OUP user menu

Homology effects in Neurospora crassa

Caterina Catalanotto, Tony Nolan, Carlo Cogoni
DOI: http://dx.doi.org/10.1111/j.1574-6968.2005.00037.x 182-189 First published online: 1 January 2006


It has become clear in the past few years that eukaryotic organisms possess different genetic systems to counter viruses, transposons and other repeated elements such as transgenes that could otherwise accumulate in the genome. In addition to serving as a model organism for genetic, biochemical and molecular studies, Neurospora crassa has proved to be a paradigm for the study of gene-silencing mechanisms. Indeed, its genome can be protected from expansion of selfish nucleic acids by a variety of mechanisms that inactivate duplicated sequences. Studies of these mechanisms have made a fundamental contribution to the understanding of the gene-silencing field.

  • Neurospora
  • repeat-induced point mutation (RIP)
  • quelling
  • meiotic silencing by unpaired DNA (MSUD)
  • posttranscriptional gene silencing (PTGS)
  • siRNA


In the past century, genetic studies focusing on the consequences of nucleic acid homology effects were largely centred on recombination and segregation in meiosis, where the pairing of homologous chromosomes is an essential step. The repair of DNA double-strand breaks and dosage compensation because of X-inactivation in mammals are also based on homology effects. These considerations suggest that the detection of sequence identity is a powerful mechanism to control correct genome transmission and gene expression in the progeny (Gloor, 2002; Lee, 2002). More recently, the genetic transformation of many eukaryotes has revealed a number of homology-dependent gene-silencing (HDGS) phenomena that serve to preserve the overall structure of the genome from the expansion of transposable elements and from infection by RNA and DNA viruses and viroids. These defence mechanisms may either operate at a level where transcription is blocked (transcriptional gene-silencing [TGS]) or at a level where mRNA is degraded after transcription (posttranscriptional gene-silencing [PTGS]) (Pickford & Cogoni, 2003). The plethora of related gene-silencing mechanisms that act at different levels suggests an evolutionary radiation from an ancestral defence mechanism with a specialization in different functions. The filamentous fungus Neurospora crassa represents a valuable model to study the functional relationship between different gene-silencing phenomena, as three different gene-silencing phenomena have been found to operate in diverse phases of the Neurospora life cycle (see Fig. 1) (Selker, 1990). In the vegetative phase (Fig. 1a), the introduction of transgenes homologous to an endogenous gene highlighted the existence of a reversible PTGS mechanism known as ‘quelling’ (Romano & Macino, 1992). Furthermore, during the sexual phase, two other gene-silencing mechanisms have been, in part, characterized in Neurospora: repeat-induced point mutation (RIP) (Cambareri , 1989), a TGS mechanism that occurs during the premeiotic phase of sexual reproduction (Fig. 1b), and meiotic silencing by unpaired DNA (MSUD), a PTGS mechanism occurring after karyogamy in the zygote cell (Fig. 1c) (Shiu , 2001). In addition to these mechanisms, and similarly to plants and mammals, N. crassa possesses a DNA methylation system, usually superimposed on RIP (Selker , 2003), which is able to render repetitive sequences inactive. At this point, it is interesting to ask why Neurospora possesses so many HDGS mechanisms and what the implications of these mechanisms are for genome evolution in Neurospora. This review will attempt to respond to these questions, discussing each silencing mechanism in this light.

Figure 1

Silencing pathways in the Neurospora cell cycle. (a) Quelling occurs in the vegetative phase when Neurospora grows as a haploid mycelium. In conditions of nitrogen starvation, each mating type (A and a) initiates the sexual phase and is fertilized by the opposite mating type. Fertilization begins with cytoplasmic fusion between a specialized female hypha and a male vegetative cell. (b) During plasmogamy, the haploid nuclei are confined in a specialized hook-shaped structure (crozier), whereas repeat-induced point mutation takes place on duplicated sequences. Red C : G pairs represent targets of methylation by RID (a putative DNA methyltransferase). Putative deamination of these pairs can lead to a T : A conversion (blue). Unmutated C : G pairs (shown in green) are subsequently targeted by DIM-2 DNA methylation. (c) After DNA replication, the fusion of the two nuclei (karyogamy) occurs with the formation of the zygote. In this cell, unpaired DNA sequences are identified and silenced by the meiotic silencing by unpaired DNA pathway.

