Cre/lox-mediated chromosomal integration of biosynthetic gene clusters for heterologous expression in Aspergillus nidulans

Building strains for stable long-term heterologous expression of large biosynthetic pathways in filamentous fungi is limited by the low transformation efficiency or genetic stability of current methods. Here, we developed a system for targeted chromosomal integration of large biosynthetic gene clusters in Aspergillus nidulans based on site-specific recombinase mediated cassette exchange. We built A. nidulans strains harbouring a chromosomal landing pad for Cre/lox-mediated recombination and demonstrated efficient targeted integration of a 21.5 kb heterologous region in a single step. We further evaluated the integration at two loci by analysing the expression of a fluorescent reporter and the production of a heterologous polyketide. We compared chromosomal expression at those landing loci to episomal AMA1-based expression, which also shed light on uncharacterised aspects of episomal expression in filamentous fungi. This is the first demonstration of site-specific recombinase-mediated integration in filamentous fungi, setting the foundations for the further development of this tool.


Introduction
Filamentous fungi are prolific producers of metabolites and enzymes with biotechnological applications in pharmaceutical, agricultural and food industries. 1,2 Importantly, fungal secondary metabolites (SMs) remain a promising source of novel drug leads. 3 The genes required to produce a SM are usually colocalised in the genome forming biosynthetic gene clusters (BGCs), which can be easily identified bioinformatically. However, a large fraction of BGCs remain silent or lowly expressed under standard culture conditions, which limits their analysis. 3 As novel BGCs are often found in fungal species that are difficult to cultivate or engineer, heterologous expression in hosts with more genetic tools available is preferred. 4 Filamentous fungi hosts present several advantages for the heterologous expression of BGCs from other filamentous fungi, such as increased compatibility of promoters and intron splicing, and their natural capability for producing SMs. In particular, Aspergillus nidulans has been widely used to produce SMs by chromosomal integration or episomal expression of heterologous BGCs. [4][5][6] Episomal systems based on the replicator AMA1 have facilitated initial testing of BGC products due their high transformation efficiency compared to integrative vectors. [7][8][9] However, the phenotypic stability of AMA1-vectors has been shown to be limiting, even under selective conditions. 10 Therefore, chromosomal expression is preferred for largescale long-term stable bioproduction in industrial settings. 11 Additionally, strains with chromosomally encoded genes can be grown in low-cost complex substrates, such as spent grains, molasses or other agri-food residues 12 , as they do not require maintaining selection pressures.
Currently, targeted chromosomal integration in A. nidulans is pursued through homologous recombination (HR) facilitated by strains deficient in the non-homologous end joining pathway (nkuA∆). 13 However, HR efficiency drops when integrating large constructs. 11 As a result, HR-mediated integration of large BGCs relies on the laborious step wise integration of smaller BGC fragments. 14,15 To leverage the increasing amount of BGCs identified in novel fungi, new methods are needed for the fast creation of strains for heterologous BGC expression. 16 Here, we develop a Cre/lox site-specific recombinase system for one-step chromosomal integration of BGCs in the heterologous host A. nidulans. Site specific recombinases are well suited for the integration of large DNA regions, as they mediate the strand exchange between the recombination sites in a size independent manner. 17 As Cre/loxP recombination is reversible, strategies for irreversible integration rely on mutated lox sites, either on its 8-bp asymmetric core or 13-bp palindromic arms. Heteromeric lox sites contain nucleotide variations in the left or right arms, respectively named LE and RE. 18 LE/RE integration relies on a single recombination event between a chromosomal lox site and the donor vector, but results in the integration of the complete donor vector (Figure 1). Another strategy for integration relies on two sequential recombination events between two pairs of heterospecific lox sites (harbouring mutations in the core) named Recombinase Mediated Cassette Exchange (RMCE). In RMCE the chromosomal landing pad is flanked by two lox sites that are incompatible between themselves but compatible with the sites flanking the genes of interest located in the donor vector. In the presence of the Cre recombinase, the donor vector is integrated at the landing pad while in a second recombination event the donor vector backbone and the first landing pad cassette are excised, resulting on the irreversible integration of the genes of interest ( Figure 1). RMCE has been used for the integration of large constructs in a wide range of cell factories and model organisms. [19][20][21][22][23] However, to this date there is no system for site-specific recombinases-mediated chromosomal integration in filamentous fungi.
Cre/lox recombination has been widely used in filamentous fungi for the excision of marker genes. [24][25][26][27][28][29][30][31] Here, we demonstrate Cre/lox-mediated integration is an efficient alternative for the integration of large heterologous constructs in A. nidulans, with potential to be expanded to other filamentous fungi.

