Introduction

Filamentous fungi have a long-standing reputation as industrial producers of enzymes and metabolites, such as Hypocrea jecorina (Anamorph Trichoderma reesei). H. jecorina plays an important role in the production of plant cell-wall degrading enzymes1. Initially, scientists selected production strains from natural isolates that fulfilled both microbiological and technical criteria for economical production. Subsequently, genetically modified strains with novel properties and higher enzyme yields were obtained through traditional random mutagenesis2. In recent years, however, recombinant technologies have contributed significantly to improved enzymes production and system design of genetically manipulated strains.

Gene replacement via disruption, deletion and overexpression of target genes is a significant genetic tool for gene functional analysis. However, it is limited by the number of available selectable marker genes. In recent years, a Cre/loxP recombination system has been widely used for marker rescue in various organisms. With the Cre/loxP system, genetic modifications were successfully introduced into different organisms by recycling a single selectable marker. Cre-mediated recombination depends on the expression of cre. Retransformation with a cre gene that remained in the fungal genome3,4 is associated with the risk of chromosomal rearrangement and Cre toxicity5. Another method is transient expression of Cre recombinase by Florea et al.6. The disadvantage of this method is low efficiency (marker was eliminated in 0.5–2% of the colonies) and two-step processes including fungal transformation and transient transfection. A third option is direct introduction of Cre protein, which is still a time-consuming two-step process7. A fourth option is PCR-based fusion of self-excision cassette, which includes a series of line DNA fragments fused by PCR and PCR-generated errors might be introduced into the genome8. Another potential disadvantage of this method is that the gene disruption cassettes they generated are linear DNA fragments by PCR. They are not circular plasmids which can easily be transformed into bacterial cells and purified away from them.

Furthermore, previous methods above cannot easily transplant from one fungus to another one, because different systems using different inducible promoters. Native fungal inducible systems with an inducible promoter from one fungus strain are not always well recognized or regulated by another system. Therefore, engineered regulatory systems are good choices for cre expression in a wide range of eukaryotic cells. Engineered regulatory systems to control fungal protein expression include chemically regulated gene expression systems that respond to orthogonal molecules such as estrogen (Gal-ER-VP16)9 or doxycycline (Tet-OFF)10 and light-inducible transcription systems11,12,13. Chemical inducers of protein dimerization have been extensively used to allow inducible control of signal transduction pathways and transcription in live cells. However, the systems are affected by compound uptake or drug pump issues, which can confound studies. For example, using Gal-ER-VP16 system protein levels are controlled by the addition of β-estradiol to the media9 resulting in considerable differences in the rate of β-estradiol uptake, with different strains or gene deletions that affect membrane transporters or drug efflux pumps drastically altering reporter protein expression levels. For instance, the Gal-ER-VP16 system does not work in a strain that cannot use β-estradiol. In contrast to chemicals, light is an ideal inducer of gene expression because it is highly tunable and has high spatiotemporal resolution11,12,13.

Homologous recombination (HR) and non-homologous DNA end-joining (NHEJ) both play critical roles in repairs of double-strand DNA breaks (DSBs). In the unicellular eukaryote Saccharomyces cerevisiae, HR is the main role involved in repair of DSBs, whereas NHEJ is the most active damage response mechanism in multicellular eukaryotes14. HR is, in principle, the most efficient method of disrupting, modifying, or replacing a target gene. Unfortunately, in most multicellular eukaryotes, the gene targeting efficiency is quite low because of the predominance of NHEJ over HR pathway15. We considered that these two repair pathways work independently and competitively16. Consequently, a strategy to eliminate or inhibit the NHEJ pathway is needed to achieve highly efficient gene targeting in multicellular eukaryotes.

In eukaryotes, the NHEJ system is composed of the DNA-dependent protein kinase catalytic subunit (DNA-PKcs), the DNA ligase IV-Xrcc4 complex and the KU70-KU80 heterodimer (KU complex)15,17. The NHEJ occurs in bacteria, fungi and mammals, indicating evolutionary conservation17. In Neurospora crassa, the HR efficiency increased to 100% in the KU disruption strains, compared to 5 to 20% in the wild-type species18. This approach has proven to be a major gene targeting breakthrough in most filamentous fungi. Deletion of the KU70 or KU80 gene to eliminate the NHEJ activity has dramatically increased the frequency of HR in numerous filamentous fungi such as Aspergillus nidulans, A. sojae, A. oryzae, A. niger and H. jecorina (Anamorph Trichoderma reesei)19,20,21,22.

