Heterodimerization between two classes of homeodomain proteins in the mushroom Coprinus cinereus brings together potential DNA-binding and activation domains

Gene, 172 (1996) 25-31 0 1996 Elsevier Science B.V. All rights reserved.

GENE

25

037%1119/96/$15.00

09774

Heterodimerization between two classes of homeodomain proteins in the mushroom Coprinus cinereus brings together potential DNA-binding and activation domains (Transcriptional

activation;

transcriptional

Rachel N. Asante-Owusua,*, Lorna A. Casseltona

repression)

Alison H. Banhama,*

,

Heidi U. Biihnert”,

E. Jane C. Mellorb and

“Department ofPlant Sciences, University of Oxford, Oxford OX1 3RB, UK; and bMicrohiology Oxford OX1 3QU, UK. Tel. (44-1865) 175109

Unit, Department of Biochemistry,

Received by J.R. Kinghorn:

1995; Received at publishers:

20 November

1995; Revised/Accepted:

19 December/24

December

University of Oxford,

22 February

1996

SUMMARY

The A mating type-genes of the mushroom, Coprinus cineveus, encode two classes of homeodomain-containing distinguished as HDl and HD2 on the basis of conserved, but distinctly different motifs. Compatible mating

proteins partners

bring together versions of the proteins that can heterodimerize, thereby generating an active transcription factor complex that commits mated cells to sexual development. We have previously described a rare mutation in which an HD2::HDI gene fusion generates a ‘fused dimer’ lacking much of HDl including the homeodomain yet capable of constitutively promoting development [Kties et al., EMBO J. 13 (1994b) 4054440591. Here, we exploit this mutation to help identify contributions made by each protein class to normal heterodimer function. We show that the HD2 homeodomain is essential; deletion within the HDl homeodomain can be tolerated in a normal heterodimer, as well as in the mutant fusion protein, but not substitution of a critical amino acid. We define, by deletion analysis, an essential C-terminal region of the HDl and demonstrate its potential activation function by the ability to activate transcription in yeast when fused to the GAL4 DNA-binding domain. We also identify a potential role in transcriptional repression for the predicted C-terminal helix of HDl proteins.

INTRODUCTION

Mating in the basidiomycete fungus Coprinus cinereus (Cc) triggers early steps in sexual development. Provided mates have different alleles of genes at the A and B mating-type loci, somatic cell fusion is sufficient to convert the asexual monokaryon with uninucleate cells into

Correspondence to: Dr. L.A. Casselton, Department of Plant South Parks Road, Oxford OX1 3RB, UK. Tel. (+44-1865) Fax (+44-1865) 275074; e-mail: [email protected] * Equal first authors.

Sciences, 275109;

Abbreviations: aa, amino acid(s); bp, base pair(s); BGal, p-galactosidase (product of lacZ); Cc, Coprinus cinereus; GAL4, SC galactose regulatory PII SO378-1119(96)00177-1

a binucleate-celled dikaryon on which mushroom fruit bodies develop. A characteristic feature of the dikaryon is the formation of structures known as clamp connections which play an essential role in maintaining the binucleate condition of the cells. Both sets of genes play a role in forming these structures; different A genes promote clamp cell formation and different B genes

gene encoding Gal4 regulatory protein; kb, kilobase or 1000 bp; Hd, homeodomain (s); HD, Hd-containing protein; HD, gene (DNA, RNA) encoding HD; NLS, nuclear localization signal(s); ::, novel junction (fusion or insertion); nt, nucleotide(s); PCR, polymerase chain reaction; SC, Saccharomyces cerevisiae; Urn, U&ago maydis; XGal, 5-bromo-4chloro-3-indoyl-p-D-galactopyranoside.

