd toxin

band in the Tris elute, lane B.

1

-

H4a-KT pL7 H21 pL3l8 ABABABAB ~

'"" ,I •

~..... .' ,- . ..; ..... ' t-I4a-KT HE11 A BAB

~

-~

ZH22 A B

HE3 A B

•

ZH23 A B

... • Il 4 9kd

HE6 ZH1 A BAB

_. , ~"

-

... 9kd

:,,-- . _.~ .~~.,'h

showed this phenotype, as did pL336, localized at the central

hydrophilie region of the a subunit (Boone et al.. 1(86). lIowl'vl'f. a

77

large group of mutations (pL 7, ZH23, 1I21, I-IE II, pL318. R-l--32, Z1122,

and R4-35) localized to the a subunit Icd to mutant to\ins which falkd

to kill both cells and sphcroplasts, suggesting these mutations affected

at least IOn channel formation. To test whether thesc mutations abo

affected cell wall receptor binding, we assaycd the toxins fOI ahility to

bind to the glucan column. Four of thesc u. s,~bunit mutatIons (;IEII.

ZH22, R4-35 and R4-32) gave mutant toxins unable to htnd ln the

column, implicating these regions in cell wall receptor hinding.

Putative cell wall reeeptor binding defective mutants in the P

subunit, (HE3, HE6 and ZHl), also failed to bind to the glllcan column.

Our results demonstrate that both the a and ~ subunits arc rcquircd for

cell wall receptor binding, and that the A B model of ricin is not

applicable to killer toxin.

KI toxin kill~ sensitive celb by forming ion channcls (Chapter J).

From the deduced amino acid sequence of the preprotoxin, it was

suggested that the two hydrophobie regions f1anking a central

hydrophilic region in the a subllnit may be responsible for forming thc

transmembrane channels (Bostian et al., 1(84). Our stùdy provides

evidence to support this idea. Mutant toxins unable l" form Ion

channels would be unable to kiJl spheroplasts, and such phenotype": arc

caused by mut:üions in the Tt'gions eneoding the two hydrophohle

domains, (ZH23, H21, pL318, ZH22, R4-35, R4-32, Tahlc J and 1:lg 1 ().

One distinct character of these mutations i~ that a chargcd atlllno acid,

such as Lys, Asp, and Arg, or an a-helix breaker rc\iduc Clly or Pro, wa ...

introduced into the hydrophobie regionl) (Tablc 3). The ... e change\ will

,

78

reduce the hydrophobicity , or penurb the ex. -helical structure of the

two rydrophobic ion channel-forming regian.

In addition, mutation pL7, localized at ti,'e hydrophilic N-tenninus

of the ex. subunit makes a mutant toxin incapabk of ki lling spheropla' . .;ts

indicating this rcgion also involved in ion-channel formation. Two other

mutations, ZH20 and ZH21, al the corling region 0) the C-terminus of the

ex. subunit, alIow secretion of arpraximately 80% 0\ toxin cornpared with

the wild type, but whose toxin activity toward celh and spheroplasts is

reduced 90%, suggesting a role of this region in ion ~hannel formation.

This h,ts been shown directly as an altered channel ar:tivity was

dctected by patch-clamp when ZH20 mutant toxin 'N.\) incorporated

into artifical liposomes (Chapter3). It is not known wh ~ther the regions

defined by pL 7, ZH20 and ZH21 are directly involved in IOn channel

formation, or these mutations caused extcnded folding perturbations

and indirectly 'liter the channel îorming d(Jmait~. None ot the mutations

in the region encoding the ~ subunit affect spheroplast killing activity,

arguing that this polypeptide 15 not involved in i0n channel formation.

The pheilomenon of toxin immunity adds complexity ta the killer

toxin system in toxin producing cells. Immunity occurs at the plasma

membrane level, and is conferred by the product of the toxin precursor

gene. where mutations map to the region encoding the Cl subunit of the

toxin (Boone et aL, 1986, Sturley et aL, 1986; Chapter 2). We have

testcd the ability of aIl mutants to confer immunity, and the results are

consistent with and extend previous findings. Mutations altering the

two hydrophobie regions of Cl were found to be defective in both ion

channel formation and immunity, as were ZH20 and ZH21 at the C­

terminal hydrophilic region of this subunit. The overlapping nature of

7 ()

these domains suggests they rn~ly compete or interact in lhe imlllunity

process, perhaps with the precursor preventing functlOnal channel

formation (Boene ct al, 1986; Bussey el ,tl .. 19(0).

In nature the use of toxie protcms is w\(kSpreal~: ~()me kili

sensitive cells by formmg ion channels, othcrs enter the l~lrget l'l'II 10

inhibit e5.sential processcs such as DNA or protcin synthe~i~. Pl'rtlllcnl

here are the weil studied colicins, many of which form IOn l'hannels;

and diphtheria toxin which like killer toxin consi'its of two disulphide­

linked peptide chains. The structural arrangement of eolicins and ot

killer toxin are very different, with the colicin'i conslsting of li ~lIlgle

toxie polypeptide with three distinct domains. These arc, il central

receptor binding domain necessary for binding at the outer IllCmh, ~\Ile,

a NH2-terminal translocation domain to get the protetrl through the

outer membrane; and a COOH-terminal itthal channel formlllg dOlllain

(Pattus et al., 1990). The irnmunity function of colicin~ IS conferrcJ by a

separate protein encoded on an immunity gcne (Konisky, 1 <JX2).

Diphtheria toxin consists of A (action ) and B ( reccptor hi nding )

peptide chains. The B chain is responsible for binding to a ccII surface

receptor (Neville and Hudson, 1986) and for translocating the toxill

through the plasma membrane to reach it~ cytosollc target. The A chain

has the toxin activity domain, which catalyzcs the ADP-riomylatlon of

elongation factor 2, inhibiting protein synthesis (Ncville and Il ud'-.oJl ,

1986; Van Ness et al., 1980). Our results indicate that the organl/allon

of functionaI domains of yeast KI toxin is dj~tinct l'rom tht.: cllphthefla

and the abrin and ricin classes of A/B tOXIllS. In the yea\t KI lox in the

hydrophilic ~ subunit does appear 10 be a " B" type \ubunit Jnvolved III

receptor binding. The a subunit in contnl,t i~ multlfunclJonal havlllg

..

.' 80

rcgJOn~ nccc..,sary for ion charmel formation. immunity and cell wall

rcccptor bmtiJng • that appear to overlap in the polypeptide.

Both the colicin group and diphtheria toxin contain a

"translocat:on" domé.lm nccc~sary Lo effect pas~age throueh a membrane

to dcliver the toxin molecule to lhclr appropriate cellular sites of action.

ln the case of the KI toxin. there IS no requirement for passage through

a memhrane fur the ~oxin to rt>ach its site of action at the plasma

membrane. and wc have not identified such a translocation domain.

The phenotype of mutants in s'Ich a domain would be an inactive toxin,

that could btnd to the cell wall receptor and kilt spheroplasts. The way

in which the toxin is translocated from the ~-glucan wall receptor to the

site of action at the plasmu membrane remains unknown. It is possible

that no formaI translocation step occurs, and that only a subset of ~­

glucan ft ..:eptor:-; are functional. for example, new]y synthesized glucan

near the bud tip, where the receptor may be in clo:-,e proximity to the

plasma membrane. Such :10 explanation is consistent with binding data

where only a fraction of binding sites are necessary for toxin action

(Bussey et al.. 1979) and with the early observation (Woods anù Bevan,

19(8) that the toxin is most active on growing cells.

Our reslJlts provide the first detaiIed study of the functional

domains of a l'ungal killer toxin. Protein toxins are widespread among

yeasts (y ung and Yagiu. 1978) and although for some the genes have

been sequenced IKluyvermnyces lactis (Stark and Boyd, 1986), Ustilago

maydis (Tao el al.. 1990) 1 for many thciT structure or mode of action

remams unknow(~. The K2 killer toxin of S cerevisille acts

physiologically in a similar way to the KI toxin (Bussey et al.. 1990). as

dors the toxin frorn PiC/lia kluy\. en which forms ion channels in vitro in

~ 1

Iipid bilayers (Middelbeek et al.. 1980; Kagan. 19ln). Th~Sl' tn\.ins Illa)'

be structurally or mechanistically similar to the KI to'\in. allhough

recent sequencing of the K2 protoxin gcne rcvcal'i no similarity \Vith

the K~ toxin at the amino acid sequence Icvel (Whitcway CI al..

unpublished resul ts).

