MAPKKK are kinases involved in cell signaling. In fungi, these kinases are known to regulate development, pathogenicity, and the sensing of external conditions. We show here that Podospora anserina strains mutated in PaASK1, a MAPKKK of the MEK family, are impaired in the development of crippled growth, a cell degeneration process caused by C, a nonconventional infectious element. They also display defects in mycelium pigmentation, differentiation of aerial hyphae, and making of fruiting bodies, three hallmarks of cell differentiation during stationary phase in P. anserina. Overexpression of PaASK1 results in exacerbation of crippled growth. PaASK1 is a large protein of 1832 amino acids with several domains, including a region rich in proline and a 60-amino-acid-long polyglutamine stretch. Deletion analysis reveals that the polyglutamine stretch is dispensable for PaASK1 activity, whereas the region that contains the prolines is essential but insufficient to promote full activity. We discuss a model based on the hysteresis of a signal transduction cascade to account for the role of PaASK1 in both cell degeneration and stationary-phase cell differentiation.
IN eukaryotes, intracellular and intercellular signaling is involved in developmental processes and adaptation to the environment. Key players in signaling pathways are mitogen-activated protein kinase (MAPK) modules (Widmannet al. 1999), which are composed of three successive kinases: the MAP kinase kinase kinase (MAPKKK) that phosphorylates the MAPKK that phosphorylates the MAPK. These cascades are well suited for some signaling events because they allow ultrasensitivity; i.e., the outputs are not proportional to the inputs, but display an all-or-none activation (Huang and Ferrell 1996), which permits the cells to make clear-cut decisions. However, the presence of mitogen-activated protein (MAP) kinase phosphatases, by dephosphorylating the MAPK, modulates this all-or-none behavior (Bhallaet al. 2002).
In animals and plants, differentiation may be driven by emergent properties of regulatory networks, including MAP kinase signal transduction cascades, which can create sequences of differentiated cellular “states” without modification of nucleic acid sequences (Ferrell 2002). In fungi, phenomena akin to differentiation in higher eukaryotes are present, but for most the underlying mechanisms are unknown, although easily detected because the variations they promote segregate in a non-Mendelian fashion. These phenomena are designated as “phenotypic switches” leading to various morphological, physiological, and virulence “states” described in several yeasts including Saccharomyces cerevisiae (Clemonset al. 1996; True and Lindquist 2000; Lockshon 2002; Kuthanet al. 2003), Candida albicans (Soll 2002), Candida glabrata (Lachkeet al. 2002), and in the basidiomycete dimorphic fungus, Cryptococcus neoformans (Frieset al. 2002). The determinism of these morphological switches is known only for S. cerevisiae in which the [PSI+] prion has been shown to be the causal factor (True and Lindquist 2000). Such differentiation-like processes are also widespread in filamentous fungi, in which they affect morphology, fertility, and other cellular processes and it is estimated that about one-third of the species harbors at least one of these differentiation processes (Silar and Daboussi 1999). We have described one such process in the ascomycete Podospora anserina (Silaret al. 1999) and named it crippled growth (CG). The crippled growth cell degeneration process leads to several phenotypes, including an inability to differentiate fruiting bodies and aerial hyphae, slow growth, and alteration of pigmentation.
We have shown that CG is caused by a cytoplasmic and infectious factor named C (Silaret al. 1999). Elements such as C can trigger degenerative processes in P. anserina because P. anserina's thallus presents a coenocytic structure that permits interhyphal propagation to a large area of the thallus via septal pores or natural cell fusion (Esser 1974). Infectious particles are thus able to spread to a large area of the mycelium. The properties of C suggest that it is an epigenetic factor rather than a nucleic acid (Silaret al. 1999). First, C is induced by “stationary phase” because it appears in the oldest cells of cultures grown on a petri plate; these cells are rapidly starved for nutrient and enter a state similar to stationary phase. Some of these cells are nonetheless able to mobilize storage materials to build aerial hyphae, to produce pigments, and to complete the sexual cycle, which requires differentiation of the multicellular fruiting bodies (Esser 1974). We showed that C is expressed in these stationary-phase cells even if they display a healthy appearance (Silaret al. 1999). Second, C is cured by stresses and by environmental conditions that promote growth. Remarkably, one of the mechanisms involved in the restriction of C during growth is activated by translation errors. In wild-type strains that display a basal error level, C is efficiently removed from growing cells and growth almost always proceeds normally. In contrast, in cells with a low error level, i.e., in cells that carry a so-called “antisuppressor” mutation (AS strains; Picard-Bennoun 1976), C is not removed after passage in stationary phase and triggers the CG alteration. In usual petri plate growth conditions, C rarely appears spontaneously in AS strains, but the probability of its appearance is greatly increased by cultivation in race tubes (Silaret al. 1999). When cultivated in the same conditions, AS strains can thus be obtained in two alternate growth regimens, Normal and Crippled, a condition named bistability. Such conditions mimic a simple differentiation process (Ferrell 2002). In the case of CG, bistability is especially easy to study since AS strains can be switched from Normal to Crippled by passage in stationary phase and switched from Crippled to Normal by application of various stresses, such as culture at high temperature or in high osmolarity medium.