Results
Design and construction of a recombinase-mediated integration system in Aspergillus nidulans Cre/lox-mediated irreversible integration can be achieved by different strategies such as RMCE or LE/RE, making use of different engineered lox sites. 18 To simultaneously evaluate the feasibility of RMCE and LE/RE based chromosomal integration in A. nidulans we designed a combined strategy ( Figure 1). We designed lox sites that contained both the heterospecific mutation of lox2272 32 and either of the heteromeric mutations of lox71 and lox66 33 , creating the sites lox2272-71 and lox2272-66 (Table S1). In principle, lox2272-71 and lox2272-66 recombination is irreversible as it results in the double LE/RE mutant site lox2272-72 and lox2272 (Table S1). Additionally, those sites are incompatible with loxP for recombination. By flanking the donor cassette and landing pad by loxP and either lox2272-71 or lox2272-66, the system can also be used for RMCE ( Figure 1).

Figure 1
Overview of the strategy for Cre/lox-mediated chromosomal integration. 1) The landing pad (LP) containing the bar marker gene flanked by loxP and lox2272-71 is integrated in the destination locus of Aspergillus nidulans genome by homologous recombination. The resultant strain A. nidulans LP is used to prepare protoplasts that can be stored for subsequent transformations. 2) A. nidulans LP protoplasts are transformed with the donor vector that contains loxP and lox2272-66 flanking the marker gene pyrG, a fluorescent reporter and the genes of interest, along with a second vector for transient expression of cre recombinase. Stable integration can be achieved by LE/RE in 1-step recombination or in 2-steps by RMCE. 3) Selection of the recombinant colonies in minimal media for pyrG complementation.
First, we validated the capability of Cre to recombine the sites lox2272-66 and lox2272-71 by an assay in vitro ( Figure S1). To create the recipient fungal strain, we chromosomally integrated the floxed (flanked by lox sites) landing pad (LP) by homologous recombination in the strain A. nidulans LO8030. 34 The landing pad consisted of the sites loxP and lox2272-66 Table S1) flanking the marker bar for glufosinate resistance. We selected as a first landing locus the sterigmatocystin (stc) biosynthetic gene cluster boundaries located in A.
nidulans chromosome IV, as we previously used this locus for chromosomal expression of other genes. 35 After glufosinate selection of the transformant colonies and PCR verification, the strain A. nidulans landing pad 1 (LP1) was isolated for future tests.
For transient Cre recombinase expression, we created a helper vector unable to replicate in A. nidulans. The helper vector encodes a cassette for constitutive expression of Cre under the promoter gpdA (PgpdA) and the terminator trpC ( Figure 1). As the recipient strain A. nidulans LP1 carries nkuA∆, which minimises random integration events, the helper vector presumably would be lost during fungal growth. 13 To test the feasibility of RMCE/LE-RE integration at LP1, we built different donor vectors containing the fluorescent reporter mCherry and the pyrG marker flanked by loxP and lox2272-71 ( Figure 1, Figure S2A). Donor vector 1 is 6.6 kb long while donor vector 2 is a 12.2 kb shuttle vector that supports optional transformation-associated recombination cloning in Saccharomyces cerevisiae. Donor vector 2 also contains within the floxed region four cloning sites for the expression of biosynthetic genes under strong alcohol inducible promoters, derived from the vector pYFAC-CH2 8 ( Figure S2A). As the four-promoter multiple cloning site is flanked by EcoRI sites, donor vector 2 can also be used for cloning genomic fragments containing a BGC. To demonstrate the utility of donor vector 2, we used it to clone an 18 kb region of Aspergillus burnettii by isothermal assembly, resulting in a 21.7 kb floxed donor region. This vector, named donor vector 2-bue contains six genes responsible for the biosynthesis of the polyketide compound 1, a burnettiene precursor which will be described elsewhere ( Figure S2A). As burnettienes are produced in A. burnetti (bue cluster is not silent), we hypothesised that if the bue genes were chromosomally integrated at a good expression locus in A. nidulans we would observe the production of compound 1 as proof-of-concept ( Figure S2B). 36