26 their

fusion

to the adjacent

cell (Swiezynski

1960). The A locus of Cc, the subject several

genes that encode

1

of this report,

two classes

tran-

(Kties et al., 1992; 1994a). In the simpler b locus of the hemibasidiomycete fungus

Ustilago maydis (Urn) there are just two genes encoding analogous proteins of each class (Gill&en These basidiomycete fungi have multiple There

are some 33 versions

estimated mediated

et al., 1992). mating-types.

of the Urn b locus

classes, plays an essential role in self nonself recognition. A compatible mating brings together different allelic forms of HDl and HD2 proteins and in both Cc and (Banham et al., proteins encoded

that these can form heterodimers

1995; Kamper et al., 1995) whereas by genes within the same locus, and

there may be as many as five in Cc, are unable to heterodimerize. We have described a mutation in the A6 locus of Cc that has generated a chimeric gene, part HD2 and part HDI, that can be translated into a ‘fused dimer’ which constitutively promotes A-regulated development in the absence of any other A protein (Kties et al., 1994b). The chimeric protein no longer needs to recognise a compatible dimerization partner, and by deleting the 5’ end of the fusion gene, we showed that the discriminating N-terminal dimerization domain is not essential for heterodimer regulation of clamp cell development (Banham et al., 1995). The integrity of the fused dimeric complex is, however, showing that heterodimerization plays other critical roles in regulating the activity of the HDl and HD2 proteins. In this report we investigate some of the contributions made by each protein class to the function of the heterodimer.

RESULTS

N

::

II

/

1

632

/

HO2

\

,,,/

_ 'N ’N

l,,,,,

VAWrnEl 1

A6m

,,,,,,,,,, I ,,,,,,

:::

11,

A6md4

A6md5

A6md6

I

:::

1

597

381 ,,,, ,,,,,, ,,,,,,,,, ,, ,,l l,, l,,

MwDsr

1

N N

1

I

687

and an

160 of the Cc A locus. Heterodimerization via an N-terminal domain in both protein

Urn, it has been shown

520

\

contains

of putative

scription factors distinguished as HDl and HD2 on the basis of conserved but distinctly different homeodomain (Hd) motifs mating-type

N

HOI

HOZ

and Day,

AND DISCUSSION

Several of the A mating-type Cc genes have been sequenced and whilst different versions of the proteins have as little as 55% aa sequence similarity, the overall predicted structures are very conserved (Kties et al., 1994a). Fig. 1 illustrates the essential domains of each protein class, the derivation of the constitutive fusion protein and several deletions that we have generated by engineering the fusion gene. We will describe the effects of these deletions on A-regulated development following transformation of these constructs into suitable host strains.

?~a-hel,cal ?

regmns

OSer

(S), Thr CT), Pro (P) enrlched

Hod

? S,T,P ?

enriched.

posltlvely

charged

N NLS

fl S.T.P

ennched,

negatively

charged

1

Fig. 1. Predicted structural features of the HD2 and HDI A mating type proteins of Coprinus cinereus (Cc) and the constitutive fusion proteins generated

by gene fusion and engineering.

aa are numbered

from

the N terminus. The fusion gene, A6m, in pWFR1 (Kties et al., 1994b) was deleted internally using naturally occurring restriction sites, StuI in the HD2 sequence

derived from d-1

and either BglII or AsuII in the

HDl sequence derived from dl-1 to generate proteins A6md4 and A6md5, respectively. The C-terminal 94 aa of A6m were removed to generate A42mdl

protein A6md6 by digesting the genomic is encoded by an artificially generated

sequence with PstI. fusion between the

HDZgene (G-1 and the HD1 gene hl-1 which occurs in the A42 locus. The point of fusion in the naturally occurring mutant was reproduced exactly because hl-1 and dl-1 have a conserved sequence in this region. ~2-1 and hl-I were first fused using available restriction sites, and reverse PCR then used to delete the requisite number of bp to make the correct fusion point. pAMT2 (Tymon et al., 1992) was used to provide the hl-I sequence because this contains a truncated gene encoding a protein lacking the C-terminal 38 aa. Plasmids containing the genes encoding the proteins illustrated were introduced into Cc strain LT2 (A6B6 trp-l.l.l.6) by co-transformation with pCclOO1 which contains the Cc ~rp-I + gene. pFGC5, containing the a?-l:bJ-l fusion was tested in an A42 host (LN118, A42B42 trp-1.1,1.6) which represented the exact self background in which the self-compatible phenotype could be monitored. At least 50 tvp’ transformants were isolated from each experiment and screened microscopically for the presence ofclamp ceils. Media for Cc were those described by Mutasa et al. (1990). The transformation (1989).