Infectious epigenetic elements that trigger bistability, such as C, can result from several mechanisms that involve a positive feedback regulatory loop, including the presence of a prion-like protein or a self-sustained regulatory circuitry (Silar and Daboussi 1999). Because C may not be easily purified and because injection into P. anserina hyphae is not easily feasible, we have undertaken a molecular genetic approach to unravel the mechanisms of C production. Here, we describe three mutants that are altered in a gene required for the expression of C. These mutants are also unable to differentiate some characteristics of the stationary phase, i.e., pigmentation, aerial hyphae, and fruiting bodies. We cloned the corresponding gene and show it to encode PaASK1, a MAPKKK of the MEKK family. A model relying on emergent properties of a signal transduction cascade is proposed to explain how this MAPKKK controls crippled growth and stationary-phase differentiation.
MATERIALS AND METHODS
Strains and culture conditions: The P. anserina mutant strains used in this study were derived from the S strain, ensuring a homogenous genetic background (Rizet 1952). The AS4-44 and AS4-43 strains contain an amino acid change in the translation elongation factor eEF1A (Silar and Picard 1994), and the AS6-5 strain contains a mutation located in a gene encoding an unidentified ribosomal protein (Dequard-Chablatet al. 1986). The 193 mutant strain has a mutation in a gene involved in a green pigment biosynthesis and lacks color at every stage of its life cycle (Picard 1971); the 193 mutant strain is used to measure in vivo translation accuracy (Picard 1973). Auxotrophic mutations for leucine and lysine are leu1-1 and lys2-1, respectively (Marcouet al. 1984). The phenotypes of the IDC118, IDC172, and IDC507 mutants are described in the results. The geographic races used were B, D, E, M, U, V, and Z (Belcouret al. 1997).
Standard culture conditions, media, and genetic methods for this fungus have been described (Esser 1974). Because P. anserina is a pseudohomothallic, ascospores yielding heterokaryotic mycelia can be obtained after sexual reproduction. Alternatively, heterokaryons form spontaneously and readily in this species when mycelium fragments are mixed.
Mutant screens: Protoplasts of strains AS4-44 and AS6-5 were prepared and regenerated as described (Brygoo and Debuchy 1985) and spread on regeneration plates at 104/plate and irradiated with UV254nm at 250 J/m2; the plates were incubated at 27° in the dark. After 2 days, 100 regenerating clones/plate could be observed under the binocular (1% survival). The mutagenized clones were transferred on fresh M2 medium at nine clones/plate. The plates were incubated for 10 days at 27° to permit entrance into stationary phase. The clones were then replica plated on M2 plates supplemented with 50 mm ammonium acetate because this compound stabilizes CG (Silaret al. 1999). The growth of resulting mycelia was checked. Several mutants that did not display CG were obtained and their full genetic analysis will be described in a forthcoming article. Among them, we selected IDC118, IDC172, and IDC507 that mapped at the same locus (see results) for further analysis. IDC118 and IDC172 were obtained in the AS6-5 background and IDC507 in the AS4-44 background and the mutants were thus at first available as IDC118 AS6-5, IDC172 AS6-5, and IDC507 AS4-44 strains.
Nucleic acid manipulations: Standard methods were used for nucleic acid manipulation (Ausubelet al. 1987). RNA was extracted as described (Lecellier and Silar 1994) or with the RNeasy plant mini kit from QIAGEN (Chatsworth, CA), except that liquid cultures were made in Roux flasks and the cells were collected, frozen in liquid nitrogen, and mechanically broken in a mortar before hot phenol extraction.
Cloning of PaASK1: Protoplasts from the IDC118 mutant were sequentially transformed with pools of 50 cosmids from a P. anserina wild-type genomic bank (provided by Michelle Dequard-Chablat) as described (Brygoo and Debuchy 1985). The female fertility of the transformants was then checked, and in pool no. 36 some were fertile. The 50 cosmids from this pool were purified and transformed independently. With cosmid 36E1, nearly 100% of transformants with a wild-type phenotype were obtained, indicating that it likely contained the wild-type allele of the gene mutated in IDC118. Various restriction enzymes were used to subclone 36E1 in pCB1004 (Carrollet al. 1994). The smallest ClaI-BamHI fragment, which complemented in an ectopic fashion, measured ∼6.5 kb and was contained in plasmid pSKI3.
Complete sequence for this fragment was established. It contains a large CDS (coding sequence) interrupted by one intron at its C terminus. Position of the intron was established by RT-PCR using the Titan one tube kit from Roche Diagnostics. The primers K5, 5′-GTCGTCGCGTCTCTGACTCG-3′, and K12, 5′-TGTGTAACAAACGCTCAAACGCT-3′, amplified a 670-pb product in PCR reactions. Sequence of this product with the primer K26, 5′-GACTCTCCCTTTATCCATTA-3′ permitted us to define intron/exon junctions and established that the gene was expressed; the gene was designated PaASK1.
Sequence of the PaASK1 alleles that are mutant in IDC118, IDC172, and IDC507 was established by sequencing PCR amplification products obtained by two or more independent reactions with the mutant DNA as template. The primers for the amplification in the case of IDC118 and IDC172 were K23, 5′GGTGTACCTGCCGACGCGGCCGCCA-3′, and K26. Primers for amplification in the case of IDC507 were K54, 5′-ATCGATT TTACACTCC-3′, and K12, 5′-TTCTCGCCATACTTATAGCG TCT-3′.
