Identification of a mitotic recombination hotspot on chromosome III of the asexual fungus Aspergillus niger and its possible correlation with [corrected] elevated basal transcription.

Genetic recombination is an important tool in strain breeding in many organisms. We studied the possibilities of mitotic recombination in strain breeding of the asexual fungus Aspergillus niger. By identifying genes that complemented mapped auxotrophic mutations, the physical map was compared to the genetic map of chromosome III using the genome sequence. In a program to construct a chromosome III-specific marker strain by selecting mitotic crossing-over in diploids, a mitotic recombination hotspot was identified. Analysis of the mitotic recombination hotspot revealed some physical features, elevated basal transcription and a possible correlation with purine stretches.

Introduction

Aspergillus niger fermentation products have many uses in food applications, for example, as a producer of organic acids and enzymes (Archer 2000). Production processes have been thoroughly optimized by improvements in process technology (Naidu and Panda 1998; Schügerl 2000) and by strain breeding (Punt et al. 2002). Classical improved strains have been obtained using random mutagenesis, recombination and genetic modification based on transformation of protoplasts.

Because of its haploid nature, A. niger production strains can be improved by subsequent rounds of mutation and selection. Genetic recombination is more troublesome because A. niger lacks a sexual cycle. An alternative is recombination using the parasexual cycle, which first was described for Aspergillus nidulans (Pontecorvo et al. 1953). The parasexual cycle starts with a heterokaryon, which can be obtained via anastomosis or protoplast fusion. Using double markers, a heterokaryon can be selected and maintained on synthetic medium. A heterozygous diploid, formed by the fusion of the two different nuclei, can be selected as a prototroph. Through chemically induced loss of chromosomes, this diploid is reduced to an unstable aneuploid, finally giving rise to stable haploids. Essentially in this way complete chromosomes are exchanged between the starting strains. Markers with less than 25% recombination are believed to be located on the same chromosome. Using the parasexual cycle, 89 genetic markers were assigned to a linkage group and several marker strains were constructed (Bos et al. 1993).

Although in A. niger mitotic crossing-over of homologous chromosomes occurs at low frequencies, it is the only method used to determine the gene order on a linkage group. In a heterozygote diploid, all markers distal to a cross-over become homozygous. Debets et al. (1993) used several genetic markers to select such partially homozygous diploids, thereby constructing a genetic map of A. niger which shows the gene order of 60 markers.

For the improvement of strain properties, such as enzyme production, the construction of a diploid may result in improved isolates (Khattab and Bazaraa 2005; Loera and Cordova 2003; Montiel-Gonzalez et al. 2002). By completing the parasexual cross, also recombinants with higher enzyme production than the parental strains have been isolated (Ball et al. 1978; Das 1980; Das and Ghosh 1989). This strategy was also successfully used for genetically modified production strains (Bodie et al. 1994). Furthermore, the parasexual cycle can be used for recombination of less compatible strains via protoplast fusion. Heterokaryons that would not be formed otherwise can be obtained this way, and haploidization followed by selection has resulted in the isolation of improved strains (Rubinder et al. 2000).

Usually recombinants are selected randomly, after which their performance is compared to their parents. By first assigning mutations or disruptions to a linkage group, recombination of unlinked properties can be achieved in an orchestrated manner (Swart et al. 1990; van den Hombergh et al. 1997).

Recombination of linked mutations is troublesome because of the low rate of mitotic crossing-over in A. niger. Meiotic crossing between mitotically linked markers is much more frequent and provides a means to establish reliable genetic maps. Possibly an increased rate of mitotic crossing-over could approach the advantages of meiotic recombination. In several organisms, mitotic crossing-over has been induced using heat shock, chemicals or irradiation (Becker et al. 2003; Davies et al. 1975; Hilton et al. 1985; Jansen 1964; Klinner et al. 1984; Sermonti and Morpurgo 1959; Whelan et al. 1980). In order to examine the use of increasing mitotic recombination with effectors in A. niger, a chromosome specific tester strain would be very useful.

