Nov 27, 1995 - Miunsterberg,A., Vivian,N., Goodfellow,P. and Lovell-Badge,R. ... Laity,C., Giasson,R., Campbell,R. and Kronstad,J. (1995) Heterozygosity.
The EMBO Journal vol.15 no.7 pp.1632-1641, 1996
The pheromone response factor coordinates filamentous growth and pathogenicity in Ustilago maydis H.Andreas Hartmann, Regine Kahmann" and Michael Bolker Institut fur Genetik und Mikrobiologie der Universitat Munchen, Maria-Ward-Strasse la, 80638 Munchen, Germany 'Corresponding author
In Ustilago maydis, the a and b mating type loci regulate cell fusion, filamentous growth and pathogenicity. The a locus encodes a pheromonebased cell recognition system, and the b locus specifies two homeodomain proteins. The expression of all genes in the a and b loci is induced by pheromone. We have identified a HMG protein (Prfl) that binds sequence specifically to pheromone response elements present in the a and b loci. prfl mutants do not express the a and b genes and are sterile. The disruption of prfl in pathogenic haploid strains results in a loss of pathogenicity. The constitutive expression of the b genes restores pathogenicity and induces filamentous growth in the absence of the pheromone signal. These results provide evidence that pheromone signalling, ifiamentous growth and pathogenic development are linked through Prfl. Keywords: fungal dimorphism/HMG domain/mating type/ pheromone response/transcriptional regulation
Introduction The complex life cycle of the corn smut fungus Ustilago maydis has attracted the attention of phytopathologists and geneticists alike for many years. Pathogenic development is initiated when two compatible haploid cells fuse and form the infectious dikaryon. This process is genetically controlled by the a and b mating type loci. Haploid cells grow yeast-like by budding. They respond to pheromone secreted by cells of opposite a mating type by forming conjugation tubes that fuse at their tips (Snetselaar, 1993; Banuett and Herskowitz, 1994a; Spellig et al., 1994; Figure 1). If the two nuclei in the resulting dikaryon carry different alleles of the multiallelic b locus, filamentous growth is initiated. In the natural situation, this filamentous form of the fungus is able to infect corn plants and undergo pathogenic development (Banuett, 1992; Kamper et al., 1994). Genetic studies involving artificially created diploid strains have revealed different contributions of the a and b mating type loci to these distinct stages of development (Figure 1). While the a locus alone controls cell fusion, and the b locus alone is responsible for regulating pathogenicity, both loci are required for the maintenance of filamentous growth (Rowell and DeVay,
1954; Rowell, 1955; Holliday, 1961; Puhalla, 1970; Day et al., 1971; Banuett and Herskowitz, 1989). Molecular studies have demonstrated that the biallelic a locus contains structural genes for a pheromone-based cell recognition system. Each allele encodes a lipopeptide pheromone precursor (mfa) and a receptor (pra) that recognizes pheromone secreted by cells of opposite mating type (Bolker et al., 1992; Spellig et al., 1994; Urban et al., 1996a). The pheromone receptors belong to the family of the seven transmembrane class that are coupled to heterotrimeric G proteins. The pheromone signal is believed to be transmitted by a MAP kinase cascade, whose complete set of components remains to be identified in U.maydis (Banuett and Herskowitz, 1994b). The transcription of all genes located in the a and b loci is induced upon pheromone stimulation (Urban et al., 1996b). This leads to amplification of the pheromone signal during mating and guarantees that the expression of the b genes is increased prior to fusion. A short DNA sequence, termed the pheromone response element, is found in the vicinity of all pheromone-inducible genes (Figure 6A) and is both necessary and sufficient for pheromone induction (Urban et al., 1996b). The multiallelic b mating type locus encodes the bE and bW homeodomain proteins that are involved in intracellular recognition through combinatorial interactions (Kronstad and Leong, 1990; Schulz et al., 1990; Gillissen et al., 1992). Recently it has been shown that the bE and bW polypeptides from the same allele are unable to interact, whereas the bE and bW gene products from different alleles can form heterodimers (Kamper et al., 1995). These heterodimers are thought to be transcription factors that switch on genes required for pathogenicity (Figure IC). The switch from yeast-like to filamentous growth associated with dikaryon formation requires the autocrine stimulation of the pheromone response pathway (Bolker et al., 1992; Spellig et al., 1994; Figure IC). Haploid strains engineered to express an active bE-bW heterodimer are pathogenic but grow by budding in vitro, illustrating that the active b gene complex is sufficient for tumour induction (Gillissen et al., 1992; Bolker et al., 1995a; Figure ID). These strains can be induced by pheromone of opposite mating type to form filaments, which illustrates that the activation of the pheromone response pathway is necessary for this morphological transition (Gillissen et al., 1992; Spellig et al., 1994). There is a complex interplay between the a and b loci during development (Figure 1), and discovering the molecular level at which pheromone signalling is connected with the control exerted by homeodomain proteins has become an interesting problem. Here we show that a single protein, Prfl, plays a pivotal role in coupling the pheromone response and b gene expression.