procedure

was based on Casselton

and de la Fuente

Herce

(a) The HD2 but not the HDl homeodomain (Hd) is essential for dimer function The most striking feature of the fusion protein is that it has retained only the Hd of the HD2 as its potential DNA binding domain. This prompts the question of whether only a single Hd is necessary for the function of the heterodimer formed by individual HDl and HD2 proteins, and if so, whether the HDl and HD2 Hd are equally effective. The Hd comprises three M helices, the second two of which adopt a helix-turn-helix (HTH) structure (Kissinger et al., 1990; Wolberger et al., 1991). Crystal structures show that most of the invariant aa which are critical for making contacts with the major groove of

27 DNA

helix.

one of the resident

genes leads to A-regulated

The Cc HD2 has a typical Hd with the conserved WFXNXR residues in the recognition helix whereas the

ment; the formation

of unfused

HDl

occur

in helix

3, the so-called

Hd is classified as atypical

recognition

with WFXDXR

(Btirglin,

1994) (Fig. 2). The Trp residue (W) is the first invariant aa of the recognition helix and is situated at the hydrophobic

core where,

although

it does not make

with DNA, it plays a critical

stabilizing

contact

role in determin-

ing how helix 1 packs against helix 3. Using site directed mutagenesis we changed the conserved Trp residue in both

Hd to Ala (A). We also used PCR

to delete 9 aa

from the recognition helix of the HDl Hd and 8 aa from the HD2 Hd. The latter deletion was introduced into the protein

that contributes

fusion protein. the mutated

the single Hd of the constitutive

By replacing sequence

the normal

gene sequence

into

a

we were able to test the effect of

between

the proteins

suitable

host.

encoded

Heterodimerization

by the introduced

gene and

Clamp homeodomain

modification

cells

A5 host

HO1 gene bl-1

KDIDAWFIDARRRIGW

bl-lWa

KDIDAAFIDARRRIGW

bl-lhd

KDIDA -GW

HO2

b2-3 a2- 1

A

gene

b2-3/a2wa hd

1

RQIEVWFQNHRRRAR RQIEVAFQNHRRRAR RQIEV

AR I

I A

HD2:HDl

fusion

gene

A6m

RQIEVWFQNHRRRAR

A6m hd

RQIEV,

would normally

, AR A

Fig. 2. Mutational analysis of the HDl and HD2 Hd. A Trp to Ala change was effected using the CLONTECH transformerTM site-directed mutagenesis kit to convert the TGG codon in bl-f and b2-3 to GCG and generate bl-1”” and b2-3”“. Plasmids used for subcloning mutagenized regions were pAMT2 (Tymon et al., 1992) for bl-1 creating pB1 W and ~A625 for b2-3 (Banham et al., 1995) creating pB2W. Deletions were effected using reverse PCR; deleting nt 4955520 in the bl-1 gene to give bl-lhd and nt 580-651 in the a2-1 gene to give a2-lhd and A6mhd. Mutated regions were sequenced and then subcloned to replace the appropriate wild-type sequence in bl-1 (pBHl), ~2-1 (pAH1) and the mutant fusion gene A6m (pA6d3). Genes were tested for their ability to promote clamp cell development in the host FA2222 (ASB6 trp-1.1,1.6) by cotransformation with pCclOO1 as described in the legend to Fig. 1.

clamp cells (Tymon

et al.,

genes together with wild-type into a host strain in which they

meet a compatible

dimerization

partner

(a host with an A5 locus) and promote clamp cell development (Fig. 2). The control genes were able to promote clamp cells as expected. lated

HD2

coding

None

domains,

fusion gene, promoted

of the genes with manipuincluding

the constitutive

clamp cell development.