Repeat-induced point mutation

As the first HDGS mechanism to be discovered in N. crassa more than 15 years ago, RIP is a process that can detect and inactivate DNA sequences that are duplicated during the haploid dikaryotic stage of the sexual cycle. As a result of RIP, duplicated sequences greater than ∼400 bp and having a nucleotide identity greater than ∼80% undergo DNA methylation and are irreversibly mutagenized via G : C to A : T transitions (Fig. 1b) (Galagan , 2003). Several lines of evidence support the idea that RIP results from DNA–DNA interactions: RIP occurs strictly in a pairwise manner with both copies being RIPed at the same time; linked repeated sequences are more readily RIPed than nonlinked repeats; the action of RIP is limited to a single nucleus in the heterokaryotic tissue of the crozier where RIP takes place, suggesting that there is no diffusible signal (Selker, 1997). The mechanism of RIP mutation is not clear, but it has been proposed that it involves de novo methylation of cytosines in premeiotically paired sequences, followed by their deamination, which would result in a cytosine to thymidine conversion (Fig. 1b) (Selker, 1990; Cogoni & Macino, 2000). Supporting this hypothesis, the rid-1 (RIP- defective) gene, which is required for RIP, encodes a putative DNA methyltransferase protein (Freitag , 2002). Moreover, RIPed sequences also become elicitors for further methylation by another DNA methyltransferase, DIM-2, during the vegetative phase (Rountree & Selker, 1997). It was shown that the positive signal able to promote DNA methylation consists of high densities of TpA dinucleotides and of A : T-rich sequences, which are consequences of the mutations created by RIP (Miao , 2000; Tamaru & Selker, 2003). As a result of DNA methylation, the RIPed sequences exhibit a strong reduction in transcription (Rountree & Selker, 1997). Nuclear run-on experiments suggested that DNA methylation blocks transcription elongation causing the RNA polymerase II to stall. On the contrary, methylation in promoter regions does not affect transcription, highlighting a difference between N. crassa and plants and mammals, where methylation of promoter sequences frequently functions to block the initiation of transcription (Cogoni & Macino, 2000). However, as shown in in vitro assays, DNA methylation alone is not sufficient to block transcription, indicating a requirement for other downstream factors, which possibly function to induce further chromatin modifications that form transcriptionally silent heterochromatin (Rountree & Selker, 1997).

As a genome defence mechanism, RIP appears to be highly effective as the N. crassa genome is littered with RIP-mutated relics of a wide range of transposon families and does not reveal the presence of a single active transposon (Galagan , 2003).

It is also possible to attribute to RIP a role in genome organization. The high efficiency and permanent nature of RIP in inactivating repetitive sequences probably explains the low percentage of highly similar gene families in N. crassa (Galagan & Selker, 2004). Gene duplication is proposed as a major source of evolutionary innovation, because it allows selection to mutate one copy of a gene freely while the other copy continues to fulfil the original function. The availability of the recently sequenced genome of Neurospora has revealed a meagre number of gene repeats. Thus, studying this small number of repeated sequences can allow us to gain insight into the mechanisms that a duplicated gene might use to evade RIP. Some repeated sequences escape from RIPing owing to their small size, like the 5S rRNA, the tRNAs and the histone H4 genes, which are all under the threshold length required to activate RIP. Moreover, and again probably to avoid RIPing, the ∼75 copies of the Neurospora 5S rRNA are not tandemly arranged as they are in other eukaryotic organisms, but are scattered throughout the genome in order to avoid small tandemly repeated sequences being recognized as larger repeated blocks and activating RIP. It is possible that Neurospora has evolved this strategy to prevent the loss of these genes through successive sexual cycles (Galagan , 2003). Analogous natural selection as a consequence of the presence of RIP in Neurospora could explain why the histone genes for H2A, H2B and H3 in Neurospora are present in single copy and contain intronic sequences, whereas, despite their conserved amino-acid profile, other eukaryotic genomes contain several intronless versions of these genes (Hays , 2002). In this case evolution might have selected only those individuals in which the insertion of introns in the histone genes was able to fragment the continuous stretch of homology to below the RIP threshold. However, it is important to note that some sequences repeated in the Neurospora genome are protected from RIP, despite their large size. For example, the 9 kb rDNA tandem repeat, although present as ∼200 copies in the Neurospora genome, produces the large ribosomal rRNAs. To the contrary, when introduced by random transformation into the Neurospora genome, these sequences were RIPed normally, suggesting that the particular nucleolar organization of the endogenous rDNA, and not its nucleotide composition, is able to protect them from RIP silencing (Selker, 1990). It is possible that some components of the nucleolar organiser region can interfere with some component of the RIP machinery or prevent their access to the nucleolar sequences.

Although the above observations make Neurospora an interesting model to study the strategies of the genome evolution alternative to gene duplication, it is possible that RIP is highly constraining within the evolutionary context, perhaps explaining why it is unique to certain fungi (Galagan , 2003; Galagan & Selker, 2004).