With the A. niger genomic sequence available (Pel et al. 2007), the genetic map can be improved by cloning a number of markers, thus linking the genetic map to the physical map. Using such linked markers, we investigated mitotic crossing-over via selection of partially homozygous diploids. For this purpose we chose to study markers on chromosome III for several reasons; many markers are available on both arms, on the left arm the cloned markers pyrA and areA are present and in parasexual crosses the linked markers bioA1 and lysA7 were found to have recombination frequencies around 15% (Bos et al. 1988). Furthermore, the sequence information for chromosome III, located on supercontigs 12 and 15 adds up to 4.5 Mb (Pel et al. 2007), while the estimated size by CHEF gels is 4.1 Mb (Verdoes et al. 1994a). This indicates that the sequence information for this chromosome must be close to complete. Using the improved genetic map of chromosome III, we constructed specific marker strains with which we studied the occurrence and frequency of mitotic crossing-over.

Materials and methods

The A. niger strains used were derived from CBS 120.49 and were kindly provided by the Laboratory of Genetics, Wageningen University, The Netherlands. The strains used to construct diploids and new master strains are listed in Table 1. Strains used for the complementation of genetic markers are N495 (adeB2; nicA1) for adeB, N521 (fwnA1; adeE8) for adeE8, N658 (lysA7 argD6; nicA1) for argD and lysA, N660 (bioA1 argL2; nicA1; pabA1) for argL, N687 (fwnA1; proC3 lysA7 cysA2) for cysA and lysA, N705 (fwnA1; cnxD6; leuA1; nicA1) for cnxD, N733 (hisD4; adeG13) for adeG13, N884 (bioA1 lysE28) for bioA and lysE, N885 (bioA1 hisH8; nicB5) for hisH, N901 (olvA1; bioB2; argI15 metB10) for bioB. All strains carry the cspA1 mutation, conferring low conidiophores and dense sporulation.

Table 1

Strains used for the construction of diploids and two new marker strains

Diploid	Strain	Markers on III	Other markers
dp 1	N705	cnxD9	metB10
	N932	lysA7 argD6	fwnA25; nicA1
dp 2	N430	argD6
	N720	cnxD6	fwnA1; leuA1; nicA1
dp 3	N893	pyrA5 hisH8 bioA1	fwnA1; pheA1
	N933	argD6 cnxD6	nicA1
dp 4	N934	lysA7 cnxD9	fwnA25; nicA1
	N935	pyrA5 hisH8 bioA1	phenA1
New marker	N936	pyrA5 argD6 cnxD6	nicA1
Strains	N937	pyrA5 hisH8 bioA1 cnxD6	fwnA1; nicA1

PCR cloning and complementation of genetic markers

Open reading frames (ORFs) selected for complementation of markers were cloned by using a forward primer 200 bases 5′ of the start codon, and a reversed primer 200 bases 3′ of the stop codon. The PCR was performed on genomic DNA of A. niger CBS 513.88, using pfu-polymerase (Promega). After the last cycle of amplification, Amplitaq (Applied Biosystems) and dATP were added, incubated at 72°C and 1 μl of the PCR mix was used in the pGEMT easy cloning kit (Promega). Of twenty different clones plasmid DNA was isolated and pooled. Transformations were performed essentially as described by Kusters-van Someren et al. (1991). A mixture of 1 μg of the pooled PCR-clones and 5 μg of an ama1 containing plasmid (Verdoes et al. 1994b) was used for PEG-mediated transformation of 106 protoplasts. Transformants appearing were tested for their residual markers. The successful transformations were repeated using plasmid DNA from single clones.

Selection for mitotic crossing-over and segregation of diploids

Diploids were selected as described (Bos et al. 1989). Selection of cnxD homozygous diploids was done by growing single cell colonies of diploids on minimal medium (MM) with sodium nitrate as single nitrogen source and stab inoculation of these colonies on selective complete medium (CM) containing 150 mM KClO4 and proper supplements for all markers present in the diploids. These diploids were analyzed by haploidization on CM plates with 0.88 mg/l d/l-benomyl followed by replica plating of the progeny to establish their fenotype.

Microarray data

The microarray data of all ORFs examined in this study are averaged data of days 3 and 5 expression levels using GeneChips for a glucose fed-batch culture of CBS 513.88 (Pel et al. 2007). The expression levels were normalized to an average of 100.