Construction of overexpression plasmid: To overexpress PaASK1, we used the pBC-HA vector (a gift from H. Lalucque) that contains the AS4 promoter (Silar and Picard 1994) cloned into the NotI site of pBC-hygro that carries a hygromycin-resistant marker (Silar 1995). Downstream of the AS4 promoter, several unique restriction sites permit the introduction of the desired CDS. The BamHI fragment that surrounds PaASK1 was purified from cosmid 36E1, filled with T4 DNA polymerase, and cloned in pBluescript (Stratagene, La Jolla, CA). In one of the resulting plasmids, a ClaI fragment that included the complete CDS could be excised and cloned directly in pBC-HA; the fragment starts 60 nucleotides before the start codon and ends ∼300 bp after the stop codon. The resulting plasmid was called pSUR-PaASK1.
The strain IDC118 was transformed with pSUR-PaASK1 and numerous transformants were obtained, most with a wild-type phenotype; PaASK1-sur1 and PaASK1-sur2 were selected for further analysis. A cross with wild type was performed and analysis of the progeny showed that the transgenes inserted at single loci in the genome that were unlinked to PaASK1.
Construction of partially deleted alleles: To construct a modified PaASK1 gene that encoded a protein lacking the polyglutamine stretch, the 5′ of the gene containing the promoter region and the proline-containing domain were amplified by PCR, using plasmid pSKI4 as a substrate. This plasmid contained a T4 blunt-ended RcaI-RcaI fragment surrounding PaASK1 that was obtained during subcloning of cosmid 36E1 into the SmaI site of pBC-SK+ (Stratagene). The primers used were 4503 (5′-GTAAAGGGAACAAAAGCTG-3′), which is located in the cloning vector in a region upstream from the multicloning site, and K41 (5′-CTGCTGTTGCTGTGTCAT GAGCGGAGGTG-3′), which is located in a region just upstream from the polyglutamine stretch. The region 3′ of the polyglutamine stretch was amplified by PCR with primer K47 (5′-CAAAGTGATCTCATGACGGCGCTGACAAGT-3′), which is located in a region just downstream from the polyglutamine stretch, and primer K28 (5′-AATATTCCTCACCAACCTCC-3′), which is located in PaASK1 just downstream from a NsiI site. The first fragment was digested by XbaI and RcaI and the second by RcaI and NsiI and both fragments were purified and ligated in a single step with pSKI4 digested with XbaI and NsiI. To construct the pASK1-ΔPQM plasmid that carried the PaASK1 catalytic domain fused to the PaASK1 promoter, the catalytic domain and promoter regions were amplified by PCR using paired primers K49 (5′-GGGTTGGGGATCATGACGTC TATCCGTGAG-3′) and 3382 (5′-CACTATAGGGCGAATTGG-3′) and 4503 and K40 (5′-GGGAGGGTTGGTCATGATGTCT TCCTC-3′), respectively. Primer 3382 is located downstream from the multicloning site of pBC-SK+. The fragments were purified and digested with XbaI/RcaI and RcaI/HindIII, respectively, and ligated with the pBC-hygro vector digested with XbaI and HindIII. This yielded, after Escherichia coli transformation, plasmid pASK1-ΔN. This resulted in the making of pASK1-ΔQ, a plasmid carrying the desired modified gene. To construct the pASK1-ΔPQ plasmid that carried a deletion of the proline domain and polyglutamine region of the PaASK1 gene, we ligated the XbaI-RcaI fragment from plasmid pASK1-ΔN that contained the PaASK1 promoter region with the RcaI-HindIII fragment from plasmid pASK1-ΔQ that contained the PaASK1 coding sequence downstream from the polyglutamine stretch into the pBC-hygro vector cut with XbaI and HindIII. To construct the pASK1-ΔQM plasmid that had the polyproline region fused with the catalytic domain, we had to use the pBR322 vector (Biolabs) because the construct was highly unstable in the pBluescript derivative. The catalytic domain was excised from pASK1-ΔPQM by RcaI and HindIII. The promoter region, along with the beginning of the gene containing the polyproline region, was amplified by PCR from cosmid 36E1 and cloned into pGEM-t (Promega, Madison, WI). This process resulted in the creation of an EagI site upstream from the promoter. This promoter + polyproline fragment was then excised using EagI and RcaI and cloned along with the RcaI and HindIII fragment of pASK1-ΔPQM containing the catalytic domain in pBR322 cut with EagI and HindIII to yield pASK1-ΔQM. All constructed plasmids were sequenced to ensure that no undesirable mutation occurred in the coding sequence.
Mutant IDC118 was transformed with pASK1-ΔPQM and pASK1-ΔPQ and cotransformed with pBC-hygro and pASK1-ΔQ for one set and pBC-hygro and pASK1-ΔQM for the second set. No restoration of female fertility was observed among the 50 transformants analyzed in the experiments with pASK1-ΔPQM and pASK1-ΔPQ, showing that the mutated alleles were not functional. Numerous female fertile transformants were obtained with pASK1-ΔQ. Two were selected and crossed with wild type. Progeny analysis showed that both had a transgene inserted in a unique locus unlinked to PaASK1; both transgenes lead to the same phenotypes. With pASK1-ΔQM, several transformants complementing the pigmentation defect of IDC118 were obtained. Two were selected and crossed with wild type. Progeny analysis showed that both had a transgene inserted in a single locus unlinked to PaASK1; both transgenes led to the same phenotypes.