PTGS phenomena in Neurospora crassa

The introduction of transgenes or double-strand RNAs (dsRNAs) into a variety of eukaryotic cells can trigger a series of PTGS mechanisms in which dsRNA intermediate molecules, after being processed into short interfering RNA molecules (siRNA), were identified as strong elicitors of mRNA degradation. Because dsRNA is an intermediate molecule of the replication of several viruses and mobile genetic elements, it has been suggested that the presence of dsRNAs is interpreted by eukaryotic cells as an ongoing viral infection (Pickford & Cogoni, 2003). Supporting this hypothesis, the pathways responsible for both transgene-induced PTGS in plants (cosuppression) (van der Krol , 1990) and RNA-interference (RNAi) via the introduction of dsRNA in animals (Fire , 1998) have been shown to limit the expansion of transposable elements and viral propagation (Chicas & Macino, 2001). Two PTGS mechanisms have been identified in Neurospora: quelling and meiotic silencing by unpaired DNA (MSUD). Forward and reverse genetic approaches as well as biochemical studies have indicated that both quelling and MSUD share common features with PTGS phenomena in other eukaryotes, suggesting that the molecular bases of all the PTGS phenomena are similar in different organisms (Waterhouse , 2001). Interestingly, the two Neurospora PTGS pathways appear to have their own complete silencing machinery. The alignment of the molecular components of quelling and MSUD to their orthologues in plants, animals and Schizosaccharomyces pombe indicates that the Neurospora paralogues are sufficiently diverse from each other to be clustered into two separate clades, suggesting an ancient origin of the two PTGS pathways (Borkovich , 2004). These observations are consistent with the fact that the two mechanisms are active in different phases of the life cycle and respond to completely different triggers (either the presence of repetitive sequences or that of unpaired DNA during meiosis).


Quelling occurs during the vegetative phase of the life cycle and was the first PTGS mechanism characterized in this organism (Fig. 1a). Several components of the quelling machinery have been identified using either forward- or reverse-genetic approaches. The identification of genes required in the silencing process together with findings from other organisms has led to a current model for quelling. The introduction of transgenes has been proposed to lead to the transcription of a silencing signal, namely a transgenic RNA, which in turns triggers the silencing cascade. Importantly, transgenes per se are not sufficient to induce silencing and it appears that silencing is triggered only when the transgenic sequences are organized in large tandem arrays. It has been proposed that these large repeated sequences can potentially form secondary structures such as cruciform arrangements that can be recognized by the QDE-3 RecQ DNA helicase (Cogoni & Macino, 1999a). It has therefore been suggested that either QDE-3 can facilitate the production of a qualitatively different, or aberrant, RNA resulting from this transcription, or the presence of QDE-3 at the transgenic locus is able to attract QDE-1, an RNA-dependent RNA polymerase (RdRP) (Cogoni & Macino, 1999b). In either case, QDE-1 converts a single-stranded RNA into a dsRNA molecule that is then processed by one of two redundant DICER-like ribonuclease proteins (DCL-1 and DCL-2) into siRNAs of 21–25 nucleotides (Catalanotto , 2004). These siRNAs are incorporated into a large multiprotein complex known as the RNA-induced silencing complex (RISC) containing the QDE-2 protein (Catalanotto , 2002). qde-2 encodes a homologue of the Argonaute protein that provides a nuclease activity to the mammalian form of RISC (Liu , 2004). The siRNAs are used by RISC as guide molecules to degrade complementary mRNA targets. In addition to guiding RISC, data from Caenorhabditis elegans suggest that siRNAs can also function as primers to synthesize new dsRNA from complementary mRNA, thereby amplifying the reaction (Cogoni & Macino, 2000; Sijen , 2001). In Neurospora, in vitro studies have shown that the QDE-1 RdRP can convert mRNA into dsRNA in both a primer-dependent and a primer-independent fashion. The preference or capability of this RdRP in vivo is not clear (Makeyev & Bamford, 2002).

The dsRNA is a necessary intermediate in the quelling pathway and the efficiency of silencing appears to be linked to its quantity. This is evidenced by two observations: (1) the introduction by the transformation of a construct able to transcribe a hairpin dsRNA leads to a high frequency of quelling and the strength of silencing seems to be proportional to the level of transcription of the dsRNA hairpin (Goldoni , 2004). (2) The direct expression of dsRNA bypasses the requirement of qde-1 and qde-3 suggesting that these two genes are involved in steps upstream of the production of dsRNA. Moreover, the number of transgenes required to induce and to maintain quelling was reduced in transformed strains, where the overexpression of QDE-1 allowed a more efficient conversion of RNA into dsRNA, suggesting that conversion of transgenic RNA into dsRNA is a limiting step in the quelling pathway (Forrest , 2004).

