In vivo assembly of DNA-fragments in the moss, Physcomitrella patens

Direct assembly of multiple linear DNA fragments via homologous recombination, a phenomenon known as in vivo assembly or transformation associated recombination, is used in biotechnology to assemble DNA constructs ranging in size from a few kilobases to full synthetic microbial genomes. It has also enabled the complete replacement of eukaryotic chromosomes with heterologous DNA. The moss Physcomitrella patens, a non-vascular and spore producing land plant (Bryophyte), has a well-established capacity for homologous recombination. Here, we demonstrate the in vivo assembly of multiple DNA fragments in P. patens with three examples of effective genome editing: we (i) efficiently deleted a genomic locus for diterpenoid metabolism yielding a biosynthetic knockout, (ii) introduced a salt inducible promoter, and (iii) re-routed endogenous metabolism into the formation of amorphadiene, a precursor of high-value therapeutics. These proof-of-principle experiments pave the way for more complex and increasingly flexible approaches for large-scale metabolic engineering in plant biotechnology.

transformation with a linearized vector. The short length requirement for efficient homologous recombination is a significant technological advance, since short (12)(13)(14)(15)(16)(17)(18)(19)(20) base overlaps are convenient to incorporate into PCR primers. Complex in vivo assembly of multiple DNA fragments is a routine procedure using S. cerevisiae, contributing greatly to its extensive use as a synthetic biology and biotechnology host [19][20][21] , therefore the development of a similar methodology for P. patens can further increase its utility as a plant biotechnological chassis. Here, we explore direct PEG transformation of P. patens using PCR products containing short overlapping sequences, and show that it is a robust method for genome engineering.

Results and Discussion
To demonstrate uptake, assembly, and correct genomic integration of multiple DNA fragments in moss, we focused on three examples of P. patens transformation related to plant biotechnology. We utilized our published vector pBK3 that was established in order to study terpenoid biosynthesis through the knock-out of the native P. patens bifunctional copalyl diphosphate/kaurene synthase (CPS/KS) gene 7,22,23 . The vector replaces CPS/KS with the gene for the enhanced yellow fluorescent protein variant SEYFP-F46L (Venus) 24 under hygromycin selection (Fig. 1a). Using this vector as a template, we amplified three DNA sequences (Fig. 1b) that cover the whole insert region, each having a sequence overlap with the one next to it. Four different overhang lengths (12,20,50, and 100 base pairs) were examined.
A correct transformation event results in stable P. patens lines that have three distinct phenotypes. They are resistant to hygromycin, express Venus from the CPS/KS promoter, and lack ent-kaurene derived diterpenes (Fig. 1c,d). We transformed P. patens with linear pBK3 vector and with equimolar amounts of the PCR fragments produced from the same vector. All the PCR fragments were also transformed individually as negative controls (Supplemental Fig. 1). As the coding sequences of both the Venus reporter and the hygromycin resistance are split on different DNA fragments, and none of the fragments contain a full-length coding sequence, the DNA repair system of P. patens must assemble the three fragments via homologous recombination to gain all the phenotypes described above. We recovered 12-21 (Table 1) antibiotic resistant lines for the different transformation approaches, the linear vector-transformation and the in vivo assembly (Fig. 1c). We then screened all of these resistant lines for fluorescent protein expression. Transformations using single fragments did not result in plants expressing Venus or surviving antibiotic selection. Accumulation of Venus was observed in both protonemal tissue (Fig. 1c) and gametophores (Supplemental Fig. 2). However, accumulation of Venus was only observed in rapidly dividing cells, such as protonemal tips and the core of the gametophore, tissue giving rise to the sporophyte. This suggests that the CPS/KS enzyme and ent-kaurene related metabolites are associated with cell division and differentiation, providing further insights in the functionality of ent-kaurenes in P. patens and other bryophytes. We performed metabolite profiling of the fluorescent lines, and confirmed the lack of detectable diterpenoids in 6-11 lines using the different transformation approaches (Table 1, Fig. 1d). In order to verify the correct insertion of the transgenes, we genotyped the lines that were both fluorescent and lacking ent-kaurenes, (Fig. 1e). Sequence information was obtained for the regions where the DNA assembly occurred. Two fully genotyped lines were recovered showing the appropriate phenotypes with linear vector-transformation and 4 lines of the different in vivo assembly approaches, the fragments were correctly assembled and integrated in the targeted genetic locus, and lacked the native CPS/KS gene. The correctly integrated and assembled transformants (for both linear vector and multi-fragment transformation) correspond to 10-22% of the antibiotic resistant lines. The transformation results and the number of obtained lines are summarized in Table 1.