(

(

82

Summary

The C-termim of the a and p ~ubunits of KI toxin have been

determined hy protein sequencing and amino acid analy~is of peptide

fragment ... generated fiom the mature, ~ecreted tOXIn. Il was revealed

that the (1. and 13 ~uhunJts onglnate l'rom the preprotoxin amino acid

reslLÏue ... 45-147 and 234-316, respectively. The location of the a and p

suhllnit~ wtthm the preprotoxin are, thereforc. completcd. and the

preprotoxin configura!ion can be repre~ented as prepropeptide-pro­

J\rg-(1.-Arg-J\rg-y-Ly~-Arg-p, where y represents the interstitial

glycosylatcd peptide. Thi~ configuration in which pairs of basic residues

f1ank mature polypeptides defined a specifie processlOg pathway

widesprcad among ellkaryotes in the processing of many secreted

hormones and neurotransmitters. which arc known to be synthesized

as precllrsors (Lynch and Snyder, 1986). This processll1g pathway

involv~~; an endoproteolytic cleavage at either mono- or dibasic

re~,idues, followed by a carboxypeptidase B like enzyme removlI1g the

COOII-terrlllnal basic amino acids (Lynch and Snyder, 1986).

ln yeast, the K EX 2 gene product is a dibasic endoprotease which

cleaves al the COOH terminal side of the paired basic residues of both

prepro-a-factor and preprotox1l1 (Julius, et al., 1984). Our work

(Chapter 1) suggested that the K EX 1 gene product may be a

carboxypcptidase B like enzyme, removing the basic residues from the

COOII tcr1ll1l1US of the a subunit of killer toxin, following the action of

the K EX.:? gcne prodllct. Cloning and sequencing of the K EX 1 gene

rcvcalcd that it has sig!1ificant homology at the amino acid level with

the carboxypeptidase Y (CPY)-a yeast vacuolar serine protease.

Functional studIt: ... of the K FX 1 gene produl't ln prncl""lIlg 01 preplO U

factor demon ... traled that il "'pcl'Iflcally Cled\'l'~ thl' ba,il' Il.·..,ldllC~ l'rom

the carhoxy-tcnl11nal end (Dlllochow,l-..a et al. Il)X7) Lata. Il \\'a, .11"0

shawn biocheTl1lcallv that the 1\ I:X 1 gene prodlll't l'ka\'l" the Arg

residue Iwm the COOIl-tenllln,t1 01 BPJ\,\ (l1el1l0yl phl'nylalanillL'

al.wInc-argifllne), Fulther more. cOlllpetltl\'l' InhlhItlOn .lIlaly"t'; 01 thl'

KEX 1 gcne product u'\lIlg BPJ\A a.., a ,\uh..,tratc III the pl e'l.' Il l"l' 01 N

blocked peptide" ~howed that only peptide.., \VIth a ha,ll' le'ldul' at Ihl'

COOH tcrmmu~ were capable of mhibitll1g K F ... X 1 product aCllvily ('UOPl'l

and Bussey, 1989).

Previou~ physlOlogical studie'\ of the action of KI 10\ III ,uggc..,tcd

that the to\in kills the scn"iitivc ccII" hy perturhlllg an l'Ilel gl,cd

plasma membrane ~tatc causing Ion leakage and "'Ub'CqllCllt L'cil dcath

(de la Pena et al. 1981), Hy u'\lI1g the patch-clamp Icl'lllllqUC. wc arc

able to show the toxin induced IOn channel" /1/ \'l\'(} wlth "'CIl,ttIVl' yca ... t

spheroplasts and ln vllro wlth artiflcwl a'\olectin Itpo ... ollle.., (Chapla .'"

The toxin induced ion channeb round both III Vl\'() and III \'lfI(I ... IJOW

the same fingerprint with Unit conductance of 118pS appeaflllg ln pair,

(sec Fig.4 and 5). They are voltage indcpcndcnt and prekr Illonovalent

cations. The finding that the KI tOXIll C~ln fmm Ion charlllcl ... III ... ell"tIVl'

yeast spheroplasts and that ~uch channel" are capahle of pa"'''IJlg Kt-, "

consistent with the ob~erved K+ Icakage dunng the ("noce ...... of cel!

intoxication (Skipper and Bu~\ey, 1(77). Wc cOIl"dcr that tht..: ... e

channels are likely the ha"l\ nt' klllJng hy thc KI tOXlll Such challnel"

would cause a leaky pathway for major phy ... rologrcal 1011 .... "uch a\ K-f

and H+, and dissipate the cellular Ionie gradlcnt Illcludlng the proton

gradi..-!nt necessary for physiological transport across the plasma

membrane.

84

The fact that such channels can be demonstrated in vitro in

artificial asolcctin liposomes does not nccessaflly imply that there are

no additional components involved in vivo toxin action at the plasma

membrane. As it was ~hown lJ1 chapter 4 that K! toxin can kill

sphcroplasts made from insen~Jtive cells of Candida albicans, Candida

utilis, Kluyveromyces laetls. and SchWannLOln)'CeS alluvius but fail to

kill spheropla~ts made l'rom the yeasts Candida hUlfler.SI.\', Hansenllia

mrakli, and P klu.vven, suggesting that there are other factors

neccssary for the toxin action with these spheroplasts. These factors

could include receptors, the membrane Iipid composition and proteases.

Whether thcre is a toxin specifie receptor at the plasma membrane to

mediatc toxin action is still obscure, A kre3 mutant- however, was

rcported to be resistant to the KI toxin at the spheroplJst level (Al­

Airdoor and Bussey 1978). This KRE3 gene has not been obtained.

Biologically, there is additional complexity in the process of toxin

immunity in toxin-producmg cells. immunity acts at the cytoplasmic

membrane levcl and is conferred by the preprotoxin. Site-directed

mutagenesls (Boone et al.. 1986) has defined the immunity domain

largcly from Val-86 to Ala-147 of the precursor. Determination of the

C-terminus of cr subunit has located this immunity domain completely

within the a subunit (Chapter 2). Although Ît is !.ne..vn that the

proccssing of \he precursor is not a prereq:Jisite for immunity as

cvidcnced by the fact that unprocessed toxin precursor retain full

immunity (Boone ct al.. 1986; Bussey et al., 1982; 1983; Lolle et al.,

1984: Wickncr. 1984). In chapter 5. we descri bed a further mutational

analysis of KI toxin and the results are consistent with and extend

previous findings. Several mutations (pL318, H21. ZII2J. Z1I22, R4-32.

R4-35) located at the twu hydrophobIc region,. a'; weil .lS lllutatioJ1,

ZH20 and ZH21 at the C-terminal hydrophillc regiol1 of the ex ,ubunil.

are defective In both ion channel forming and immunIty (sec Fig. 10).

This distinct overlapping nature by ion channel formation .1I1d

immunity domains suggests that they may compete for binding to the

putative membrane receptor or interact directly in the illlll1unity

process, perturbing the proper functional channel formatIon at the

plasma membrane level (Boone et al., 1986; Bussey et al., 19(0).

ln bacteria, the colicin producing strains havc an immunity

specificity similar to that of fung:tl yeast tOXIn producing strams. For

example, the colicin El producing strain is only immune to the colicin

El, but sensitive to other colicins which have the same modc 01 action

as colicin Eland vice versa (Konisky, 1982). It is found that each

immunity component is encoded by a different genc from that cncoJing

the colicins (Konisky, 1982). The immunity protein encoding gcncs for

colicin la, lb (Mankovich et al., 1986), El (Goldman et al, 1985), and A

(Lioubes et al., 1984) have been obtained and scqucnced. It was

revealed that these immunity proteins are cytoplasmlc memhrane

spanning proteins. In the case of colicin h· lb and E], the domain WhlCh

is involved in interaction with the immunity protcin is localil"cd al the

COOH terminus of these colicins, which also contains a ptJtative Ion

channel formation domain (Mankovich ct al., 1984; ) 986; Bi"hop et al,

1985). It is therefore, proposed that the immunIty may he cmfcrred hy

the immunity proteins intcracting directly wlth thc hydrophohie reglon

.... ' ............ ----------------------------------------.-------

86

of the colicin molecules, preventing proper ion channel formation by

the colicins.