Evaluation of recombination efficiency across different donor vectors
To evaluate the efficiency of the recombination system, we transformed protoplasts of A. nidulans LP1 by a small-scale PEG mediated transformation in 2 mL microtubes (approximately 5×10 6 protoplasts) (Figure 2A). We tested different amounts and ratios for each donor vector and the helper vector, and we also transformed the donor vector alone as control.
We consistently obtained transformant colonies in the strains cotransformed with the helper vector for cre expression (3-13 colonies per transformation event) ( Table 1). The variation in the number of transformant colonies was independent of the size of the transformant vector, which is expected for recombinase-mediated integration (Table 1). 21 In the control strains without cre helper vector we mostly observed only small "abortive" colonies that did not support further growth. Abortive colonies arising from residual nonintegrated vector encoding pyrG have been previously reported in A. nidulans ∆nkuA. 13 Interestingly, we observed a higher number of abortive colonies when transforming the smaller donor vector 1 compared to the larger donor vector 2 and donor vector 2-bue. Additionally, we did not observe viable colonies in the controls when transforming the larger donor 2 and donor 2-bue (Table 1).
To evaluate the mechanism of integration in the transformant pyrG+ colonies, we analysed the landing locus by PCR amplification of the recombination junction regions ( Figure   2, Figure S3). The screening strategy consisted in four PCR tests per sample, designed so one primer would bind either neighbouring region at the 5′ or 3′ from the landing locus and the other primer would bind inside of the floxed region from either the original landing pad or recombination product ( Figure 2B, Figure S3A). The expected amplicon size is ~1.5 kb in all PCRs and is indicative of recombination or presence of the original landing pad for each lox site ( Figure 2C, Figures S3C-D). We verified representative PCR amplicons by Sanger sequencing, which confirmed in vivo that the recombination product between lox2272-66 or lox2272-71 is the double LE/RE mutant site lox2272-72 ( Figure 2D).
Analysing the total of transformant colonies per experiment, the frequency of successful recombinase-mediated integration ranged 29-100% across experiments (Table 1). We observed that lower recombination efficiency was found in transformations carried out with ≤0.9 pmol of helper vector. At higher amounts of helper vector transformed (≤1.5 pmol), Cremediated recombination efficiency for donor vector 1 ranged 60-83% (Table 1). Importantly, the recombination efficiency of the larger vectors donor 2, and donor 2-bue ranged a 90-100% at the highest helper vector concentrations (Table 1, Figures S3C-E). The lower frequency of false positives seems to indicate that larger donor vectors are less prone to random integration.  Figure 2E, Figure S3B).
Overall, around half of the colonies that recombined at LP1 presented complete RMCE (N=17), while the rest were the intermediates LE/RE (N=6) and loxP (N=14) (  Figure S4). These results could imply that random integration of the cre helper vector occurs in A. nidulans or that traces of the residual vector might be present at later growth stages.
Our screening strategy consisted in testing for the presence of recombination product along with presence or absence of the original landing pad (  Evaluating expression at landing pad 1 with a fluorescent reporter and metabolite production To evaluate the strains with recombinase-mediated integration at LP1, we first analysed the expression of the fluorescent reporter mCherry encoded in the donor cassette. To benchmark chromosomal expression at LP1, we compared this expression system to strains expressing mCherry from episomal AMA1-pyrG vectors. We consistently observed fluorescence in mycelia of colonies with mCherry integrated at LP1 compared to the negative control ( Figure 3A, Figure S5). However, fluorescence in strains with mCherry integrated at LP1 was low compared to the AMA1-encoded counterpart under selective conditions. We also observed that when hyphae from strains harbouring AMA1-pyrG encoded PgpdA-mCherry were grown in non-selective conditions, some hyphae retained higher fluorescence than their LP1integrated counterpart ( Figure 3A, Figure S5).