We con-

clude that the HD2 Hd is essential for A protein function. Changing the critical W residue to A in the HDl Hd led to loss of function whereas removal of the entire recognition

by

this deletion on the constitutive protein (Fig. 2). The activity of a cloned A gene can be tested by transformation

1992). The manipulated controls were introduced

develop-

helix

left a fully functional

protein.

As with

the

mutant fusion protein, the potential DNA-binding properties of the HDl Hd are not the essential contribution of the HDl

to the heterodimer.

The W to A change within

the Hd is likely to cause a steric effect that could prevent a normal interaction with DNA. The target site of the HDZHDl dimer is, as yet, unknown and we do not know how dimerization affects DNA binding activity. Our data allow us to predict that in a normal heterodimer both Hd normally contact DNA but that the binding of the HD2 Hd is more critical. Our conclusions are different from those of Luo et al. (1994) who tested the effect of more extensive deletions in the HDl Hd of an A mating type protein (AclZ3) from another mushroom species, S. commune. As with the deletion we describe, these had no detectable effect on protein activity. The authors concluded that the HDl function relies simply on its dimerization with its HD2 partner and that its Hd has no role in DNA-binding. However, no single aa substitutions analogous to the one we describe were made. Relevant to our experiments are those with the homeodomain mating type proteins al and a2 of Saccharomyces cerevisiae (SC). These proteins, encoded by genes in the alternative versions of the MAT locus, have Hd motifs analogous to the basidiomycete HD2 and HDl motifs, respectively. Subunits al and a2 heterodimerize following mating and generate a new transcription factor complex that binds unique sites upstream from developmentally regulated genes. Neither Hd alone has strong affinity for DNA, but interaction between a C-terminal domain of the a2 protein with the al Hd increases both the affinity for DNA-binding and the specificity of operator site selection (reviewed by Johnson, 1995). Interestingly, mutational alteration of critical aa residues in the recognition helix of the a2 Hd had no effect on dimer function (Vershon et al., 1995) suggesting that the contacts of the al Hd are more important. This is further suggested by the fact that the specificity conferred by cooperative binding with ~12 could be reconstituted in vitro if the al Hd was fused to just the C-terminal 22 aa tail from a2 (Stark

28 and Johnson,

1994). The recent

resolution

of the crystal

structure of the al/a2 heterodimer bound to DNA confirms that both Hd contact DNA (Li et al., 1995). It reveals

that the C-terminal

an amphipathic this interaction permits

tail of the a2 protein

helix that contacts introduces a bend

protein-protein

forms

the al Hd and that into the DNA that

and protein-DNA

contacts

that

Tymon

et

C-terminal

al.

(1992)

together

dimerization

with

domain.

the

potential

In this study we have

not tried to assess the importance It is interesting to note, however,

of this internal region. that we are unable to

predict NLS in the HD2 sequence and it may be that heterodimerization plays an important role in targeting a functional HDZ/HDl heterodimer to the nucleus and

are not possible on straight DNA. The artificial reconstitution of DNA-binding site specificity with just the

that both NLS are required

HD2-type al Hd may well reflect what has occurred in the fused Cc A protein dimer. The regions just C-terminal

final 38 aa which are predicted to form a C-terminal helix. This has previously been shown to be non-essential for

to the homeodomains

HDl

in both HDl

and HD2 A proteins

Of the two C-terminal

function

constitutive

ization

C-terminal

domain

that

serves

a similar

function

to the

mutant protein.

