A novel endogenous selection marker for the diatom Phaeodactylum tricornutum based on a unique mutation in phytoene desaturase 1

Phaeodactylum tricornutum is a well-developed model diatom for both marine ecology and microalgal biotechnology, which has been enabled by the sequenced genome and the availability of gene delivery tools, such as biolistic transformation and E. coli-mediated conjugation. Till now, these tools have mainly relied on two selectable markers of bacterial origin which confer resistance to antibiotics Zeocin and nourseothricin. An alternative cost-effective and preferably endogenous selectable marker would facilitate gene stacking efforts through successive transformation or conjugation. We performed UV-mutagenesis of P. tricornutum to obtain mutations in the phytoene desaturase (PDS) gene, conferring resistance to the bleaching herbicide norflurazon. Two mutants displaying high tolerance to norflurazon and carrying unique mutations in PtPDS1 (PHATRDRAFT_45735) were selected. These mutants revealed novel point mutations at a conserved residue Gly290 to Ser/Arg. Homology-based structural modeling of mutated PDS1, over a resolved crystallographic model of rice PDS1 complexed with norflurazon, suggests steric hindrance by bulkier residue substitution may confer herbicide resistance. We report the characterization of PtPDS1 mutants and the development of the first endogenous selectable marker in diatoms suitable for industrial strain development, with the added benefit of biocontainment. The plasmid carrying the mutated PDS1 as a selection marker and eGFP as a reporter was created. An optimized biolistic transformation system is reported which allowed the isolation of positive transgenic events at the rate of 96.7%. Additionally, the ease of in vivo UV-mutagenesis may be employed as a strategy to create PDS-norflurazon-based selectable markers for other diatoms.


Results
Determination of an optimal norflurazon selection concentration. For the development of an endogenous selection marker, suitable for genetic engineering in P. tricornutum, we focused on the resistance to the herbicide norflurazon. The sensitivity of WT P. tricornutum to increasing concentrations of norflurazon (0.25-5 μM) was tested by plating cells on RSSE-agar and incubation for 12 days followed by microscopy of Aniline Blue stained cells to assess cell damage as indicated by the cellular permeabilization of the stain (Fig. 1a,b). Uniform Aniline Blue permeabilization and staining of treated WT P. tricornutum cells was observed at 1.5 µM and higher norflurazon concentrations, which was associated with a breakdown of cellular membranes, a desintegration of the chloroplast and a loss of cell viability ( Supplementary Fig. S1). Norflurazon-treated cells that were re-plated for recovery on the RSSE-agar medium free of norflurazon displayed 100% mortality at 2.5 µM and higher herbicide concentrations after 15 days of incubation (Fig. 1c). Based on these observations, selection was tested at 3 and 5 µM norflurazon concentrations in biolistic transformation experiments.

Isolation of P. tricornutum norflurazon-resistant PtPDS1 mutants and their characterization.
Based on the results above and a preliminary assay of WT P. tricornutum sensitivity to norflurazon in liquid media (not shown), we chose 15 µM norflurazon concentration to enable the selection of highly resistant mutants and to limit escapes. Accordingly, UV-mutagenized colonies, obtained after exposure to different durations of UVB (see materials and methods), were selected on solid media containing 15 µM norflurazon. After 15 days, the two fastest growing norflurazon-resistant mutants (M1 and M2) were recovered from a 2-min UVB exposure.
A preliminary characterization of the M1 and M2 mutants was done in a 12-well plate, where mutants were exposed to a range of norflurazon concentrations between 10 and 40 µM. At the highest concentration of 40 µM norflurazon, M1 displayed chlorosis, whereas M2 was completely bleached at the end of day 20 of incubation ( Fig. 2a). Both mutants displayed tolerance to norflurazon concentrations lower than 30 μM. Next, we compared the growth of mutants and WT in a column photobioreactor, with and without 5 and 10 µM norflurazon. Mid-log phase cultures were diluted to chlorophyll 2 mg L −1 at the start of experiment; chlorophyll and cell density were estimated by recording A680 and A720, respectively. By the end of experimental period, norflurazon-treated WT cells displayed greater Aniline Blue staining than their treated M1 and M2 counterparts (Fig. 2b), revealing greater WT sensitivity to herbicide and tolerance in the mutant lines. In the presence of 10 µM norflurazon, M1 and M2 mutants displayed a specific growth rate of 0.211 and 0.141 Abs-680 day −1 , respectively, whereas chlorophyll production in WT cultures was completely arrested at 48 h, in the presence of 5 μM norflurazon (Fig. 2c). Untreated WT displayed slightly better chlorophyll accumulation than the M1 or M2 cultures in each experimental repetition. A similar trend was observed for cell density (A720) in norflurazon-treated and untreated cultures (Fig. 2d).