Inspired by comparative studies of chromosomal and episomal expression in yeast by
Jensen et al. 37 , we analysed the spores of recombinant colonies by flow cytometry to test phenotypic stability. We observed a unimodal distribution of fluorescence in the strains with mCherry integrated at LP1, distinguishable from the negative control ( Figure S6A). The strains with AMA1-based episomal expression of mCherry presented a much wider multimodal distribution ( Figure S6A), which is expected due to the phenotypic instability of AMA1-encoded genes on spores even under selective conditions. 10 While these results indicated that chromosomal expression at LP1 results in a more homogeneous cell population, expression at LP1 is at least one order of magnitude below the signal from the best performer spores of AMA1-episomal expression. Overall, our analysis of the fluorescent reporter at LP1 by flow cytometry and microscopy supported phenotypic stability, which is expected of chromosomally integrated gene.
To assess compound production at LP1, we cultivated different transformant strains of Cre/lox-mediated integration of donor vector 2-bue (N=9) ( Figure S7). However, we did not observe detectable compound production when the genes were integrated at LP1, while we consistently observed production of compound 1 in the strains with episomal expression (N=8) ( Figure S7).
Overall, when evaluating fluorescent reporter mCherry and compound 1, expression at LP1 was lower than its episomal counterpart. Furthermore, the lack of compound production at LP1 makes it not an ideal locus for BGC expression for natural product discovery. To make our recombinase-mediated integration system more useful, we decided to evaluate integration at a landing locus previously validated for production of heterologous compounds. Strains with mCherry chromosomally integrated at LP1 show lower fluorescence than the AMA1-based episomal counterpart under selective conditions. Selection pressure is needed to support high AMA1based episomal expression. Samples with similar mycelial growth were observed under mCherry filter and brightfield (BF) by fluorescence microscopy. Scale bar 50 µm. Biological replicates are found in Figure S5. B. Overview of the strategy to evaluate the site landing pad 2 at Aspergillus nidulans IS1 locus. C. Analysis of colony growth by fluorescence photography shows a patchy pattern in the fluorescence of colonies with AMA1-based expression under selective conditions, compared to more homogeneous fluorescence in strains with chromosomal expression at LP2. Extended information at Figure S9. D. Analysis of spores by flow cytometry shows more compact and homogeneous fluorescence in samples with chromosomal integration of mCherry at LP2 compared to AMA1-based episomal expression. A proportion of the spores expressing mCherry episomally from AMA1 vectors can reach fluorescence levels one order of magnitude higher than spores with chromosomal expression at LP2 (see red arrow). Biological replicates, event counts and gating strategy can be found in Figure  S6.
Construction and evaluation of landing pad 2 The site IS1 located in A. nidulans chromosome I has been used as target for integration of heterologous biosynthetic genes by homologous recombination, for example the geodin BGC. 14 We proceeded to integrate our floxed landing pad at IS1 by HR, creating the strain A.
At this stage we also evaluated the use of a replicative AMA1-based vector containing the floxed pyrG cassette as donor vector 3 ( Figure S8). However, we did not obtain evidence of recombination with this prototype, which highlights the relevance of selecting for the recombination event and not just DNA uptake ( Figure S8).
To To further analyse the profile of mCherry expression at LP2, we analysed spores from different transformant colonies by flow cytometry. We observed a similar pattern than the strains with mCherry integrated at LP1 ( Figure 3D, Figure S6). These results indicate that P gpdA-mCherry integrated at either LP2 or LP1, failed to achieve the fluorescence levels comparable to the best performer spores in the multicopy AMA1-encoded PgpdA-mCherry. We also observed lower fluorescence in mycelia from strains with chromosomal expression at LP2 compared to AMA1-based episomal counterpart ( Figure S5).
Next, we evaluated the production of compound 1 in the A. nidulans strains with bue genes integrated at LP2. After cultivation, like the strains with bue genes integrated at LP1, we did not observe production of compound 1 in the recombinant strains (N=5) when analysing media extracts ( Figure S10A). For troubleshooting, we verified that the end of culture fungal pellets expressed mCherry ( Figure S10A). As a next troubleshooting step, we investigated two transformant strains with bue genes integrated at LP2 and one at LP1 by whole genome sequencing ( Figure S11). We confirmed the expected recombination with no mutations that could explain the lack of compound production. Thus, the lack of compound production could be due to the bue genes being silent when integrated chromosomally.
Activation of chromosomally integrated bue genes by transcription factor overexpression As the strains with chromosomal integration of bue genes at LP1 or LP2 did not produce compound 1, we hypothesised that overexpression of the bue cluster specific transcription factor (TF) bueR could activate the cluster in chromosomal context. and TF overexpression in media supplemented with uracil and uridine. It should be noted that the AMA1-based vector for TF expression is under the selection of pyridoxal auxotrophy with the pyroA marker and the medium used lack pyridoxine. As expected, the uracil and uridine selection pressure was not needed for compound production ( Figure 4B). Interestingly, we observed 3-fold higher production in the strains grown in non-selective conditions ( Figure 4C).
Positional effects arising for insufficient expression of the pyrG marker integrated in some loci have been observed in A. nidulans before, however changes in compound production in uracil/uridine supplemented media could be due other effects. 39,40 To sum up, overexpression of the cluster specific TF allowed to activate the expression of bue genes at two different chromosomal contexts. These results demonstrate that recombinase-mediated integration is a feasible strategy to build strains for heterologous compound production, but that alternative strategies might be needed for BGC activation it the cluster remains silent. Overall, chromosomal expression permitted cultivating in non-selective media without compromising yields.