In Neurospora, as in other organisms, it would seem that quelling serves to limit the expansion of transposons, since an introduced Tad element, a LINE-1-like retrotransposon, has an elevated expansion in the absence of the quelling components QDE-2 and DICER (Nolan , 2005). In addition, Tad is also susceptible to RIP (Kinsey , 1994). Moreover, siRNAs against RIPed relics of several classes of DNA transposons have also been detected, suggesting that, at the time of invasion, their active versions likewise would have been subject to quelling (Chicas , 2004).

Why does Neurospora have two mechanisms, RIP and quelling, to work on the same sequences? One reason why this might be so could lie in the advantage that co-operation between RIP and quelling offers: transposons invading during the vegetative phase would be limited in their expansion by quelling, thus rendering RIP more efficient by reducing the number of the copies to be mutated in a single sexual cross. RIP can inactivate only a percentage of duplicated sequences; therefore, maintaining a restricted set of Tad elements during the vegetative phase might lead to a faster elimination of these sequences in successive sexual cycles (Nolan , 2005).

Meiotic silencing by unpaired DNA

It has been shown that during meiosis each allele in the zygote (the only diploid cell in the Neurospora lifecycle) can ‘sense’ the presence of its partner on the homologous chromosome (Aramayo & Metzenberg, 1996). If a DNA sequence cannot be detected on the opposite chromosome, the looped unpaired region produces a silencing trigger able to silence all homologous sequences in the genome, whether they are paired or not (Shiu , 2001). In principle, MSUD is not restricted to any specific sequences. However, because of its reversible and transient nature, its phenotypic effect is only seen on those genes required for meiosis or ascospore development. On this basis it has been proposed that the ascus dominance of many Neurospora mutants could be explained with MSUD (Shiu , 2001). More recently, mutational studies have revealed at least two genes responsible for MSUD: sad-1 and sms-2. As these two genes are paralogues of the quelling components qde-1 and qde-2, respectively, it seems likely that quelling and MSUD are highly related processes (Shiu , 2001; Lee , 2003). The starting point of meiotic silencing should be the transcription of RNA from unpaired regions that is converted into dsRNA by SAD-1 (Fig. 1c). However, as in quelling, the presence of a promoter is not required to induce silencing, suggesting that the transcription of the unpaired DNA does not occur via a conventional polII mechanism (Lee , 2004). MSUD, like quelling, acts at a posttranscriptional level as MSUD is activated only by the nonpairing of transcribed coding sequences, and not intronic sequences or a promoter sequence. In analogy with quelling, it is likely that siRNAs produced by DICER are loaded onto a downstream RISC-like effector complex containing the Argonaute homologue SMS-2 (Fig. 1c) (Lee , 2003). As yet, however, siRNAs have not been detected in association with MSUD, probably owing to the difficulty of manipulating the sexual structures where MSUD occurs. As quelling and MSUD present many similarities, it is not clear why Neurospora has evolved and conserved two virtually identical PTGS mechanisms. Neurospora could have developed a meiotic PTGS mechanism as a powerful defence specialized in counteracting transposons similar, for example, to P element in Drosophila (Bingham , 1982), some elements in C. elegans (Vastenhouw & Plasterk, 2004) or the LINE-1 element in humans (Trelogan & Martin, 1995), which all jump predominantly in germinal cells where meiosis takes place. Recently it has been demonstrated that RIP-induced methylation of repeated sequences can increase the efficiency of sensing this sequence and triggering MSUD (Pratt , 2004). This would suggest that, again, there could be a co-operation between silencing mechanisms in Neurospora to ensure an efficient shutting off of duplicated genes.

The ability of MSUD to detect and inactivate unpaired DNA during the meiotic prophase could have important evolutionary consequences on promoting genome stability by preventing transmission of genomic rearrangements such as deletions, duplications and translocations. In addition, it has been proposed that MSUD could constitute a powerful isolation mechanism for speciation in Neurospora. Microheterogeneity accumulated within two reproductively isolated populations could be tolerated normally by MSUD during intrapopulation crosses, whereas build-up of this heterogeneity might supercede the detection threshold for meiotic silencing during interpopulation crosses, rendering them barren. Support for this idea comes from the observation that usually barren inter-specific crosses within the Neurospora genus could became fertile in an MSUD defective context (for example in a Sad-1 background) (Shiu , 2001).