Phylogenetic and sequence analysis of PtPDS1 mutants. The plastid-targeted P. tricornutum PDS1 displayed the highest homology to P of other diatoms (Fig. S2). Within the phylogenetic tree for chloroplastic phytoene desaturases, P. tricornutum PDS1 clustered more closely to early divergent green microalgae www.nature.com/scientificreports www.nature.com/scientificreports/ (Prasynophytes), which form a clade separate from higher green algae (Chlorophytes) and land plants. Brown algal PDS peptide sequences were the most divergent and were used to root the phylogenetic tree.
Alignment of the PtPDS1 and OsPDS1 (GenBank accession number AF049356) peptide sequences showed a 59.51% sequence identity and a sequence coverage of 76%. Alignment between secondary structures of PtPDS1 and OsPDS1 showed highly similar topologies (Fig. 3a). The homology-based modeling of PtPDS1 over the OsPDS1 crystallographic structure (PDB: 5MOG.A, Fig. 3b) had a QMEAN value of −1.98 indicating high structural similarity and high accuracy of the model. Phe145 and Arg283 (Fig. 3c) residues are known to interact with norflurazon and allow its stable binding and inhibition of the phytoene desaturase enzyme. The PtPDS1 modeled structure overlaid on norflurazon-inhibited OsPDS1 indicated a clash between Gly290Ala (M1) and the backbone carbonyls of Arg283 and Glu287, which may interfere with residues stabilizing norflurazon in WT phytoene desaturase-1 enzyme (Fig. 3c), which may explain the basis of norflurazon resistance in mutants. optimization of P. tricornutum transformation with the pPtPDS1-M1 construct. Next, we used the PDS1 sequence from mutant M1 to construct a pPtPDS1-M1vector (Fig. S4), harboring eGFP as a reporter (for details, please refer to Materials and Methods). An efficient norflurazon concentration, suitable for transformant selection, would maximize the selection of transgenic events and yet minimize the occurence of false www.nature.com/scientificreports www.nature.com/scientificreports/ positives. Transformation efficiency and the rate of false positive events were compared at 3 and 5 µM norflurazon selection post-biolistic delivery of pPtPDS1-M1 expression vector (Fig. 4).

Validation of transgene integration and expression.
Evaluation of long-term expression stability of transgenic lines. Transgenic lines (T1, T2 and T3), maintained without selection for about 48 months, representing a range of transgene copy number integration were characterized for norflurazon sensitivity and eGFP expression. Transgenic lines cultured in column photobioreactor, with or without 5 µM norflurazon displayed robust growth over three repetitions of the experiment (Fig. 6a,b). In the presence of 5 µM norflurazon, transgenic lines displayed minimal inhibition of growth when compared to untreated controls, including wild-type. Furthermore, transgenic lines displayed tolerance over a range of norflurazon concentrations (10-120 µM) and remained uninhibited in a plate-based assay of norflurazon www.nature.com/scientificreports www.nature.com/scientificreports/ sensitivity (Fig. 6c). Integration of eGFP and PtPDS1-M1 was validated by PCR ( Supplementary Fig. S5a). Stable eGFP expression in all three transgenic lines was validated by fluorescence microscopy ( Supplementary Fig. S6b).