Identification of three mutants that impair the development of crippled growth: AS4-44 and AS6-5 strains are two antisuppressor strains that present CG after passage in stationary phase (Silaret al. 1999). To begin to unravel the molecular mechanism responsible for CG, we screened for mutants that impair the development of crippled growth (IDC mutations) in these two strains after passage in stationary phase (see materials and methods). Briefly, we selected for mutants that would not display a different shape before and after stationary-phase incubation. Here we focus on three mutants recovered during these screens, IDC118, IDC172, and IDC507. In the genetic screens, IDC118 and IDC172 were obtained in the AS6-5 background and IDC507 in the AS4-44 background and the mutants were thus at first available as IDC118 AS6-5, IDC172 AS6-5, and IDC507 AS4-44 strains. These three strains displayed the same properties in that they not only lacked CG (see below), but also were unable to make aerial hyphae and pigmentation and to reproduce sexually (Figure 1). The inability to reproduce was due to a defect in female organ formation; the ascogonia that are the progenitor cells of the female organs were present, but the protoperithecia that are the mature female organs were not differentiated. The mutants made male gametes like wild type; they grew slightly slower than wild type and produced a dense, flat mycelium. Therefore, these mutants seem not to be significantly modified during growth, but presented some stationary-phase differentiation defects. The color, shape, and fertility phenotypes permitted an easy genetic analysis.
When crossed with wild type, the three mutations segregated 2:2, suggesting that they were caused by a single mutation. Lack of color, aerial hyphae, and female fertility cosegregated, indicating that the three phenotypes were due to the same mutation. Ascus analysis showed that the three mutations segregated with a second-division segregation frequency of 90%; they also segregated independently of AS6-5 and AS4-44, showing that they were not secondary mutations within these genes. In the progeny of these crosses, we recovered the IDC118, IDC172, and IDC507 single mutants that were not associated with AS6-5 and AS4-44. These strains had the same phenotype as the parental strains, i.e., lack of CG, color, aerial hyphae, and female organs.
Heterokaryons between the IDC118, IDC172, and IDC507 mutants and the 193 tester strain of opposite mating type were constructed; the 193 strain lacks a green pigment at all stages of its life cycle (Picard 1971). The heterokaryotic mycelia readily differentiated small and unpigmented fruiting bodies that typically developed from 193 ascogonia and larger dark-green ones that originated from ascogonia, which carried the IDC mutation. This observation indicated that the wild-type product(s) necessary to develop fruiting bodies and produced only by 193 nuclei could rescue the fertility defect of the IDC mutants and, hence, that the mutations were recessive. It also suggested that some anastomoses occurred for cytoplasm to mix and that the inability to develop CG was not trivially due to lack of natural cell fusion. To confirm this assumption, we first confronted the IDC118 mutant with the D strain and observed a barrage reaction at the confrontation zone. The barrage results from the death of cells that have fused and mixed their cytoplasms with the incompatible strain (Rizet 1952). We also measured life span of the three mutant strains. It is known from previous experiments that longevity reflects the speed at which the cytoplasmic factor responsible for senescence spreads through cell fusions (Marcou 1961). Therefore, life-span measurement provided a way to check if cell fusions occurred with wild-type efficiency. None of the three IDC mutations significantly modified P. anserina life span, suggesting that cell fusions were likely not affected in the mutants. Finally, we checked if heterokaryons could be formed with the IDC118 mutant. We constructed IDC118 leu1-1 and IDC118 lys2-1 strains; leu1-1 and lys2-1 are auxotrophic mutations that prevent growth on minimal medium. Heterokaryons were then made by mixing together and spreading the mycelia of the tested strains on minimal medium. Leu1-1 × lys2-1, IDC118 leu1-1 × lys2-1, and IDC118 lys2-1 × leu1-1 heterokaryons were formed with the same rate, but the formation of IDC118 lys2-1 × IDC118 leu1-1 was delayed and less efficient as judged by the number of prototrophic sectors. Overall, the data suggest that cell fusion occurred. Note that the delay in the formation of heterokaryons could be explained by the slightly reduced growth of the IDC118 mutant or by a slowing in a step subsequent to cell fusion, since in P. anserina anastomoses are not immediately followed by nucleus mixing (Marcou 1961).
The three IDC mutations abolish the production of C: IDC118 AS6-5, IDC172 AS6-5, and IDC507 AS4-44 never presented CG after passage in stationary phase, unlike AS6-5 and AS4-44 that exhibited CG in 100% of the cases. A cross between IDC118 AS6-5 (IDC172 AS6-5) and the AS4-44 strain allowed constructing the IDC118 AS4-44 strains (IDC172 AS4-44). A cross between IDC507 AS4-44 and AS6-5 allowed constructing the IDC507 AS6-5 strains. The IDC118 AS4-44, IDC172 AS4-44, and IDC507 AS6-5 strains never presented CG after passage in stationary phase, showing that the IDC mutations were not specific to a particular AS background.
An obvious reason why CG was abolished could be the restoration of a wild-type level of translation errors in the IDC strains. To discount this possibility, the single mutant strains IDC118, IDC172, and IDC507 were crossed with the 193-translation accuracy tester strain as indicated (Picard 1973). The color of the ascospores obtained from these crosses showed that the three mutations did not increase translation error rate.
