In vivo aggregation of the HET-s prion protein of the fungus Podospora anserina

In vivo aggregation of the HET-s prion protein of thefungus Podospora anserina

Virginie Coustou-Linares,1† Marie-Lise Maddelein,2†

Joel Begueret2 and Sven J. Saupe2*1Laboratoire de Parasitologie Moleculaire, UMR 5016

CNRS/Universite de Bordeaux 2, Bordeaux, France.2Laboratoire Genetique Moleculaire des Champignons,

Institut de Biochimie et de Genetique Cellulaires,

UMR 5095 CNRS/Universite de Bordeaux 2, 1 rue Camille

St Saens, 33077 Bordeaux cedex, France.

Summary

We have proposed that the [Het-s] infectious cyto-

plasmic element of the filamentous fungus Podospora

anserina is the prion form of the HET-s protein. The

HET-s protein is involved in a cellular recognition

phenomenon characteristic of filamentous fungi and

known as heterokaryon incompatibility. Under the

prion form, the HET-s protein causes a cell death

reaction when co-expressed with the HET-S protein,

from which it differs by only 13 amino acid residues.

We show here that the HET-s protein can exist as two

alternative states, a soluble and an aggregated form in

vivo. As shown for the yeast prions, transition to the

infectious prion form leads to aggregation of a HET-

s–green fluorescent protein (GFP) fusion protein. The

HET-s protein is aggregated in vivo when highly

expressed. However, we could not demonstrate HET-s

aggregation at wild-type expression levels, which

could indicate that only a small fraction of the HET-s

protein is in its aggregated form in vivo in wild-type

[Het-s] strains. The antagonistic HET-S form is

soluble even at high expression level. A double

amino acid substitution in HET-s (D23A P33H), which

abolishes prion infectivity, suppresses in vivo aggre-

gation of the GFP fusion. Together, these results

further support the model that the [Het-s] element

corresponds to an abnormal self-perpetuating aggre-

gated form of the HET-s protein.

Introduction

The term prion qualifies the proteinaceous infectious agent

causing mammalian spongiform encephalopathies such

as Creutzfeld–Jakob disease and mad cow disease

(Prusiner, 1982). This infectious agent corresponds to an

abnormal form of a cellular protein termed PrP. Prions

propagate by converting the normal form of the protein into

an altered b-sheet-rich conformation (Prusiner, 1998).

Infectious proteins have also been found in lower

eukaryotes, namely yeast and the filamentous fungus

Podospora anserina (Wickner et al., 1999a). In these

fungi, the identified prions correspond to previously known

non-Mendelian elements whose genetic behaviour was

not reconcilable with nucleic acid-based inheritance.

[URE3] was the first of these non-Mendelian elements to

be identified as a fungal prion (Wickner, 1994). [URE3] is

the prion form of Ure2p, a protein involved nitrogen

repression. Similarly, Wickner (1994) proposed that the

long known [PSI] element of yeast corresponds to the

prion form of Sup35p, a subunit of the translation release

factor. Both proteins have a modular organization and

comprise an N-terminal prion domain responsible for

propagation and a C-terminal catalytic domain (Doel et al.,

1994; Ter-Avanesyan et al., 1994; Masison et al., 1997).

The N-terminal prion domains of Sup35 and Ure2p are

characterized by a strong bias in amino acid composition

with a high glutamine and asparagine content. Aggrega-

tion upon transition to the prion state was evidenced in

living cells. A Sup35–green fluorescent protein (GFP)

fusion protein gives a diffuse cytoplasmic signal in [psi–]

cells but forms dot-like fluorescent foci in [PSI1] cell

cultures (Patino et al., 1996). It has also been demon-

strated biochemically that, upon transition to the [PSI1]

prion state, the aggregation state of the Sup35 protein

increases (Patino et al., 1996; Paushkin et al., 1996).

Similarly, it has been reported that a Ure2p–GFP fusion

protein forms fluorescent dots in [URE3] cells (Edskes

et al., 1999). However, Fernandez-Bellot et al. (2000)

reported no correlation between GFP coalescence and

expression of the [URE3] phenotype. In vitro, purified

Sup35p and Ure2p proteins undergo self-seeded polym-

erization into amyloid-like fibres (Glover et al., 1997; King

et al., 1997; Taylor et al., 1999; Thual et al., 1999). A

combination of genetic and biochemical analyses has

shown that the ability to propagate [PSI] in vivo and

aggregation properties in vitro are intimately linked (Glover

et al., 1997; DePace et al., 1998; Liu and Lindquist, 1999;

Santoso et al., 2000). Recently, overexpressed Ure2p was

detected by electron microscopy in [URE3] strains as a

sort of filamentous network (Speransky et al., 2001).

Accepted 10 September, 2001. *For correspondence. [email protected]; Tel. (133) 5 56 99 90 27; Fax(133) 5 56 99 90 67. †The first two authors contributed equally to thiswork.

Molecular Microbiology (2001) 42(5), 1325–1335

Q 2001 Blackwell Science Ltd

These different observations have led to the conclusion

that the [PSI] and [URE3] prions propagate as ‘heritable

amyloidose’ (Wickner et al., 2000).

We have proposed that the protein encoded by the het-s

allele of the het-s heterokaryon incompatibility locus of the

filamentous fungus Podospora anserina is a fungal prion

(Coustou et al., 1997). Filamentous fungi have the rather

unique property of being able spontaneously to undergo

vegetative cellular fusion (anastomoses). A vegetative

heterokaryon is formed if such cell fusion involves two

genetically distinct isolates. The viability of such vegetative

heterokaryons is controlled by the heterokaryon incompat-

ibility loci (het loci). het loci exist as two (or more) allelic

variants that lead to deleterious effects (growth inhibition

and cell death) when co-expressed in the same cell (Glass

et al., 2000; Saupe, 2000). As a consequence, a viable

heterokaryon can only be formed between strains of

identical het genotype. It has been proposed that het loci

exist to prevent the exchange of deleterious cytoplasmic

replicons (mycoviruses, senescence plasmids…) between

unlike strains (Caten, 1972). The het-s locus is one of nine

heterokaryon incompatibility loci of P. anserina. It exists as

two incompatible alleles termed het-s and het-S (Rizet,

1952). Upon fusion, mixed het-s/het-S heterokaryotic cells

die. When het-s and het-S strains are confronted on solid

medium, this cell death reaction translates macroscopi-

cally into the formation of an abnormal contact line, termed

a ‘barrage’. Compatibility between strains can therefore be

assayed very conveniently. The HET-s and HET-S

proteins are 289 amino acids in length, differ by 13

amino acid residues and do not resemble any known

protein (Turcq et al., 1990). Inactivation of the het-s locus

by gene replacement does not lead to any defect apart

from the loss of the incompatibility function (Turcq et al.,

1991). A single amino acid replacement at position 33 in

HET-S yields a protein of HET-s specificity. In other words,

a single amino acid difference between the HET-s and

HET-S products is sufficient to trigger incompatibility

(Deleu et al., 1993).