Results

Linking the genetic map to the physical map

A first goal was to identify genes complementing known genetic auxotrophic marker genes on Chromosome III and to link the position of these genes on the physical map with the genetic map. Candidate genes for the corresponding biosynthetic routes were selected from supercontigs 12 and 15, comprising Chromosome III specific contigs (Pel et al. 2007) by using the FunCat annotation (Ruepp et al. 2004) (Table 2). A second criterion for the selection of genes was presence of their transcripts in a GenChip hybridization experiment using RNA isolated from a glucose fed-batch fermentation (Pel et al. 2007). With two of the markers, pyrA and areA, already identified (Goosen et al. 1987; Lenouvel et al. 2001), 14 remaining candidates were cloned as genomic PCR clones. Using these clones, complementation was found for the genetic markers bioA, bioB, lysA, hisH, argD, and cnxD. In all cases, the auxotrophic mutant was complemented by only one of the candidate genes (Table 1). With 8 cloned markers we compared the physical map of chromosome III with the genetic map described before (Fig. 1, Debets et al. 1993). This comparison adds extra information to the genetic map of chromosome III and is in accordance with the previously published genetic map (Debets et al. 1993), except for the position of the lysA marker.

Table 2

Complementation of genetic markers with selected genes and their position on supercontig 12 and 15

ORF name	Possible corresponding markers	Complementation found of mutation	No complementation of mutations	Position of the ORF (kb)
End of An12				2562
An12g08960	areA	areAd		2131
An12g07690	lysA or lysE		lysA7	1845
An12g03570	pyrA	pyrA6		907
An12g01280	argD or argL		argD6 L25	354
An12g01110	cysA		cysA2	319
An12g00320	adeB E or G		adeB2 E8 G13	76
Centromere				0
An15g00350	lysA or lysE	lysA7		94
An15g00610	hisH	hisH8		174
An15g01980	bioA or bioB	bioB2	bioA1	534
An15g01990	bioA or bioB		bioA1 B2	535
An15g02000	bioA or bioB	bioA1	bioB2	539
An15g02340	argD or argL		argD6 L25	612
An15g02360	argD or argL	argD6	argL25	618
An15g05170	cysA		cysA2	1220
An15g05720	cnxD	cnxD6		1334
End of An15				1877

Fig. 1

Linking the genetic and physical map of chromosome III. a shows the genetic map of linkage group III as published by Debets et al. (1993). Markers with unknown gene order are shown in parentheses and/or above each other. b shows the physical map drawn to scale, consisting of supercontigs 12 and 15 of the A. niger genome sequence (Pel et al. 2007). The location of identified genetic markers is indicated. The centromere (c) is located in a gap of unknown size between the positions 0 of supercontig 12 and 15. For convenience the size of the gap is depicted as 0 bases

Recombination of the genetic markers

The genetic map of Debets et al. (1993) gives limited information on the order of the markers because they were positioned individually, with the markers on the right arm at a crossing-over frequency of 4% from the centromere. This positioning of markers was established by selection for mitotic crossing-over in heterozygous diploids. The strains available at the beginning of our study contained only a few of the cloned genetic markers. Therefore, additional marker strains were constructed containing several auxotrophic mutations, which allowed the determination of genetic distances between these markers. Recombination of linked markers requires crossing-over, which is a rare event. Selection for crossing-over events is possible when using a recessive marker located more closely to the telomere than the other markers, as is the case for cnxD (Debets et al. 1990). The first recombination performed was the cnxD9 mutation with argD6 using diploid 1 (Table 1). From this heterozygous diploid, 15 independent homozygous cnxD9 diploids were selected on chlorate. The colonies appearing on the chlorate plates were carefully checked for a sector-like growth, indicating that the homozygous diploids originated from independent crossover events. These diploids were analysed by haploidization on benomyl, and 14 of them appeared to have had a mitotic crossover between lysA7 and argD6, one diploid had a crossover between lysA7 and the centromere, but no crossover between argD6 and cnxD9 was found (Fig. 2). In a second recombination experiment using diploid 2 (Table 2) with cnxD6 and argD6, we found one diploid with a crossover between argD and cnxD. The other 12 selected homozygous cnxD6 diploids resulted from crossovers between the centromere and argD6 (Fig. 2). Of the first crossed-over diploid N933 was isolated, which was subsequently used for the construction of diploid 3.

