Genome sequencing of evolved aspergilli populations reveals robust genomes, transversions in A. flavus, and sexual aberrancy in non-homologous end-joining mutants
Aspergillus spp. comprises a very diverse group of lower eukaryotes with a high relevance for industrial applications and clinical implications. These multinucleate species are often cultured for many generations in the laboratory, which can unknowingly propagate hidden genetic mutations. To assess the likelihood of such events, we studied the genome stability of aspergilli by using a combination of mutation accumulation (MA) lines and whole genome sequencing.
We sequenced the whole genomes of 30 asexual and 10 sexual MA lines of three Aspergillus species (A. flavus, A. fumigatus and A. nidulans) and estimated that each MA line accumulated mutations for over 4000 mitoses during asexual cycles. We estimated mutation rates of 4.2 × 10−11 (A. flavus), 1.1 × 10−11 (A. fumigatus) and 4.1 × 10−11 (A. nidulans) per site per mitosis, suggesting that the genomes are very robust. Unexpectedly, we found a very high rate of GC → TA transversions only in A. flavus. In parallel, 30 asexual lines of the non-homologous end-joining (NHEJ) mutants of the three species were also allowed to accumulate mutations for the same number of mitoses. Sequencing of these NHEJ MA lines gave an estimated mutation rate of 5.1 × 10−11 (A. flavus), 2.2 × 10−11 (A. fumigatus) and 4.5 × 10−11 (A. nidulans) per base per mitosis, which is slightly higher than in the wild-type strains and some ~ 5–6 times lower than in the yeasts. Additionally, in A. nidulans, we found a NHEJ-dependent interference of the sexual cycle that is independent of the accumulation of mutations.
We present for the first time direct counts of the mutation rate of filamentous fungal species and find that Aspergillus genomes are very robust. Deletion of the NHEJ machinery results in a slight increase in the mutation rate, but at a rate we suggest is still safe to use for biotechnology purposes. Unexpectedly, we found GC→TA transversions predominated only in the species A. flavus, which could be generated by the hepatocarcinogen secondary metabolite aflatoxin. Lastly, a strong effect of the NHEJ mutation in self-crossing was observed and an increase in the mutations of the asexual lines was quantified.
KeywordsAspergillus Aflatoxin Mutation accumulating lines Genome stability Non-homologous end-joining ku70
The taxon of filamentous fungi comprises organisms of great importance for human daily life, as friends or foes. As friends, we take advantage of the capabilities of filamentous fungi to produce enzymes and compounds of interest to humans, such as pharmaceuticals. Some of these enzymes have proved to be highly useful in the production of bio-energy [1, 2] and fungal natural products impactful in many applications such as the antifungal echinocandins . Within the filamentous fungi group, aspergilli are widely used for biotechnological applications in industry. In particular, they have become model organisms to study different aspects of the bio-energy production, such as the production of cellulolytic enzymes useful for bioethanol production [4, 5, 6], the carbon-regulation of these enzymes [4, 7] and natural product synthesis as drug sources [3, 8, 9]. These studies rely on two important aspects of the biology of aspergilli: their fantastic features for fermentation purposes and the genetic amenability of the genus. The availibilty of many genome sequences of Aspergillus species makes this group of fungi very useful for comparative genomics [10, 11, 12]. As foes, some Aspergillus species are pathogens of humans, animals and/or plants, or appear as dangerous contaminants of crops before or after the harvest. They are capable of producing an ample diversity of secondary metabolites, such as the carcinogen aflatoxin [13, 14, 15, 16].