The particular feature of the het-s locus resides in the

fact that strains of the het-s genotype can display two

alternative phenotypes termed [Het-s] and [Het-s*]. [Het-s]

strains are incompatible with [Het-S] strains, whereas

[Het-s*] strains are neutral in incompatibility (i.e. they are

compatible with both [Het-S] and [Het-s]). [Het-s*] strains

are obtained in the progeny of a [Het-s]� [Het-S] cross.

The [Het-s] phenotype is maternally inherited in [Het-

s*]� [Het-s] crosses. [Het-s*] strains are systematically

converted to the [Het-s] phenotype upon contact with a

[Het-s] strain. In P. anserina, as in many filamentous fungi,

cross-walls (septa) of filaments are incomplete. As a

consequence, there is a cytoplasmic continuity throughout

the mycelium. Therefore, the [Het-s*] to [Het-s] phenotypic

conversion spreads from the point of anastomosis to the

whole mycelium. This infectious conversion travels at up to

several mm h21 (10 times the linear growth rate of the

fungus) (Beisson-Schecroun, 1962). [Het-s*] strains also

spontaneously acquire the reactive [Het-s] phenotype at a

low frequency. In practice, upon prolonged subculture, all

[Het-s*] strains ultimately acquire the [Het-s] phenotype.

Therefore, at least in laboratory conditions, the [Het-s]

element is nearly ubiquitous in het-s strains. The

frequency of spontaneous transition to the [Het-s] state

is strongly increased by overexpression of the het-s gene.

[Het-s] propagation requires a functional het-s gene, and

het-s strains can be reversibly cured of [Het-s] (Coustou

et al., 1997). Based on these observations, we have

proposed that the [Het-s] element is the prion form of the

HET-s protein (Coustou et al., 1997). A genetic dissection

of het-s has identified several key residues that are critical

for the appearance of the [Het-s] element (Coustou et al.,

1999). HET-s lacks the N- and Q-rich regions found in the

prion domain of Ure2p and Sup35.

In the present paper, we analyse the modification of the

aggregation state of the het-s-encoded protein upon

transition to the [Het-s] phenotype by various methods. We

show that transition from [Het-s*] to [Het-s] involves

aggregation of the HET-s protein. However, this aggrega-

tion is not detected when the HET-s protein is expressed at

a low level.

Results

Aggregation state of the HET-s and HET-S proteins in

crude extracts

In order to compare the aggregation state of the HET-s

protein in wild-type [Het-s*] and [Het-s] strains, crude

protein extracts obtained from protoplasts under non-

denaturing conditions were submitted to size exclusion

chromatography and ultracentrifugation. We first deter-

mined the amount of HET-s protein in crude extracts from

[Het-s*] and [Het-s] strains. It appears that the amount of

HET-s protein detected by Western blot is systematically

lower in [Het-s] than in [Het-s*] strains (Fig. 1A). The het-s

mRNA level is the same in [Het-s] and [Het-s*] strains (C.

Deleu, unpublished results). We then verified that this

difference in the amount of HET-s protein was not the

result of incomplete extraction of the HET-s protein from

the [Het-s] strains (prion state). Protein extraction from

protoplasts was carried out under strongly denaturing

conditions (8 M urea, 2% SDS at a 1008C). In these

conditions, the amount of HET-s protein detected in [Het-s]

was again significantly lower (at least threefold) compared

with the [Het-s*] extract (Fig. 1B). We conclude that the

steady-state level of HET-s protein is reduced in [Het-s]

strains. It also appears that the amount of HET-S protein is

significantly higher than that of HET-s (Fig. 1A).

1326 V. Coustou-Linares, M.-L. Maddelein, J. Begueret and S. J. Saupe

Q 2001 Blackwell Science Ltd, Molecular Microbiology, 42, 1325–1335

The HET-S protein was eluted from the size exclusion

column as a globular protein with a molecular mass of

about 35 kDa (Table 1). This is in agreement with the

calculated molecular mass of the HET-S monomer

(31.9 kDa). HET-S is thus soluble and monomeric in

crude extracts under these buffer conditions. The same is

true for the HET-s protein, whether it is extracted from a

[Het-s*] or a [Het-s] strain. In each case, the protein eluted

from the column with an apparent molecular mass of about

35 kDa. Thus, by this method, transition to the [Het-s] state

does not lead to detectable aggregation of the protein in

wild-type strains. It should, however, be noted that the

amount of HET-s protein detected in the 35 kDa fractions

is at the limit of the detection level (not shown). Thus, if a

small amount of HET-s protein was present in the high-

molecular-weight fractions, we could fail to detect it by this

method. When the same extracts were submitted to

ultracentrifugation for 1 h at 100 000 g, the HET-s protein

extracted from [Het-s] and [Het-s*] strains was found in the

supernatant, as was HET-S (Fig. 1C). So, as in the size

exclusion chromatography experiment, the presence of

HET-s aggregates in wild-type [Het-s] strains could not be

demonstrated.

The same experiments were performed with crude

extracts from strains that strongly express either HET-s or

HET-S. These strains (het-s8::pGs and het-s8::pGS)

contain a null het-s8 allele at the het-s locus and an

ectopic copy of het-s or het-S under the control of the

strong GPD promoter of Aspergillus nidulans (Punt et al.,

1988). In these strains, the amount of HET-s or HET-S

protein is increased at least 10-fold compared with the

wild-type strains (Coustou et al., 1997). The HET-S protein

was again eluted from the column with an apparent

molecular mass of about 35 kDa, whereas HET-s was

detected in the fractions corresponding to the exclusion

volume of the column (. 106 Da). An ultracentrifugation

experiment (1 h at 100 000 g ) was performed with the

same extracts. The HET-s protein was found in the pellet

fraction, whereas HET-S remained in the soluble fraction

(Fig. 1C). The HET-s aggregates were not solubilized in

high-salt buffers (1 M KCl), non-ionic detergents (1%

Triton X-100 or Tween 20) or reductants, but could be

solubilized in 6 M GmdHCl. Thus, when highly expressed,

HET-s forms aggregates, whereas HET-S remains

soluble. Increased expression of HET-s leads to a rapid

spontaneous transition to the [Het-s] prion state (Coustou

et al., 1997). Thus, these gel filtration and ultracentrifuga-

tion experiments could not be carried out with extracts

from [Het-s*] strains overexpressing HET-s, as these

strains spontaneously acquire the [Het-s] phenotype

before enough mycelium can be recovered to carry out

protein analyses.