To check if the C element was still produced in these mutants during stationary phase as observed in wild type, contamination tests were set up (Silaret al. 1999). In the contamination tests, mycelium explants from donor cultures were deposited at the growing edge of AS recipient cultures. If the donor cultures contain C, it should be transmitted through cell fusion to the AS recipient strain and create a sector of CG in the mycelium downstream from the explants (Figure 2). We have optimized this test by using ammonium-acetate-containing medium on standard petri plates (3 g/liter of CH3COONH4). Under these conditions, the C element rarely appears spontaneously (zero sector on 20 petri plates when using AS6-5), ensuring that the sectors observed downstream from the explants are due solely to transfer of C from the donor to the recipient strain. Table 1 and Figure 2 report the data obtained by using various combinations of donors and recipients. The donor explants were taken from the stationary-phase mycelium to guarantee that C is present in the donor explants. C was not detectable during stationary phase in AS6-5 IDC118, AS6-5 IDC172, and AS4-44 IDC507 as no sector was obtained when using these strains as donor. Additionally, no sector among 30 trials for each mutant was observed when IDC118 AS6-5, IDC172 AS6-5, and IDC507 AS4-44 were used as recipient, showing that these strains were unable to propagate C. Controls with wild type as a recipient and IDC118 AS6-5, IDC172 AS6-5, and IDC507AS4-44 as donors did not show any sector formation as expected (Silaret al. 1999). Overall, these data indicate that the gene mutated in IDC118, IDC172, and IDC507 is absolutely required for the expression and propagation of C.
The gene mutated in IDC118, IDC172, and IDC507 encodes a MAPKKK of the MEKK family: We cloned the mutated gene in IDC118, IDC172, and IDC507 by functional expression of the wild-type allele in the IDC118 mutant strain and by looking for complementation of its female sterility (see materials and methods). One cosmid, 36E1, was able to complement the fertility defect and restore normal pigmentation and aerial hyphae production. Additionally, 36E1 was also able to reestablish the production of C during stationary phase as 8 of 30 explants taken from stationary-phase culture could induce CG in contamination experiments. Upon transformation, this cosmid integrated at the IDC118 locus in three of the six tested transformants, indicating that it likely contains the corresponding wild-type gene (Picardet al. 1987). The cosmid was subcloned and a ClaI-BamHI 6.5-kb fragment fully able to complement all the phenotypes when present at an ectopic position was recovered. In particular, 12 sectors were observed in the contamination assay using AS6-5 as a recipient and the complemented strain as donor, showing that this fragment was sufficient to restore production of C.
Complete sequence was established for this fragment and reveals that the fragment contains a large 1832-codon CDS that was interrupted by one intron (Gen-Bank accession no. AY077730; Figure 3). Search in the data banks reveals that the deduced protein is homologous to the MAPKKK of the MEKK family, especially with Mkh1p from Schizosaccharomyces pombe and Bck1p from S. cerevisiae, for which an identity of 50 and 48% and a similarity of 67 and 68% are observed, respectively, when comparing the kinase catalytic domain. This conserved catalytic domain is located at the C terminus of the protein and is fused at its N terminus to a very large region that is much less conserved (Figure 3).
The primary sequence of the protein suggests that it may be composed of five domains (Figure 3). The first domain is rich in proline. Comparison with the homologous protein of Neurospora crassa obtained by searching the complete genomic sequence of this organism (available at http://www-genome.wi.mit.edu/annotation/fungi/neurospora/) reveals that all the prolines are conserved, arguing for an important role in the function of the protein. This domain is terminated by a glycine stretch. The second domain consists of a 60-glutamine stretch whose position is not conserved in N. crassa. Interestingly, the protein from N. crassa also presents a small polyglutamine stretch, but in the fourth domain. The third domain contains a first motif that is loosely conserved in the fungal proteins and is terminated by a second glycine stretch. The end of the fourth, and largest, domain contains a second motif loosely conserved between the fungal proteins and ends with a small glycine stretch. The last domain of the protein is the catalytic domain.
A human member of this family of MAPKKK is ASK1, an MAPKKK involved in cell death (Ichijoet al. 1997). Because the MAPKKK that we identified controls cell degeneration, we called it PaASK1, the corresponding gene PaASK1, and the three mutant alleles PaASK1-IDC118, PaASK1-IDC172, and PaASK1-IDC507.
Although Northern analysis did not detect a messenger RNA corresponding to PaASK1 (data not shown; see Figure 4 for a slot-blot analysis), RT-PCR irregularly amplified a product corresponding to the reverse transcription of the mature mRNA in extracts from both growing and stationary-phase cultures, indicating that the gene is weakly expressed in all growth conditions.