We also generated two novel constructs that were designed for integration in the well-established neutral locus Pp108 that allows efficient genomic integration in P. patens 3 . Here, we amplified the overlapping DNA fragments, adding 27-30 base overhangs via PCR primers for transformation without pre-assembly of vectors in E. coli. As inducible systems are valuable for tuning of heterologous expression, we studied the P. patens sodium inducible promoter, PpENA1 25 with Venus as a reporter gene. For selection, this three-fragment assembly encoded a geneticin (G418) resistance (nptII) cassette (Supplemental Fig. 3). We obtained 32 fluorescent lines, of which six were assembled correctly. The promoter activity was measured in plants growing on media supplemented with NaCl (0-100 mM). We observed NaCl induced Venus expression (Fig. 2a), which reached a plateau at a concentration of 60 mM NaCl (Fig. 2b). These results are consistent with previously reported data on this promoter 25 .
Finally, we expressed the gene for the amorphadiene synthase AaADS, a sesquiterpene synthase from Artemisia annua that catalyzes the first committed step of the biosynthesis of the antimalarial drug artemisinin 26 . AaADS was expressed coupled to Venus using the LP4 linker (a linker peptide from a polyprotein precursor of Impatiens balsamina, which results in expression of two separate proteins 27,28 ). The promoter driving the fused genes was the UBIQUITIN1 promoter from Zea mays (ZmUBI1), and the selection was again based on geneticin resistance. Collectively, these PCR products gave a four-fragment assembly (Supplemental Fig. 4). We analyzed the 10 lines showing antibiotic resistance for yellow fluorescence (Fig. 2c), and we subjected the resulting 9 fluorescent lines to GC-MS analysis to confirm biosynthesis of amorphadiene by the AaADS (Fig. 2d). Correct insertions in the Pp108 locus and assembly of both the PpENA1 and the AaADS constructs were confirmed for three independent lines via genotyping. Assembly and integration were tested by PCR-based genotyping and sequence-verification (Supplemental Figs 5 and 6). In these cases, assembly was demonstrated by appropriate phenotypes and PCR confirmation of internal junctions, however genomic integration could not be verified. This suggests the possibility that assembled DNA was maintained episomally, without integration to the genome.
In all three transformation experiments, we acquired plant lines that contain the desired genomic sequence, assembled correctly and inserted in the targeted genomic locus ( Table 1). The evaluation of the transformant phenotypes was performed sequentially, i.e. resistant lines were screened for fluorescence, followed by metabolite profiling using GC-MS, before genotyping. The transformation efficiency and correct integration were similar for the multi-fragment transformations and the traditional transformation using a linearized vector. Previous studies indicated that the specific targeted locus and the presence of repeating elements are possibly more important determinants of the transformation efficiency than the length of the homology regions 17,29 .
Previous studies also reported incomplete and non-targeted integration. Episomal concatamers and tandem insertions are common events, and were suggested as result from non-homologous end-joining. In addition, one-sided integration and random insertions were described 18,29,30 . Consistent with such non-targeted and incomplete integration events, we observed antibiotic resistant plant lines lacking one or more of the independent phenotypes used (Table 1). However, the three examples of transformation presented here demonstrate the ability of P. patens to take up, assemble and correctly integrate up to 9.2 kilobases of DNA directly into the genomic target site. This new transformation technology provides an easy way to assemble DNA fragments in a plant system, minimizing the need for laborious sub-cloning and gene construction in E. coli or S. cerevisiae vectors. Moreover, using this technique allows for flexible genome editing, combinatorial assembly of DNA parts, and engineering of complex metabolic pathways, which are all important for high-throughput synthetic biology studies.
Improved techniques for targeted, controlled, and large-scale genome manipulation of plant systems are highly valuable and could facilitate understanding and improving genetic structure and function at the chromosome and genome levels. P. patens is already extensively used for both basic research and applied biotechnology. The implementation and optimization of in vivo DNA assembly methodologies can significantly improve the potential of P. patens as a biotechnological host species and as a photosynthetic chassis for synthetic biology.

Transformation of P. patens protoplasts.
Protoplasts were prepared and transformed using a standard PEG method 14,15 . More precisely, approximately 1.5 g in fresh weight (three Petri dishes) of 5 days cultured P. patens was digested with a 0.5% Driselase ® enzyme solution in 8.5% mannitol (Sigma D9515), using one ml of solution for every 40 mg P. patens tissue and incubated with gentle agitation for 30-60 minutes at room temperature. The digested tissue was filtered through mesh with pore size of 100 μ m. Protoplasts were pelleted by centrifuging at 150-200 × g for 5 minutes with slow breaking. The supernatant was discarded and the pellet was washed twice with protoplast wash solution (8.5% mannitol, 10 mM CaCl 2 ). Protoplast density was measured using a hemocytometer, followed by centrifuging and resuspension in MMM solution (9.1% D-mannitol, 10% MES and 15mM MgCl 2 ) giving a protoplast concentration of 1.6 × 10 6 cells/mL. 300 μ L of protoplast suspension and 300 μ L of PEG solution were added to a 15 mL tube containing the DNA. For both linearized plasmid and multi-fragment transformations, 20 μ g total DNA was used. For the multi-fragment transformations, fragments were added in an equimolar ratio.