1 2 Oxford University Press 1632
Pheromone response factor in Umaydis
A BurI4 ncg
D
B
Dimerization of homeodomain proteins
Pheromone signalling
al
a2
bW2c= bEl
a
;~
0hec -s at: s- r' ci r, -:ri. r. s
r
c- s
Pathocer ic cv
cCuAe
C
bWl bE2 bW2:. bEl
a2
al v
....
-e
P
ec: ty
umrFl ou>s orowYtk
Fig. 1. Schematic illustration of the genetic control of filamentous growth and pathogenic development in Umavdis. (A) Haploid cells grow yeast-like by budding and are non-pathogenic. (B) Reciprocal pheromone stimulation between haploid cells of opposite a mating type induces the formation of conjugation tubes leading to cell fusion. (C) After cell fusion, autocrine pheromone stimulation is necessary to maintain filamentous growth in conjunction with an active bE-bW heterodimer. (D) The active bE-bW heterodimer is sufficient to induce pathogenicity genes in haploid cells without the need for pheromone stimulation.
Results The observation that the transcription of the b genes is pheromone inducible (Urban et al., 1996b) prompted us to ask whether this represents a functionally important interaction between the a and b pathways. To analyse this connection, we have attempted to clone the gene that mediates the pheromone response. The core sequence of the U.maydis pheromone response element (ACAAAGGGA; Urban et al., 1996b) is remarkably similar to the consensus sequence recognized by proteins of the HMG domain family, suggesting that it may be recognized by a member of this class of DNA binding proteins (Grosschedl et al., 1994).
Identification of an HMG domain protein The DNA binding domain of HMG proteins shows elements of sequence conservation (Grosschedl et al., 1994) which we used to design degenerate primers for PCR amplification. Using primers HMG1 and HMG4 (Figure 2), we were able to identify a PCR product of 214 bp that could encode a peptide with significant similarity to the HMG domain (Figure 2B). With the help of this PCR product we identified genomic and cDNA clones. Sequencing revealed the presence of a single open reading frame coding for a protein product of 840 amino acids (Figure 2A; Genbank accession number U40753). We called this gene pfl (pheromone response factor; see below). Comparison
of the cDNA sequence with the genomic sequence revealed no evidence of introns. The presumed translational start site matches the fungal consensus sequence for translation initiation (Ballance, 1990). The HMG domain located in the N-terminal half of the protein belongs to the transcription factor class (see Discussion) and is most similar (32 identical residues in a stretch of 74 amino acids) to that of ROXI of Saccharomyces cerevisiae (Figure 2B). ROXI is involved in the repression of genes under aerobic conditions (Balasubramanian et al., 1993). Sequences located outside the HMG domain in Prfl displayed no similarities to other known proteins. Potential nuclear localization signals are found at positions 126 (KRPR) and 654 (KRRR). Interestingly, two perfect pheromone response elements are located -700 bp upstream of the presumed translational start site of prfl (Figure 2A). By Northern blot analysis, we were able to demonstrate that prfl is expressed in haploid cells; upon pheromone stimulation, expression levels were induced -20-fold (Figure 3).