fusion

protein

and

in the

Hd

deleted

HDl

(b) The HDl C terminus contains an essential region which enables the mutant fusion protein to constitutively promote clamp cell development The constitutive fusion protein contains the first 387 aa of an HD2 fused to the C-terminal 394 aa of an HDl. The HDl sequence is essential for constitutive function (Kties et al., 1994b). Here we define more precisely the essential regions of this sequence by making two internal and two C-terminal deletions of the fusion protein (Fig. 1). The constitutive protein function was tested by introducing the gene into a host strain that has an A6 locus. The mutation arose in this locus and an A6 host contains no A proteins that could be compatible dimerization partners for the manipulated protein. This is important because the HD2 region of the fusion protein can still interact with a compatible HDl (Kties et al., 1994b). One of the constructs (see legend to Fig. 1) was an artificially derived fusion between two genes from the A42 locus and this was tested in an A42 host. The two internal deletions in the fusion protein lacked an additional 6 aa from the C-terminal domain of the HD2 (a2-1) that have previously been shown to be nonessential (Kties et al., 1994a) and either 116 aa or 210 aa from the HDl (dl-1). The shorter deletion resulted in a protein that retained its constitutive function but the clamp cell phenotype was poorly expressed with only a few cells producing these structures. We conclude that this protein is very inefficient at binding its target sites. The longer deletion led to complete loss of function. This allowed us to define the region beyond aa 503 as being absolutely essential for HDl function. The deletions removed either one or both of the bipartite nuclear localization signals (NLS) predicted by

using the clamp

the first removed

cell assay (Tymon

1992) and we now show that it is also inessential

are predicted to form amphipathic helices and we have previously suggested that these might constitute a dimerC-terminal interaction of al and a2 (Kties et al., 1994a). Significantly these helical regions are retained in both the

for efficient transport.

deletions,

activity

of the fusion protein.

94 aa, however,

removed

Removal

a negatively

the

et al., for the of the charged

domain adjacent to the terminal helix and lack of clamp cell development shows that this must be essential for heterodimer function. (c) The essential region of the HDl C terminus contains a region which can activate transcription in yeast Deletion analysis essential C-terminal

of the fusion protein identified an domain in normal HDl proteins

between aa 386 and 593. We have asked whether this region contributes an essential activation domain. We fused regions of an HDl gene (bl-1) to sequences encoding the DNA binding domain of the SC Gal4 protein. The translated fusion proteins bind the GAL4 target site and by using different reporters (Fig. 3) we assessed whether

Yeast

Integrated

strain

Repotter

,

310bp GGY1:171 UAS G

40bp

72bp I

SS38-G4

17mers

site

NJAS,l I

ss19-8

I HISIUAS

I

l-l GCN4

1

48bp

,

3Zbp

,

A

17mer DJASG]

Fig. 3. Succharomyces cereoisiae (SC) strains used to assay transcriptional regulation by Gal4 fusion proteins. GGYl:171, a strain for assaying transcriptional activation, was described by Gill and Ptashne (1987). Plasmids SS38’-G4 and SS19-8, for measuring transcriptional repression (Saha et al., 1993), and strain MLY530/18 were gifts from S. Saha and M. Ptashne. The plasmids were linearized with StuI and integrated into the URA3 locus of strain MLY530/18 (MATa, d&4, Agal80, leu2, uru3, trp, his) to give strains SS38-G4 and SS19-8. UAS, is the upstream activation site or binding site for the Gal4 regulatory protein. The 17.mers represent consensus Gal4-binding sites and are designated [UAS,]. HIS4 UAS is the binding site of the HIS4 regulatory protein.

29 the fusion proteins and Ptashne, et al., 1993). The fusion

had any ability

1987) or repress constructs

to either activate

(Gill

gene transcription

illustrated

dieted C-terminal

(Saha

helix which gave negative

results in the

activation assay and is dispensable in vivo. The C-terminal domain is highly conserved

in Fig. 4 were trans-

proteins

for which the sequence

in all HDl

has been predicted

(Kiies

formed into the SC strain GGY1:171 and assayed for their ability to activate transcription of the 1ac.Zreporter gene.

et al., 1994b; Gieser and May, 1994; Gottgens, 1994) as is the HDl Hd. Our deletion analysis of an HDl, how-

The data, summarized in Fig. 4A, indicated that certain regions of the HDl protein could activate transcription

ever, showed that both these domains

of lacZ. The different

levels of BGal activity

detected

regulating the phenotype function of the HDl/HD2

in

transformants are probably not significant because problems may arise due to different stability and conformation

opment).

of partial

is required

proteins

(Fields

and

Sternglanz,

constructs used defined a potential between aa 486 and 593.