However, deficient KU70/KU80 orthologues in different fungal species triggers varying susceptibility to different DNA damaging agents. For example, A. fumigatus showed mild sensitivity towards methyl methanesulfonate (MMS)23. In contrast, N. crassa showed increased sensitivity towards phleomycin and MMS24. A higher UV sensitivity was reported for A. niger21 and H. jecorina Δtku70 strains22. Such strains deleted in KU70/KU80 orthologues lack an important DNA repair mechanism. Furthermore, KU complex affected fundamental cellular aspects such as telomere maintenance, nuclear spatial organization or mitotic recombination25,26. To minimize the effect of a ku70/ku80 deletion, it is very advantageous to restore the NHEJ pathway by reintroducing the ku70/ku80 gene. The Δku70 strain can be retransformed with endogenous ku70. But, another genetic modification is necessary to screen for integration of the ku70 gene retransformation. Another possible method involves conventional genetics by crossing out the ku70/ku80 deletion, especially in strains already subjected to sexual crossing. However, sexual crossing is time-consuming and most industrially relevant fungi do not have a sexual cycle. A third option presented by Nielsen et al.27 used a transient disruption of nkuA, KU70 orthologue in A. nidulans, which was rectified by removing the bifunctional marker gene (pyrG) flanked by a direct repeat and re-establishing a functional nkuA gene spontaneously. A last option presented by Janus et al.28 involves transient silencing of the Pcku70 gene based on an autonomously replicating AMA1-based RNAi plasmid. However, the targeting efficiency obtained was less than in a ΔPcku70 strain28. In addition, AMA1-bearing self-replicative plasmids undergo rearrangement and multimerization at higher frequencies in some species29.

For efficient and sequential gene modification, we leveraged the widely used Cre/loxP recombination system, which can switch on gene expression by DNA excision and inversion30. We designed a one-step method to combine self-excisable marker rescue and an on-off control NHEJ pathway in industrial eukaryotic microorganisms.

In our system, cre was transiently expressed to generate marker-free transgenic fungi. Disrupted KU70 function was rapidly switched repeatedly using a simple selection scheme for the desired genetic manipulations. This controllable genetic manipulation is a powerful and labor-saving tool in fungal strains and easily fulfills the essential safety requirements to obtain marketing approval.

Results

Design of LML 2.0

Currently, several methods have been reported for unmarked genetic modification of eukaryotic microorganisms. However, these methods were tedious and entailed a two-step process3,4,6,7,8. Our objective was to upgrade from the previous LML1.03,4 to a new self-excisable gene disruption system, LML 2.0, which spliced two loxP sites, a Cre expression cassette and a selectable marker together to generate multiple fungal mutants through efficient one-step recycling of markers. This method obviates the need for a heterologous Cre expression cassette or Cre protein. It is much faster and more efficient than current gene deletion methods in fungi and reduces Cre toxicity. It could also be used sequentially to delete fungal genes without introducing selectable markers. Unfortunately, no stable plasmid was available for LML 2.0f design (Fig. 1A) in an E. coli host. We observed sufficient Cre activity from the xyn1-cre construct to catalyze the recombination of loxP plasmids in E. coli, confirming that many eukaryotic promoters were active in bacterial cells.

Figure 1
figure 1

Construction of LML cassettes.

(A) LML 1.0 and 2.0 cassettes. (B) Multicolor labeling cassette LML 2.0s and fluorescence micrographs of marker self-excision course using fluorescence microscopy in H. jecorina transformed with LML 2.0s. Bars represent 20 μm.

Full size image

To prevent the bacterial expression of cre driven by fungal promoters, we initially constructed three chimeric cre genes (z1cre, z2cre and z3cre). Three randomly selected predicted genes (z1, z2 and z3) from the H. jecorina were fused with the cre coding sequence by a 2A peptide, respectively. This 2A peptide could generate two mature proteins by self-cleavage. One protein is z1, z2 or z3 encoding peptide. The other one is Cre recombinase. Second, we constructed two modified cre genes (w1cre, w2cre), in which the coding sequence of native cre was interrupted by the two introns from the H. jecorina cbh1 gene, respectively. The resulting fungal transformation vectors LML 2.0a–e are shown in Fig. 1A. Restriction analysis and PCR demonstrated the absence of any excision from these plasmids in both E. coli and A. tumefaciens demonstrating that intron insertion abrogated all enzymatic activity of the cre gene in bacterial cells.

Using the Agrobacterium-mediated transformation, the above constructs were introduced into H. jecorina Qm6a, N. crassa and A. niger. Eight hygromycin B-resistant transformants for each host (H. jecorina Qm6a, N. crassa and A. niger) were randomly selected. These transformants were confirmed to have an intact LML 2.0 by PCR and a single copy of LML 2.0 by qPCR. The modified cre genes were drived by the xyn1 promoter which was induced by xylose. Introns were recognized and excised during the transcription to produce Cre recombinase that affected two loxP sites in the same orientation. For z1cre, z2cre and z3cre, the 2A peptide underwent self-cleavage to generate the mature H. jecorina protein Z1 (Z2, or Z3) and Cre recombinase, respectively.