The stability of genomes highly depends on the lifestyle and the complexity of each organism. Whereas viruses have very dynamic genomes, eukaryotic genomes are highly stable and eukaryotes do not cope well with insults to the genome. Yet cells are continuously exposed to DNA damaging agents. Therefore, the stability of the genome relies on a complex molecular machinery that repairs the damage. One such example of DNA repair machinery is the non-homologous end-joining (NHEJ), which joins two DNA molecules that had suffered double-strand breaks (DSBs) [22, 23, 24]. Thus, a doubt arose with the discovery of the NHEJ genes in filamentous fungi and the use of the NHEJ deletions as genetic tools . The underlying fundamental question behind the doubt is what is the role of the NHEJ system in the stability of an eukaryotic genome. Although this is a fundamental question in biology, the knowledge has direct consequences on biotechnology with the use of NHEJ mutants to generate new fungal strains with desired genetic modifications. The fungal homologues of the KU70 and KU80 required for NHEJ were first identified in Neurospora crassa . Following the spectacular results of N. crassa ∆ku70 strains as genetic tools, NHEJ mutants have been constructed in many other fungal species (for some examples, see [26, 27, 28, 29, 30, 31]). N. crassa NHEJ mutants are more sensitive to bleomycin, a compound producing double-strand DNA breaks , although an A. nidulans NHEJ mutant did not appear to show increase sensitivity to this drug . However, there is an additional body of data generated in fungi that suggests that the KU70/KU80 heterodimer plays additional roles. For example, N. crassa and Aspergillus fumigatus mutants, but not A. nidulans, are more sensitive to agents producing point mutations [25, 33, 34], which were presumed to get repaired using different mechanisms. They are telomere-associated proteins contributing to the formation of heterochromatin in the telomeres . In A. nidulans, NkuAKU70 appears to regulate telomere position effects . Both ku70 and ku80 homologues were found to be overexpressed during asexual reproduction (conidiation) in a histone acetyltransferase mutant, suggesting that they are involved in the maintenance of genome stability in conidia, DNA repair or the responses to DNA stress . In haploid cells, as Aspergillus species normally propagate, DNA repair by homologous recombination (HR) of double-strand breaks is only possible in S and G2 phases of the cell cycle, when a sister chromatid is available. Therefore, the repair of double-strand breaks is expected to be more error-prone in a haploid because of the higher likelihood of using NHEJ repair.
As these additional roles of the KU70/KU80 heterodimer on fungal biology and potentially DNA stability have emerged, it is prudent for the academic community and industry to determine if the NHEJ mutants display concerning genome instability properties. A stable genome is especially important for biotechnology applications, where much time and money is spent to optimize processes and a small change in yield can have large financial consequences. In addition to insights on Aspergillus genome evolution and stability, we expect our results will be widely applicable to other fungi where the use of NHEJ mutants is expanding [28, 38, 39]. Therefore, and considering that strains lacking KU70/KU80 are commonplace in biotechnology, we decided to test the global consequences of such mutants for the stability of the genome. We studied the genome stability of aspergilli by using a combination of MA lines and whole genome sequencing (WGS). We found that Aspergillus genomes are very robust and the deletion of the NHEJ machinery results in a slight increase in the mutation rate, but at a rate, which we postulate, is still safe to use for biotechnology purposes.
MA lines, sequencing and identification of the mutations
Three different aspergilli species have been selected based on their different characteristics and relevance in the aspergilli community: A. nidulans, A. flavus and A. fumigatus. All species were already sequenced before this work started, and they have genome sizes ranging between 29.4 and 36.8 Mb [11, 40]. Ten independent MA lines for each strain (wild type and ∆ku70 of each species) were set up by picking ten individual isolated colonies as described in the “Materials and methods” section, and the strains were allowed to accumulate mutations by growing in complete media for 60 passages (Fig. 1b). Based on the nuclear division time in A. nidulans and previous calculations [19, 41] we estimated that nuclei underwent 4032 mitotic divisions. This strategy allowed the 10 populations of each strain to accumulate mutations independently from each other along the passages (Fig. 1b). Samples were taken at time 0 (controls) and at the end of the MA experiment for WGS.