Aprfl mutants are sterile To study the function of prfl in pheromone signalling we generated a prfl mutant allele, Aprfl-l. This was accomplished by deleting the region encoding amino acids 22-196, including the entire HMG domain, and replacing this with the hygromycin resistance cassette. Aprfl-i was introduced into the compatible haploid strains FB1 (al bl) and FB2 (a2 b2) by gene replacement (data not shown). To analyse mating behaviour, the mutant strain FB2prfl- was co-spotted with strains FB 1 (al bl), FBD 123 (al a2 bi bl) and FBD11-7 (al al bi b2). These tester strains were chosen to distinguish between loss of function in pheromone production and pheromone recognition. All combinations of the wild-type strain FB2 with the three tester strains produced aerial hyphae seen as a white fuzziness (Fuz+ phenotype; Figure 4). In combination with FB1 and FBD12-3, this fuzziness is caused by cell fusion and the development of dikaryotic hyphae. In the FB2 and FBD 11-7 combination, the Fuz+ reaction results from pheromone stimulation of the diploid strain that is heterozygous for b but does not involve cell fusion (Spellig et al., 1994; Laity et al., 1995). Pheromone mutants are able to fuse with FBD12-3 (al a2 bi bl; because this strain provides both pheromones), and as receptor mutants produce pheromone, they can stimulate the filamentous growth of FBD11-7 (al al bi b2; Bolker et al., 1992; Figure 4). FB2prfl- was unable to elicit a Fuz+ reaction with any of the tester strains (Figure 4). This suggests that Aprfl-I mutants are sterile because they are defective in both pheromone and receptor function. Consistent with the observed mating defect on plates, mutant strain FB2prfl- is non-pathogenic in combination with FB 1 (Table I).
Prfd regulates the transcription of pheromone-inducible genes To understand the basis for the sterility of the Aprfl-i mutants better, we have analysed the expression patterns of four genes that are induced by pheromone: mfal, pral, bEl and bWJ (Urban et al., 1996b). Total RNA was isolated from FB 1 and FB lprfl-, and from the same strains exposed to pheromone. For pheromone induction, 1633
H.A.Hartmann, R.Kahmann and M.BbIker
A ..... ...CCCTTTGT ..
.
TCC~~~~~~~~~~~T~~~~~TGT..
...
.
......~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~. ..
r17F,--
S tel M atl1-2 MT
X
*M
wN
I
a-liw rtr j
ivIH
Fig. 2. Nucleotide and protein prf1 sequences. (A) Nucleotide sequence and deduced amino acid sequence of the U.maydis prf1 gene. Pheromone response elements in the region preceding the open reading frame are indicated in bold, underlined letters. Restriction sites used for the construction of the Apr-fl-i disruption mutant are indicated. The location of primers (HMG 1 and HMG4) used for PCR amplification is shown. Asterisks mark the polyadenylation sites observed in three independent cDNA clones. (B) Amino acid sequence alignment of the Prf 1 HMG domain with members of the Sry subgroup of the HMG superfamily. The Prf I HMG domain (residues 126-199) was compared with the fungal HMG domains of S.cerevisiae ROXlI (residues 10-87; Balasubramanian et al., 1993), Spombe Ste 1I (residues 16-84; Sugimoto et al., 199 1), Cochliobolus heterostrophus Matl1-2 (residues 131-203; Turgeon et al., 1993), Neurospora crassa MT a- I (residues I116-187; Staben and Yanofsky, 1990), S.pombe Matl-Mc (residues 103-175; Kelly et al., 1988), Podospora anserina Fprl (residues 132-204; Debuchy and Coppin, 1992) and the HMG domain of murine Sry (residues 5-77; Gubbay et al., 1990). The multiple sequence alignment was performed using the program CLUSTAL W (Thompson et al., 1994). Highlighted amino acids represent identity between P41 and one or more of the other sequences. Asterisks indicate three highly conserved amino acids in the Sry class of sequence-specific HMG domains (Travers, 1995).