Fig. 2

Frequency of crossing-over between centromere and cnxD. In four diploids, homozygous cnxD6/9 diploids were selected on chlorate. The regions of crossover are delimited by the markers used in the four diploids and are positioned between the centromere (CEN) and cnxD. The frequency of the crossing-over in a region correlates with the surface of the box depicted above that region

Delimiting the recombination hotspot

Apparently, in A. niger there is a preference for mitotic crossover in the genomic region between lysA and argD compared to the region between argD and cnxD (Fig. 2). First mentioned region overlaps the region between lysA and bioA where the previously reported recombinations were found (Bos et al. 1988). To further examine the distribution of mitotic crossovers we analysed 26 homozygous cnxD6 diploids selected from diploid number 3 and 18 homozygous cnxD9 diploids selected from diploid number 4. In the first case, 25 crossovers occurred between the centromere and hisH8 and only 1 between argD6 and cnxD6. In the latter case, only 1 crossover occurred between the centromere and lysA7, 10 between lysA7 and hisH8, 2 between hisH8 and bioA1 and 5 between bioA1 and cnxD9 (Fig. 2). This clearly shows that there is a recombination hotspot in the 80 kb region between hisH and lysA.

Analysis of the recombination hotspot

Possibly the region between hisH and lysA contains genetic elements that increase crossing-over. We searched for microsattellite repeats (Majewski and Ott 2000) and polyA stretches, using the pattern search on the Biomax server (Frishman et al. 2003). Furthermore, the expression profiles of all ORFs on supercontig 15 were examined, using the microarray data published by Pel et al. (2007) (Fig. 3a, b) and the Genome Atlas of supercontig 15, kindly provided by Dr. D. Ussery using previously developed tools (Skovgaard et al. 2002; Wanchanthuek et al. 2006) (Fig. 3c, d). We found a short polyA stretch and three GA stretches and at the telomeric side of the hisH gene a CT repeat. Looking at expression across supercontig 15 on a scale of 0–3,000, there is no apparent abnormality near the mitotic recombination hotspot (Fig. 3a). There is a large region from 990 to 1,240 kb where gene expression seems to be low. With the expression plotted on a scale from 0 to 100, it becomes clear that within the mitotic recombination hotspot, particularly around hisH, the lowest transcribed ORFs have a higher transcription compared to the lowest transcribed ORFs outside this region (Fig. 3b). Another feature deviating along the complete lysA–hisH region is a more negative stacking energy, indicating a more stabilized region (Skovgaard et al. 2002) and a low AT percentage. The position preference, a measure for flexibility (Baldi et al. 1996) is only slightly higher near hisH. There are a few simple repeats in the lysA–hisH region and there are multiple purine or pyrimidine stretches of minimal 10 bases ((Y)10 vs. (R)10) around hisH, coinciding with an elevated GC skew (G-C)/(G + C) and AT (A-T)/(A + T) skew. A zoom-in of the lysA–hisH region is shown in Fig. 3d.

Fig. 3

Genome atlas and expression profile of supercontig 15. Expression of all genes on supercontig 15 are plotted in two scales a and b. The genome atlas of supercontig 15 according to c shows repeats, structural parameters and parameters directly related to the base composition. The region studied for mitotic recombination is located between the centromere (left of 0 kb) and cnxD (1334 kb). The recombination hotspot is located between lysA (94 kb) and hisH (174 kb)

Occurrence of purine stretches in supercontig 15

Purine stretches inside ORFs are known to block transcription via the formation of an intramolecular triple helix (Bidichandani et al. 1998; Grabczyk and Fishman 1995) while purine stretches were also reported to occur more often in fungal genomes than could be expected from random sequences (Ussery et al. 2002). As the genome atlas shows the occurrence of purine stretches to be different in the genomic region of interest, we had a closer look at them. Inside the coding DNA located in the region between lysA and hisH, the largest purine stretches are 17 and 16 bases, respectively. Outside the coding DNA there are 6 purine stretches of 32–65 bases long, and 2.7 kb at the centromeric side of hisH, an 80 bp purine stretch containing only two pyrimidines.