1994). The

activation

domain

SS38-G4

(Fig. 4B)

or

strain

SS19-8

which we currently

lower than

those observed

for the vector

control

(p13H). This occurred when placed downstream (SS19-8) and more significantly, upstream (SS38-G4) of the promoter. This region of the HDl protein encodes the pre-

A bl-1

I

protem

tragment

:olour 1, :: v632

pAB32

++

IJ 1500

that

HDl

target

have

~12protein

site upstream

cells. This it does in the general transcrip-

have no phenotypic

assay.

C

plasmid pAB44 pALI

l-163

0 pAB2

pAE21 pAtI pAtl29 pAB22

a different

B

cDNA~ pl99mla

therefore,

(e) Conclusions between the HDl and HD2 Hd (I) Heterodimerization mating-type proteins of Cc brings together two proteins with potential DNA binding properties. Our mutational analysis of the Hd sequences indicates that the HD2 domain is critical for heterodimer function and that the HDl Hd is inessential but probably participates in binding the target site. Our observations suggest that there may be a parallel with the analogous al and a2 Hd

These

strains constitutively express PGal activated by the promoters illustrated in Fig. 3. Only one construct, pAB29, was able to repress transcription of 1acZ to levels significantly

for binding

for

tion factor MCMl (reviewed by Johnson, 1995). Like a2, HDl proteins might bind other DNA target sites to regulate genes not involved in the clamp cell pathway and for

gene the coninto SC strain

(Fig. 4C).

possible,

In SC the Hd of the HDl-like

from genes expressed in unmated association with another protein,

(d) The non-essential C-terminal helix from the HDl protein bl-1 can repress transcription in vitro To test whether any of the fusion proteins were able to repress transcription of a 1acZ reporter structs shown in Fig. 4A were transformed

It seems

other functions.

are dispensable

that enables us to assess the heterodimer (clamp cell devel-

593

?632 ?

++

1300

+

0.5 0

164m262

pAB31

30.5

-’

pAB33

305~381

kva

pAB30

366ma632

pAB34

486

632

++

1

pAB24 pAB29 pAB22

0

pAB3 1

4800

pAB33

0

pAB34

++

NT

++

2300

pl3H “One

ma632

0

0 ::

z !z

pGal activity

8 ” (units)

8

: N

z “7

:: r-

:: pGal activity

z z

(units)

Fig. 4. Identification of putative activation and repression domains in the HDl protein bl-1. (A) Transcriptional activation by Ga14::bl-1 fusion proteins. The restriction sites in 62-l cDNA are represented by: C, ClaI; M, MU; N, NruI: Pv, PuuII; P, PstI and aa in protein fragments by numbers, cDNA sequences were fused in frame to the region of the GAL4 gene encoding the DNA binding domain in p13H (the ADH promoter driving expression of amino acids l-147 of the Gal4p). Plasmids were transformed into SC strain GGY:171 and cells assayed for ability to produce l3Gal both on XGal plates and by enzyme assay. The expression product of pAB30 was toxic and although blue colour was observed on XGal plates the transformants were not tested (NT) for BGal activity because they did not grow sufficiently. (B) and (C). Transcriptional repression by Gal4::bl-1 fusion proteins. Constructs illustrated in (A) were transformed into SC strains SS38-G4 (B) and SS19-8 (C) and assayed for BGal synthesis. /3Gal activity was quantitated using a permeabilized cell assay (Guarente, 1983). Three transformants were assayed in duplicate for each plasmid transformed and results are expressed as mean activity obtained. ADH = gene encoding alcohol dehydrogenase.

30 mating-type correct

proteins

DNA

target

from SC. The al Hd is sufficient site selection

providing

for

it is associ-

ated with an essential part of the ~12protein which affects its conformation (Stark and Johnson, 1994; Vershon et al., 1995). The ability of one class of protein

to function

without an Hd implies that protein-protein interactions in a heterodimeric association can be more critical than contacts

made to DNA.