Globally in the MA lines, we identified a total of 122 single nucleotide variants (SNVs) and 17 insertions/deletions (indels) in A. flavus, 27 SNVs and 10 indels in A. fumigatus and 75 SNVs and 29 indels in A. nidulans. Out of those mutations, the wild-type MA lines accumulated 58 SNVs and 5 indel in A. flavus, 10 SNVs and 2 indels in A. fumigatus and 32 SNVs and 18 indels in A. nidulans. Aspergillus species like other filamentous fungi are multicellular and multinucleated organisms, i.e. the hyphal filament consists of connected cell compartments, each containing several nuclei (Fig. 1a). Consequently, for the estimation of the mutation rate, we need to consider nuclear divisions (mitoses) rather than cell divisions as previously employed for unicellular eukaryotes [42, 43, 44, 45, 46, 47]. As mentioned above, the estimation of the number of mitoses is complicated in the case of filamentous fungi. Our calculations of the number of mitoses are based on previous cell cycle studies performed by Bergen and Morris on liquid media , which did not differ substantially from studies by Kaminskyj and Hamer  on solid media. The number of nuclear divisions occurring during conidiophore development was not taken into account because the division times and nuclear dynamics in the stalk and vesicle cells are not well characterized. This is partially compensated by the fact that the samples for the cultures passages were taken from the conidiation edge of the colony, which is far behind from the real growing colony front. This strategy to estimate the number of mitoses during the growth passages has been previously carried out by Schoustra et al. . This approach gave a total mutation rate of 4.2 × 10−11 ± 5.6 × 10−12 (A. flavus), 1.1 × 10−11 ± 4.1 × 10−12 (A. fumigatus) and 4.1 × 10−11 ± 6.4 × 10−12 (A. nidulans) per base per mitosis and an average mutation rate for all three species of 3.1 × 10−11 ± 5.4 × 10−12 per base per mitosis revealing that indeed Aspergillus genomes are very robust in comparison to yeasts and higher eukaryotes. The distribution of mutations across the MA lines was consistent with a uniform distribution (Kolmogorov-Smirnov test (K-S test); P = 0.93, 0.31 and 0.91 for A. flavus, A. fumigatus and A. nidulans, respectively).
The ∆ku70 deletion mutants of the three species accumulated a slightly higher number of mutations: 64 SNVs and 12 indels in A. flavus, 17 SNVs and 8 indels in A. fumigatus and 43 SNVs and 11 indels in A. nidulans (Fig. 2b), which gives an estimate of a total mutation rate of 5.1 × 10−11 ± 4.7 × 10−12, 2.2 × 10−11 ± 3.8 × 10−12 and 4.5 × 10−11 ± 8.1 × 10−12 per base per mitosis for A. flavus, A. fumigatus and A. nidulans, respectively, and an average mutation rate for all three species of 3.9 × 10−11 ± 5.5 × 10−12 per base per mitosis. The distribution of mutations across the MA lines of the ∆ku70 mutants was also consistent with a uniform distribution (K-S test; P = 1.00, 0.77 and 0.97 for A. flavus, A. fumigatus and A. nidulans, respectively). The differences between the wild-type and the ∆ku70 mutant strains for each species are not statistically significant (two-way ANOVA with Tukey’s post hoc test, P = 0.675 for each pair of strains; two-tailed t test, P = 0.33, P = 0.14 and P = 0.67 for A. flavus, A. fumigatus and A. nidulans, respectively). However, if we did not consider the different species, and we take into account only the genetic background combining the data from the three wild-type strains and the data from the three ∆ku70 strains, then the differences between the wild-type and the mutant backgrounds are statistically significant (two-way ANOVA with Tukey’s post hoc test, P = 0.025), suggesting that the NHEJ has a role in the genome stability of aspergilli.
Distribution of mutations
The concept of an individual in filamentous fungi is a matter of debate and the definition of a physical individual has been even considered a “hopeless task” . Therefore, we considered our MA lines as populations of nuclei. Thanks to the high coverage of the sequencing, it was possible to detect mutations present in as low as 10% of the nuclei. A boxplot representation of the frequency of all the mutations in the six strains showed that there were a larger proportion of mutations that were not fixed in the ∆ku70 mutants than in the wild-type strain, suggesting that the mutations in the ∆ku70 mutants keep appearing at a slightly faster rate than in the wild-type strain (Additional file 1: Figure S1).
The distribution of single nucleotide mutations in the eight chromosomes appeared to be random and not statistically different from the expected number according to their sizes (K-S test; P = 0.92 and 0.92 for A. fumigatus wild-type and ∆ku70 mutant, and 0.92 and 1.00 for A. nidulans wild-type and ∆ku70 mutant, respectively) (Additional file 2: Figure S2). In the particular case of A. flavus, this analysis was complicated because the 138 contigs are not assigned to specific chromosomes. We tried to map these contigs to the chromosomes of the closely related species A. oryzae with little success. Therefore, we performed the analysis asking the question whether the mutations were uniformly distributed in the 138 contigs using the K-S test. In this case, we found that P < 0.001, meaning that the mutations were not uniformly distributed. Further analysis revealed that all the mutations were located in 20 contigs, which account for 99.25% of the total length of the genome, while the other 118 contigs only provide the remaining 0.75% of the genome, and consequently represent a minor portion of the genome. We repeated the test using only the 20 big contigs accounting for 99.25% of the genome and containing the mutations. In this case, P = 0.68 suggests that the distribution of mutations is random in these 20 contigs and in the genome of A. flavus.