As shown in Table 1, LML 2.0a–e auto-excised in these 3 strains and the average self-excision efficiencies were 60 to 85% for a, b and c, 20 to 90% for d and e. This xyn1-regulated LML 2.0 system works in A. niger as well and is also successfully transferred into N. crassa, leading to self-excision of the loxP maker cassette. These findings point to the general applicability of this system in filamentous fungi. Furthermore, the self-excision efficiencies of LML 2.0a–c seem to be slightly better than 2.0d and e due to stable self-excision efficiency in the 3 strains. The introns, obtained from H. jecorina cbh1, inserted into w1cre and w2cre, were not native, while the introns in z1cre, z2cre and z3cre were originally derived from z1, z2 and z3.

Table 1 Self-excision colony counts of LML2.0a–e cassettes in three different filamentous fungi.
Full size table

The xyn1 promoter is active in the presence of xylan and D-xylose and virtually silenced in the presence of glucose. Therefore, the xyn1 promoter controlled LML 2.0 system well. However, previous reports31 showed that industrial cre1-deficient strain RUT C-30 exhibited a basal level of xynl expression on glucose much higher than that of cre1 intact strain QM6a and QM 9414. To investigate if LML 2.0 was stable enough for industrial cre1-deficient strains, we analyzed the self-excision average efficiencies of 2.0a with different carbon source in cre1 deficient strain RUT C-30. The results (Table 2) showed that LML 2.0a had much higher self-excision efficiencies with D-xylose than other D-xylose downstream metabolites, xylitol, L-arabinose and L-arabitol. These metabolites failed to induce cre expression under the control of xyn1 promoter as the self-excision efficiencies were lower than that of glycerol, a repressing carbon source. Our data showed that the induction efficiency of L-arabitol on cre expression under the control of xyn1 promoter was lower than that of glycerol. This result is, however, in conflict with published reports32,33, which suggested that L-arabitol as well as D-xylose also induce xyn1 promoter transcription in H. jecorina. Transcript analysis was based on a single time point under induction by D-xylose downstream metabolites and was therefore over-estimated. The xyn1 promoter used to control cre expression was truncated (0.6 kb) and not in xyn1 locus.

Table 2 Total self-excision efficiencies of LML2.0a cassette with different carbon sources in cre1-deficient strain Hypocrea jecorina RUT C-30.
Full size table

The results (Table 2) also showed that LML 2.0a was quite stable under repressed conditions (with glucose as carbon source) in wild-type strain Qm6a and even industrial cre1-deficient strain RUT C-30. However, LML 1.0 is not so stable under glucose conditions in RUT C-30 (self-excision strains (total strains): 14(96)). To investigate the factors underlying the different stabilities under repression, a red fluorescent protein was fused behind z1cre and cre, which were integrated into xyn1 locus instead of xyn1 gene, respectively (Fig. S-1). The fluorescence intensity data (Fig. S-1) showed that the expression of z1cre-rfp in cell nucleus was much lower than that of cre-rfp although they were under the control of the same xyn1 promoter. A reasonable explanation for this is that z1 gene fused before cre had a negative effect on the expression levels of functional Cre recombinase. A lower Cre recombinase leads to lower Cre toxicity, lower efficiency of self-excision and higher stability under repressed condition.

In order to visualize the process of self-excision, multicolor labeling cassette LML 2.0s was constructed and shown in Fig. 1B. The H. jecorina transformants were used to analyze fluorescence during the growth after 24, 36 and 48 hours on Mandels’ medium (MA) containing D-xylose. The time course of changes in color from red to green, represented the course of self-excision (Fig. 1B). The excision above was tested with PCR and gene sequencing for the related gene locus. Finally, we used phosphinothricin acetyltransferase and sulfonylurea resistance allele expression cassette instead of hygromycin B phosphotransferase to construct plasmids, LML 2.0a-bar and 2.0a-sur. We confirmed that this xyn1-regulated LML 2.0a system works in M. anisopliae, which led to the ~83% self-excision of the LML cassette.