Fig. 1. Immunodetection of HET-s and HET-S incrude P. anserina extracts and afterultracentrifugation.A. Crude extracts of [Het-S], [Het-s*] and [Het-s]strains (20mg of total protein) were submitted toSDS–PAGE and analysed by immunoblotting.B. The same experiment was repeated with[Het-s*] and [Het-s] strains, but proteins wereextracted under denaturing conditions (8 Murea, 2% SDS at 1008C); molecular mass sizemarkers are given in kDa. The dotted linecorresponds to the limit between the stackingand resolving gel.C. Crude extracts from [Het-S], [Het-s*] and[Het-s] strains were fractionated byultracentrifugation (1 h at 100 000 g ) intosupernatant (s) and pellet (p) fractions andanalysed by immunoblotting after SDS–PAGE.

Table 1. Determination of aggregation state by size exclusion chromatography.

Strain

Calculated monomericmolecular mass(kDa)

Apparent mass by size exclusionchromatography(kDa)

Deducedstate

het-S 31.9 35 Monomerhet-s [Het-s*] 31.9 35 Monomerhet-s [Het-s] 31.9 35 Monomerhet-s8::pGS [Het-S] 31.9 35 Monomerhet-s8::pGs [Het-s] 31.9 . 1000a Aggregate

Molecular mass standards were thyroglobulin (670 kDa), bovine gamma globulin (158 kDa), chicken ovalbumin (44 kDa), equine myoglobin(17 kDa).a. Exclusion limit of Superose-12 gel filtration column.

In vivo aggregation of the HET-s prion protein of P. anserina 1327


These two approaches gave similar results, indicating

that, unlike HET-S, the HET-s protein aggregates in vivo.

However, we could only detect these aggregated species

at increased protein concentration. At the wild-type

expression level, the transition to the [Het-s] phenotype

correlates with a decrease in the steady-state amount of

HET-s protein.

Visualization of HET-s–GFP and HET-S–GFP fusion

proteins at low expression level

Vectors were constructed allowing the expression of

HET-s and HET-S tagged at their C-terminus with GFP.

Expression was driven by the native promoters (psGFP,

pSGFP constructs). Transformants expressing the HET-

S–GFP fusion produced a ‘barrage’ reaction when

confronted with [Het-s] tester. Most transformants expres-

sing HET-s–GFP initially displayed the [Het-s*] pheno-

type, whereas a small proportion (about 5%)

spontaneously expressed the [Het-s] phenotype. This is

similar to what is observed when a het-s8 strain is

transformed with a vector containing the wild-type het-s

gene (Coustou et al., 1997). The GFP tag apparently does

not alter reactivity in incompatibility and ability to

propagate the [Het-s] state.

[Het-s*], [Het-s] and [Het-S] transformants expressing

these GFP fusions were examined by fluorescence

microscopy. The HET-S–GFP fusion protein had a

homogeneous diffuse cytoplasmic distribution (Fig. 2A).

Based on this experiment, HET-S is a cytoplasmic protein

that is apparently not associated with any particular

subcellular structure. The distribution of the fluorescence

signal observed for the HET-s–GFP fusion was essen-

tially the same in [Het-s] and [Het-s*] strains (Fig. 2B and

C). In most cases, the protein had a diffuse cytoplasmic

distribution. However, interestingly, in [Het-s] strains in a

fraction of cells (about 10–20%), the fusion protein had a

non-homogeneous distribution and appeared as patches

of variable shapes and sizes that co-localized with the

large vacuoles when those were visible. Conversely,

HET-S–GFP and HET-s–GFP in [Het-s*] strains were

generally excluded from the large vacuoles. So, at low

expression level, there was again no clear evidence of

HET-s–GFP aggregation upon transition to the prion

state. The vacuolar distribution of the fusion protein in a

fraction of the [Het-s] strains together with the detected

decrease in the amount of HET-s protein in [Het-s] strains

(Fig. 1A) could suggest that, upon transition to the [Het-s]

prion state, the protein is preferentially degraded.

In vivo distribution of HET-s–GFP and HET-S–GFP

fusion proteins at increased expression level

As we found evidence of HET-s aggregation only in strains

overexpressing the protein, we chose to drive expression

of the GFP fusions by a strong constitutive promoter, the

GPD promoter from A. nidulans. Vectors were constructed

allowing the expression of HET-s–GFP and HET-S–GFP

fusion proteins under the control of the GPD promoter

(pGsGFP and pGSGFP) and introduced into a het-s8

recipient.

The het-s8::pGSGFP transformants expressed the [Het-

S] phenotype (i.e. produced a ‘barrage’ when confronted

with [Het-s]). The fluorescence signal was greatly

increased compared with transformants expressing the

fusion under the control of the native promoter. The fusion

protein had a diffuse cytoplasmic distribution (Fig. 3A). In

agreement with the ultracentrifugation and gel filtration

experiments, the HET-S–GFP protein apparently remains

soluble even at high expression levels.

As expected, among the het-s8::pGsGFP transformants,

the proportion of transformants spontaneously expressing

the [Het-s] phenotype was higher (about 60%) than

when expression was driven by the native promoter (about

5%; see above). In contrast to het-s8::pGs transformants

Fig. 2. Visualization of HET-S–GFP and HET-s–GFP fusion proteins.A. Vegetative mycelium from het-s8::pSGFP [Het-S] transformants.B. Vegetative mycelium from het-s8::psGFP [Het-s*] transformants.C. Vegetative mycelium from het-s8::psGFP [Het-s] transformants.Strains were observed by fluorescence microscopy (left) and inbrightfield microscopy (right) (scale bar¼ 3mm). Black and whitetriangles indicate vacuoles. Note the exclusion of the fluorescencesignal form the vacuole in [Het-s*] and [Het-S] strains and the co-localization of the signal with the vacuoles in [Het-s] strains.