Single nucleotide mutations and spectrum
Out of the 280 total number of mutations, 56 indel mutations were identified in the six strains of Aspergillus, 19 were deletions and 37 were insertions, which on average accounts for 1.9 times more insertions than deletions (Fig. 4b). The distribution varied between species, being A. nidulans the species accumulating more indel mutations (7 deletions and 11 insertions in the wild-type strain). This is 1.8 times lower than the number of base-substitution mutations. The low proportion of indel mutations found in aspergilli (ranging from 0.08 to 0.36) is consistent with other organisms. In general, indels were observed to be increased in ∆ku70 mutants compared to the corresponding wild-type strain. However, insertions in the ∆ku70 were only significantly increased in A. flavus (one-tailed t test; P = 0.038) and deletions in A. fumigatus (one-tailed t test; P = 0.033). In A. nidulans, the number of indel mutations was even higher in the wild type (18 indel mutations) than in the ∆ku70 mutant (11 indel mutations), with a particular but not significant increase in the number of deletions (7 in the wild type vs 2 in the ∆ku70 mutant strain) (one-tailed t test; P = 0.09). Based on these data, we estimated an indel mutation rate of 3.4 × 10−12 ± 1.5 × 10−13 (A. flavus), 1.8 × 10−12 ± 1.2 × 10−13 (A. fumigatus) and 1.5 × 10−11 ± 4.4 × 10−13 (A. nidulans) per base per mitosis for the wild-type strains and 8.1 × 10−12 ± 2.2 × 10−13 (A. flavus), 7.1 × 10−12 ± 2.2 × 10−13 (A. fumigatus) and 9.1 × 10−12 ± 1.5 × 10−13 (A. nidulans) per base per mitosis for the ∆ku70 mutant strains.
Cessation of cleistothecia development in the sexual MA lines of the A. nidulans NHEJ mutants
Filamentous fungal species are excellent cell factories for the production of biotech goods such as organic acids, recombinant enzymes and pharmaceuticals. This makes them worth billions of euros every year. This is the first report that estimates the mutation rate based on WGS in filamentous fungi. We allowed 10 independent MA lines of each strain to accumulate mutations for over ~ 4000 mitoses with 60 growth passages. Under these conditions, we assumed that the nuclei with deleterious mutations were lost during the experiment, and only nuclei with mutations that can grow in complete media will progress. For this reason, we chose complete media rather than minimal media, which would have resulted in a lower accumulation of mutations due to the loss of many auxotrophic mutants. Globally, our experiment considered a total of ~ 242,000 mitoses and we sequenced the whole genomes with a 100× depth on average. After the analysis of ~ 1.9 × 1011-bp sequences, we have estimated that the total mutation rate is 3.1 × 10−11 ± 5.4 × 10−12 per base per mitosis on average for the three Aspergillus strains with small differences between the different species (4.2 × 10−11 ± 5.6 × 10−12 per base per mitosis in A. flavus, 1.1 × 10−11 ± 4.1 × 10−12 per base per mitosis in A. fumigatus and 4.1 × 10−11 ± 6.4 × 10−12 per base per mitosis in A. nidulans). Only the mutation rate of A. fumigatus seems to be statistically different from the other two Aspergillus species (two-way ANOVA, Tukey’s post hoc test P = 2.9 × 10−5 and 2.5 10−3 with A. flavus and A. nidulans, respectively). We considered two possible scenarios for these differences. First, the genome of A. fumigatus is more robust. However, it is hard to argue that this is due to its pathogenic capabilities, as it is an opportunistic pathogen and A. flavus is also a pathogenic species. The second and more likely possibility is that many calls in regions of poor/spurious read alignments in A. fumigatus were filtered out, resulting in a lower number. This second possibility is supported by the coverage depth of 97% in A. nidulans and only 90% in A. fumigatus.