Design of LML 2.1 cassette

One more loxP site was added to the genome during each round of genetic modification and subsequent marker self-excision using the LML (1.0 or 2.0) system, which was associated with the risk of chromosomal rearrangement between the remaining loxP sites when carrying out the marker recycling process34. Therefore, we upgraded from LML2.0a to LML2.1, which employed the LE/RE mutant strategy using LE mutant lox carrying mutations in the left-inverted repeat region and RE mutant lox carrying mutations in the right-inverted repeat (Fig. 2). Recombination between a LE mutant lox and a RE mutant lox resulted in the generation of a double mutant lox site with mutations at both ends (Fig. 2). The Cre recombinase cannot efficiently catalyze recombination between the double mutant lox sites. The recombination reaction between double mutant lox sites was unavailable based on transient expression of Cre proteins by our LML 2.1 for only a limited time before its self-excision. Therefore, chromosomal rearrangement rarely occurred between the double mutant lox sites in the genome during further rounds of marker rescue.

Figure 2
figure 2

Construction of LML 2.1, 2.11, 2.12 and 3.0 cassettes.

Full size image

We selected three LE lox sites (loxJT15, lox44 and lox75) and three RE lox sites (loxJTZ17, lox43 and lox76)35. To test whether these mutant lox sites hindered recombination efficiency or promoted stability in our system, we used the same strategy as LML 2.0a to build LML 2.1a–k (Fig. 2).We constructed nine integration cassettes, LML 2.1a–i, harboring different LE/RE mutant lox pairs. H. jecorina transformation was followed by comparison of self-excision frequencies (Table 3). The average frequency of the LML 2.1a, loxJT15/loxJTZ17 pair, was 92%, which was higher than that of 2.0a (~80%). The frequency of LML 2.1b, loxJT15/lox43 pair was 26% and the remaining pairs showed similar and lower frequencies of about 2–10%. We then synthesized two double mutant lox sites, lox32 (recombination between loxJT15 and loxJTZ17) and lox58 (recombination between loxJT15 and lox43) and established LML 2.1j and k with them, respectively. The average frequency of the LML 2.1j was 2% and that of 2.1k was 0%, which were much lower than that of wild-type loxP (60%). The results show that LML 2.1a and b were suitable for sequential gene modification and too stable to induce recombination when carrying out further marker recycling processes.

Table 3 Total self-excision efficiencies of LML2.1a–k cassettes in RUT C-30.
Full size table

The tku70 deletion cassette was used to transform H. jecorina Qm6a. The Qm6aΔtku70 strain with correct transformation and marker excision was confirmed using PCR and qPCR. We used Qm6aΔtku70 as a host strain to delete several genes. We performed six additional deletions and determined homologous integration and self-excision frequencies (Table 4). PCR and gene sequencing of the gene-targeting locus in the excision strains revealed one remaining double mutant lox32 at the locus.

Table 4 Homologous integration and self-excision efficiencies of additional seven genes deleted in Qm6aΔtku70.
Full size table

As expected, the Qm6aΔtku70 strain showed similar homologous integration rates (Table 4). However, the hygromycin B excision frequencies were quite different between genes (Table 4). No self-excision was detected in Δxyr1 strain for the general activator Xyr1 as essential for xyn1 transcription. The expression of z1cre was under the control of xyn1 promoter, which was governed by transcription factors xyr1 and ace2. However, few self-excision strains were detected in Δxyr1 strain using 2.11 and 2.12 cassettes (Table 4). Very few self-excision strains of LML 2.0a or 2.1a in Δxyr1 strain were detected after two or three rounds of excision (Table 4). Consistent with earlier reports36, our results clearly demonstrate that the lox-FRT fusion sequences as recognition sites dramatically enhance the Cre-mediated excision efficiency. Luo et al.36 speculate that the fused loxP-FRT sequences may enhance the alignment of the recognition sequences, DNA bending or cleavage, or the formation of a Holliday junction or DNA–recombinase complex, resulting in improved efficiency. A 86-bp-length lox-FRT site scar was left after each round of self-excision of their ’GM-gene-deletor’ system36. Similar results were obtained with our LML 2.11. The LML 2.12 cassette, including one loxJT15-FRT site and one FRT-loxJTZ17 site, has never been reported before, with excision efficiency similar to that of LML 2.11. Furthermore, self-excision resulted in a scar of 34-bp-length lox32, shorter than 86-bp-length lox32-FRT of LML 2.11. The results encourage us to upgrade our one-step marker introduction/self-excision system for fungal strains, with disrupted xylanase regulators, or in which H. jecorina xyn1 promoter was not recognized or regulated.

Design of LML 3.0 cassette

Compared to LML 2.1, 2.11 and 2.12, we anticipate that the LML 3.0 is able to efficiently excise in other fungi subjected to sequential gene modification. Native fungal inducible systems from one fungus strain are not always well recognized or regulated by another system, just as xyn1 promoter was incompatible in Δxyr1 strain above. Therefore, we upgraded the LML 2.12 to 3.0 using a light regulator to control cre expression. Light-inducible transcription systems compatible with LML 3.0’s that work with different types of fungal hosts (other than 2.0 and 2.1) need to be designed.