(lacking the GFP tag), het-s8::pGsGFP transformants

can be maintained for a few days in the [Het-s*] state

before they acquire the [Het-s] phenotype. Thus, the

addition of the GFP tag at the C-terminus of HET-s

does not affect the incompatibility function per se but

apparently reduces the frequency of spontaneous tran-

sition to the [Het-s] state. In het-s8::pGsGFP transformants

displaying the [Het-s*] phenotype, the signal was strong

and showed a diffuse cytoplasmic distribution (Fig. 3B).

When such strains were converted to the [Het-s] state

by confrontation with a [Het-s] tester, the HET-s–GFP

fusion protein coalesced into a small number of strongly

fluorescent dots (Fig. 3C). The number of dots was

usually one per article. The dots did not co-localize with

large vacuoles. The HET-s–GFP fusion protein is

pelletable by ultracentrifugation in these strains (not

shown). This aggregation process starts in the region of

the mycelium that fuses with the [Het-s] tester and then

spreads to the rest of the mycelium over the course of a

few hours.

Specific HET-s–GFP aggregation in [Het-s] strains

could also be demonstrated in protoplasts. In protoplasts

generated from [Het-s*] mycelium, fluorescence was

diffuse. In protoplasts generated from [Het-s] mycelium,

the fluorescence was detected as brightly fluorescent dots

(Fig. 3D).

It should be emphasized that, in a [Het-s] mycelium, the

bright discrete dots were not the only type of fluorescent

signal that was observed. In a fraction of the cells (about

20%), the fluorescence signal was composed of vesicle-

like structures of various shapes, sizes and cellular

distribution. They were either round or tubular (Fig. 4A).

These structures were reminiscent of the vacuolar network

described in filamentous fungi. In filamentous fungi,

vacuoles are highly dynamic and can appear as tubules,

multiple small vesicules or large ‘typical’ vacuoles visible in

Fig. 3. Visualization of HET-S–GFP and HET-s–GFP fusion proteins at increased expression levels.A. Vegetative mycelium from a het-s8::pGSGFP [Het-S] transformant.B. Vegetative mycelium from a het-s8::pGsGFP [Het-s*] transformant.C. Vegetative mycelium from a het-s8::pGsGFP [Het-s] transformant.D. Protoplasts from a het-s8::pGsGFP [Het-s] transformant. The insert at the bottom left corresponds to a protoplast from the same strain inthe [Het-s*] state.Strains were observed by fluorescence microscopy (left) and in brightfield microscopy (right) (scale bar¼ 5mm).



brightfield (Cole et al., 1998). In Fig. 4B, examples are

given of vacuolar labelling in P. anserina using a

fluorescent dye. Direct co-localization of HET-s–GFP

and vacuoles could not be performed in this experimental

setting, as the vacuolar dye and the GFP were detected

using the same filter. However, based on the similarity

between the subcellular structures detected in both cases,

we propose that these alternative structures visualized in

[Het-s] het-s8::pGsGFP transformants, which are clearly

distinct from the very bright dots observed in the majority of

the filaments, could reflect vacuolar localization of the GFP

fusion. These observations suggest that, in addition to the

aggregation process, specific degradation of the HET-s–

GFP fusion protein occurs in [Het-s] strains.

In vivo distribution of a HET-s(D23A P33H)–GFP fusion

protein

Under high expression, HET-s aggregates specifically in

[Het-s] strains, whereas HET-S remains soluble. Among

the 13 differences between HET-s and HET-S, two

positions (23 and 33) were found to be sufficient to define

the het-s and het-S specificities in incompatibility (Deleu

et al., 1993; Coustou et al., 1999). A mutant HET-s protein,

in which these two residues were replaced by those found

in HET-S, acquires the biological properties of HET-S. In

other words, a strain expressing the het-s(D23A P33H)

mutant allele is incompatible with a wild-type [Het-s] strain

and compatible with [Het-S] strains. In order to determine

whether expression of [Het-s] and aggregation are

correlated, a construct allowing the expression of a HET-

s(23A 33H)–GFP fusion under the control of the GPD

promoter was introduced into a het-s8 strain. As expected,

the het-s8::pGs2333GFP transformants produced a ‘bar-

rage’ reaction when confronted with [Het-s] testers and

were unable to convert a [Het-s*] strain to the [Het-s]

phenotype. The distribution of the fluorescence signal of

the HET-s(23A 33H)–GFP fusion was identical to that of

HET-S–GFP, namely it was cytoplasmic and homo-

geneous (Fig. 5A). Therefore, these two amino acid

substitutions at positions 23 and 33 are sufficient to

prevent aggregation of the fusion protein.

In vivo distribution of a HET-S(A23D H33P)–GFP fusion

protein

The reciprocal substitutions in a het-S allele, het-S(A23D

H33P), led to the expression of the [Het-s] phenotype

(Deleu et al., 1993). Strains expressing this mutant allele

can display the alternative [Het-s*] and [Het-s] phenotypes

and are able to convert wild-type [Het-s*] strains to the

[Het-s] phenotype (Deleu et al., 1993). We constructed a

vector allowing the expression of the HET-S(A23D

H33P)–GFP fusion protein under the control of the strong

GPD promoter and introduced it into a het-s8 recipient. All

transformants were initially neutral in incompatibility

(compatible both with het-s and het-S ). After confrontation

with a wild-type [Het-s] strain, only a small proportion

(about 10%) of the transformants acquired the active

Fig. 4. Examples of additional types of distribution of the fluorescencesignal in het-s8::pGsGFP [Het-s] transformants and vacuolar labelling.A. Vegetative mycelium from het-s8::pGsGFP [Het-s] transformantsobserved by fluorescence microscopy. These images illustrate thevarious examples of the type of fluorescence signal observed. In themajority of cases, fluorescence appears as discrete very bright dots(top) but, in a fraction of cases, fluorescence also appeared as tubularstructures or multiple vesicules of various shapes and sizes.B. Examples of vacuolar labelling in live P. anserina hyphae from awild-type het-s strain. Vacuolar labelling identifies tubular structuresand multiple vesicles.



[Het-s] phenotype. Based on fluorescence intensity, the

transformants that were able to express the [Het-s]

phenotype were those with the highest expression level.