The rate of non-homologous integrations of recombinant constructs during transformation is usually very high in filamentous fungi, which can make the construction of knock-out mutants, for example, a very laborious work. For this reason, most laboratories across the world have established the use of NHEJ mutants to increase the rate of the homologous integration [25, 28, 32, 53, 54, 55, 56, 57, 58, 59, 60, 61]. However, the genomic stability of these strains has not been tested up to date and raised concerns of the appropriateness of using these mutants in industry or to assess virulence in pathogenic fungi. Here, we found that the ∆ku70 deletion mutants accumulated an average of 1.2 times higher number of mutations: 76, 25 and 54 in all the MA lines. We estimated a total mutation rate of 5.1 × 10−11 ± 4.7 × 10−12 (A. flavus), 2.2 × 10−11 ± 3.8 × 10−12 (A. fumigatus) and 4.5 × 10−11 ± 8.1 × 10−12 (A. nidulans) per base per mitosis. The average mutation rate for all three species is 3.9 × 10−11 ± 5.5 × 10−12 per base per mitosis. This number, although statistically significant, did not increase drastically for mutations accumulated in ∆ku70 strains compared to the wild-type strains. Bruggeman et al. estimated a base-substitution mutation rate of 2.26 × 10−10 per base per mitosis for asexual passages of A. nidulans based solely on fitness assays . Our estimate using WGS is 10× lower (2.5 × 10−11 ± 4.2 × 10−12 per base per mitosis). When we compared the Aspergillus average base-substitution mutation rate with unicellular fungi (Fig. 3), we found that it is one order of magnitude lower than the reported mutation rates for yeasts: 1.70 ± 0.13 × 10−10 or 2.00 ± 0.10 × 10−10 per base per generation in Schizosaccharomyces pombe [42, 43], 1.67 × 10−10 per base per generation in Saccharomyces cerevisiae  and 1.90 × 10−10 per base per generation in the basidiomycete yeast Rhodotorula toruloides . Even more important, the base-substitution mutation rate in the ∆ku70 mutants is still 5.3–6.4 times lower than in the yeasts. S. cerevisiae is frequently employed as cell factory and also is very important in fermentation processes to make bread, beer or wine. Another cell factory bacterium Escherichia coli has a base-substitution mutation rate of 2.6 × 10−10 per base per generation based on reporter constructs , the same range as yeasts but an order of magnitude higher than the wild-type Aspergillus and 8.3 times higher than the ∆ku70 mutants. Taking altogether, this suggests that Aspergillus genomes are the most robust genomes of the cell factories analysed so far and that the ∆ku70 mutants have genomes that are robust enough for use directly in industrial fermentation processes.
The fact that the number of mutations is slightly higher in the ∆ku70 than in the wild-type strains raises some questions about alternative roles of the KU70/KU80 dimer in genome stability. After a DNA double-strand break, two possible repair mechanisms can act: the NHEJ and the HR [63, 64]. Many filamentous fungi are haploids during most of their life cycles, although diploids can be constructed and maintained in the lab. Therefore, in the absence of a sister chromatid and a homologous chromosome, the NHEJ is responsible for the repair of DSBs during the G1 and early S phases [63, 64]. However, in the absence of KU, cells are still capable of DNA resection in G1 at DSBs prior to repair by homologous recombination, which activates the checkpoints. Bergen and Morris  estimated that the G1 phase lasts 15 out of 75 min of the complete nuclear cycle of A. nidulans. Therefore, Aspergillus species spend most of their time in S or G2 phases (55 min) where sister chromatids are present and HR is possible. Indeed, the number of base-substitution mutations was on average 24% higher in the ∆ku70 than in the wild-type strain (ranging from 10 to 70%). This is in agreement with the time that Aspergillus spends in G1, which is approximately 20% of the total nuclear cycle. Furthermore, it has been shown in yeasts that the replication fork is stalled at fragile sites containing CCG and CAG trinucleotides that lead to DSBs . However, when we analysed the trinucleotide context of the mutations found in our MA lines, we found that only in the case of A. flavus mutations in the CCG trinucleotides are statistically significant (χ2; P = 0.001; df = 1). Actually in A. flavus MA lines, mutations in the CpG context are particularly enriched (χ2; P = 0.00; df = 1). In most of the reports describing the spontaneous mutation rates in eukaryotic organisms, mutations in this context are associated mainly to the oxidation of guanine to give 8-oxo-guanidine, which produces a GC→AT transition. These mutations are significantly increased in A. fumigatus and A. nidulans accounting for 32% and 39% of all the base-substitution mutations (χ2; P = 0.028 and 0.000; df = 1, respectively), but not in A. flavus (χ2; P = 0.517; df = 5). This can be easily explained because the strains were cultivated on the surface of solid media and continuously exposed to an air interphase for 8–9 months. Methylation of cytosines results in the spontaneous deamination of cytosines provoking a GC→AT transition. However, the oxidation of guanine is probably more parsimonious taking into account that 5-methylated cytosines (5-mC) have not been detected in neither A. nidulans nor A. flavus yet [52, 66, 67, 68], despite the existence of DNA methyltransferases in the aspergilli genomes.