(2) We have demonstrated terminus

of the HDl

protein

experimentally contains

that the C

a region

between

aa 386 and 593 that can activate transcription in SC. This region has been shown to be essential in vivo for the constitutive activity of the mutant fusion protein in promoting clamp cell development. Our data is thus consistent with the conclusion that in the HD2/HDl heterodimer,

the

HD2

contributes

binding domain and the HDl activation domain.

an

provides

essential

DNA

a transcriptional

(3) Two features of the Cc HDl/HD2 heterodimer are puzzling in the light of our deletion analysis; the extreme conservation of the apparently dispensable HDl Hd and the presence of an equally conserved and apparently dispensable C-terminal domain in HDl proteins that has a potential role in transcriptional repression. This C-terminal domain is also conserved in the corresponding HDl mating-type proteins of another mushroom S. commune (Stankis et al., 1992). This leads us to suggest that HDl proteins might bind more than one target site and, by having a different transcriptional role, employ functional domains not required by the HD2/HDl heterodimer.

compatible

partners

room Coprinus Btirglin,

and initiate Plant

cinereus.

T.R.: A comprehensive

Deboule, Casselton,

classification

D. (Ed.), Guidebook

University

sexual development

in the mush-

Cell 7 (1995) 773-783. of homeobox

to the Homeobox

genes. In:

Genes.

Oxford

Press, 1994, pp. 27771.

L.A. and de la Fuente

sion in the basidiomycete

Herce, A.: Heterologous

fungus Coprinus

gene expres-

cinereus. Curr. Genet.

16

(1989) 35-40. Fields, S. and Sternglanz, R.: The two-hybrid system: an assay for protein-protein interactions. Trends Genet. 10 (1994) 2866292. Gieser,

P.T. and May, G.: Comparison

mating-type

of the

(1994) 167-176. Gill, G. and Ptashne, tion function. Gillissen,

of two bl alleles from the A

basidiomycete

M.: Mutants

Coprinus

of GAL4 protein

B., Bergemann,

J., Sandmann,

in Ustihgo

maydis. Cell 68 ( 1992) 6477657.

B.: Analysis

Guarente,

of Mating

L.: Yeast promoters

Johnson,

of cloned

A.D.:

budding

Type Protein and

lacZ

mechanisms

yeast. Curr. Opin. Genet.

in Coprinus

Oxford,

fusions

M. and

recognition

Interactions

of Oxford,

genes in yeast. Methods

Molecular

B., Balker,

system for self-non-self

D.Phil. Thesis, University

expression 181&191.

altered in an activa-

C., Schroer,

R.: A two component

cinereus.

146

Cell 51 (1987) 121-126.

Kahmann, Giittgens,

Gene

cinereus.

UK, 1994.

designed

to study

Enzymol.

of cell-type

101 (1983)

determination

in

Dev. 5 (1995) 552-558.

Kamper, J., Reichmann, M., Romeis, T., Bblker, M. and Kahmann, R.: Multiallelic recognition: nonself-dependent dimerization of the bE and bW homeodomain

proteins

in Ustilago

maydis. Cell 81 (1995)

73-85. Kissinger,

CR.,

C.O.: Crystal

Liu, B., Martin-Blanco, structure

E., Kornberg,

of an engrailed

plex at 2.8 A resolution:

a framework

for understanding

main-DNA interactions. Cell 63 (1990) 5799590. Kties, U., Richardson, W.V.J., Tymon, A.M., Mutasa, B., Gaubatz,

S., Gregoriades,

tion of dissimilar proteins ment

contain

T.B. and Pabo,

homeodomain-DNA

A. and Casselton,

comhomeodo-

E.S., Glittgens,

L.A. The combina-

alleles of the Ar and A/I gene complexes. homeo

in the mushroom

domain Coprinus

motifs, determines cinereus.