Discussion

To link the genetic map to the physical map, thirteen ORFs were selected to possibly complement eleven markers. Six of those genes indeed complemented a corresponding marker, which is indicative for the high quality of the genome annotation. Six of these markers and two already cloned markers areA and pyrA, were positioned on the genetic map by Debets et al. (1993; see also Fig. 1). The position found for lysA on the right arm is not in agreement with the expected position left of the centromere. The experimental evidence for the left arm position is mainly provided by selection for mitotic crossing-over on fluoro-orotic acid (FOA) in a diploid heterozygous for pyrA5 and lysA7 (F. Debets, personal communication). FOA is very mutagenic compared to chlorate and possibly induced additional crossovers that led to a false positioning of lysA. In our results with cnxD selected crossovers in diploids 1 and 3, we found further evidence for lysA to be located on the right arm (Fig. 2). The resulting physical map can be seen as an improved genetic map because more markers are placed in the correct physical order. The genetic distance of these markers, however, is still unclear.

A series of mitotic recombinations to combine multiple genetic markers into one single master strain proved much more difficult than expected because of the presence of a mitotic recombination hotspot in the region between lysA and hisH (Fig. 2). There are several examples of increased recombination in genomic regions, both during mitosis and meiosis. Meiotic recombination hotspots can be caused by increased double-strand breaks in defined genomic regions (Nishant and Rao 2006). In yeast, mitotic recombination in general occurs more often near the centromere (Minet et al. 1980) and a mitotic recombination hotspot is also proposed to act via double-strand breaks (Neitz and Carbon 1987), while HotI stimulates mitotic recombination in adjacent sequences (Stewart and Roeder 1989). In A.nidulans, there is a position dependent difference between meiotic recombination and mitotic recombination. Using sexual crosses, genetic distances near the centromeric region appear larger than the physical distances (Aleksenko et al. 2001; Espeso et al. 2005), while the opposite is found when using mitotic recombination (Käfer 1977). Although there seems to be a preference for mitotic crossing-over near the centromere, we conclude that the 80 kb region between lysA and hisH contains a true mitotic recombination hotspot, because most mitotic crossover events occur in that region and not in the 94 kb centromere-lysA region (Fig. 2).

The expression graphs and the genome atlas (Fig. 3) visualize some interesting features of the mitotic recombination hotspot. Over the whole lysA–hisH region there is a high basic transcription, indicating an open chromatin structure (Fig. 3b). A high AT percentage compared to the rest of the supercontig causes an increased stacking energy, which indicates a more stable helix (Ornstein et al. 1978). Around hisH, both GC-skew and AT-skew are higher, coinciding with purine stretches (Ussery et al. 2002). In some plants and fungi, an elevated GC-skew near the transcription start sites of highly transcribed genes was found (Fujimori et al. 2005). As mentioned before, purine stretches can have a drastic effect on transcription. The lysA–hisH region contains purine stretches outside coding DNA, whereas inside the ORFs no apparent purine stretches are found. Additionally, the region of low transcription from 990 to 1,240 kb contains very few purine stretches (Fig. 3) and is situated between the markers argD and cnxD, which show very little recombination. Ussery et al. (2002) reported that in microbial eukaryotic chromosomes, the purine tracks occur more often than would be expected from random sequences, and are localized mainly in non-coding regions. Also, a relation between mitotic recombination frequencies and transcription levels (Saxe et al. 2000; Aguilera 2002; Grewal and Elgin 2002) may be relevant in the lysA–hisH region.

All together, it is not clear whether the deviating physical properties of the DNA cause the mitotic recombination hotspot or influence the level of transcription, and if so, which of them is most important. However, the connection between noted properties could be through elevated basic transcription, causing or caused by an open chromatin structure that could lead to an increased mitotic recombination in the lysA–hisH region. Especially the presence of purine stretches outside coding DNA, and the absence of purine stretches inside coding DNA seem to correlate with basic transcription and mitotic recombination. However, we cannot exclude that the mitotic recombination hotspot could also be caused by a yet unidentified genetic element.