The type of mutation that most called our attention was the high number of GC→TA transversions found in A. flavus MA lines, which accounts for nearly half of all the base-substitution mutations (χ2; P = 0.000; df = 1). As mentioned earlier, A. flavus produces the carcinogenic secondary metabolite aflatoxin, the most potent natural environmental mutagen known. Aflatoxin is infamous for causing human hepatocarcinomas by inducing mutations in the p53 oncogene [50, 51]. Specifically, an epoxide derivative of aflatoxin intercalates into DNA and the epoxide moiety reacts with the guanine residue to produce GC→TA transversions . A. fumigatus, which is unable to produce aflatoxin, accumulated only 11% of GC→TA transversions. On the other hand, A. nidulans produces an intermediate in the aflatoxin pathway, sterigmatocystin, which is also a carcinogen and shows demonstrable mutagenic properties in vitro . Yet evidence does not support a mutational role for sterigmatocystin in nature, likely due to the non-pathogenic nature of A. nidulans as well as the low levels of sterigmatocystin synthesized by this species in comparison to aflatoxin output in many A. flavus strains (including the strain used in this study which produces aflatoxin in the medium used to grow the strains, ). Indeed, in A. nidulans, the number of GC→TA transversions accumulated is not significantly increased (χ2; P = 0.500; df = 1). Considering the A. flavus data, this raises interesting questions about the consequences and advantages of producing mutagenic secondary metabolites for endogenous producers.
During this study, it was shown that the absence of KU70 lead to an early cessation of cleistothecia development after repetitive selfing crosses in A. nidulans. This observation was not reported in a previous article that employed 40 generations of MA lines . However, their setup was slightly different from our setup and this observation could have been overlooked. They observed a reduction in fitness after the 40 generations, which they argued was due to the accumulation of mutations . Another study in which Xu employed the heterothallic mushroom Agaricus bisporus  reported an inbreeding depression, as observed with the percentage of heterokaryons forming primordia, density of fruiting bodies and other parameters. Because A. bisporus is heterothallic, the inbreeding coefficient was 0.5 and yet they observed a significant decreased in the percentage of heterokaryons forming primordia . In our setup, because A. nidulans is homothallic, the inbreeding coefficient was 1.0. We observed that 30% of the wild-type MA lines ceased to form cleistothecia suggesting an inbreeding depression. In case of the ∆ku70 MA lines, all ceased to form cleistothecia before the 10 passages. The reason is unclear. Both strains are veA1, which made us to consider that this allele could not be responsible for the phenomenon. However, we cannot exclude possible genetic interactions between veA1 and ∆ku70 leading to the cessation of cleistothecia formation. Theoretically, chromosomal re-arrangements are more likely to be the cause than telomere shortening, because it appears to be specific to meiosis, and therefore the more drastic depression could be due to increased genomic instability in the absence of KU heterodimer in the ∆ku70 strain during meiosis. However, this does not explain why cleistothecia development was halted. The phenomenon resembles epigenetic regulation. Several studies have reported aberrancies in sexual development in A. nidulans (or A. flavus) chromatin modifying mutants including DmtA (a cytosine methyltransferase, ), SntB (a histone reader, ), Set3 (a chromatin-associated PHD domain protein, ) and HosA (a histone deacetylase, ).