Discussion
In this work we designed, built, and tested a system for Cre/lox-mediated chromosomal integration of long DNA fragments for the heterologous expression of BGCs in A. nidulans.
Site-specific recombinase mediated integration represents a relevant expansion to the synthetic biology toolbox for filamentous fungi, where recombinases had only been used for gene deletion or inversion. 31 The vector set developed in this work has the potential to be easily adapted for integration at different chromosomal landing loci or related fungal chassis. 41 We demonstrated targeted one-step integration of long DNA regions of up to 27 kb by LE/RE and 21 kb by RMCE in A. nidulans. We obtained high transformation efficiency (up to 100%) using a small-scale transformation protocol with optimised amounts of helper vector.
We additionally observed that the false positive rate and the presence of abortive colonies diminished when transforming large donor vectors (≥12 kb), making this system particularly advantageous for integration in the size range where HR efficiency drops. We also demonstrated that the resulting strains with chromosomally encoded genes presented a more uniform fluorescence phenotype in spores and mycelia compared to AMA1-encoded genes, evidencing genetic and phenotypic stability.
The developed Cre/lox-mediated integration system presents an advantageous alternative to HR for the single-step integration of large constructs. However, it also faces the same constraints as HR for heterologous expression featuring a single gene copy in a chromosomal context. When integrating the heterologous genes bueA/B/C/D/E/F/R under their native promoter at LP1 or LP2 initially we did not observe compound production, unlike the episomal counterpart. Nevertheless, when overexpressing the cluster specific TF bueR we restored compound production from biosynthetic genes at LP1 and LP2. Strategies for BGC activation, such as TF overexpression, promoter replacement or CRISPRa could be used to activate chromosomally integrated genes that remain silent. 35 Alternatively, integration in a better locus for expression could also result in improved performance. Even though our results indicate that the phenotypic stability of AMA1-encoded genes in spores is limited, we observed that during mycelial growth in liquid culture there is a more prevalent phenotype for strong expression, as evidenced by fluorescence and compound production.
Lastly, the Cre/lox-mediated integration platform in A. nidulans can be expanded in future works. Cre-mediated integration could be particularly relevant for building strains for heterologous expression of large genomic regions containing BGCs cloned by genome capture. 9 The current system could also be upgraded for simultaneous integration in different loci by using multiple landing pads with different heterospecific recombination sites. Different strategies could also be evaluated to optimise the delivery of Cre recombinase, such a selfexcising Cre expression cassettes. 22 The use of a split pyrG marker between recombination sites could be evaluated to minimise the background by random integration and abortive colonies, or to further evaluate AMA1-based donor vectors. 45 To sum up, Cre/lox-mediated integration of BGCs has the potential to speed up the process of constructing strains to produce heterologous metabolites in A. nidulans. The capability to uptake and maintain complex exogenous DNA is a key requirement for a good chassis organism for bioproduction 46 . Thus, this system can be used to upgrade other filamentous fungi chassis.

Vector Construction
All the oligonucleotides used for vector construction are listed in Appendix 1

Flow cytometry
Spores were collected with a sterile loop from colonies grown for 3 days in solid GMM u-p+r+ and suspended in water. The spore suspension was carefully filtered through a syringe containing sterile cotton to remove residual mycelia and diluted to a concentration of ~1.10 6 spores/mL. Data acquisition was performed immediately after spore suspension using a FACSCalibur (BD Biosciences) flow cytometer operated with filtered water as shear fluid.
mCherry signal was observed with a 488 nm excitation laser and the filter FL3 ( ≥670 nm) and 35,000 events were detected per measurement or 4 minutes of run were collected for the water control (no spore sample was run more than 4 minutes).
The data was processed using FlowJo V10 software (TreeStar). The output was gated according to FCS size to limit to the size range of spores, an example of the gating strategy is indicated on the Figure           We also observe more colonies in the plates with no selection pressure (~30 colonies) than in the plates with selection pressure (21-25 colonies) , which is indicative of the genetic stability of AMA1-pyrG. Plates were incubated at 37 °C for three days.