Null alleles of PaASK1 impair crippled growth: To obtain the sequence of the PaASK1 mutant alleles, we transformed IDC118, IDC172, and IDC507 mutants with various DNA fragments encompassing wild-type PaASK1. The fertility defect of IDC118 and IDC172 was rescued in some transformants by a 4-kb DNA fragment covering the last two-thirds of the gene, including the catalytic domain, whereas in IDC507 no such transformants were recovered. Genetic analysis showed that all the transformants with a wild-type phenotype had the transformed DNA fragments integrated at the PaASK1 locus. This indicated that restoration of the phenotype was triggered by a recombination event that restored a wild-type gene and hence that the mutation was contained in the 4-kb fragment that was amplified by PCR from IDC118 and IDC172 genomic DNA. Sequence analysis of the PCR products indicated that PaASK1-IDC118 and PaASK1-IDC172 differed from wild type by single mutations. In PaASK1-IDC172, a single base difference changed the conserved asparagine 1670 of the catalytic domain to a tyrosine. In PaASK1-IDC118, a two-consecutive-base modification changed lysine 865 to asparagine and created a stop codon. Because the 4-kb DNA fragment did not rescue IDC507, we surmised that the mutation was not located in this region. Sequence of a DNA fragment covering the beginning of the gene and amplified by PCR from the IDC507 genomic DNA revealed a single base insertion after codon 60 that created a +1 frameshift, leading to a truncated product of 141 amino acids. Therefore, in both PaASK1-IDC118 and PaASK1-IDC507, only a truncated protein without the catalytic domain could be produced. These data indicate that the defects in stationary-phase differentiation and production of C, presented by the IDC, result from a complete loss of function of PaASK1.
Overexpression of PaASK1 permits wild type to present crippled growth: To further establish the role of PaASK1 in the control of CG, we overexpressed the PaASK1 mRNA in wild-type and AS4-43 backgrounds. We fused the PaASK1 coding and terminator sequences with the promoter of the AS4 gene (Silar and Picard 1994), which encodes translation elongation factor eEF1A, one of the most abundant proteins of the cell; expression with the AS4 promoter should be very high. The IDC118 strain was transformed with the chimeric gene, and PaASK1-sur1 and PaASK1-sur2, two transformants with restored fertility, were selected for further analysis. Because these primary transformants might be heterokaryotic, both were crossed with wild type. In the progeny, we recovered the PaASK1-sur1 PaASK1+ and PaASK1-sur2 PaASK1+ strains. Slot-blot analysis verified that PaASK1 mRNA level was increased in these two strains (Figure 4). PaASK1-sur1 PaASK1+ and PaASK1-sur2 PaASK1+ displayed a wild-type phenotype, showing that overexpression of PaASK1 was not toxic. However, the strains displayed a weak and reversible CG after stationary phase (Figure 4). When in stationary phase, PaASK1-sur1 PaASK1+ and PaASK1-sur2 PaASK1+ contained C since they triggered sectors in contamination assays with AS6-5 as a recipient in 29/30 and 21/30 cases, respectively. To confirm this, PaASK1-sur1 PaASK1+ and PaASK1-sur2 PaASK1+ were crossed with the AS4-43 mutant because the AS4-43 mutation promotes a very weak CG (Silaret al. 1999) and can be used to detect synergic effect. Among the progeny of this cross, we recovered strains with the following genotypes: PaASK1-sur1 PaASK1+ AS4+, PaASK1-sur1 PaASK1+ AS4-43, PaASK1-sur2 PaASK1+ AS4+, and PaASK1-sur2 PaASK1+ AS4-43 with equal frequencies showing that the two transgenes were not linked to AS4. PaASK1-sur1 PaASK1+ AS4-43 and PaASK1-sur2 PaASK1+ AS4-43 presented a strong CG (Figure 4). Therefore, PaASK1 overexpression not only permitted wild type to present CG, but also increased the severity of CG in the AS4-43 mutant.
The glutamine stretch is dispensable, but the proline domain is essential for activity of PaASK1 and domains III and IV are important for CG: Because CG is caused by a cytoplasmic and infectious element, a plausible explanation is that C is a prion. In the yeast S. cerevisiae, prions can be generated by proteins rich in glutamine (DePaceet al. 1998). We investigated the glutamine stretch in the function of PaASK1, especially its role in the control of CG. The glutamine stretch was conserved among wild-type P. anserina strains although it varied in length from 46 to 71 (Figure 5). The strains with the longest stretch were unable to present CG as our wild-type reference S strain. The AS6-5 strain that presents CG had the same glutamine number as the S strain from which it was derived.
We constructed in vitro a mutant allele with a complete deletion of the glutamine stretch (Figure 6; PaASK1-ΔQ) and introduced this allele in IDC118 by transformation. ASK1-ΔQ1 and ASK1-ΔQ2, two independent transformants able to complement the sterility defect of the mutant strains, were selected for further studies and crossed with wild type. Analysis of the progeny showed that:
ASK1-ΔQ1 and ASK1-ΔQ2 had integrated at a single position in the genome.
The integration in ASK1-ΔQ1 and ASK1-ΔQ2 did not occur at the PaASK1 locus since progeny with an IDC phenotype were obtained. Therefore, expression did not result from the reconstitution of a wild-type gene by recombination.
The progeny that carried the transgenes were indistinguishable from wild type; they did not present CG and they were able to transmit C to an AS6-5 recipient during contamination assays in 19 of 30 trials. Therefore, the glutamine stretch did not play any detectable role in the function of PaASK1.
Various additional internal deletions were made in vitro in the PaASK1 gene to evaluate the role of other domains of the protein (Figure 6); the modified genes were introduced by transformation in IDC118. When the first proline-rich domain was absent, no phenotypic restoration was detected among the transformants. In randomly selected transformants with constructs lacking the proline-rich region, we could not detect C during stationary phase by contamination assays with AS6-5 as recipient (in 30 trials for each construct).