Aspergilli are widely used in the biotechnology industry and are emerging as the best model systems for fungal genetics and comparative genomics with species that are pathogens of plants and/or animals. The increasing capacity of DNA sequencing has allowed us to evaluate the consequences of evolution in a control experiment in the laboratory for the first time in a multicellular eukaryote under standard non-selective conditions. The widespread use of the NHEJ strains has impelled this work. We have found that the genomes of the Aspergillus ∆ku70 strains are very robust, more than other cell factories employed in the biotech industry, and therefore they can be considered safe enough for production purposes. A stable genome is especially important for biotechnology applications, where much time and money is spent to optimize processes and a small change in yield can have large financial consequences. Therefore, as shown in this work, Aspergillus and their corresponding NHEJ mutants also offer robustness in addition to the other previously known advantages for industrial applications. The comparative studies on Aspergillus genome evolution and stability observed in this work can be extrapolated to other Aspergilli, and maybe also applicable to other fungi, where the use of NHEJ mutants is growing.
Materials and methods
Strains and media used in this study
There are several genes involved in the NHEJ DNA repair mechanism; for this work, only mutants in the homologues of human KU70 were employed to allow for comparisons between species. All the strains used in this study are listed in Additional file 9: Table S5. Strains were grown in Aspergillus complete media .
Establishment of MA lines and growth passages
The MA experiment was performed as previously described [19, 76] with small modifications as depicted in Fig. 1b. Appropriate dilutions of one single stock of spores of each strain were plated on 10 plates per strain containing complete solid media to give single isolated colonies. Ten independent colonies of each strain were point inoculated in the centre of individual plates (one plate per colony). Then, the 10 independent MA lines of each strain were established. Plates were incubated for 3 days at 37 °C (30 °C for A. flavus), and the spores were harvested and were kept frozen as controls.
Plugs were taken from the conidiating edge of each single colony, i.e. from the farthest point from the colony centre that was conidiating, just before there was only vegetative mycelium. Plugs were placed in a microcentrifuge tube with 1000 μl of tween 0.05% (0.02% for A. fumigatus) to resuspend the spores. For the next growth passage (generation), 2 μl of the suspension was inoculated in the centre of the plate and the plates were incubated at 37 °C (30 °C for A. flavus). Two passages were made per week, which accounted for bottlenecks every ~ 58 and ~ 77 nuclear mitosis (3 and 4 days, respectively).
In the case of A. nidulans, sexual passages were also performed. The procedure was the same, with the difference that one cleistothecium from each plate was picked, cleaned of spores and Hülle cells, and ascospores were resuspended in 200 μl of tween 0.05% and 0.02% respectively.
After all passaging was completed, gDNA of populations at passage 0 and after 60 passages was isolated using standard techniques. For standard fragment Illumina libraries, 500 ng–1 μg of genomic DNA was sheared using the Covaris E210 (Covaris) and sized selected for 475 bp using Agencourt Ampure Beads (Beckman Coulter). The DNA fragments were treated with end repair, A-tailing and adapter ligation using the TruSeq DNA Sample Prep Kit (Illumina) and purified using Agencourt Ampure Beads (Beckman Coulter). These were sequenced on HiSeq to produce 2 × 150 bp reads. These reads were aligned to the reference genome using BWA , and on average, 100× coverage was sequenced for each population. SNPs and small indels were called using bbtools callVariants.sh (https://sourceforge.net/projects/bbmap/) and allele frequencies estimated. The SRA accession numbers for the genome sequences are listed in Additional file 10: Table S6.
Filtering of mutations
To remove false-positive calls and identify the point mutations that appeared during the growth passages, we removed all calls that had less than 3 supporting reads. We found that in some areas, the sequenced strain was highly divergent from the reference genome deposited in the database. In these cases, only a small percentage of the reads from these regions were able to align, and none of the calls in these regions were reliable. To filter out these unreliable calls, they were also removed if they occurred in regions with less than 10% of the average depth to avoid false positives in highly diverged regions. Finally, since point mutations will be unique to only a single strain, all calls that were shared by multiple strains were removed.