Discussion
The diatom P. tricornutum has developed into the primary model microalga for biotechnological exploitation, specifically for lipids, LC-PUFA and recombinant protein production, and a wide range of molecular tools used in developing this diatom into a valuable synthetic biology chassis have been reported 15,39 . UV-mutagenesis is a simple method for the introduction of dominant genic mutations in a wide range of microorganisms. In P. tricornutum, UV-mutagenesis has been successfully employed to enhance EPA content up to 44% 40 . Here, we report a novel dominant endogenous selection marker in diatoms for the transformation of P. tricornutum, based on a novel amino acid substitution in PtPDS1 (PHATRDRAFT_45735; P. tricornutum JGI database v2.0) conferring resistance to the bleaching herbicide norflurazon.
Through random or targeted mutagenesis of the PDS enzyme, norflurazon resistance has been successfully developed as an endogenous selectable marker in several algae [41][42][43][44][45][46][47] and plant species 48,49 . With respect to the PtPDS1 sequence, five amino acid substitutions at highly conserved residues in the PDS enzyme that are known to confer resistance to norflurazon have been reported so far: Phe145Val 43 , Arg283Pro/Ser/Thr/Cys 49-52 , Leu403Pro 34 , Val505Gly 34,53,54 , and Leu550Arg/Phe 41,42,[44][45][46][47] . These residue substitutions were all shown to occur in the vicinity of the reaction center comprising the FAD-binding site and the plastoquinone/norflurazon binding pocket in the Rice PDS (OsPDS1) crystal structure 33,55 . In vivo UV mutagenesis and stringent selection allowed isolation of norflurazon-resistant mutants of P. tricurnutum with novel substitutions at a highly conserved residue, Gly290. Homology modeling of the PtPDS1 peptide over the OsPDS1-elucidated structure, suggests that Gly290Ala/Ser substitutions introduce bulky residue that clash with the backbone carbonyls of Arg283 and   www.nature.com/scientificreports www.nature.com/scientificreports/ Glu287, which may destabilize norflurazon interaction with phytoene desaturase enzyme and likely result in norflurazon resistance.
Nuclear transformation in P. tricornutum is routinely achieved by two selectable markers of bacterial origin, Sh-ble-Zeocin ™ and nat-nourseothricin. Apart from Sh-ble and nat, nptII-geneticin and cat-chloramphenicol are the only other selectable markers that have been used with less success in transformation of P. tricornutum and therefore not used routinely. Use of alternate selectable markers such as nptII and nat are fraught with difficulties, such as being effective only at higher concentrations and at reduced salt strength in the selection medium. Moreover, an array of commonly used antibiotic selection agents, such as kanamycin, streptomycin, spectinomycin, and hygromycin are ineffective as selection agents in P. tricornutum 22,56 . Commercial herbicides and their corresponding resistance genes provide a cost-effective and efficient selection scheme when genetically encoded antibiotic resistance markers are unavailable. However, widely used herbicides, such as glyphosate, bialaphos, phosphinothricin, sulfometuron-methyl, chlorosulfuron, and imazapyr, have proven to be ineffective in killing P. tricornutum 22 . Due to this dearth of efficient selectable markers, PDS-norflurazon resistance based selection system was explored in P. tricornutum. In contrast to selectable markers of bacterial origin, endogenous PDS1 gene, resistant to norflurazon provides additional benefits of efficient selection, biocontainment and cost-effectiveness.
When compared on a cost per liter basis for producing one liter of selective medium, norflurazon, Zeocin ™ and nourseothricin, respectively, are ca. USD 2 (Merck, USA), 20 (InVivoGen, San Diego, USA) and 3000 (Merck, USA). In order to develop PtPDS1-M1 into an efficient selectable marker, optimization of the selection pressure on biolistic-transformed cells was carried out at 3 and 5 µM norflurazon concentrations. Selection at 5 µM norflurazon allowed significant improvement in selection of positive transgenic lines at the rate of 96.7% of total colonies www.nature.com/scientificreports www.nature.com/scientificreports/ obtained. Similarly, non-transgenic escapes were limited to 3.8% of the total colonies screened. With 5 µM norflurazon selection, on an average each microprojectile coating preparation produced 35 transformants as compared to ~100 with sh ble-Zeocin ™ ; this may be attributed to the PtPDS1-M1 CDS length of 1875bp, which is five times that of sh-ble CDS.