Genes

whose

sexual developDev. 6: (1992)

568-577.

ACKNOWLEDGEMENTS

We thank Dr. S. Saha and Dr. M. Ptashne for sending us plasmids SS19-8, SS38-G4 and SC strain MLY530/18. We thank Drs. Ian Connerton and Ursula Kties for help with PCR strategies and site directed mutagenesis, respectively. H.U.B. thanks the Gatsby Charitable Foundation for a summer studentship and E.J.C.M. the Welcome Trust for a Senior Research Fellowship. This work was supported by Biotechnology and Biological Sciences Research Council grant No. PG043/0564 to L.A.C. and E.J.C.M., a research studentship to R.N. A.-O., and a postdoctoral fellowship to L.A.C.

Kties, U., Asante-Owusu, R.N., Mutasa, ES., Tymon, A.M., Pardo, E.H., O’Shea, S.F., Giittgens, B. and Casselton, L.A.: Two classes of homeodomain proteins specify the multiple A mating types of the mushroom Coprinus cinereus. Plant Cell 6 (1994a) 146771475. Kties, U., Gottgens, B., Stratmann, R., Richardson, W.V.J., O’Shea, SF. and Casselton, compatibility Coprinus

L.A.: A chimeric and constitutive

cinema

EMBO

homeodomain

protein

sexual development

causes self-

in the mushroom

J. 13 (1994b) 4054-4059.

Li, T., Stark, M.R., Johnson, A.D. and Wolberger, C. Crystal of the MATal/MATa2 homeodomain heterodimer bound Science 270 (1995) 2622269. Luo, Y., Ullrich, R.C. and Novotny, Schizophylhm

commune,

C.P.:

Acx mating-type,

Only

structure to DNA.

one of two paired

homeobox

genes encodes

a homeodomain essential for Ace-regulated development. Mol. Gen. Genet. 243 (1994) 2233229. Mutasa, ES., Tymon, A.M., Giittgens, B., Mellon, F.M., Little, P.F.R. and Casselton, L.A.: Molecular organization of an A mating type factor of the basidiomycete fungus Coprirzus cirrereus. Cm-r. Genet. 18 (1990)223-229. Saha, S., Brickman, J.M., Lehring, N. and Ptashne, M.: New eukaryotic transcriptional repressors. Nature 363 (1993) 648-652.

REFERENCES

Banham, A.H., Asante-Owusu, Kingsnorth, dimerization

R.N., Gottgens, B., Thompson, S.A.J., C.S., Mellor, E.J.C. and Casselton, L.A.: An N-terminal domain permits homeodomain proteins to choose

Stankis, M.M., Specht, C.A., Yang H.. Giasson, L., Ullrich, R.C. and Novotny, C.P.: The Act mating type locus of Srhizophyllum commune encodes two dissimilar, multiallelic, homeodomain proteins. Proc. Natl. Acad. Sci. USA 89 (1992) 716997173. Stark, M.R. and Johnson, A.D.: Interaction between two homeodomain

31 proteins

is specified

by a short

C-terminal

tail. Nature

371 (1994)

4299432. Swiezynski,

lacking K.M. and Day, P.R.: Heterokaryon

formation

in Coprinus

lagopus.Genet. Res. Camb. 1 (1960) 114-128. Tymon,

Vershon,

A.M., Kties,

fungal mating

U., Richardson,

type protein

opment contains 1805-1813.

that regulates

a POU-related

sexual and asexual

domain.

EMBO

L.A.: A devel-

J. 11 (1992)

specific side chains

transcriptional Wolberger,

W.V.J. and Casselton,

A.K., Jin, Y. and Johnson repression.

C., Vershon,

A.D.:

A homeodomain

protein

of helix 3 can still bind DNA and direct Genes Dev. 9 (1995) 182-192.

A.K., Liu, B., Johnson,

A.D. and Pabo,

CO.:

Crystal structure of a MAT ct2 homeodomain-operator complex suggests a general model for homeodomain-DNA interactions. Cell 67 (1991) 517-528.