Putative single nucleotide variants and small indels were identified using callVariants (from the bbmap package v37.06) with default parameters and adding “rarity = 0.05 minallelefraction = 0.05”. A series of filters were then applied, only calls unique to a single strain were retained, calls with less than 10% of the average depth were discarded after noting an excess of these calls in regions of poor/spurious read alignment especially in A. fumigatus and calls with less than 3 reads supporting the alternate allele were also discarded. The remaining calls were then manually reviewed using the Integrative Genomics Viewer (IGV)  genome browser to remove calls that were obviously bad, typically due to sequencing errors. Allele frequency counts were also manually adjusted during the review due to some issues with allele frequency estimates in that version of callVariants. No variants of higher complexity than SNVs and small indels were attempted.
The statistical tests described in the text for the analysis of the data were performed using Excel, PAST or STATGRAPHICS software.
D.C. would like to thank Pablo Huertas for comments on the manuscript, Berl Oakley for the helpful discussions on nuclear division timing and Luka Pavageau for the help with the DNA isolation.
IAE, CS, JWB and ATM performed the growth passages. IAE, CS, JWB, ATM, SC and DC isolated the DNA from the MA lines. HN, MA, AL and KB sequenced the MA lines. WS, JM and IVG processed and analysed the DNA sequences. GG and DC performed the statistical analysis. IVG, NPK, GHB and DC wrote and edited the manuscript with the help of CS, JWB, GG and WS. All authors read and approved the final manuscript.
The work conducted by the U.S. Department of Energy Joint Genome Institute, a DOE Office of Science User Facility, was supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. The work was supported by the Spanish Ministry of Economy and Competitiveness (MINECO, grant number BIO2015-67148-R) to DC, NIH R01AI065728 to N.P.K., NIH SBIR R44AI118086 to J.W.B. and by the Deutsche Forschungsgemeinschaft (DFG: BR1502/11-2 and SFB860) to GHB.
Ethics approval and consent to participate
Consent for publication
The authors declare that they have no competing interests.
- 2.Berka RM, Grigoriev IV, Otillar R, Salamov A, Grimwood J, Reid I, Ishmael N, John T, Darmond C, Moisan MC, et al. Comparative genomic analysis of the thermophilic biomass-degrading fungi Myceliophthora thermophila and Thielavia terrestris. Nat Biotechnol. 2011;29(10):922–7.PubMedCrossRefPubMedCentralGoogle Scholar
- 10.Andersen MR, Salazar MP, Schaap PJ, van de Vondervoort PJ, Culley D, Thykaer J, Frisvad JC, Nielsen KF, Albang R, Albermann K, et al. Comparative genomics of citric-acid-producing Aspergillus niger ATCC 1015 versus enzyme-producing CBS 513.88. Genome Res. 2011;21(6):885–97.PubMedPubMedCentralCrossRefGoogle Scholar
- 33.da Silva Ferreira ME, Kress MR, Savoldi M, Goldman MH, Hartl A, Heinekamp T, Brakhage AA, Goldman GH. The akuB(KU80) mutant deficient for nonhomologous end joining is a powerful tool for analyzing pathogenicity in Aspergillus fumigatus. Eukaryot Cell. 2006;5(1):207–11.PubMedPubMedCentralCrossRefGoogle Scholar
- 36.Palmer JM, Mallaredy S, Perry DW, Sanchez JF, Theisen JM, Szewczyk E, Oakley BR, Wang CC, Keller NP, Mirabito PM. Telomere position effect is regulated by heterochromatin-associated proteins and NkuA in Aspergillus nidulans. Microbiology. 2010;156(Pt 12):3522–31.PubMedPubMedCentralCrossRefGoogle Scholar
- 38.Catalano V, Vergara M, Hauzenberger JR, Seiboth B, Sarrocco S, Vannacci G, Kubicek CP, Seidl-Seiboth V. Use of a non-homologous end-joining-deficient strain (delta-ku70) of the biocontrol fungus Trichoderma virens to investigate the function of the laccase gene lcc1 in sclerotia degradation. Curr Genet. 2011;57(1):13–23.PubMedCrossRefGoogle Scholar
- 40.Nierman WC, Yu J, Fedorova-Abrams ND, Losada L, Cleveland TE, Bhatnagar D, Bennett JW, Dean R, Payne GA. Genome sequence of Aspergillus flavus NRRL 3357, a strain that causes aflatoxin contamination of food and feed. Genome announcements. 2015;3(2):e00168–15.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.