1634
Pheromone
al
b2
al b2 pra2
factor in U.maydis
Table I. Prf 1 is essential for pathogenicity
-1 pI-t
response
Number of infected plants
Number of plants with tumours
20 30 39 48 8
20 0 31 0 8
I
FBI XFB2
FBIXFB2prfl CL13 CL13 prflP CL13 prfl- bcon
cbx
pheromone induced FBI FBI IFBI FB prf I prf I -
Fig. 3. Pheromone induction of the prfl gene. The transcription of prfl was investigated by Northern blot analysis. RNA was prepared from haploid strains FB6b (a] b2) and FB6b/pra2 (al pra2 b2). In FB6b, the pheromone response is not activated; in FB6b/pra2, the pra2 gene has been introduced by transformation (Boilker et al., 1992), resulting in the constitutive activation of the pheromone response. The probes used are indicated to the left. The cbx gene serves as a loading control.
I
mfa A.el,
\X
.
I
\$ r:
pral
bEl
aI2 b2
wu. -gama
a2 b2
Iirnfa-
bWl
a12 b2 pra2actinl
a2 b2 prfl Fig. 4. prfl mutants are sterile. Relevant genotypes of the strains used are indicated to the left and on top. The respective strains were spotted alone and in the indicated combinations on charcoal-containing plates. The white fuzziness (Fuz+ phenotype) indicates the formation of aerial hyphae. The following strains were used: FBI, al bl; FBD12-3, a] a2 bl bl; FBDII-7, al al bI b2; and FB2. a2 62. The mfa2 mutant strain is RK2232 (a2 b2 mnfa2::hvg), the pra2 mutant strain is RK2234 (a2 b2 pra2::hvg) and the ptfl mutant strain is FB2prfl- (a2 b2 Apt-]f-I).
FB1 and FBlprfl- were cocultivated with RK2176 (a2 Ab), which provides a2 pheromone and should stimulate the transcription of pheromone-inducible genes. For the wild-type strain FB1, Northern blot analysis revealed the expected pattern of enhanced expression of all four genes under conditions of pheromone stimulation (Urban et al., 1996b; Figure 5). In contrast, the expression of these genes was undetectable in the FB lprfl- mutant, even under conditions of pheromone stimulation (Figure 5). This shows that prfl is required both for the basal level of transcription and for the pheromone-induced transcription of the pheromone and receptor genes. For the b genes, there is clearly an effect of Prf on the induced level of transcription. However, the low level of
Fig. 5. Transcription of pheromone-inducible genes is affected in ptfl mutants. The expression of the genes indicated on the left was analysed in strains FBI (a] bl) and FBlprfl- (al bi Aprfl-J). To analyse pheromone induction, strains were cultivated alone or together with RK2176 (a2 Ab) which provides pheromone (designated pheromone induced). Strains and mixtures of strains were plated on CM charcoal medium for 48 h and total RNA was analysed by Northern blotting. Probes for the genes indicated on the left are described in Materials and methods.
b gene expression in haploid cells does not allow us to assess whether prfl also affects the basal transcription of these genes.
Pdfl binds to the pheromone response element assay the interaction of Prfl with the pheromone response element, a His-tagged version of Prf 1, comprising amino acids 1-289 and including the HMG domain, was overexpressed in Escherichia coli, purified and used for DNase I footprint analysis (Figure 6). The recombinant Prf 1 1-289 protein bound specifically to all seven pheromone response elements in the al and b2 loci that perfectly match the consensus sequence (Figure 6A and B). Elements with mismatches are bound less efficiently (some examples are To
1635
X4,;:i.__+
H.A.Hartmann, R.Kahmann and M.Bolker
A
444 -4444*
.4 /e, ,~/a
B;vM
-
...4-...4- 4 K 4
-4
B
44 : p. m.
e *.4^
6162
f/ \t .)