To determine the functional domains of KI killer toxin we have

analyzed the phenotypes of a set of mutations throughout regions

encoding the CI. and J3 subunits that allow secretion of mutant toxins. We

have used a range of techniques to examine the ability of these mutant

toxins to bind to a (l~6)-J3-D-glucan cell wall receptor, and to form

lethal ion ehannels. Our data demonstrate that the hydrophobie ex

subunit tS multifunctional with ion channel forming cell wall receptor

binding and immunity domains, partially overlapped with each other.

The p subunit, on the other hand, is identified to contain only the cell

wall receptor binding domain, apparently localized at the NH2- and

CO OH termini of the P subunit. The A, B model of ricin and abrin,

therefore, is not applicable to killer toxin.

The set of mutations throughout the encoding regions of the ex and

J3 subunits produced man y modified killer toxins and such toxins will

be usefui for further study of various aspects of toxin function

including more detailed characterization of toxin-induced ion channels

both in vivo and in vitro, and interaction with the cell wall receptor.

. . X7

References

AJ-Aidroos, K., and Bussey, H. (1978). Chromosomal mutanls of Saccharomyces cerevisiae affecting the cell wall binding sitc for kilk. toxin factor. Cano J. Microbiol. 24:228-237

Baty, D., Frenette, M., Lloubes, R., Geti, V., Howard, S. P., PailliS, ..... and Lazdunski, C. (1988). Functional dom ai ns of colicin A. Mol. M /('rohlol :2: 807-811

Baty, D., Knibiehler, M., Verheij, H., Pattus, F., Shire, D., BCllladac. A., and Lazdunski, C. (1987). Site-directed mutagenesis of thc cooh­terminal region of colicin A: effect on secretion and vollagc-dcpl'Ildcnt channel activity. Proc. Natt. Acad. Sel. USA. 84: 1152-1 156

Bevan, E. A., Herring, A. J., and Mitchell, D. J. (J 97)). Prcliminary characterization of two species of ds RNA in yeast and thcir relalionship to the "killer" character. Nature 245:81-86

Bevan, E. A., and Somers, J. M. (1969). Somatie segregation of the killer (k) and neutral (n) cytoplasmic genetic determinants in yeast. Ge lIet

Res. 14: 71-77

Bishop, L. J., Bjes, E. S., Davidson, V. L., and Cramer, W. A. (1985). Localization of the immunity protein-reactive domain in unmodificd and chemically modified COOH-terminal peptides of colicin El . .1 Bacteriol. 164: 237-244

Boone, c., Bussey, H., Greene, D., Thomas, D. Y., and Vernel, T. (1 <JX6). Yeast killer toxin: site-directed mutations implicate the prccur~()r

protein as the immunity compont. Cel! 46:105-113

Boone, C., Sommer, S. S., Hensel, A., and Bussey, I-I. (1990). Yea~t KRE genes define a pathway of cell ail ~-glucan assembly . .1. Cell Inol 110: 1833-1843

Bostian, K. A., Bum. V. E., Jayachandran, S., and Tippcr. D. J. (19X3a). Yeast killer dsRNA plasmids are transcribed in vivo to produce full and partial-length plus-stranded RNAs. Nue/cie Acid Res 1 1 : 1077 - 1 O<J7

Bostian, K. A., Elliot, Q. A., Bussey, H., Hurn, V., Smith, A., and TipPCI, D . .J. (1984). Sequences of the preprotoxin ds RNA gene of type 1 killcr

-------- -- --- -------------

(

88

ycast: multiple processtng events produces a two component toxin. Ce Il 36:741-751

Bostian, K. A., Hopper, J. E., Rogers, D. T., and Tipper, D. J. (1980a). Translational analysis of the killer-associated virus-like particle dsRNA genome of Saccharomyces cerevisiae : M-ds RNA encodes toxin. r:ell 19:403-414

Bostian, K. A., Jayachandran, S., and Tipper, D. J. (1983b). A glycosylated protoxin in killer yeast: models for hs structure and maturation. Ccli 32: 169-180

Bostian, K. A., Sturgeon, J. A., and Tipper, D. 1. (l980b). Encapsidation of ycast killer double-stranded ribonucleic acids: dependence of M on L . .1. Bacteriol. 143 :463 -470

Brandt, W. F., Henschen, A., and Von HoIt, C. (1982). The nature of non­specifie peptide bond cleavage during the isothiot.-yanate degradation of proteins. Methods in protein sequence analysis (Elzinga, M., ed) pp. 101-110, Humana Press, Clifton, NJ

Bruenn. J., Bobek, L., Brennan, V., and Held, W. (1980). Yeast viral RNA polymerase is a transcriptase. Nucleie Aeid Res. 8 :2985-2997

Bruggeman, E. P., and Kayalar, C. (1986). Determination of the molecularity of the colicin El channel by stopped-flow ion flux kinetics. Proc. Natl. Acad. Sei USA 83: 4273-427(;

Buck, K. W., Lhoas, P., Border, D. J., and Street, B. K. (1973). Virus particles in yeast. Bioehem. Soc. Trans. 1:1141-1142

Bullock. J. O., Cohen, F. S., Dankert, 1. R., and Cramer, W. A. (1983). Comparsion of the macroscopic and single channel coundance properties of colicin El and its COOH-terminal tryptic peptide . .1. Biol. Chem. 258: 9908-9912

Bussey, H. (1972). Effects of ye.ist kiIler factor on sensitive cells. Nature New Biolop,y 325: 73-75

Bussey. H. (1981). Physiology of killer factor in yeast. Adv. Mierob. Physiol. 22:98-122

Bussey, H., Boone, C., Dmochowska, A .. Greene, D .. Zhu. Il .. Lolle. S . .1 .• Vernet, T., Dignard, D. And Thomas. D. Y. (19X8). secretion and action of yeast KI killer toxin, in: Viruses of" FUIlp,i mu/ SUI/l'Il' EIIAaryO!Cs. Mycology series. 7: 161-178. Eds Y. Koltin and M. J. Lci-howlt/. Marcel Dekker, New York 1988.

Bussey, H., Boone, C., Zhu, H., Vernet, T., Whiteway, M .. and Thomas, D. Y. (1990). GenetÏC and molecular approaches ta synthesis and actIOn of the yeast killer toxin. Experientia (Switzerland) 46: 193-200

Bussey, H., and Meaden, P. (1985). Selection and stability of ycast transformants expressing cDNA of an Ml killer toxin-ill1lllunity ~cnl'.

Curr. penet. 9 :285 -291

Bussey, H., Sacks, W., Galley, D., and Sdville, D. (1982). Yeast killer plasmid mutations affecting toxin secretion and activity and toxin immunity function. Mol. Ce Il. Biol 2: 346-354

Bussey, H., Saville, D., Greene, D., Tipper, D. J., and Bostian, K. A. (1 ~83). Secretion of yeast killer t0xin: processing of the glycosylated prccursor. Mol. Cell. Biol. 3: 1362-1370

Bussey, H., Saville, D., Hutchins, K., and Palfree, R. G. E. (1979). Binding of yeast killr!" toxin to a cell wall receptor of sensitive Sacclwromy('cs cerevisiae. 1. Bacteriol. 140: 888-892

Bussey, H., and Skipper, N. (1976). Killing of Torulopsis p,lahrata hy Saccharomyces cerevisiae killer factor. Antimicroh. Ap,ents Chemothcr. 9:352-354

Bussey H., and Sherman, D. (l973a). Yeast killer factor: ATP Icakage and coordinate inhibition of macromolecular synthesis in sensitive cclls. Biochim. Biophys. acta 298: 868-875

Bussey H., Sherman, D., and Somers, 1. M. (1973b). Action of yeast killer factor: a resistant mutant with sensitive spheroplasts . .1 BactcrlOl 1 13: 1193-1197

Bussey, H., Vernet, T., and Sdicu, A.-M. (1988). Mutual antagoni\1ll among killer yeasts: competition between KI and K2 killcrl\ and a novel cDNA-based K 1-K2 killer strain of Saccharomyces ccrevisUlc. Can .1 Microbiol.34:38 -44

90

Casadaban, M., and Cohen, S. N. (1980). Analysis of gene control signaIs by DNA fusion and cloning in E. coli. J. Mol. Biol. 138: 179-207

Cavard, D. and Lazdunski, C. (1981) FEMS Microbiol. Lett. 12: 311-316

Christie, D. L., and Gagnon, J. (1982). Isolation, characterization and N­terminal sequences of the CNBr-cIeavage peptides from human complement factor B. BLOchem . .1. 201 :555-567

Cleveland, M., Slatin, J., Finkelstein, A., and Levinthal, C. (1983). Structure-function relationships for a voltage-dependent ion channel: properties of COOH -terminal fragments of cotid . El. Proc. Nat!. Acad. Sei USA. 80: 3706-3 7 10

Chung, D. W. and Collier, R. J. (1977). The mechanism of ADP­ribosylation of elongation factor 2 catalyzed by fragment A from Diphtheria toxin. Biochim. Biophys. acta. 483: 248-257

Collier, R. J., and Kandel, J. (1971). Interaction of diphtheria toxin. 1. Thiol-dependent dissociation of a fraction of toxin into enzymatically active and inactive fragments . .l. Biol. Chem. 246: 1469-1503.