Recently, we described a blue light-mediated regulation of DNA transcription in filamentous fungus T. reesei37. Using light-switchable transactivator G1V and its binding promoter 5Up37, we constructed LML 3.0 (Fig. 2) and LML 3.0s (Fig. S-2). The two active sites included loxJT15-FRT and FRT-loxJTZ17 fusion sequences. Using Agrobacterium-mediated transformation, the LML 3.0 constructs were introduced into H. jecorina Qm6a, Qm6aΔxyr1, N. crassa, A. niger and M. anisopliae. Using hygromycin B, phosphinothricin or chlorimuron ethyl, transformants for each cassette were randomly selected. These strains express Cre recombinase under the control of light. After the light-inducible self-excision, the phenotype was tested on plates containing corresponding antibiotics. As shown in Table 5, approximately 40–70% of the transformed strains had successfully excised the LML 3.0 cassette and did not show growth on corresponding antibiotic-containing plates. LML 3.0 excised in all the five strains, including Δxyr1 strain and even yeast strains (unpublished work). The LML 3.0 cassettes were widely used in more hosts than LML 2.0/2.1 due to artificial light-inducible control of self-excision.

Table 5 Self-excision colony counts of LML3.0 cassette.
Full size table

Design of OFN 1.0

NHEJ-deficient strains may negatively influence genome stability and fitness of the respective strains. For example, this mutation causes telomere shortening38, increased sensitivity to DNA damage26, higher susceptibility to various toxins and irradiation21. To avoid such drawbacks and to minimize the effect of a ku70 deletion, four main strategies are used currently to restore the NHEJ pathway: (1) ectopic or homologous integration of the endogenous ku70 gene; (2) sexual back-crossing; (3) transient disruption based on direct-repeat recombination27; and (4) transient expression system with an inducible RNAi vector28. The disadvantages include: (1) time-consuming genetic modification to screen for integration of the ku70 gene; (2) sexual back-crossing leads to additional undesired phenotype effects and is time-consuming. In addition, sexual crossing experiments cannot be performed with some asexual strains; (3) transient systems from Nielsen et al.27 have a low screening frequency and the strain restored pyrG negative, requiring uridine for cultivatation; (4) transient silencing system from Janus et al.28 have a targeting efficiency lower than that of a Δku70 strain. The AMA1-bearing plasmids undergo rearrangement and multimerization at higher frequencies in some species29.

Clearly, a new strategy is needed for the rapid and accurate on/off control of NHEJ pathway. Here we describe a foolproof strategy to on/off control the expression levels of tku70 in H. jecorina using the Cre/loxP system. DNA sequence flanked by loxP sites can be inverted or excised in the presence of the Cre recombinase, which depends on the orientation of the two loxP sites. We deduced that the orientation of tku70 gene would continuously be converted by Cre recombinase catalyzing, because tku70 gene was flanked by two inverted loxP sites. If Cre recombinase was removed from the cell, tku70 gene in one cell nuclei would be in one stable orientation. By adding the Cre recombinase self-excision system, Cre recombinase was removed from the cell. Finally, we can obtain two kinds of cell nuclei with two orientation of tku70 gene. We can get two strains including two orientation of tku70 gene by single spore isolation, respectively.

We developed and tested cassettes (OFN 1.0 system, Fig. 3) that incorporated two 2-state designs where tku70 adopted either A–C (LOW and HIGH) or D (OFF and ON) constitutive expression states after “inversion” with transient Cre expression of LML 2.0 or more advanced system. The loxP sites of OFN 1.0 and LML system may lead to irreversible recombination exacerbated by additional genetic modification. To address this issue, we utilized orthogonal loxP sites (commonly referred to as loxN and TATAlox) (Fig. 3A) that are capable of recombining with themselves, but do not interact with loxP sequences or each other.

Figure 3
figure 3

An on-off control protocol of nonhomologous end-joining (NHEJ) pathway using the Cre-lox system.

(A) Construction of OFN 1.0A–D cassettes. Genomic PCR of multiple strains using a given primer configuration (left bottom) shows inversion only after Cre induction. (B) Comparative transcript ratio analysis of tku70. Transcript ratios for ON and OFF state were calculated using ABI Stepone plus software. Values above 1 indicate higher transcription in the ON state strain compared to QM6a and values below 1 indicate lower transcription. Error bars represent 95% confidence intervals. ***means not done.

