Intracellularly expressed Cas9 led to unexpected off-target genome editing
The cas9 gene has been expressed constitutively or induced successfully in T. reesei, in both cases which can be used in successful genome editing in QM6a and RUT-C30 . The codon-optimized cas9 gene  was synthesized, ligated downstream of the strong, constitutive pdc1 promoter, and transformed into T. reesei QM9414. Nine positive transformants were obtained, among which strain C5 had the highest cas9 transcript level (determined by RT-qPCR, data not shown) and was therefore selected for subsequent gene disruption.
The ura5 gene has been used as the target when the CRISPR/Cas9 technology was established in T. reesei . The in vitro transcribed gRNA targeting the same ura5 locus was transformed into C5. Nineteen 5-fluoroorotic acid (5-FOA) resistant transformants were obtained. While no mutation was found in the expected editing locus for all these transformants, seven had an unexpected insertion of 12-bp at the 100-bp downstream of the expected editing locus (Fig. 1, Additional file 1: Figure S1). One (QM9414-C5/T18) had an insertion of 9-bp, while another (QM9414-C5/T13) had a deletion of this 9-bp fragment, at the 70-bp downstream of the expected editing locus. Interestingly, the inserted short DNA fragments were a direct repeat of the nucleotides downstream of the insertion site. Moreover, two direct repeats (7- and 11-bp, respectively) were found in both sites (Fig. 1), suggestive of biased off-target editing.
Designing gRNA for cbh1 gene disruption
Direct transformation of Cas9/gRNA complex may be an alternative for genome editing. Indeed, this technique has already been successfully used for gene disruption in various filamentous fungi including Mucor circinelloides , Aspergillus fumigatus , and Fusarium oxysporum . For those filamentous fungi that are difficult to manipulate, the transformantion of Cas9/gRNA complex provides a means to speed up genome editing. With this advantage, nevertheless, there was no report of using such a technique in T. reesei. CBH1 is the major cellulase secreted by T. reesei, accounting for above 50–60% in the fermentation broth . Disruption of cbh1 will result in loss of the major band on SDS-PAGE gel, facilitating identification of the strains with successful genome editing. Therefore, for ease of verification, cbh1 was selected as the target gene. Three different gRNAs on cbh1 were designed and synthesized by in vitro transcription. Cas9 and a certain gRNA was incubated with the BcuI-linearized pT3cbh1 plasmid (4717-bp) containing the cbh1 gene. gRNA-3 was the best one to guide Cas9-mediated digestion of pT3cbh1 into two DNA fragments with expected sizes (Fig. 2).
Disruption of cbh1 involved insertion of large DNA fragments
Twenty-seven transformants were obtained in two transformations of Cas9/gRNA with pSKpyr4 in TU-6. Through SDS-PAGE analysis, it was suggested that eight among the 27 transformants could have the cbh1 gene disrupted since the band corresponding to CBH1 disappeared (Fig. 3A). Interestingly, in these strains, expression of other proteins apparently increased (Fig. 3A), similar to improved secretion of a heterologous lipase in a cbh1-silenced T. reesei strain . Using the primers specific for cbh1 (Additional file 1: Table S1), a DNA fragment of 1676-bp was amplified from the parent strain TU-6 (Fig. 3B), as well as from the nineteen transformants still secreting the CBH1 protein (data not shown). Sequencing of all these DNA fragments from these transformants indicated no mutation in the cbh1 gene. Using the same PCR condition, no DNA fragment corresponding to the 1676-bp could be amplified from the transformants (except T1) that did not secret CBH1. It has been pointed out that in Nodulisporium sp., transformation of a linear marker plasmid into the cas9-expressing host cell leads to insertion of the marker gene in the gRNA targeting locus . It was hypothesized that similar integration of large DNA fragments could take place. Therefore, the PCR condition was modified, mainly by lengthening the extension time, which would allow amplification of larger DNA fragments. Using this method, we successfully amplified DNA fragments from five transformants (Fig. 3B). The sizes of the DNA fragments from four transformants (T2, T7, T8, and T9) were from 5- to 8-kb, while that for one transformant (T1) was 1.9-kb.
Further sequence analyses indicated that there were gene editing events at the expected cbh1 locus. Deletion of 91~705-bp in cbh1 was observed in T2, T7, and T9 (Fig. 3D). The inserted fragments varied in length and sequence by containing: i) the transformed plasmid (T2, T7, and T8); ii) a chromosome fragment (T1); or iii) mixed chromosome and plasmid sequences (T9, Fig. 3C&D). Although T2 and T7 had different deletions, they had the same inserts. It seemed that the pSKpyr4 DNA fragments starting from nucleotide 23 were favored in these gene rearrangement events (Fig. 3D). The underlying reason for this is not known. All five had another fragment (“other insert” in Fig. 3C&D) which was not from pSKpyr4 or T. reesei chromosome but from E. coli chromosome (T1) or an unidentified plasmid(s) (T2, T7, T8, and T9), which was likely contaminants from pSKpyr4 preparation.