Cooper, A., and Bussey, H. (1989). Characterization of the yeast K EX 1 gene product: a carboxypeptidase involved in processing secreted precursor proteins. Mol. Celi. Biol. 9: 2706-2714

Cramer, W. A., Dankert, J. R., and Uratnai, Y. (1983). The membrane channel-forming bacteriocidal protein, colicin El. Biochim. Biophys. acta. 737:173-193

Cramer, W. A., and Phillips, S. K. (1970). Response of Esci1erichia coli­bound fluorescent probe to coHcin El. J. Bacteriol. 104: 819-825

Davidson, V. L., Brunden, K. R., and Cramer, W. A. (1984). Dependence of the activity of colicin Elin artificial membrane vesicles on pH, membrane potential, and vesicle size. J. Biol. Chem. 259: 594-600

Davidson. V. L., Brunden, K. R., Cramer, W. A., and Cohen, F. S. (1984). Studies on the mechanism of action of channel-forming colicins using artificial membrane. J. Membr. Biol. 79: 105-118

Davidson, V. L., Brunden. K. R .• and Cramer, W. A. (1985). Acidic pH requiremcnt for insertion of colicin El into artificial membrane vesicIes:

91

relevanee to the mechanism of colicins and certain toxins. Pro('. Nat/. Acad. Sei. USA. 82: 1386-1390

Davis, R. W., Thomas, M., Cameron, J., John, St. 'l'., Scheler, S., and Padgett, R.A. (1980). Rapid ON A isolations for cnzymatic and hybridization analysis. Meth. Enzymol. 65: 404-41 1

de la Pena, P., Barros, F., Gascon, S., Ramos, S. and Lazo, P. S. (1980). Primary effects of yeast killer toxin. Biochem BlOphys Res. Commull 96:544-550

de la Pena, P., Barros, F., Gascon, S., Lazo, P. S. and Ramos, S. (1981). Effeet of yeast killer toxin on sensitive cells of Saccharomyces cerevisiae. J. Biol. Chem. 256: 1 0420-1 0425

Deleour, A. H., Martine, B., Adler, J., and Kung, C. (1989). A modified reconstitution method used in patch-clamp studies of Esche!ïchia ('011

ion ehannels. Biophys. J. 56:631-636

Dmoehowska, A., Dignard. D., Henning, D., Thomas, D. Y., and Bussey, II. (1987). Yeast KEX 1 gene encodes a putative protease wilh a carboxypeptidase B-like funetion involved in killer toxin and ((-factor preeursor processing. Cel! 50: 573-584

Donovan, J. J., Simon, M., Draper, R. and Montai, M. (1981). Diphthcria toxin forms transmembrane channels in planar lipid bilayers. Pro(' Nat/. Acad. Sei USA.78: 172-176

Feingold, D. (1970). The mechanism of eolicin El action . .1 Memhrane Biol. 3: 372-386

Fink, G. R., and Styles, C. A. (1972). Curing of a killer factor ln Saccharomyces cerevisiae. Proe. Natl. AClld Sel USA 6<): 2846-2849

Frene tte , M., Knibiehler, M., Baty, D., Geli, V., Pattus, F., Verger, R., and Lazdunski, C. (1989). Interaction of coliein A domain ... with phmpholipid monolayers and liposomes: relevance lo the mcehanism of action. B iochemist ry 28: 2509-2514

Fried, H. M .. and Fink, G. R. (1978). Electron mlcro ... copic hetero-duplex analysis of "kil 1er" double stranded RNA spccic"i l'rom yca~t. Pro(' Natl Acad. Sei. USA 75 :4224-4228

(

92

Fu lier, R. S., Sterne, R. E., and Thomer, J. (1988). Enzymes required for yca~t rrohormone processing. Ann Rev. PhysLOI. 50:345-362

Gcli, V., Baty, D., Crozel, V., Morion, J., L1oubes, F. P. and Lazdunski, C. (19X6). A molecular genetic approach to the functioning of the immunity protein to colicin A. Mol. G€fl. Genet. 202: 455-460

Gill, D. M., and Dinius, L. L. (1971). Observation on the structure of diphtheria tOXIrl . .1. Biol Clzem. 246: ] 485-1491

Goldman, K., Suit, J. L., and Kayalar, V. (1985). Identification of the plasmid-encoded immunity protein for colicin Elin the mner membrane or Escherichia coll. FEBs Lett. 190: 319-323

Gopal, P. K., Shepherd, M. G., and Sullivan, P. A. (1984). Analysis of wall glucans from yeast hyphal and germ-tube forming cells of Candida alhicalls . .J Gen. Microbiol 130:3295-3301

Gould, J. M., and Cramer, W. A. (1977). Studies on the depolarization of the Escherichia coli cell membrane by colicin El. J. Biol. Chem. 252: 5491-5497

Gross, E., and Witkop, B. (1962). Nonenzymatic cleavage of peptide bounds: the methionine residues in Bovine Pancretic ribonuclease. J. Biol. Chem 237: 1856-1860

Guerry-Kopecko, P., and Wickner, R. B. (1980). Isolation and characterization of temperature-sensitive mak mutants of Saccharomyces cerevisiae. J. Bacteriol. 144: 1113-1118

Gunge, N., Tamaru, A., Ozawa, F., and Sakaguchi, K. (1981). Isolation and charactcrizat;on of linear deoxyribonuleic acid plasmids from K luyveromyces Laclis and the plasmid-associated killer character. J. Bacferio/. 145: 382-390

Gustin. M. C .. Martinac, B .. Saimi, Y., Culbertson, M., and Kung, C. (1986). Ion channels in yeast. Science 223: Il 95 -1197

Gustin. M. C., Zhou. X. L., Martinac, B .• and Kung, C. (1988). A mechanosensitive ion channel in the yeast plasma membrane. Scie nee 242:762-765

•

Halvorson, H. C. (1958). Studil'!S of protein and nuclcic acid turnovcr 111

growing cultures of yeast. Bioc/llIll BlOplz\'s clcta 27: 267-27~)

Hamill, O. P., Marty. A., Neher, E., Sakmann, B., Sigwolth, F. J. (19XI). Improved patch-clamp techniques for high-rcsolution ClIrll:nt recording from cells and cell-free membrane patehes. P(lIlI~(,IS ,\rel! 39:85-100

Hanes, S. D., bum, V. E., Sturley, S. L., Tipper, D. J., and Bostian, K. 1\. (1986). Expression of a cDNA derived l'rom the ycast kilicl prcprotoxlI1 gene: implications for processing and immunity. PrOf Nat! Ac(/c/ Sn USA 83: 1675-1679

Hanning, E. M., Williams, T. L., and Leibowitz, M. J. (19X6). The internaI polyadel1ylate tract of yeast killer virus MI double-strandcd RN 1\ is variable in length. Vlro!ogy 152: 149-158

Hermodson, M. A. (1982). Chemical cleavage of proteins. 111 Met/lOds ill protein sequence analysis (Elzinga, M., ed) pp: 313-323, Ilumana Press, Clifton, NJ

Herring, A. J., and Bevan, E. A. (1974,. Virus-like p:trticles associated with the double-stranded RNA ~pecies found in killcr and sensitive strains of the yeast Saccharomyces cerevisiae . .1. Gen. \lll·O!. 22: .1X7-394

Herring, A. J., and Bevan, E. A. (1977). Yeast virus-like particlcs posscss a capsid-associated single-stranded RN A polymerase. Nature 26 X :464-466

Hewick, R. M., Hunkapiller, M. W., Hood, L. E., and Drever, W. J. (19HI). A Gas-liquid phase peptide and protein sequonator. .1 Bw! Chern 256: 7990-7997

Hinnen, A., Hicks, J. B., and Fink, G. R. (l97X). Transformation of yeast. Proc. Natl. AClld. Sei. USA 75: 1929-1933