S..1%: 0-
34
]l4b .
-v k5 ]b2 - 7
s _
t__-.
w_ --^S
-3
w _
__ .
p. ____ _
P',
*>.* _
...
..
l ]b2-2
._.,
j al -4
..
._
__ -
- ,.,
a
a1
_
_ ......
al -5
.
_-
1al -6
joD
b-4
___
Jb2 -6
ja1 -1 0 al
-7
J *-rT .11
A AG AC A
n-
C
*.. : A
.::
LI
..
._.V GA
TGTTTC OCT
-_
_
..::
::..
_-
Fig. 6. Recognition of pheromone response elements by Prf (A) The location of the Prf binding sites in the al and b2 alleles. Triangles denote pheromone response elements (ACAAAGGGA, with one mismatch allowed). Of these sequences, the following match the consensus perfectly: al-2, al-6, al-7, b2-4, b2-5, b2-6 and b2-7. Prfl1289 has been shown to bind to all elements indicated by filled triangles. Exons of the mating type genes are denoted by open bars. Arrows indicate the direction of transcription. Borders of the a] allele are indicated by thick lines. (B) DNase I footprinting of recombinant Prfll259 protein on fragments containing pheromone response elements. (I) Prfl binding sites in the mfal promoter region; (II) binding sites in the pral promoter region; (III) binding sites in the bW2 intron; and (IV) binding sites in the region downstream of bE2. DNase I digestion patterns were analysed on 6% polyacrylamide gels. Lane M is a Maxam-Gilbert G>A sequence reaction. In (I), lanes 1 and 5 contained no protein, and lanes 2-4 contained 25, 100 and 200 ng Prfll289 protein, respectively. In (II)-(IV), lane I contained no protein, and lane 2 contained 200 ng Prfll289 protein. Brackets indicate the protected regions; arrowheads indicate enhanced DNase I cleavage. The dashed bracket in (II) indicates weak binding to the sequence (TGAAAGGGT) which deviates from the consensus at three positions. (C) Common features of those sites protected from DNase I digestion by Prfll2895 Protected regions on the upper and lower strands of the consensus pheromone response element are indicated by brackets; DNase I hypersensitive sites are marked by arrowheads. 1.
shown in Figure 6B). The protected region comprised 18 bp and extended 2 bp on the 5' side and 7 bp on the 3' side on the upper strand of the consensus sequence (ACAAAGGGA). For the lower strand a similar protection 1636
was observed. In some cases, protection was accompanied by DNase I hypersensitive sites located immediately adjacent to the 3' end of the protected region (Figure 6B and C).
pattern
Pheromone response factor in Umaydis
( 1- It -"'
CT !, (1-i":t pr !'I p r I' )1', -
-
h"k
hEI
_w_
have generated a b allele (bcon) in which transcription of the bEl and the bW2 genes is driven by the constitutive U.maydis promoters derived from the hsp7O and the tefl gene (Holden et al., 1989; M.Dahl and R.Kahmann, unpublished results). The bcon allele was introduced into CL13prfl- by gene replacement. Northern blot analyses showed that in this strain both b genes are expressed at detectable levels without pheromone stimulation (Figure 7). On charcoal plates CL13prfl-bcn exhibited a strong Fuz+ phenotype (Figure 7) comparable with that of a dikaryon derived from the mating of two compatible haploid strains (Figure 4). When injected into maize plants, CL13prfl-bcon was fully pathogenic (Table I). These data demonstrate that under these experimental conditions prfl affects filamentous growth and pathogenicity by regulating the level of b gene expression.
Discussion (luIlII
coo.