The [Het-s] element was lost upon subculture of the

transformants. It therefore appears that, as noted for the

wild-type HET-s–GFP fusion, the addition of the GFP tag

apparently partially inhibits transition to the [Het-s] prion

state. Although het-s and het-S(A23D H33P) could not

be phenotypically distinguished previously (Deleu et al.,

1993), the addition of the GFP tag now reveals differences

between the allele products. In het-S(23D 33P)–GFP

transformants expressing the [Het-s] phenotypes, fluor-

escence was diffuse, and massive aggregation observed

in [Het-s] strains expressing the HET-s–GFP fusion was

not detected (Fig. 5B). These results suggest that, among

the 13 amino acid differences between HET-s and HET-S,

residues other than those at positions 23 and 33 contribute

to the propensity of the HET-s protein to aggregate in vivo.

In order to determine whether the wild-type HET-s

protein could force aggregation of the HET-S(A23D

P33H)–GFP fusion, we introduced the same construct

into a het-s8::pGs strain in which the HET-s protein is

highly expressed and aggregated (see above). In this

background, the fluorescence signal corresponded to

bright dot-like foci (Fig. 5C). This suggests that the

HET-S(A23D P33H)–GFP fusion protein is able to co-

aggregate with wild-type HET-s protein.

HET-s–GFP aggregation in a [Het-S] background

Co-expression of HET-s (under its prion form) and HET-S

triggers the incompatibility reaction. Strains that are

engineered to co-express both proteins are termed self-

incompatible and display a sublethal phenotype charac-

terized by very slow growth and an aberrant morphology

(Coustou et al., 1999). In order to determine whether

co-expression of HET-s and HET-S could affect the

aggregation state of HET-s, we introduced the pGsGFP

construct, leading to strong expression of the HET-s–GFP

fusion, into a het-S recipient strain. Most transformants

had normal growth and produced a barrage reaction when

confronted with [Het-s] testers but not with het-S testers.

Fig. 5. Visualization of HET-s(D23A P33H)–GFP and HET-S(A23DH33P)–GFP fusion proteins.A. Vegetative mycelium from a het-s8::pGs(23A33H)GFP [Het-S]transformant.B. Vegetative mycelium from a het-s8::pGS(23D33P)GFP [Het-s]transformant.C. Vegetative mycelium from a pGPD-het-s::pGsGFP [Het-s]transformant.In a background in which wild-type HET-s is overexpressed andaggregated, the mutant fusion protein forms large bright dots. Strainswere observed by fluorescence microscopy (left) and in brightfieldmicroscopy (right) (scale bar¼ 5mm).

Fig. 6. Phenotype of a het-s-GFP/het-S self-incompatible transformant and visualization ofHET-s–GFP in a het-S background.A. Growth phenotype of a self-incompatiblehet-S::pGsGFP transformant, overexpressingHET-s–GFP in a het-S background (top, self-incompatible transformant; bottom, het-Srecipient). Strains were grown for 4 days onsynthetic medium.B. Microscopic phenotype of self-incompatiblehet-S::pGsGFP transformants and visualizationof the GFP fusion protein. Note the abnormalcellular shapes, the presence of ‘empty’ articles(dead cells) and the small vesicles or droplets(scale bar¼ 5mm).



These transformants presumably correspond to strains

that did not switch to the [Het-s] prion state and are thus

maintained in the [Het-s*] state. In the het-S genetic

background, transition to the prion state is strongly

counterselected, as transition to the prion state will trigger

the incompatibility reaction (Coustou et al., 1999). In the

transformants displaying normal growth, the fluorescence

signal was diffuse, and no dot-like GFP aggregates

were detected (not shown). Transformants displaying

the characteristic self-incompatible phenotype were also

recovered (Fig. 6A). These self-incompatible transformants

are characterized by an abnormal cellular morphology:

many cells have a round shape; septation and hyphal

branching are increased, but most branches abort in a

bud-like state. Vacuolization is intense in many cases.

Cells often contain small round vesicles or droplets. These

morphological alterations are similar to those caused by

another incompatibility interaction in Podospora (Saupe

et al., 2000). Despite the very slow growth of the selected

het-s–GFP/het-S sublethal transformants, enough

mycelium could be cultured to allow examination by

fluorescence microscopy. In these self-incompatible

transformants, the HET-s–GFP fluorescence signal was

partly cytoplasmic and diffuse but also vacuolar (Fig. 6B).

The characteristic brightly fluorescent HET-s–GFP dots

were not detected. This observation suggests that

aggregation of the HET-s–GFP fusion is at least partially

inhibited in the presence of HET-S in a self-incompatible

background.

Discussion

So far, the hypothesis that the [Het-s] element is a fungal

prion has been based largely on genetic evidence

(Coustou et al., 1997; 1999). We now show that in vivo

transition to the [Het-s] prion state could be associated

with aggregation of the HET-s protein. When strongly

expressed, a HET-s–GFP fusion protein specifically

aggregates upon transition to the [Het-s] state. Thus,

as described for the yeast system, in these conditions,

emergence of the [Het-s] prion is accompanied by a

protein aggregation process. In wild-type [Het-s] strains

that express the protein at a low level, the transition from

[Het-s*] to [Het-s] leads to a decrease in the steady-state

level of HET-s protein and to vacuolar localization of the

HET-s–GFP fusion in some cases. However, we did not

detect aggregated HET-s protein in such wild-type [Het-s]

strains. We suggest that, upon transition to the [Het-s]

state, the HET-s protein forms aggregates, but that

this transition also leads to preferential degradation of

the protein (Fig. 7). Misfolded or misassembled proteins

are generally targeted for degradation, and aggregate

formation is considered as a failure of this quality control

process (Wickner et al., 1999b). Presumably, HET-s

aggregate formation upon transition to the [Het-s] state

is kept under control as long as the expression level of

HET-s is not too great to overcome the capacity of the

degradation machinery. At low expression levels, HET-s

degradation would prevail, so that no aggregates were

detected while the steady-state level of HET-s greatly

decreased. In this hypothesis, the soluble HET-s protein

detected in wild-type [Het-s] strains would correspond to

unconverted protein rather than to converted soluble

species.