Holmes, D. S., and Quigley, M. (1981). A rapid boiling method for the preparation of bacterial plasmids. Anal Biochem 114: 1 <)3-1 <)7

Hopper, J. E., Bostian, K. A., Rowe, L. B., LlI1d Tlpper, D. J. (J <)77). Translation of the L-species dsRNA genome of the killcr-a ...... ociated

(

(

virus-like particles of Saccharomyces cerevisiae. 1. Biol. Chem. 252:9010-9017

94

Huang, Il. V., Bond, M. W., Hood, L. E., and Drever, W. J. (1981). Cleavage at Tryptophanyl rcsidus with Dimethyl Sulfoxide-Hydrochloric Acid and C'yanogcn Bromide. Melh Enzymol. 91: 318-324

lIutchins, K., and Bussey, Il. (1983). Cell wall receptor for yeast killer toxin: involvcment of a (1--"7 6 )-~-D-glucan. J. Bacleriol. 154: 161 -169

Icho, T., and Wickner, R. B. (1989). The double-stranded RNA genome of ycast virus L-A encodes its own putative RNA polymerase by fusing two open reading frames. 1. BlOl. Chem. 264: 6716-6723

Ittelson, T. R., and Gill, D. M. (1973). Diphtheria toxiil: specifie ~omretition for cell receptor. Nature 242: 330-332

Iwai, K., and Ando, T. (1967). N-O acyl rearrangement. M ethods EnzynlOl Il :263-282

ho, H., Fukuda, Y., Murata, M., and Kimura, A. (1983). Transformation of intact yeast cells with alaki cations. J. Bacteriol. 153:163-168

Jacks, T., Madh,.ni, H. D., Masiarz, F. R., and Varmus, H. E. (1988). Signais for ribosomal frameshifting in the Rous Sarcoma Virus gag-pol region. Cell 55: 447-458

Julius, D., Blair, L., Brake, A., Sprague, G., and Thomer, J. (1983). Yeast a factor is processed from a large precursor poplypeptide: the essential role of a membrane-bound dipeptidyl aminopeptidase. Cell 32: 839-852

Julius, D., Brake, A., Blair, L., Kunisawa, R., and Thomer, J. (1984). Isolation of the putative structural gene for the lysine-arginilte­cleaving endopeptidase required for processing of yeast prepro-a­factor. Cell 37:1075-1089

Kagan, B. L. (l9R3). Mode of action of yeast killer toxin: channel formation in lipid bilayer membrane. Nature 302:709-710

Kagan, B. L.. Finkelstein, A .• and Colombini, M. (1981). Diphtheria toxin fragment fors large pores in phospholipid bilayer membranes. P roc. Natl Acwi Sei USA 78: 4950-4954

Kalderon, D., Richardson, W. D., Markham, A. F .. and Smith, A. E. (1 ()~q). Sequence r~quirements for nuc\ear location of silllian VHUS --l-O large- r antigen. Nature 311: 33-38

Klapper, D. G .. Wilde, C. E., and Capra, J. D. (197X). Âutomatl'd amino acid sequence of small peptides utilizing polyhrl'nl'. Allal nlO('IIl'1I/ 8): 126-13l.

Konisky, J. (1982). Colicins and other bacteriocins with e~tahl ished modes of action. Annu Rev Microhiol. 36: 125-144.

Konisky, J., and Tokuda, H. (1978). Mode of actlon of l'olicin la : l'ffect of colicin on the Escherichia L:oli proton electrochcmical gradient. P, (1('

Natl. Acad Sci. USA.75: 2579-2583

Kunkel, T. A. (1985). Rapid and effIcient site-specifie mutagencsis without phenotypic selection. Proc Natl Acad Sei. USA H2: 4XH-492

Laemmli, U. K. (1970). Cleavage of structural prolcins dunng the assembly of the head of bacterophage T4. Nature 227:6HO-685.

Lazdunsky, c., Baty, D., Geti, V., Cavard, D., Morion, J., Lloublc ... , R., Howard, P., Knibiehler, M., Chartier, M., Varenne, S., Frenelle, M., Dasseux, J -L., and pattus, F. (1988). The membrane-channel forming colicin A: synthesis, secretion, structure, action and immunlty. BIOClzilll Biophys. acta 947: 445-464

Landon, M. (1977). Cleavage at Aspartyl-Prolyl BO/lds. M et/lOcI.\· Enzymol. 47:145-149

Leibowitz, M. J., and Wickner, R. B. (1976). A chromosomal gene required for killer plasrnid expression mating and sporulation in Saccharomyces cerevlsiae. PrOf Natl Acad Sei US A 73: 2061-2065

Leppla, S., Dorland, R. B., and Middlebrook, J. 1.. (1980). Inhl bit IOn of diphtheria toxin degradation and cytotoxic action by chloroqulfle . .1 IJIOI Chem.255: 2247-2250

LIoubes, R., Chartior, M., Journet, A .. Varene, S" and Ll/dun,kl, C'. (1984). Nucleotide sequence of the gene for the immunlty prolclfl to colicin A. Eur. J Biochem. 144: 73-7g

(

96

Lolle, S. J., and Bus~ey, H. (1986). In vivo evidence for signal c1eavage of the killcr preprotoxin of Saccharomyces cerevisiae. Mol. Celi. Biol. 6:4274-4280

LolIc, S. J .• Skipper, N.,Bussey, H., and Thomas, D. Y. (1984). The exprcssion of cDNA clones of yeast Ml double stranded RNA in yeast confcr!'l both killer and immunity phenotypes. EMBO J. 3:1383·1387

Lynch, D. R., and Snyder, S. H. (1986). Neuro;Jeptides: multiple molecular forms, metabohc pathways, and receptors. Annu. Rev. Bi()chem. 55: 773-799

Maniatis, 1'., Fritsch, E. F., and Sambrook, J. (1982). MolecuJar c1oning: A laboratory manual. Cold Spring Harbor Laboratory. Cold Spring Harbor, New York.

Mankovich, J., Lai, P.-H., Gokul, N., and Konisky, J. (1984). Organization of the colicin lb gene: promoter structure and immunity domain. J. Biol Chem 259:8764-8768

Mankovich, J., I-Isu, C. H. and Konisky, J. (1ge6). DNA and amine acid sequence analysis of structural and immunity genes of colicins la and lb. J. Bacleriol 168: 228-236

Martinez, M. C., Lazdunski, C., and Pattus, F. (1983). Isolation, rnolecular and functional properties of the C-terminal domain of coIicin A. EMBO .1. 2: 1501-1507

Meaden, P., Hill, K., Wagner, J., Slipetz, D., Sommer, S. S., and Bussey, H. (1990). The yeast KRE5 gene encodes a probable endoplasmic reticulum protein required for (1-6)-~-D-glucan synthesis and normal cell growth. Mol Cell BIOL 10: 3013-3019

Middelheck, E. J., Hermans. 1. M. H., and Stumm, C. ( 1979). Production, purification and properties of a Pichia kluyveri killer toxin . .1. Microhiol. Serol 45: 437-450

Middclheek, E. J., Stumm, c., and Vogels, G. D. ( 1980). Effects of Pichia kluyven klllcr toxin on ,ensitive cells. Antonie Van Leeuwenhoek. J. MlCro/JÎo/. Serai. 46: 205-220.

1

Mitchell, D. J .• Bevan. E. A., and Herring, A. J. (197 J). The L"ollclalillIl between dsRNA in yeast and the "killer character'. Ile/l'du\" 31: 133-134

Mizuno, K .• Nakarnura, T .. Oshima. T .. Tanaka, S., and l\1ahllo. II. (!<)SS). Yeast KEX2 genc encodes an endopeptida-.;e homologolls to 'lIhllll<,lIl­like serine proteases. B IOchem B iophys Res COmfllllll. 15{): '} i6-254

Mizuno, K., Nakamura, T .. Takada. K .. Sakaklbara. S., and MalslIo. II. (1987). A rnembrane-bound. Clic lum-depl'IHlent protease 1 Il yea"l ('(­cell c1eavaing on the earboxyl side of paircd ba.,lc rC'\ldues. UiodIl'11l

Biophy~ Res Commun. 144: 807 -8 J 4

Neville, Jr., D. M., and Hudson, T. H. (1986). Translllcmbl ane tran ... port of diphtheria toxin, related toxins. and eolicins. Allllll Rl'\' [hoc/I('m. 55: 195 -224.