On the contrary, the proline-rich domain fused to the catalytic domain was sufficient to restore pigmentation and aerial hyphae, but not female fertility, in the transformants (Figure 6). The transformants expressing the PaASK1-ΔQM chimeric protein also exhibited a characteristic growth with periodic waves. The sterility and growth phenotypes were recessive, suggesting that the mutant allele encoded a protein displaying an incomplete activity (Figure 6). To check if the C element was produced by these strains, contamination tests were set up with AS6-5 as recipient and PaASK1-ΔQM as donor. Explants taken from the stationary phase of PaASK1-ΔQM resulted in the formation of sectors (59/144 trials). Explants taken from the growing margin did not trigger sectors (0/30 trials). This indicated that the partial phenotype exhibited by PaASK1-ΔQM correlated with a reestablishment of C production during stationary phase. We evaluated whether the PaASK1-ΔQM mutant was able to sustain CG when associated with an AS mutation by crossing it with the AS4-43 and AS6-5 strains. PaASK1-ΔQM AS4-43 and PaASK1-ΔQM AS6-5 strains were recovered in the progeny and tested for the development of CG. In both cases, CG was abolished (Figure 6), indicating that domains III and IV, lacking in this mutant, are important for fruiting body development and for sustaining CG.
Crippled growth, a cell degenerative process of the filamentous fungus P. anserina, is under the control of PaASK1, a MAP kinase kinase kinase of the MEKK family; two major findings support this notion. First, null alleles in PaASK1 entail an inability to produce the cytoplasmic and infectious C element associated with crippled growth. The mutant strains are unable to present CG, are not infectious for the AS6-5 strain that can present crippled growth, and cannot be infected by strains carrying C. Second, overexpression of PaASK1 allows the wild-type strain to present CG in normally unreceptive conditions.
In addition to the inability to make C, IDC mutants of PaASK1 display defects in a stationary-phase differentiation program leading to sexual reproduction. PaASK1 defines a MAP kinase module homologous to the Bck1p/Mkk1p-Mkk2p/Mpk1p(Slt2p) module of S. cerevisiae (Heinischet al. 1999). In other filamentous fungi, the homologous module is known by the corresponding MAP kinases (Meyet al. 2002). Mutations in these MAP kinases seem to lead to cell wall defects, but these defects trigger different outcomes at the cell and organism level. For example, in the plant pathogen Magnaporthe grisea, deletion of the Mps1 kinase leads to a decrease of aerial hyphae and conidia production and to autolysis of the stationary-phase region of the thallus and female sterility (Xuet al. 1998); this results in decreased virulence of M. grisea mutant strains. In Fusarium graminearum, the mgv1 deletion mutant has a reduced hyphal growth, is female sterile, undergoes some autolysis, and is unable to make heterokaryons with wild-type MGV1 strains (Houet al. 2002). In another plant pathogen, Claviceps purpurea, deletion of CPMK2 leads to a decrease in conidiation and a hyperbranched mycelium, but not to autolysis (Meyet al. 2002). Finally, deletion of MPKA in Aspergillus nidulans decreases apex extension speed and germination of conidia in low osmolarity media (Bussink and Osmani 1999).
We show that the PaASK1 mutants are unable to differentiate aerial hyphae, pigmentation, and fruiting bodies, which are three hallmarks of stationary phase in P. anserina. These phenotypes resemble those observed for the deletion of Mps1 in M. grisea, the most closely related fungus for which the role of the MAPK module has been investigated (Xuet al. 1998), but with the difference that we have not observed any autolysis. Therefore, PaASK1 seems most necessary for a stationary-phase-specific differentiation program, which is not the survival program as the IDC118, IDC172, and IDC507 strains can be stored over a very long period in stationary phase like wild type. We thus suggest that PaASK1 is activated in the cells that will give rise to the differentiated structures or in the precursors of these cells to signal stationary phase. As the related fungal Map kinase modules seem implicated primarily in cell wall regulation, it is possible that PaASK1 signal a strengthening of the cell wall that would ensure proper pigment fixation, allowing for the elaboration of aerial hyphae and fruiting bodies. However, this kinase is also involved in many other aspects of cell maintenance in S. cerevisiae, including control of the cell cycle and nutrient sensing (Heinischet al. 1999; Harrisonet al. 2001), and could signal a switch of cell physiology affecting more components than the cell wall. The fact that in the PaASK1-ΔQM mutant aerial hyphae and pigmentation, but not fertility, are restored suggests that the signaling pathway is a multi-step process in P. anserina.
How can a MAPKKK control the spreading of a cytoplasmic and infectious factor responsible for cell degeneration? We show first that restoration of a wild-type translation error level in the IDC mutants is not responsible for the lack of CG. Second, although we cannot completely eliminate the possibility that the coenocytic structure of the mycelium is modified in a way that it cannot amplify and propagate C, we present evidence that this hypothesis is unlikely. Indeed, C is never induced in the IDC mutants, whereas it should be if PaASK1 acts through the coenocytic structure, and cell fusions are likely not modified in the mutants on the basis of the facts that: (1) vegetative incompatibility with the D strain is observed in the mutants; (2) longevity is not different from wild type; (3) complementation can be achieved with the 193 strain; and (4) heterokaryons can be constructed with the IDC118 mutant, although we observe a delay in the formation of IDC118 lys2-1 × IDC118 leu1-1 heterokaryons.