The Phaeodactylum tricornutum genome contains two homologs of the phytoene desaturase enzyme viz. PtPDS1 (PHATRDRAFT_45735) and PtPDS2 (PHATRDRAFT_55102), of which only the former is functional 57 . A phylogenetic analysis of the PtPDS1 peptide sequence revealed close similarity to other diatom PDS sequences, which is in agreement with Dambek et al. 57 . Based on the endosymbiotic history of diatom chloroplasts, it would be expected that plastid-targeted nuclear-encoded proteins would share greater homology with their red algal endosymbiont ancestor; however, a recent study of the genes involved in the carotenoid biosynthesis of the chromist algae revealed that about two-thirds of these genes are closely related to the prasinophytes 58 . Retention of sequences from a prasinophyte source has been implicated in the evolution of fitness in diatoms and their capability to dominate the marine environment 59 . PtPDS1 clustered to a clade comprising diatoms, cryptomonads, and prasinophytes, which displayed divergence from the clade comprising the green algae, and the lower and higher plants, in agreement with Frommolt et al. 58 .
Norflurazon-resistant PDS mutants with instances of significant impairment of in vivo carotenogenic activity upon PDS mutagenesis have been reported 34,53,54 . In vitro characterization of mutations Leu550Phe/Arg and Phe145Val in rice PDS, under normal conditions, showed less than 5% activity compared to WT rice PDS except for Arg283Ser, which retained about 15% activity 55 . It has been proposed that such mutants may compensate for reduced carotenogenic activity by transcriptional up-regulation via retrograde signaling to the nucleus. On the other hand, norflurazon-resistant mutants without major retardation of carotene biosynthetic capability have been isolated 51 . Such mutants are preferable for exploitation as selectable markers as they would neither significantly alter cellular transcription nor adversely affect transgene expression. The two norflurazon-mutants, M1 and M2, displayed growth rates, comparable to WT under untreated conditions. The specific growth rates of WT, M1 and M2 lines displayed significant retardation in the presence of a killing concentration of norflurazon (10 µM), to 0%, 65% and 36%, respectively, when compared to their untreated controls.
Consequently, we used the PDS1 sequence from mutant M1 to develop a novel selectable marker for P. tricornutum. The established transformation vector pPha-T1 21 expressing eGFP was modified by replacement of the ShBle selectable marker with PtPDS-M1 to create a test vector for nuclear transformation, pPtPDS1-M1 (Fig. 4a). After confirmation of successful transformation, a new transformation vector, pBS-PtPDS1-M1 (Fig. 4b), was created by ligation of an excised PtPDS1-M1 selectable marker cassette into the SpeI restriction site of the pBlue-Script SK II (+) vector with extensive multiple cloning sites for insertion of additional transcriptional units.
Stable integration and expression of transgenes is essential for a transformation system to be considered reliable. To this end, three transgenic lines maintained over 48 months without selection were subjected to norflurazon sensitivity assay in plate and growth characterization in column photobioreactor. These transformants displayed enhanced tolerance range with slight inhibition at 120 µM norflurazon which is nearly three times the upper limit for mutant M1. Similar results were observed in the column photobioreactor, with minimal growth inhibition for transgenic lines in the presence of the herbicide. These results indicate that the PtPDS1-Gly290Ala is a functional enzyme, which displays resistance to norflurazon.
Identification of a novel endogenous dominant selection marker for the model species P. tricornutum is of great significance towards advancing synthetic biology approaches in this versatile and extensively researched species. Availability of an endogenous selection marker will not only facilitate the selection of successful transformants but will also allow the creation of cisgenic P. tricornutum events provided that only native sequences are repurposed for genetic engineering, so that regulatory hurdles for the cultivation and marketing of such products may be overcome.

Conclusion
We demonstrate a simple in-vivo UV-mutagenesis strategy to isolate dominant mutants, viz. norflurazon resistance in the model diatom P. tricornutum. These novel mutants were characterized for herbicide resistive strength, and the coding sequence of PtPDS1 from M1 mutant with greater resistance was used to create the first successful endogenous selectable marker in P. tricornutum which is cost-effective, efficient and free of biocontainment concerns. In the future, domestication of type-IIS restriction sites within the selectable marker would make it a preferred selectable marker for use in large-scale transcriptional assembly methods such as Golden braid 60 or Mobius assembly 61 for engineering synthetic pathways in P. tricornutum.