Niwa, O., Sakaguchi, K., and Gunge, N. (19R 1). Curing of "-iller deoxyribonueleic aeid plasrnids of Kluyveromyccs la<'lls .1 Uac(l'!ïol. 148: 988-990

Ohno-Iwashita, Y., and Imahon, K. (1982). As ... ignment of the fllllcilonai loci in the colicin El molecule by charactcnzation of It ... protcolytlc fragments. J. Biol. Chem. 257: 6446-6451

Olsnes, S., and Pihl, A. (1973). Isolation and propcrtie ... of ahnfl' a tox;c protein inhibiting protein synthesis. Evidence for dlffcrcnt hlOlogical funetions of its two constituent-peptide chain~. Fllro{Jl'{/Il.l Bw('!tclll 35:179-185

Oppenheeilner, N. 1., and Bodley, J. W. (1981). DiphtherIa tOXIIl . ..,ite apd configuration of ADP-ribosylation of diphthalllide in clongation faclo! J. Biol. Chem. 256: 8579-8581

Palfree, R. G. E., and Bussey, H. (1979). Yca<.,t killcr tOXIIl: pUrIflcatlOIl and charaeterization of the protein toxin from S(lcclzaroI1/Yc('\

cerevisweo Eur J BIOChcnz 93: 487-493

Pattus, F., Heitz, F., Martmez, M. C., Provenchcr, S. W., and Lanlllll",kl, C. (1985). Secondary 'itructure of the porc-formlng coliclIl A and II'> ('­

terminal fragment ExperImentai faet and \lructurc prediction l'ur .1 Biochem 152: 681-689

(

(

Patlu~, F., Martincz, M. c., Dargent, B., Cavard, D., Verger, R., and Lazdun!'.kJ, C. (l9X3). Interaction of colicin A with phospholipid monolayer.., and IIpo~omc..,. B LOchernlst ry 22: 5698-5707

98

Pattu!'., F., Ma..,..,ottc, D., Wilm~en, II. U., Lakey, J., Tscrnoglou, D., Tucker, A., and parker, M. \V. (1990). Cohcins: ~)rokaryotic killer-pores. h'xfJertcntw 46' IXO-192

Pfeiffer, P., and Radier, F. (1982). Purification and characterization of extracellular and Illtracellular killer toxin of Saccharomyces cerevisiac strain 2X . .1 Gen MlcrohlOl. 128:2699-2706

Pfe If Te r, P., and RadIer, F. (1984). Comparison of the killer toxin of several yeasts and the purification of a toxin of type K2. Arch. of M i (' r () ln () 1. 1 37: 357 -3 6 1

Philliskirk. G., and Young, T. W. (1975). The occurence of killer .:haractcr In yeast of various genera. A ntonie Van ' .. eellwenhoek J. MICrolJlO1 Serol.41:147-151

PiszklCwic.l, D., Landon, M., and Smith, E. L. (1970). Anomalous cleavage of Aspartyl-ProlIne peptide bounds during amino acid sequence dClerminations. BLOchem Bio

1'7hys Res. Commun. 40: 1173-1178

Proia, R .. llart. D. A., Holmes. R. K., Holmes, K., and Eidels, L. (1979). Immunoprccipitation and partial characterization of diphtheria toxin­binding glycoprpteins from surface of guinea pig cells. Proc. Natl. Acad. Sn USA 76: 685-689

Pugslcy, A. P. (1984a). The ins and outs of colicins. Part 1: production and translocation across membranes. M icrobiol. Sei. 1: 168-175

Pugsley. A. P. (19H4b). The ins and outs of colicins. Part 2: Lethal action, IllltnunIty and eco~ogical implications. MicrobLOl. Sei 1: 203-205

Pugslcy. A. P. (1987). Nucleotide sequencing of the structural ge1ke for colicin N reveals homology between the catalytic, C-terminal domains of coliclll A and N. Mol Mlcro/Jlul. 1: 317-325

Raymond. 1... Slatin. S. L., and Finkelstein, A. (1985). Channels formed by colIc\I1 Elin plannar lipid bIlayers: translocatIOn of regions outside the channel-forming c'..nnain. J Mem BlOl. 84: 173-181

•

Reeves, P. (1965). The bacteriocins. Bacler/of Rc\·. 29: 24--+5

RidIey, S. P., Sommer, S. S., and Wickncr, R. B. (1984). SlIJ1l'[ -kil Ie[ mutations in Saccharomyces cerevisiae suprc<;s l''\.c1l1sion of M2 double-stranded RNA by L-A-HN and confer cold scn<;lllVlty in the presence of M and L-A-HN. Mol Cdl. BLOl 4: 76 i -770

Rogers, D., and Bevan, E. A. (1978). Group classlÎ;~'ation of k.illcr yeasts based on cross-reactions betwecn strains of diff ... rcnt spccics and origin. J. Gen. Microhiol. 105: 199-202

Sabet, S. F., and Schnaitman, C. A. (1971). Localization a[ . .1 solubilization of colicin receptors . .1 Bacteriol 108: 422-4.10

Sandvig, K., and Olsssnes, S. (1980). Diphtheria toxin entry into cclls lS

facilitated by low pH . .1 Cell. Biol. 87: 828-832

Sanger, F. G., Nicklen, S., and Coulson, A. R. (1977). DNA scqllcncillg wlth chain-termining inhibitors. Proc Nat/ Acad Sn. USA 74: 5463-5467

Schauer, 1., Emr, S., Gross, C., and Schekman, R. (1985). Invertase "igllal and mature sequence r1lbstitutions that delay inlercompartmc!ltal transport of active enzyme . .1. Cell BlOl. 100: 1664-1675

Schein, S., Kagan, B., and Finkelstem, A. (1978). Colicin K Hcts hy forming voltage-dependent channels in phospholipid planar bilayer Il1cmhranes. Nature 276: 159-163

Schmitt, M., and RadIer, F. (1987). Mannoprolein of thc yea"t cell wall as primary receptor for the killer toxin of Saccharomyces cerCVI.\'we strain 28 . .1. Gen. Microhiol. 133: 3347-3354

Schmitt, M., and Radier, F. (1988). Moleclliar structure o. the ccII wall receptor for killer toxin KT28 in Saccharomyces ccrcvlsiae . .1 BacterLOI. 170: 2192-2196

Schmitt, M., Brendel, M., Schwarz, R., and Radier, F. (19X9). InhibItion 01 DNA synthesis in Saccharomyces cercvisiae hy yeast killer toxln KT2X. 1. Gen. Microbio/. 135: 1529-1535

Sclafani, R. A., and Fangman, W. L. (1984). Con~crvation replicatlOn of double-stranded RNA in Saccharomyces cereviswe by di..,placcmcnt of progeny single strands. Mol Cell. BLOI. 4: 1618-1626

(

100

Singh, A., Chen, E. Y., Lugovoy, J. M., Chang, C. N., Hitzman, R. A., and Secburg, P. H. (1983). Nucleic Acids Res. Il :4049-4063

Skipper, N. (1983). Synthesis of a double-stranded eDNA transcript of the killer toxin coding region of the yeast Ml double stranded RNA . Biochem. Biophys. Res. Commun. 114: 518-525

Skipper, N., and Bussey, H. (1977). Mode of action of yeast toxins: energy requirement for Saccharomyces cerevisiae killer toxin. J. BacterlOl. 129:668-677

Skipper, , N., Sutherland, M. Davies, R. W., Kilburn, D., Miller, Jr, R. c., Warren, A., and Wang, R. (1985). Secretion of a baeterial cellulase by yeast. Science 230: 958-960

Skipper, N .• Thomas, D. Y., and Lau, P. C. K. (1984). Cloning and sequencing of the preprotoxin-coding region of the yeast Ml-dsRNA. EM BO l. 3: 107-111

Slatin, S. L. (1988). Colicin El in planar lipid bilayers. lnt. J. Biûchem. 20: 737-744

Somers, J. M., and Bevan, E. A. (1969). The inheritance of the killer charaeter in yeast. Genet. Res. 13: 71-83

Sommer, S. S., and 'Nickner, R. B. (1982a). Co-curing of plasmids affecting killer dsRNAs of Saccharomyces cerevisiae : [HOK}, [NEX], and the abundance of L 3re related and further evidence that Ml requires L. J. Bacteriol. 150:: 545-551

Sommer, S S., and Wickner, R. B. (l982b). Yeast L ds RNA consists of at least three distinct RNAs: ecidence that the non-mMendelian genes [ HOK], INEX] and IEXL] are on one of these dsRNAs. Cell 31:429-441