The proteasome and the vacuole are the main sites for

protein turnover (Wickner et al., 1999b; Klionsky and Emr,

2000), but aggregated proteins cannot be efficiently

degraded by the proteasome. Conversely, as proteins

are imported to the vacuole by a vesicular transport

system, proteins can be targeted to the vacuole/lysosome

even in a highly multimeric state (Teter and Klionsky,

1999). This might explain why at least part of the HET-s

degradation appears to take place in the vacuole and

is consistent with the hypothesis that the HET-s protein is

preferentially degraded in [Het-s] strains because it is

misassembled.

The detected in vivo aggregation state of HET-s

depends on the expression level of the protein. To a

lesser extent, this also appears to be the case for the yeast

prion proteins. The number and apparent size of Sup35–

GFP aggregates in [PSI1] strains was shown to depend

on the expression level of the fusion protein (Bailleul-

Winslett et al., 2000; Wegrzyn et al., 2001). Also, for

[URE3], in contrast to what was reported by Edskes et al.

(1999), Fernandez-Bellot et al. (2000) did not detect

Ure2p–GFP aggregates in [URE3] strains expressing the

fusion protein at low levels and, to our knowledge, Ure2p

aggregation in wild-type strains in the [URE3] state has not

been reported.

One of the specificities of the [Het-s] prion system is the

existence of the HET-S natural variant, which differs from

Fig. 7. Proposed model for the interplay of HET-s aggregation anddegradation processes. In this model, we suggest that transition to the[Het-s] state results from the formation of infectious HET-s oligomers.These oligomers are recognized as abnormal and are preferentiallydegraded. They can also polymerize further into high-molecular-weight aggregates. At low expression levels, degradation prevails,whereas at high expression levels, massive aggregation is favoured.



HET-s by 13 amino acid residues and lacks the prion

properties. We show here that aggregation is specific to

the HET-s variant in its prion form. HET-S was detected as

a soluble protein even at increased expression levels. A

double amino acid substitution at positions 23 and 33,

which changes the HET-s protein to the HET-S specificity,

suppresses in vivo aggregation of a GFP fusion. This

correlates expression of the [Het-s] phenotype and

aggregation. Surprisingly, the reciprocal substitution in

an HET-S–GFP fusion does not lead to in vivo

aggregation. This indicates that other amino acid

differences between HET-s and HET-S participate in

specifying the propensity of the protein to form aggre-

gates. The fact that no foci of HET-S(23D 33P)–GFP

fusion were detected in strains displaying the [Het-s]

phenotype could suggest that [Het-s] can be propagated in

the absence of aggregation. However, it is entirely

possible that a fraction of the mutant HET-S fusion protein

is aggregated in these strains, but that the amount or the

size of the aggregates is too small to be detected by

fluorescence microscopy. Moreover, unlike strains expres-

sing HET-s–GFP, these strains do not stably maintain the

[Het-s] element. The rapid loss of [Het-s] might result from

the fact that the mutant protein has a decreased

propensity to aggregate. We have shown that this mutant

HET-S protein is able to aggregate massively in cells

overexpressing wild-type HET-s. This suggests that wild-

type HET-s protein can seed the aggregation of the mutant

protein. The detected foci presumably correspond to co-

aggregates between HET-s and the mutant HET-S–GFP

fusion.

The loss of function associated with the transition to the

prion form in the yeast prions is thought to result largely

from the fact that the proteins are aggregated and can no

longer interact with their normal cellular partners. The

[Het-s] system apparently differs from the yeast prion

models by the fact that the prion form corresponds to the

reactive form in incompatibility. Therefore, if the sole

function of HET-s is to control self/non-self recognition,

then transition to the prion form corresponds to a gain

rather than a loss of function. This could indicate that, at

the molecular level, there are fundamental differences

between [Het-s] and the yeast prions. However, we show

here that the basic molecular mechanisms of propagation

of [Het-s] and the yeast prions appear to be based on

related protein aggregation processes. This is backed up

by the recent finding that recombinant HET-s protein forms

amyloid-like fibres in vitro (S. Dos Reis and S. Saupe,

unpublished results). The challenge is now to understand

how this apparent gain of function (reactivity in incompat-

ibility) can be associated with transition of the protein to an

aggregated state. The presence of HET-s aggregates per

se does not affect growth; it is only in the presence of

HET-S that HET-s in its prion state triggers the cell death

reaction. We found that, in a self-incompatible back-

ground, HET-s–GFP aggregation is reduced. There are

two alternative hypotheses that can account for this

observation. First, this could represent an indirect effect of

the perturbation of the cellular metabolism during the

incompatibility reaction. This cellular stress might induce

chaperones and/or proteases that might lead to solubil-

ization and/or degradation of HET-s–GFP aggregates. It

is known that, during the incompatibility reaction in

Podospora, there is a strong increase in proteolytic

activity. In particular, a vacuolar serine protease is strongly

induced during incompatibility (Paoletti et al., 2001).

Alternatively, it is conceivable that HET-S is directly

responsible for the inhibition of HET-s aggregation. In this

hypothesis, HET-S might ‘poison’ the aggregation seed of

HET-s, preventing further increase in size of the

aggregate. There is increasing evidence suggesting that,

in protein deposition diseases, toxicity of amyloidogenic

proteins results from oligomeric aggregates, whereas the

very high-molecular-weight aggregates are relatively

innocuous (Lansbury, 1999). If the cell death reaction

triggered by het-s/het-S incompatibility is directly related to

protein aggregation processes, future studies on the [Het-

s] system could well shed some light on cell death

associated with protein aggregation in other systems.

Experimental procedures

Strains and DNA constructs

Podospora anserina strains used in this study have been

described previously (Coustou et al., 1997). General methodsfor growth, protoplast formation and transformation were

performed as described previously (Coustou et al., 1997). Forconstruction of the psGFP and pSGFP constructs, an Nco I–

Xba I fragment of the eGFP (Clontech) blunted at the Nco Iend was co-ligated with an Xba I fragment corresponding to

the TrpC terminator (Punt et al., 1988) into the Eco RV–Xba Iwindow of pCB1004 (Carrol et al., 1994). The 30 end of the

het-s (or het-S ) open reading frame (ORF) was amplified bypolymerase chain reaction (PCR) using the s4 (50-CGGCT

GAATGATCTCGTTTCTCGG-30) and s5 (50-GGCCGTCGACTCCCAGACCCC-30) oligos and cloned as a Sal I fragment

in the Sal I site of the recombinant pBC1004 vector describedabove. The 50 end of the het-s (or het-S ) gene was then

cloned as a HindIII–Eco RV fragment in the latter vector. Forconstruction of pGsGFP and pGSGFP allowing strong

expression of the HET-s–GFP and HET-S–GFP fusions,the Eco RI fragment from pGPD-het-s (or pGPD-het-S)