Overall, our data suggest that PaASK1 participates directly in the pathway leading to the formation of C. We can propose three levels of action for PaASK1. First, the making of C could be downstream from the signal transduction cascade containing PaASK1. Under this hypothesis, the self-regulatory mechanism responsible for C amplification would be independent of PaASK1. If this hypothesis were true, we would expect the AS6-5 IDC118, AS6-5 IDC172, and AS6-5 IDC507 strains to be infected in the contamination experiments when they are used as recipient. As we do not see this, we suggest that C is likely not downstream of the PaASK1 cascade. Second, C could be formed upstream and need PaASK1 to exert its deleterious effects. In this instance, we would expect the IDC118, IDC172, and IDC507 strains to contain C and transmit it in contamination experiments. We have not observed such transmission and thus suggest that C is likely not formed upstream from the cascade.
We are left with a third level, which is that C is generated by PaASK1. A possibility is that PaASK1 is a prion and that C is the aggregated form of PaASK1. Even though we cannot reject this hypothesis at the present time, three circumstantial evidences militate against this hypothesis. First, the strains containing the aggregated form of PaASK1, i.e., C, and the strains carrying the inactivated alleles of PaASK1 should have the same phenotype, because aggregation usually abolishes the activity of the proteins. This criteria is one of the three that establish a prion (Wickner 1994). Here, the strains containing C and those with an inactivated PaASK1 have a radically different phenotype (Figure 1). Second, we did not detect any role of the polyglutamine region, whereas this region is essential for prion conversion and propagation in yeast (DePaceet al. 1998; Uptain and Lindquist 2002). However, this region is dispensable in other prions (Uptain and Lindquist 2002). Third, there is no known prion to date whose formation is inducible by a physiological stimulus.
We thus favor the hypothesis that C originates from emergent properties of the PaASK1 signal transduction cascade, as described for related cascades, because it would explain all the properties associated with C (Bhalla and Iyengar 1999; Bagowski and Ferrell 2001; Ferrell 2002; Shaet al. 2003). In MAP kinase cascades, self-positive regulation results in hysteresis (Bhalla and Iyengar 1999; Bagowski and Ferrell 2001); i.e., the activation of the cascade is sustained even in the absence of the inducer. The return of the cascade to the inactive state can be performed only by an additional external deactivating mechanism (Bhallaet al. 2002). The role of hysteresis is to ensure clear-cut and irreversible cellular decisions through the cascade activation (Ferrell 2002). One striking property of these systems is that activation generates a cytoplasmic and infectious factor that controls the activation of the cascade. This theory has been well established in the JNK MAPK cascade in Xenopus eggs. The JNK module can be activated in the eggs by sorbitol. Cytoplasmic fractions taken from these eggs with an active cascade and transferred into a nonactivated egg are sufficient to trigger activation in the absence of sorbitol (Bagowski and Ferrell 2001). The authors named the responsible cytoplasmic factor JPF for JNK activation promoting factor.
To account for the involvement of the PaASK1 MAPKKK in the generation of the cytoplasmic and infectious factor C, we propose that the PaASK1 cascade presents a self-positive regulation and is endowed with hysteresis. C would thus be analogous to JPF and would be the cytoplasmic state in which the PaASK1 transduction cascade is active. The complex functioning of the cascade that stems from this property is schematized in Figure 7. We speculate that in our system some stationary-phase signal would trigger activation of PaASK1 to permit proper stationary-phase differentiation through the regulation of several genes acting downstream of the cascade. The role of hysteresis in the signaling cascade would be to trigger an all-or-none commitment without reversibility toward the differentiation, which would lead to pigmentation and aerial hyphae formation and culminate with sexual reproduction. The phenotype of the PaASK1-ΔQM mutant supports this notion; this mutant is partially active because it is able to restore pigment, aerial hyphae, and C production during stationary phase. However, it is not fully active because fertility and the ability to develop CG is lacking. In this mutant, the hysteresis could be abolished, hence the inability to sustain CG and the sexual cycle.
If our proposal is correct, hysteresis could be a general property of MAP kinase cascades not only in animals (Ferrell 2002), but also in all eukaryotes. To establish this definitively, this hysteresis hypothesis will require the identification of the MAP kinase kinase and MAP kinase downstream from PaASK1, as well as the other partners, which act upstream to activate or repress the PaASK1 module and to evaluate their role in the building of C. We expect that inactivation of these MAPK and MAPKK genes will also result in the inability to make C. Of prime importance is the identification of the positive feedback loop that triggers the hysteresis (Ferrell 2002). Here, our data suggest that domains III and/or IV of PaASK1 are required for hysteresis, but other molecules are likely necessary. This long task is now underway by a combination of reverse genetics on candidate genes and the cloning of the genes mutated in the other IDC mutants that we have identified in our mutant screens as mapping outside PaASK1.
We thank Fabienne Malagnac, Corinne Vierny, Hervé Lalucque, and all members of the laboratory for useful discussion, Judith Bender for a critical reading of the manuscript and K. B. B. Sobering for correcting the English. This work was supported by Action Concerteé on ESST and prions reference 2000-37, Aide aux Jeunes Equipes from Centre National de la Recherche Scientifique, and INTAS grant no. 00491. Sébastien Kicka is a recipient of a fellowship from the Ministère de la Recherche and Philippe Silar is professor at the University of Paris VII, Denis Diderot. The work was done in compliance with the current laws governing genetic experimentations in France.
Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession no. AY077730.
Communicating editor: M. S. Sachs
- Received June 16, 2003.
- Accepted November 24, 2003.
- Copyright © 2004 by the Genetics Society of America