Full size image

We constructed 4 cassettes (OFN 1.0A–D) in which the gene regions of tku70 were flanked with inverted lox sites (Fig. 3A). In OFN, 1.0A–D tku70 was flanked by inverted TATAlox or loxN sites at different loci (from TATA box to -1 of tku70 promoter), which alter its expression by inversion (Fig. 3A). The Cre recombinase transiently expressed from LML system can invert two lox sites of OFN 1.0 and transform ON state into OFF state. To test this randomization strategy, we used a PCR strategy to verify promoter inversion (Fig. 3A) and qRT-PCR to analyze tku70 transcription (Fig. 3B). We observed a 50%/50% split in tku70 ON and OFF state after excision of the loxP marker cassette and obvious tku70 transcription in ON state strains (Fig. 3B). We detected lower transcript levels of tku70 in OFN 1.0A–C OFF states (Fig. 3B); however, it was barely detectable in 1.0D OFF state (Fig. 3B). Increased tolerance of UV irradiation was seen in ON state strains compared to the OFF state strains and Δtku70 strains (Fig. 4). The results showed that tku70 were restored by inversion, though the tku70 transcriptional level in OFN 1.0D ON state is lower than that in wild type strain.

Figure 4
figure 4

Survival rate of of Qm6a, Qm6aΔtku70, Qm6a&OFN1.0D-ON and Qm6a&OFN1.0D-OFF strains following exposure to UV.

Spores were exposed to different doses of UV. Aliquots of the UV-irradiated spores were plated on potato dextrose agar plates containing the colony restrictor Triton X-100 and the surviving colonies were counted.

Full size image

Discussion

The development of a genetic system that allows highly efficient homologous integration with resilient NHEJ deficiency and simultaneous transgene removal using light, is of great significance in the regulation of transgene expression in filamentous fungi.

In this paper, we have presented a new Cre/loxP-based system to generate marker-free transgenic fungi, with high self-excision efficiency, via a single-step transformation. Cre recombinase mediated simultaneously the elimination of the selectable marker gene and cre gene. Antibiotic susceptibility, PCR screening and integration border sequencing demonstrated unequivocally that the excision led to marker-free fungi strains. Unlike the fungal Cre/loxP system described previously, this system has several characteristics as follows.

First, the structure of our LML 2.0 or upper system is integral and simplified. Our system obviates the need for introduction and elimination of a heterologous Cre expression plasmid and is much faster than current gene deletion methods in filamentous fungi. We created three chimeric cre genes (z1cre, z2cre and z3cre) to prevent the bacterial expression of cre driven by fungal promoters and constructed this all-in-one cassette (loxP-cre-Marker-loxP region). The three genes (z1, z2 and z3) were randomly selected from H. jecorina. Selection of a highly conserved gene element, such as chromodomain39 encoding gene, enables construction of chimeric cre gene, for transformation of any eukaryotic cell. The system utilizes the conventional Agrobacterium-mediated transformation. All unnecessary components including cre gene and selectable marker gene are moved from transgenic fungal genome after induction. The transient expression of the z1cre gene prior to self-excision as established here prevents accumulation of large amounts of toxic Cre protein.

Second, our system employs the LE/RE mutant loxP. Recombination of loxJT15 and loxJTZ17 sites results in a double-mutant lox32 site (Fig. 1). The lox32 site left in chromosomes inhibited excision events and thereby maintained genomic stability, since it is not effectively recognized by Cre recombinase.

Third, our system uses circular plasmids, which can be preserved in bacterial cells and are easy-to-use, as cre gene is interrupted by introns to prevent the bacterial expression of cre driven by fungal promoters. Finally, expression of cre gene is induced by blue light, which is of utmost importance since cre gene placed under the control of an inducible promoter xyn1 from H. jecorina cannot be widely recognized by other strains, such as xyr1 deficient host. The LML 3.0 system has been shown to be functional in all five filamentous fungi in this study but is likely to be applicable to other eukaryotic cells as well, such as yeasts and plants. The LML 2.0/3.0 system potentially acts with zinc finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs) or short palindromic repeat (CRISPR) system (Fig. 5).

Figure 5
figure 5

Relationship between LML 2.0/2.1/3.0 system and ZFNs (A), TALENs (B) or CRISPR (C) system.

Full size image

Furthermore, we describe an on-off control protocol of NHEJ pathway called ‘OFN 1.0’ in H. jecorina. This is a key experimental model for genetic and metabolic regulation of enzyme production40,41,42,43. To address the limitations of using NHEJ-deficient strains, we have demonstrated rapid and repeated randomization of the two expression states (ON or OFF) of tku70 in H. jecorina by flanking the gene with inverted loxP repeats. Using orthogonal loxP sites (loxN and TATAlox), the OFN and LML systems coexist in the same genetic background. Using a single-step of blue light illumination, the self-excision of marker and cre genes and the restoration of NHEJ pathway happen simultaneously. We suggest that the design of light-regulated OFN 1.0 system is also useful in other branched pathway. Using on-off or high-low control expression of proteins regulating the flux through alternate branches, the innate capabilities of the network to produce different outcomes or products can be explored.