(Coustou et al., 1999) was co-ligated with the Eco RI–Xba Ifragment from psGFP (or pSGFP) into the Eco RI–Xba I

window of the pCB1004 vector (Carrol et al., 1994). Allconstructs were introduced in the het-s8 strain in which the

het-s gene was inactivated by gene replacement (Turcq et al.,1991). The pGSmGFP and pGsmGFP vectors allowing

expression of the HET-s(D23A P33H)–GFP and HET-

S(A23D H33P)–GFP fusion proteins were constructed by



amplifying the ORF of the corresponding het-s and het-Smutant alleles (Deleu et al., 1993) using the s1 (50-ACTGCC

ATGGCAGAACCGTTCG-30) and s2 (50-CTGCAAGCTTTTATCCCAGAACCCC-30) primers. The PCR products

were digested with Nco I and HindIII and inserted into theNco I–HindIII window of pGsGFP.

Protein analyses

Crude protein extracts were obtained by osmotic lysis ofprotoplasts. Protoplasts were resuspended at 109 cells ml21

in 50 mM Na2HPO4 buffer, pH 8. Cells were lysed byvortexing, and the sample was centrifuged for 10 min at

5000 g. The supernatant represented the crude extract, andthe concentration was adjusted to 5 mg of protein ml21. For

extraction under denaturing conditions, cells were lysed in 8 Murea, 2% SDS, 100 mM Tris-HCl, pH 8, and boiled for 5 min.

The sample was then centrifuged for 10 min at 5000 g.For Western blot analyses, proteins were separated by

SDS–PAGE, transferred to polyvinylidene difluoride (PVDF)

membranes (Bio-Rad) by semi-dry transfer and treated withpurified polyclonal antibodies to HET-S and peroxidase-

coupled secondary antibodies. Secondary antibodies weredetected using the ECL1 kit (Amersham). For gel filtration

experiments, 200ml of crude extracts was applied to aSuperose-12 column (Amersham); 0.2 ml fractions were

collected through the entire run, acetone precipitatedand analysed by Western blotting with antibodies directed to

HET-S.

Fluorescence microscopy

For microscopic analyses, mycelium was grown on thinsynthetic solid medium for 12–24 h at 268C. The area of

medium containing the mycelium was cut out and transferred

to a microscope slide. The mycelium was examined under afluorescence microscope with a Leica DMRXA microscope

equipped with a Micromax CCD (Princeton Instruments). Afilter set for fluorescein isothiocyanate (FITC) was used. At

least three independent transformants were examined foreach construct. For vacuole staining, mycelia were covered

with 100ml of STC10 buffer (0.8 M sorbitol, 10 mM CaCl2,50 mM Tris-HCl, pH 7.5) containing 0.1 mg ml21 Oregon

green 488–carboxylic acid diacetate fluorescent vacuolardye (Molecular Probes) for 5 min at room temperature,

washed three times in STC10 before being observed byfluorescence microscopy using a filter set for FITC.

Acknowledgements

This work was supported by a grant from the ‘Programme de

recherche sur les ESST et les prion’ and from the CNRS‘Physique et chimie du vivant’. Virginie Coustou-Linares was

the recipient of a fellowship from the ‘Ministere del’Enseignement Superieur’. Marie-Lise Maddelein is the

recipient of a fellowship from the ‘Fondation pour laRecherche Medicale’. The authors wish to thank Martine

Sabourin for expert technical assistance, and Thomas

Laurent and Brice Roux for their help in vector construction.

References

Bailleul-Winslett, P.A., Newnam, G.P., Wegrzyn, R.D., and

Chernoff, Y.O. (2000) An antiprion effect of the anti-

cytoskeletal drug latrunculin A in yeast. Gene Expr 9:

145–156.

Beisson-Schecroun, J. (1962) Incompatibilite cellulaire et

interactions nucleocytoplamsiques dans les phenomenes

de barrage chez le Podospora anserina. Ann Genet 4:

3–50.

Carrol, A.M., Sweigard, J.A., and Valent-Central, B. (1994)

Improved vectors for selecting resistance to hygromycin.

Fungal Genet Newslett 41: 22.

Caten, C.E. (1972) Vegetative incompatibility and cytoplasmic

infection in fungi. J Gen Microbiol 72: 221–229.

Cole, L., Orlovich, D.A., and Ashford, A.E. (1998) Structure,

function, and motility of vacuoles in filamentous fungi.

Fungal Genet Biol 24: 86–100.

Coustou, V., Deleu, C., Saupe, S., and Begueret, J. (1997)

The protein product of the het-s heterokaryon incompat-

ibility gene of the fungus Podospora anserina behaves as a

prion analog. Proc Natl Acad Sci USA 94: 9773–9778.

Coustou, V., Deleu, C., Saupe, S.J., and Begueret, J.

(1999) Mutational analysis of the [Het-s] prion analog of

Podospora anserina. A short N-terminal peptide allows

prion propagation. Genetics 153: 1629–1640.

Deleu, C., Clave, C., and Begueret, J. (1993) A single amino

acid difference is sufficient to elicit vegetative incompat-

ibility in the fungus Podospora anserina. Genetics 135:

45–52.

DePace, A.H., Santoso, A., Hillner, P., and Weissman, J.S.

(1998) A critical role for amino-terminal glutamine/

asparagine repeats in the formation and propagation of a

yeast prion. Cell 93: 1241–1252.

Doel, S.M., McCready, S.J., Nierras, C.R., and Cox, B.S.

(1994) The dominant PNM2– mutation which eliminates

the psi factor of Saccharomyces cerevisiae is the result of a

missense mutation in the SUP35 gene. Genetics 137:

659–670.

Edskes, H.K., Gray, V.T., and Wickner, R.B. (1999) The

[URE3] prion is an aggregated form of Ure2p that can be

cured by overexpression of Ure2p fragments. Proc Natl

Acad Sci USA 96: 1498–1503.

Fernandez-Bellot, E., Guillemet, E., and Cullin, C. (2000) The

yeast prion [URE3] can be greatly induced by a functional

mutated URE2 allele. EMBO J 19: 3215–3222.