Stark., M. J. R., and Boyd., A. (1986). The killer toxin of KLuyverornyces Laetis: charactclIzation of the toxin subunits and identification of the gcnes which encode them. EMBO J. 5: 1995-2002

Steiner, D. F., Quinn, P. S., Chan, S. J., Marsh, J., and Tager, H. S. (1980). Processing mechanisms in the biosynthesis of protein. Ann. NY A cid. Sei. 343: 1-16

101

Stumm, C., Hermans, J. M., Middelbeek, E. J., Croes, A. F., and deVries, G. J. M. L. (1977). Killer-sensitive relationships in ycast from natural habitats. Antonie Van Leeuwenhoek .1. Microbio/ Serol 43: 125-12X

Sturley, S. L., Elliou, Q. K, LeVitre, J., Tipper, D. J., and Bostian, K. (19H6). Mapping of functional domains within the Saccharomyces ('t" ('\'/SlCll'

type 1 killer preprotoxin. EMBO J 5: 3381-3389

Sun, T., and Webster, R. E. (1986). fii, a bacterial locus rcquircd for filamentous phage infection and ils relation to colicin-tolcrant toi A and toi B. J. Bacteriol. 165: 107-115

Sun, T., and Webster, R. E. (1987). Nucleotide sequence of a gcne c1uster involved in entry of colicins and single stranded DNA of infccting filamentous bacteriophages into Escherichia coli J. 8acteriol 169: 2667-2674

Sweeney, T. K., Tate, A., and Fink, G. R. (1976). A study of the transmission and structure of double stranded RNA 'S associated with the killer phenomenon in Saccharomyces cerevisiae. Geneties 84: 27-42

Tao, J., Ginsberg, 1., Banerjee, N., Held, W., Koltin, Y. and Brucnn, J. A. (1990). U stilago maydis KP6 killer toxin: structure, expression in Saccharomyces cerevisiae and relationship to other cellular toxins. Mo/. Cel/. Biol. 10: 1373-1381.

Tarr, G. E., Beecher, J. F., Bell, M., and Mckean, D. 1. (1978). Polyquarrrternary amines prevent peptide loss from sequcnators. A na 1. Biochem. 83: 622-627

Thiele, D. J., Hanning, E. M. and Leibowitz, M. J. (1984a). Genome structure and expression of a defective interferring mutant of the killer virus of yeast. Virology 137: 20-31

Thiele, D. J., Hanning, E. M. and Leibowitz, M. J. (1 984b). Multiple L double-stranded RNA species of Saccharomyces cerevisiac: cvidencc for separate encapsidation. Mol. CeU. Biol 4: 92-100

Thiele, D. J., and Leibowitz, M. J. (1982a). Structural and functional analysis of separated strands of killer double-stranded RNA of yeast. Nue/eic. Acids Res. 10: 6903-6918

f

102

Thiele, D. J., Wang, R. W., and Leibowitz, M. J. (l982b). Separation and sequence of the 3' termini of M double-stranded RNA from killer yeast. Nucleic. Acids Res. 10:1661-1678

Thrash, C., Voelkel, K., Dinardo, S., and Sternglanz, R. (1984). Identification of Saccharomyces cerevisiae mutants deficient in DNA topoisomerase 1 activity. J. Biol. Chem. 259: 1375-1377

Tipper, D. J., and Bostian, K. A. (1984). Double-stranded ribonucleic acid killer systems in yeasts. Microbiol. Rev. 48: 125-156

Toh-e, A., Guerry, P., Wickner, R. B. (1978). Chromos omal superkiller mutants of Sacharomyces cerevisiae. J. Bacteriol. 136: 1002-1007

Toh-e, A., and Wickner, R. B. (1980). " Superkiller " mutations suppress chromosomal mutations affecting double-stranded RNA killer plasmid replication in Saccharomyces cerevisiae. Proc. Natl. Acad. Sei. USA. 77: 527-530

Van Heyningen, S. (1982). Similarities in the action of different toxins. Molecular action of toxins and viruses. Cohen and Van Heyningen ( eds.). Elsevier Biomedical Press. pp: 169-192

Van Ness, B. G., Howard, J. B., and Bodley, J. W. (1980). ADP-ribosylation of elongation factor 2 by diphtheria toxin. J. Biol. Chem. 255: 10710-10716

Vernet, T., Dignard, D., and Thomas, D. Y. (1987). A family of yeast expression vectors containing the phage fI intergenic region. Gene 52: 225-233.

Vodkin, M. H., and Fink, G. R. (1973). A nucleic acid associated with a killer strain of yeast. Proc. Natl. Acad. Sei. USA. 70: 1069-1072

Vodkin, M., Katterman, F., and Fink, G. R. (1974). Yeast killer mutants with altered double-stranded ribonucleic acid. J. Bacteriol. 116: 681-686

Weaver., C. A., Redborg, A. H., and Konisky, J. (1981). Plasmid·­determined immunity of Escherichia coli K-12 to colicin la is mediated by a plasmid-encoded membrane protein. J. Bacteriol. 148: 817-828

-,

103

Welsh, J. D., and Leibowitz, M. J. (1980). Transcription of killer vmon double-stranded RNA in vitro. Nucleic Acid Res R: 2365-2375

Welsh, J. D., and Leibowitz, M. J. (1982). Locali.lation of genes for the double-stranded RNA killer virus of yeast. Pro(' Natl. Acod. SCI USA. 79: 786-789

Welsh, J. D., Leibowitz, M. J., and Wickner, R. B. ( 19HO). Viron DNA­independent RNA polymerase from Saccharomyccs ecrc\'isiac. Nuc!de Acid Res. 8:2349-2363

Wesolowski, Algeri. A., Goffrini, P., and Fukuhara, H. (1982). Killcr DN;\ plasmids of the yeast KLuyveromyccs Lactfs. 1. Mutations effccting the killer phenotype. Curr. Genet. 5: 191-197

Wickner, R. B. (1974). "Killer character" of Saccharomyces CCI('VI.\'we:

curing by growth at elevated temperatures. 1. BacterfoL 117: 1356-1357

Wickner, R. B. (1974). Chromosomal and nonchromosomal mutations affecting the "killer character" of Saccharomyces cerevisillc. Gellc/u's 76:423-435

Wickner, R. B. (1978). Twenty-six chromosonu,1 genes needed to maintain the killer double-stranded RNA plasmid of Saccharomyces ce revisiae. Genetlcs 88: 419-425

Wickner, R. B. (1986). Double-stranded RNA replication 111 yeast: the killer system. Annu. Rev. Biochem. 55:373-395

Wickner, R. B., and Leibowitz, M. J. (I976a). Two chromosomal gcncs required for killer expression in killer strains of Saccharomyces cerevisiae. Genetics 82:429-442

Wickner, R. B., and Leibowitz, M. J. (I976b). Chromosomal gencs essentia1 for replication of a double-stranded RNA plasmid of Saccharomyces cerevisiae : the killer character of yeast. .1. M (JI BioL 105:427-443

Wickner, R. B., and Leibowitz, M. J. (1979). mak mutants of yeast: mapping and characterization. J, Bacteriol. 140: 154-1 GO

J

104

Wickner, R. B., and Ridley, S., Fried, H., and BalI, S. G. (1982). Ribosomal protein L3 is involved in replications or maintenance of the killer double-stranded RNA genome os Saccharomyces cerevisiae. Proc. Natl. Acad. Sel USA. 79: 4706-4708

Wickner, R. B., and Toh-e, A. (1982). [HOK] a new yeast nonMendelian trait, enablcs a replication-defective killer plasmid to be maintained. Genetics 100:159-]74

Williams, T. L., and Leibowitz, M. J. (198~/). Conservative mechanisms of the in vitro transcription of killer virus of yeast. Virology 158: 231-234

Woods, D. R., and Bevan, E. A. (1968). Studies on the nature of the killer factor produced by Saccharomyces cerevisiae. Genetics 82: 429-442

Wray, W., Boulikas, T., Wray, V. P., and Hancock, R. (1981). Silver staining of proteins in polyacrylamide gels. Anal. Biochem. 118: 197-203

Xu, S., Cramer, W. A., Peterson, A. A., Hernodson, M., and Montecucco, G. (1988). Dynamic prpperties of membrane proteins: Reversible insertion into membrane vesicles of a colicin El channel-forming peptide. P roc. Natl. Acad. Sei. USA.85: ;531-7535

Yamamoto, T .• Imai, M., Tachibana, K. and Mayumi, M. (1986). Application of monoclonal antibodies to the isolation anc! characterization of a killer toxin secreted by Hansenuia mrakii. FEBS Lett 195: 253 -257

Young, T. W., and Yagi u, M. (1978). A comparison of the killer character in different yeasts and its classification. Antonie Van Leeuwenhoek J. Microhiol. Seroi. 44:59-57

Zanen, J., Muyldermans, G. and Beugnier, N. (1976). Competitive antagonists of the action of diphtheria toxin in HeLa cells. FEBS Lett. 66:261-263

Zoller, M. J .• and Smith, M. (1982). Oligonucleotide-directed mutagenesis using M 13-derived vectors: an efficient and general procedure for the production of point mutations in any fragment. Nucl. Acids Res. 10: 6487-6500

10)

Contributions to original knowlcdgc

1. 1 determined the r,arboxyl-terminal sequences of the ex and p

subunits of the KI toxin and completed the localization of the mature

KI toxin within the preprotoxin and revealed a specifie processing

pathway which involves a dibasic cndoprotcasc and a carboxypcptidasl'

B like enzyme.