In summary, we presented a foolproof integrated strategy to generate genetically engineered multicellular eukaryotes in gene targeting without NHEJ deficiency and to recycle the markers during the each transformation. It is possible to perform sequential steps of genetic modification without a shortage of markers. This strategy allows functional analysis of genes in a defined genetic background, including marker-free deletion and NHEJ pathway restoration. Furthermore, this strategy enables the creation of recombinant strains without possible disturbance of marker gene residual or NHEJ deficiency on the physiology or metabolism of the strains26,42. Our results demonstrate that an acceptable transgenic manipulation has been established to produce marker-free homologous recombination in eukaryotic cells, which can be used to produce safe transgenic organisms.

Materials and Methods

Strains and media

Escherichia coli DH5α was used as a host strain for the recombinant DNA manipulations. H. jecorina strains including wild type strain QM6a, mutant RUT C-30, N. crassa wild type strain FGSC2489 and Metarhizium anisopliae ARSEF2575, belonging to the Sordariomycetes and A. niger wild-type strains44 were maintained on malt extract or potato dextrose agar. Agrobacterium tumefaciens GV3101 and AGL-1 were used in the Agrobacterium-mediated transformation system. Luria–Bertani (LB) medium was used for E. coli and A. tumefaciens culture. Mandels’ medium (MA)45 and M-100 medium46 were used for general fungal transformation. The fungal strains constructed in this study are summarized in Supplementary Table S-1.

Plasmid construction

All the plasmids used in this study are listed in Supplementary Table S-2. The vectors were built using compatible cohesive ends generated by XbaI and SpeI in pPK2-derived BioBrick base vector, as previously described47. In this study, three kinds of marker cassettes, including hygromycin B phosphotransferase, phosphinothricin acetyltransferase and sulfonylurea resistance allele expression cassette were used to construct plasmids. The construction of LML 1.0, 2.0 (a–f and s), 2.1 (a–k), 2.11, 2.12 and 3.0 is shown in Fig. 1A,B and 2.

The LML 1.0 cassette includes two repeated loxP sites with a marker in between them. The cre recombinase gene was inserted into the genome instead of xyn1 gene. In this study, cre was fused with a nuclear localization sequence (NLS; GenBank: EHK22773.1).

Compared to LML 1.0, LML 2.0a–c have an additional Cre recombinase expression sequence, which includes a xyn1 promoter, a native gene (z1, tre111731; z2, tre112680; or z3, tre112518) from H. jecorina, a 2A self-cleavage peptide48, cre gene fused with NLS and a nos terminator (GenBank: KF499077.1), between the two repeated loxP sites. In LML 2.0d and 2.0e, the Cre recombinase includes a modified cre gene (w1cre or w2cre), in which cre coding sequence is interrupted by one of two introns from the H. jecorina cbh1 gene (tre123989), respectively (unpublished work).

LML 2.0s cassette was constructed from LML 2.0a by adding a red fluorescent protein expression sequence and a xyn1 promoter and a green fluorescent protein gene with a flanking nos terminator externally, respectively.

To construct LML 2.1a–k cassettes two loxP sites were replaced by different left element/right element (LE/RE) mutant loxP sites49. To construct LML 2.11 or 2.12 cassettes the left loxP sites were both replaced by loxJT15-FRT50 fusion site and the right loxP sites were replaced by loxJTZ17-FRT or FRT-loxJTZ17 fusion sites, respectively.

Compared to LML 2.12, LML 3.0 cassettes have a light-switchable transactivator G1V gene, which is fused with the selectable marker genes by a 2A peptide and a corresponding light-switchable promoter 5Up for Cre recombinase expression instead of xyn1 promoter.

The H. jecorina tku70 deletion vector was used in the pPK2-derived BioBrick plasmid and comprised the LML 2.1a cassette flanked by 2.0-kb fragments up- and downstream of tku70. The flanking region fragments were amplified by PCR using primer pairs KU705-F/R y KU703-F/R. The construction of cassettes for the 5 deletions (cbh1, cbh2, tre108087, cre1, ace1, ace2) in the tku70 deficient strain was performed in the same BioBrick plasmid mentioned above and contained the LML 2.1a cassette flanked by 1-kb fragments up- and downstream of the target gene.

Four types of xyr1 deletions in H. jecorina tku70 deficient strain contained the LML 2.0a, 2.1a, 2.11 or 2.12 cassette, respectively, flanked by 1-kb fragments up- and downstream of xyr1. The flanking region fragments were amplified by PCR. Primers used for the 7 deletion constructs are listed in Table S-3 in the supplemental material.