The mixture was incubated in a 45 °C water bath for 5 minutes and another 5 minutes at room temperature. Samples were diluted 5 times by adding 300 μ L of 8.5% D-mannitol with 1 min pauses between dilutions, followed by an additional 5 dilutions with 1 mL of 8.5% D-mannitol. The transformed protoplasts were pelleted by centrifugation and the supernatant was discarded. The protoplast pellet was resuspended in 500 μ L of 8.5% D-mannitol and 2.5 mL of protoplast regeneration media (top layer; PRMT). 1 ml of the suspension was dispensed on each plate (three plates in total) containing protoplast regeneration media (bottom layer; PRMB) overlaid with a cellophane disc. The plates were incubated for 5-7 days. The cellophane with regenerating protoplasts was then transferred on PhyB media containing appropriate selection for two weeks.
Transformations were moved to media without antibiotics for another two weeks to allow loss of unstable and non-integrated DNA, and then plants were moved back to a final antibiotic selection for two weeks at which point they were considered stable. A detailed protocol can be found in references 14,15 . DNA parts and genes. Genomic sequences and annotations of P. patens were retrieved form cosmoss.org.
The pBK3 vector 7 was used to generate the Ppcps/ks-knockout plants. The PpENA1 promoter and the homologous recombination flanking regions of the Pp108 locus were amplified from P. patens genomic DNA. The AaADS  containing NaCl (0 mM (non induced), 20 mM, 60 mM, and 100 mM) with wild-type as a control. Micrographs were captured 24 hours after induction. Fluorescence intensity of approximately 25 replicates from each treatment was quantified with the software program Fiji (ImageJ). The Ppcps/ks knockout and PpENA1-Venus lines were further analyzed with a Leica TCS SP5 II spectral confocal laser-scanning microscope (Leica Microsystems, Heidelberg, Germany). The fluorescence was quantified with a 63x/1.2 numerical aperture water immersion objective for the detection of yellow fluorescence. A sequential scanning was performed for the PpENA1 lines, to detect the fluorescence of the Venus protein, whose expression was driven by the inducible PpENA1 promoter. The excitation was set at 488 nm allowing for the detection of chlorophyll at 590-680 nm. This was followed by excitation 515 nm that allow for the detection of Venus at 520-560 nm. The maximum intensity was set as the relative fluorescence exhibited from cells treated with 100 mM NaCl.
Metabolite profiling. The metabolite profile of the transgenic lines was examined using a Shimadzu GCMS-QP2010 Plus (GC-2010). Samples were prepared as previously described 14 . To screen for biochemical knockouts, one microliter of hexane extract was injected in split-less mode and separated with HP-5MS UI column (20.0 m × 0.18 mm × 0.18 μ m) with hydrogen as a carrier gas. The GC program was as follows: 60 °C for 3 min, 60~150 °C at 30 °C/min, 150~210 °C at 10 °C/min, 210~320 °C at 30 °C/min and was held for 5 min. The profile of AaADS lines was determined by HS-SPME (Headspace-Solid Phase Micro-Extraction) 31,32 . The identity of amorphadiene and ent-kaurene was confirmed by comparison of the obtained mass spectrum and retention index with those found in the literature 33  DNA isolation. DNA for PCR analysis was isolated as previously described 34 , with minor modifications.
Briefly, fresh P. patens tissue (approximately 10-20 mg) was collected in 1.5 ml microfuge tubes, frozen in liquid nitrogen, then ground using a plastic pestle. 400 μ L of extraction buffer (50 mM Tris·HCl pH 8.0 and 20 mM EDTA pH 8.0) was added followed by 80 μ l of 10% SDS. The mixture was gently vortexed and incubated at 65 °C for 30 min. 180 μ L of 3M NaAc pH 5.2 was added to the mixture and samples were incubated on ice for 20 min. Samples were centrifuged at 15,000 × g for 10 min to pellet debris. The supernatant was transferred to a new tube containing an equal volume of isopropanol and was mixed by inversion. DNA was pelleted by centrifugation at 15,000 × g for 30 min. The supernatant was discarded and the pellet was washed with 80% ethanol and allowed to air dry before the final resuspension in 50 μ L water.