A link between PTGS and TGS mechanisms in Neurospora

Another open question in Neurospora concerns the connection between PTGS and TGS mechanisms. The formation of transcriptionally silent heterochromatin as found in a wide range of organisms, including S. pombe, Arabidopsis and mammals, is dependent on components of the PTGS machinery (Lippman & Martienssen, 2004). At the centromeric repeats in S. pombe, for example, siRNAs corresponding to the repeats are loaded into an Argonaute-containing complex similar to the RISC called the RNAi-mediated Initiation of Transcriptional Silencing (RITS) complex. RITS is able to interact with a histone methyltransferase that leads to histone H3K9 methylation and ultimately to the binding of SWI-6, a homologue of heterochromatin protein-1. RITS is able to guide these heterochromatic modifications in a sequence-specific manner because of the loaded siRNA, which, similar to its function in RISC, acts as a guide molecule (Martienssen , 2005). This system is evident at the centromeres in S. pombe, which has numerous repeated transposon relics and satellite sequences (Volpe , 2002). Amazingly, heterochromatin formation at these regions is essential for correct kinetochore assembly and chromosome segregation, suggesting that a system initially used for genome defence has been co-opted by the cell for a role in cell regulation.

The components of transcriptional silencing in Neurospora would appear to be similar to those in other organisms, in that there it includes a histone methyltransferase, DIM-5. DIM-5 methylates H3K9, which in turn leads to the recruitment of chromodomain protein HP1 and finally the DIM-2-dependent DNA methylation (Freitag , 2004a). However, it has been demonstrated that DNA methylation and HP1 localization occur normally in mutants defective in all components of PTGS (in both quelling and MSUD) (Freitag , 2004b). Furthermore, methylation of H3K9 was similarly unaffected at transgenic loci in quelling mutants (Chicas , 2005). Likewise, just as the PTGS machinery is not required for heterochromatin formation, the reverse is also true, in that quelling can occur in the absence of the transcriptional components dim-2 and dim-5 (Chicas , 2005).

The absence of a direct link between PTGS and heterochromatin in Neurospora means that it is yet to be understood how this organism manages to direct heterochromatin formation in regions such as the centromere. The analysis of centromeric regions of Neurospora revealed that they are comprised of clustered retrotransposon-like elements (Cambareri , 1998). Interestingly, these repeated sequences present a high degree of degeneracy that is suggestive of RIP inactivation. Almost any RIP-inactivated sequence is able to function as a de novo methylation signal when re-introduced into Neurospora (Miao , 1994). It is therefore possible that, in Neurospora, methylation of repeated sequences, and the accompanying upstream histone methylation and HP1 localization, is largely signalled by previous RIP inactivation, obviating the requirement for an RNAi-directed mechanism of heterochromatin formation in this species.

Concluding remarks

The survival strategy used by a coenocytic organism like Neurospora to counteract selfish nucleic acids was to develop multiple systems of TGS and PTGS that were able to check genomic changes and alterations of gene expression. The logical consequence of this strict genomic surveillance was the almost complete absence of duplicated genes, presumably reducing the possibility for the evolution of new genes (Galagan & Selker, 2004). Perhaps this is the price Neurospora has paid for having a system of genome surveillance so efficient that there are apparently no active transposons in the genome and no known viruses of Neurospora.

It is interesting to note that, when compared with plants and animals, mutant screens in Neurospora have revealed relatively few components in each of its silencing mechanisms (Meister & Tuschl, 2004). This might simply reflect the fact that Neurospora has only a rudimentary, ‘economy class’ system of silencing. Alternatively, it might again be a result of Neurospora's restriction on gene duplication, meaning that single genes might have functions in multiple pathways, thereby making their mutant alleles more likely to be lethal.

In a similar vein, while findings from nearly all eukaryotes studied have revealed a PTGS machinery that uses endogenous, Dicer-processed, small RNA molecules called microRNAs (miRNAs) (Kim, 2005), such molecules seem to be absent in Neurospora. miRNAs are required in several pathways of developmental timing, apoptosis and cell proliferation and these molecules function in complexes similar to RISC to either block the translation (in animals) or induce the degradation (in plants) of the homologous mRNA (Kim, 2005). The apparent lack of miRNAs in Neurospora would be in accord with the observation that a double Dicer mutant, while totally blocked in transgene-induced gene silencing, had no consequences in normal development (Catalanotto , 2004). An explanation for this absence could be that the PTGS machinery was primarily evolved, in a common eukaryotic ancestor, as a defence mechanism against mobile sequences and afterwards, with the increased number of cell types of many organisms, it was converted to a regulatory role, except in those organisms with a relatively small number of cell types such as fungi.


View Abstract