Glass, N.L., Jacobson, D.J., and Shiu, P.K. (2000) The

genetics of hyphal fusion and vegetative incompatibility in

filamentous ascomycete fungi. Annu Rev Genet 34:

165–186.

Glover, J.R., Kowal, A.S., Schirmer, E.C., Patino, M.M., Liu,

J.J., and Lindquist, S. (1997) Self-seeded fibers formed by

Sup35, the protein determinant of [PSI1], a heritable prion-

like factor of S. cerevisiae. Cell 89: 811–819.

King, C.Y., Tittmann, P., Gross, H., Gebert, R., Aebi, M., and

Wuthrich, K. (1997) Prion-inducing domain 2–114 of yeast

Sup35 protein transforms in vitro into amyloid-like filaments.

Proc Natl Acad Sci USA 94: 6618–6622.

Klionsky, D.J., and Emr, S.D. (2000) Autophagy as regulated

pathway of cellular degradation. Science 290: 1717–1721.

Lansbury, P.T., Jr (1999) Evolution of amyloid: what normal



protein folding may tell us about fibrillogenesis and disease.Proc Natl Acad Sci USA 96: 3342–3344.

Liu, J.J., and Lindquist, S. (1999) Oligopeptide-repeatexpansions modulate ‘protein-only’ inheritance in yeast.

Nature 400: 573–576.Masison, D.C., Maddelein, M.L., and Wickner, R.B. (1997)

The prion model for [URE3] of yeast: spontaneousgeneration and requirements for propagation. Proc Natl

Acad Sci USA 94: 12503–12508.Paoletti, M., CastroViejo, M., Begueret, J., and Clave, C.

(2001) Identification and characterization of a geneencoding a subtilisin-like serine protease induced during

the vegetative incompatibility reaction in Podosporaanserina. Curr Genet 39: 244–252.

Patino, M.M., Liu, J.J., Glover, J.R., and Lindquist, S. (1996)Support for the prion hypothesis for inheritance of a

phenotypic trait in yeast. Science 273: 622–626.Paushkin, S.V., Kushnirov, V.V., Smirnov, V.N., and Ter-

Avanesyan, M.D. (1996) Propagation of the yeast prion-like

[psi1] determinant is mediated by oligomerization of theSUP35-encoded polypeptide chain release factor. EMBO J

15: 3127–3134.Prusiner, S.B. (1982) Novel proteinaceous infectious particles

cause scrapie. Science 216: 136–144.Prusiner, S.B. (1998) Prions. Proc Natl Acad Sci USA 95:

13363–13383.Punt, P.J., Dingemanse, M.A., Jacobs-Meijsing, B.J., Pou-

wels, P.H., and van den Hondel, C.A. (1988) Isolation andcharacterization of the glyceraldehyde-3-phosphate dehy-

drogenase gene of Aspergillus nidulans. Gene 69: 49–57.Rizet, G. (1952) Les phenomenes de barrage chez

Podospora anserina. I. Analyse de barrage entre lessouches s et S. Rev Cytol Biol Veg 13: 51–92.

Santoso, A., Chien, P., Osherovich, L.Z., and Weissman, J.S.(2000) Molecular basis of a yeast prion species barrier. Cell

100: 277–288.Saupe, S.J. (2000) Molecular genetics of heterokaryon

incompatibility in filamentous ascomycetes. Microbiol MolBiol Rev 64: 489–502.

Saupe, S.J., Clave, C., Sabourin, M., and Begueret, J. (2000)Characterization of hch, the Podospora anserina

homolog of the het-c heterokaryon incompatibility gene ofNeurospora crassa. Curr Genet 38: 39–47.

Speransky, V.V., Taylor, K.L., Edskes, H.K., Wickner, R.B.,

and Steven, A.C. (2001) Prion filament networks in [URE3]

cells of Saccharomyces cerevisiae. J Cell Sci 153:

1327–1336.

Taylor, K.L., Cheng, N., Williams, R.W., Steven, A.C., and

Wickner, R.B. (1999) Prion domain initiation of amyloid

formation in vitro from native Ure2p. Science 283:

1339–1343.

Ter-Avanesyan, M.D., Dagkesamanskaya, A.R., Kushnirov,

V.V., and Smirnov, V.N. (1994) The SUP35 omnipotent

suppressor gene is involved in the maintenance of the non-

Mendelian determinant [psi1] in the yeast Saccharomyces

cerevisiae. Genetics 137: 671–676.

Teter, S.A., and Klionsky, D.J. (1999) How to get a folded

protein across a membrane. Trends Cell Biol 9: 428–431.

Thual, C., Komar, A.A., Bousset, L., Fernandez-Bellot, E.,

Cullin, C., and Melki, R. (1999) Structural characterization

of Saccharomyces cerevisiae prion-like protein Ure2. J Biol

Chem 274: 13666–13674.

Turcq, B., Denayrolles, M., and Begueret, J. (1990) Isolation

of two allelic incompatibility genes s and S of the fungus

Podospora anserina. Curr Genet 17: 297–303.

Turcq, B., Deleu, C., Denayrolles, M., and Begueret, J. (1991)

Two allelic genes responsible for vegetative incompatibility

in the fungus Podospora anserina are not essential for cell

viability. Mol Gen Genet 228: 265–269.

Wegrzyn, R.D., Bapat, K., Newnam, G.P., Zink, A.D., and

Chernoff, Y.O. (2001) Mechanism of prion loss after

Hsp104 inactivation in yeast. Mol Cell Biol 21: 4656–4669.

Wickner, R.B. (1994) [URE3] as an altered URE2 protein:

evidence for a prion analog in Saccharomyces cerevisiae.

Science 264: 566–569.

Wickner, R.B., Taylor, K.L., Edskes, H.K., Maddelein, M.L.,

Moriyama, H., and Roberts, B.T. (1999a) Prions in

Saccharomyces and Podospora spp. protein-based inheri-

tance. Microbiol Mol Biol Rev 63: 844–861.

Wickner, S., Maurizi, M.R., and Gottesman, S. (1999b)

Posttranslational quality control: folding, refolding, and

degrading proteins. Science 286: 1888–1893.

Wickner, R.B., Taylor, K.L., Edskes, H.K., Maddelein, M.L.,

Moriyama, H., and Roberts, B.T (2000) Prions of yeast as

heritable amyloidoses. J Struct Biol 130: 310–322.



Documents

In vivo aggregation of the HET-s prion protein of the fungus Podospora anserina