Fig. 7. The constitutive expression of an active b gene complex induces filamentous growth in prfl mutants. The strains indicated on top were analysed for growth on CM charcoal medium (top row). Total RNA was prepared from the same strains grown under identical conditions and analysed by Northern hybridization for the expression of bEl and bW2. An actinl probe served as the control. The two signals observed for bW2 are seen because the bW2 message comigrates with ribosomal RNA which interferes with hybridization. The fragments used as probes are described in Materials and methods. The following strains were used: CL13 (al bW2 bEl), CL13prfl(al bW2 bEl prfl-2) and CL13prfl-bc,n (al bcoll prfl-2)
Prfl is essential for pathogenic development Because prfl mutants are sterile, the experiments described so far cannot address the question of whether prfl has additional functions after cell fusion. Therefore we introduced a prfl mutation into the haploid strain CL13 (al bW2 bE]), which carries the bW2 and the bEl gene and thus can form a heterodimer able to activate pathogenicity genes. The prfl-2 allele, in which the open reading frame is disrupted by the insertion of a hygromycin resistance cassette at a position corresponding to amino acid 195, was introduced by gene replacement (data not shown). In FB2, the prfl-2 allele causes the same mating defect as Aprfl-J (data not shown). On charcoal plates the CLl3prfl- mutant strain showed a Fuz- phenotype, even after prolonged incubation, contrasting with the phenotype of CL13 which produces some filaments under these growth conditions (Figure 7). When assayed for pathogenicity on corn plants, CL13prfl- was unable to cause tumours (Table I). This was quite unexpected because activation of the pheromone response pathway is not needed for pathogenicity (Figure 1; Banuett and Herskowitz, 1989). This result can be explained if it is assumed that loss of pathogenicity and inability to form filaments both result from a lack of b gene expression; alternatively, prfl might regulate additional genes involved in pathogenic development. To distinguish between these possibilities, we asked whether prfl-independent b gene expression could restore pathogenicity in CL13prfl-. We
In this study we have identified an HMG domain protein Prfl that mediates the pheromone response in U.maydis. Much to our surprise, this protein is required not only for cell fusion and filamentous growth, but also for pathogenicity. Recombinant Prfl protein binds specifically to the U.maydis pheromone response elements, which are present in the vicinity of all genes in the a and b loci. Of the 21 pheromone boxes analysed for Prfl binding, the highest affinity was found for the seven elements that match the consensus. No or very weak binding was observed for three elements that deviate from the consensus at positions 3 (A to G) and 6 (G to C, G to A). In all other elements that deviate from the consensus, binding affinity was detectable but reduced in comparison with the consensus. In the al allele, pheromone response elements are found predominantly in the promoter regions of the pheromone and the receptor gene. This contrasts with the situation in the b2 allele, where no such elements are found in the intergenic spacer separating the divergently transcribed b genes. Interestingly, a number of pheromone response elements are located downstream of bE2 and in the transcribed region of bW2. The b gene complex consists of a sequence region that is highly variable between different alleles. Variability is restricted to the intergenic spacer and the N-terminal protein coding regions for bE and bW that determine specificity. In this region recombination is largely prevented so that allele integrity is maintained (Kahmann et al., 1995). This could explain why conserved regulatory promoter elements are placed in the constant regions where recombination is allowed. From the location of the pheromone response elements it must be concluded that they act at a distance; however, it is unclear whether these elements can stimulate transcription from both b gene promoters or activate only one of these promoters. prfl mutants are viable but neither produce nor respond to pheromone. The transcription of pheromone-inducible genes in the a and b loci is reduced to levels undetectable by Northern blot analysis. In uninduced haploid cells, Prfl clearly affects the expression of the genes in the a locus; an effect of Prf 1 on the basal expression of the b genes is less clear because these genes are expressed at very low levels. This makes it likely that Prfl is required for
1637
~ -.
H.A.Hartmann, R.Kahmann and M.Bolker
A
Im ttl fal7
i
Prf 1
Hiaploid cell
7Id
W Lib 7 bVW?