In H. jecorina RUT C-30 strain, two chimeric genes, a red fluorescent protein gene fused behind z1cre and cre, respectively, were integrated into xyn1 locus instead of xyn1 gene (see Supplemental Material Fig. S-1).

The on-off control cassettes of NHEJ pathway, OFN 1.0A–D, are shown in Fig. 3A. To generate OFN 1.0A cassette three bifunctional TATA-lox51 sites and a hygromycin B marker self-excision cassette were inserted around tku70. The first reverse TATA-lox site was inserted 65 nt upstream of ATG of tku70 by PCR-mediated mutagenesis instead of − 65 to −32 of tku70 promoter. Therefore, the TATA box (− 52 to –38) of tku70 promoter was replaced by the first bifunctional TATA-lox site. The second TATA-lox site was set behind TGA codon of tku70. The third TATA-lox site was behind self-excision cassette.

To generate OFN 1.0B–D cassettes three loxN52 sites were inserted around tku70. The first reverse loxN sites were all inserted downstream of TATA box sequences of tku70 promoter in the three cassettes. However, they were located in different loci: −38 of tku70 promoter for 1.0B; −30 for 1.0C; and −1 for 1.0D. The second and third loxN sites and self-excision cassette were located similar to TATAlox in 1.0A cassette.

Transformation of T. reesei

Transformation of the hygromycin B resistance marker vectors was performed based on protocols described by Covert et al.53. Transformation of the phosphinothricin resistance marker vectors was performed based on protocols described by Zhang et al. (manuscript in preparation). Transformation of the chlorimuron ethyl resistance marker vectors was based on protocols described by Lin et al.54.

Characterization of the transformants

For each vector transformation, 4 to 8 transformants of Δtku70 strains or 12 transformants were collected. We used qPCR to verify that a single copy from the expression cassettes was successfully integrated into the genome. Analysis of the integration of the target gene-deleted construct was performed using diagnostic PCR and qPCR.

Excision of the loxP marker cassette

To excise the LML cassettes, transformants of LML 1.0, 2.0, 2.1, 2.11 and 2.12 were cultivated on an MA medium containing 1% (wt/vol) glycerol, glucose, D-xylose, xylitol, L-arabinose, or L-arabitol (Sigma Aldrich, St. Louis, MO) by shaking (200 rpm) at 28 °C for 2 days. LML 3.0 transformants were cultivated on an MA medium containing 1% (wt/vol) glucose exposed to 53 W/m2 constant blue light (460 nm peak) for 2 days. Conidia were harvested after 4 to 8 days under conditions similar to liquid culture before, plated on an MA agar plate containing 0.1% (wt/vol) Triton X-100 (Sigma Aldrich) and 1% (wt/vol) carbon sources. The phenotypes of the 8 to 24 fungal isolates of each transformant were monitored by plating them on an M-100 plate containing corresponding selection markers (100 μg/ml hygromycin B, 50 μg/ml chlorimuron ethyl, or 300 μg/ml phosphinothricin). No growth indicated successful excision of the LML cassettes. For further verification, the excision was examined with DNA sequencing of the related gene locus (see Supplemental Material Fig. S-3). If the fungus still retained the ability to grow on selection marker-containing medium, additional rounds of excision were conducted.

RNA preparation and quantitative real-time PCR (qPCR)

About 20 mg of T. reesei mycelium was harvested. Total RNA was extracted using the FastRNA Pro Red Kit (MPbio, U. S. A.), according to the manufacturer’s instructions. Reverse transcription was performed with 1000 ng of total RNA using TransScript All-in-One First-Strand cDNA Synthesis SuperMix for qPCR (TransGen, China), according to the manufacturer’s instructions. For qPCR, the TransStart TipTop Green qPCR SuperMix (TransGen, China) was used with 200 nM of forward and reverse primers (see Supplemental Material Table S-4) and 1 μl of 10-fold diluted cDNA in a final volume of 20 μl. For hpt transcription analysis, a SYBR green assay with reference to a small GTPase gene (sar1) was performed. Thermocycling was performed in an ABI StepOne Plus thermocycler (Applied Biosystems, USA).

Measurement of fluorescent protein

About 106 conidia were inoculated into 10 mL MA containing 1% D-xylose and cultured for 1 to 3 days. The mycelial suspension was collected and analyzed for the expression of fluorescent protein under a fluorescence microscope (Olympus BX50, Japan).

Mutagen sensitivity

Sensitivity to UV exposure was determined as described22.

Additional Information

How to cite this article: Zhang, L. et al. Light-inducible genetic engineering and control of non-homologous end-joining in industrial eukaryotic microorganisms: LML 3.0 and OFN 1.0. Sci. Rep. 6, 20761; doi: 10.1038/srep20761 (2016).