2. 1 was one of a group of investigators who have succcssfully

demonstrated that the yeast KI toxin forms ion channcls both III vivu

with sensitive yeast spheroplasts and in vitro with artifical liposomes.

The toxin induced ion channels are considered to be the eause of the

cell death.

3. 1 demonstrated that the KI toxin has a much wider killmg spectrum

at the spheroplast level than at the whole cell level, indicating that the

yeast cell wall plays an important role in determining toxin action and

can establish the apparent specificity of toxins. 1 also found KI toxin

can bind to the cell wall receptors of Candida albians, but this binding IS

not sufficient for toxin action, suggesting other factors in the ccII wall

required for toxin action.

4. 1 determined the functional domains of KI toxin, and showed that ion

channel fmmation domain is assigned exclusively to the a subunit. and

the cclI wall receptor binding domain is assigned 10 hoth (1. and p

subunits. This work ruled out a previous model that the functional

domain of KI toxin resembled ricin type toxin.

~ 1

106 Appendix

1') J~ l') ~o 70 . . ~ppr,MAA.H.vv.'.M Hr. ~C~ A.Ar, r;rA ~CC :A-A ';~A r~A ,TI ~r.~ TrI" GTC A&'T \TA. T':'A Til' ,TC Arr: ~C~ "A rn :AC r~~ ~,C r.TA ,r~

~ !,hf, !..yt PC') Thr \-;~, .al -eu '~'al \tg 3t!f ;dl "ler ae ~u Phe Phe :~C!! :nr l..eu Leu · .. u.s ...eu • .11 ,'.:11 -\...0.01

lJ 10

·

. , ) MA ._ 1 ~

2 i']

;H rGe ,-'TA rrA CAG C~T CCC ACA r.H r,c,r A-AT T'>(. ;',' ~G rrc IIrc ALC rcc CCT rCA i1'r CTA r;CG AGC GAT GCA CCT ,;rA. G7' A~r "op 'rp 'eu Lpu r;ln ~rll Ala rhr ,\,p ;1. A.n Trp 'olY c/. Ser [le Thr Trp :;1/ Ser Ph. Val Ala 'ler Aop Ua CLy Val Val Lle

.,Il '0 80

~90 11') 110 J JO 340 lsn J~ , . "TT IhT ATC MT mG TGT MG MC r';c GTG (,GT ,AG {(,T MG ';AT CAr HC IIGT IICG GAC rGC ccc MG CM ACA CT'! GCT i1'A CH ---Phe '.ly IL. A.n ,..1 [mJLY. Asnmv,.l '~ly' lu \r~ '1' A>p ",p Il. Ser Thr A.p~Gly Lv. Gin Thr Leu \la Leu Leu ':~L

91) 1110 110

17f} ~80 Ii~ W) .')() ;10 Hoall~20 :30 •• ,1 .;)

ACC ATT TT! ':TA ';CA Cil' "'CA Tce '.r.r , ... r '_Ar CTI ATA TCC GGT GGT MT ACC èëëGTG rCG CAG TCA GAr CCT MT GGC GCT ~,_~ > ••

Ser li. Ph. voll ~la Val Thr Ser ~ll lU. Hl" Leu Ile "rp Gly Gli Asn Mg Pro Val Sn Gin Ser .... p Pro Asn Gly .o.l. Thr ,al 121 IJ~ y ['III.unlt. (') 140

lùO 110 520 530 ." . ,:CT CGT CGT GAC ... rT TeT A~T --:TC ,;c ... ,,,\L LGG ..... T AIT '.0. CTC GAC rn Acr GCG TIG MC GAC ATA TTA MT CM CAT CGT ATT ACT "l .. Mi Mil A:lp lie :,er Thr 'Idl \l, "-'p Ch Asp li, Pr', :.eu ..... p Ph, Ser ,ua Leu Asn A:lp U. Leu Mn Glu Hl. Gly Ile Ser

l~;} 1~) 170

5~O ~90 blO 620

"H CT~ l'CA GC'! MC 4C' TCA CM nr '.TC MA \l.A rCA GAC ACA GeC CM tAC AC A:A \GT rrr GTA G,G ACC MC MC TAC ~CT

Ile Lau Pro ,ua ~.o.la Ser Gill :'yr Vtl lv~ ...,.~ Ser l,8p ~r .1.10 (,lu fUs ~. "lr , .. Phe Voli Val Thr Aan Œ!3)T,r Thr ;~r 181 • 'lI) 100

~40 Hl nfl ~SO ~bO 680 700 '10

TTG CAT ACC GAC CT~T CH l,cT ..... T Ge" c'u Hl. Thr A.p Leu Ile Hl. 'ill (,1,[3 '~ly

210

ACA TH KC \CG :'TT ACC ACA ' -~ ,_~. \TT crA ,;CA CTG GCe MC CGT rAT CTT TAT Thr Tu Thr Thr Ph. Thr Thr Pr l il. .le Pra Ala Val Ua Lv. Ar81Tyr 'Ial 7H

no _ 3-7oxl ~ ..

'1" 7:'J ISO 7~0 770 78'; 790 800 ~l'l · . . ~C'! ATG TroC 1,0\( CAr CGT He MG l,LC TCA r"c TCT HG CCC eTT MT CAr GCC IITC ~lG "CC G7r MT GGT MC CTG HT ~GA 'TA ~CA Pro> "'r G GLu lU. Gly lie Lys .ua Su rvr 0 '1et Ala Leu .... n \.sp A.ta ~er Val ,cr .ua "->n ~ly Mn Leu Tir Gly c.,,, \.Id

1 .. 1) ~ ')') !bO

~2G s,o 8bù 87 J d80 890 l J') · . ';M MC CTG TTT AGT GAG GU GAG GCA CM "JG .AG ACG vr TAC TH A.M rrc TA7 TCG \.r \C':' ';GC CAC TCC ATA ATG rCG \"C \.~,;

lolu Ln Leu Ph. 'io!r ~lu ..... p 'Jiu Cly Jin Trp "lu "lr .... n :--H T,r Ly. Leu Tvr Trp ,or :'hr ,Iv Gin Trp ILe '1er 1er 'er _" 110 280 ~90

HJ 920 930 ,40

rTT ATT Go\( GM AGT ATT G"T MC GCC MT MT GAC rn '.M Ph. Ile GLu ,;1" Ser ni .... p .un ,ua Mn oU" -\'op Ph. ,lu

lJO ll')

!,lùO 1'11 0 . • A"ècnCAGACCA

910 ?;) ~g 1 [ !") ?80 ~90

Ge.:: T~T GAC ACA ';GC :AC "\C ,;CCA7CGTGTCTCACCTC,r;UCCGATA-AGC GII ~ Asp Thr GI. lUs ~p --

Complete nucleotide sequence of the MI dsRNA preprotoxHI gcne from Bostlan et al. (1984).

The m-frame Ct and P tmon component N termmi are mdlcated by arrows. The speculated processmg sites for clcavage wlthm 0, by a leader pepudase, and between <l

and the central glycosylatcd rcglOn (y) are also mdlcated Cysteme resldues and potentlal asparagme glycosylauon slte~ are boxed (Bostlan el al., 1984).

The COOH termmus of the Ct subumt 15 at Alal47 of the preprotoxm, whlch IS

determllled by protem sequencmg and .muno aCld analy51s of peptide fragments generated From the secreted toxm (Chapter 2'.