Materials and Methods
Algal strain and culture conditions. An axenic culture of P. tricornutum UTEX 646 was maintained on

In vivo UV-mutagenesis of WT P. tricornutum and selection of norflurazon-resistant mutants.
The UV-mutagenesis of P. tricornutum cells was performed on a RSE-agar medium. Cells (3 × 10 7 cells) from an exponentially growing culture were spread uniformly onto a 90-mm diameter Petri dish and allowed to dry. The lid was removed, and cells were mutagenized by direct irradiation of the agar surface with UVB rays from a UV-transilluminator (312 nm; Wilber Lourmant, France) for 15 s, 30 s, 1 min, 1.5 min and 2 min. UV-irradiated plates were left for 24 h recovery at an illumination of 30 μmol photons m −2 s −1 at 22 °C before re-plating the cells on a RSE-agar plate supplemented with 15 μM norflurazon. After 15 days, obtained colonies were re-streaked onto a fresh RSE-agar plate supplemented with 15 μM norflurazon, and the two fastest growing norflurazon-resistant lines were selected for further characterization.
Preliminary determination of herbicide tolerance of norflurazon-resistant mutants. Mutants, cultured in liquid RSE medium until the mid-log phase, were diluted to 5 × 10 5 cells mL −1 and transferred to a 24-well plate and challenged with norflurazon in the range of 10-40 μM. These cultures were incubated in an incubator shaker (120 rpm) at 18 °C and illumination of 75 μmol photons m −2 s −1 , in an atmosphere enriched with CO 2 (200 mL min −1 ). After 20 days incubation, tolerance to norflurazon was determined by visual observation. Preliminary experiments under the similar setup showed that these norflurazon concentrations are lethal for WT P. tricornutum.
Comparative growth characterization of WT and mutants with and without norflurazon. WT and mutant (M1 and M2) cultures, grown to mid-log phase, were diluted to a chlorophyll concentration of 2 mg L −1 and cultivated in cylindrical glass columns in a Multi-Cultivator MC 1000-OD (Photon Systems Instruments, Czech Republic) with different norflurazon concentrations (0, 5 and 10 μM) at 22 °C, a light intensity of 50 μmol photons m −2 s −1 , and bubbling with filtered air enriched with 2% CO 2 . Chlorophyll and cell density were recorded in situ, every 10 minutes for 93 h, via optical density measurements, at 680 nm and 720 nm, respectively. The experiment was repeated three times, and the average values are reported. At the end of the experiment, cells were observed microscopically under a bright field and were also stained with Aniline Blue to assess cellular damage. phylogeny of the P. tricornutum phytoene desaturase-1 enzyme. Homologs of the P. tricornutum phytoene desaturase (PtPDS1; EEC48362.1) were searched against the NCBI database using the BLASTp algorithm 64 and chloroplastic phytoene desaturase/dehydrogenase peptide sequences, representing a diverse phylogenetic group. Alignments were made using EMBL-EBI Clustal Omega 65 . Phylogenetic trees were built and tested using the ATGC: PhyML 3.0 server 66 with Smart Model Selection 67 . Within PhyML 3.0, BIONJ 68 was used to build a bootstrap tree with Subtree-Pruning-Regrafting 69 ; tree improvements and branch support bootstrapping were set to 1000. Species with bootstrap values below 500 were iteratively dropped to arrive at the final phylogenetic tree.
Secondary and tertiary structure analysis of PtPDS1 peptide sequence. The PtPDS1 tertiary structure was modeled on Oryza sativa phytoene desaturase-1 (OsPDS1; PDB: 5MOG.A) with SWISS-MODEL 70 . Secondary structure alignments rendered above the amino acid sequence alignment, using the ESPript 3.0 server 71 . The modeled tertiary structure of PtPDS1 was superimposed on the OsPDS1 chain A complexed with norflurazon, and the effect of amino acid substitution leading to norflurazon tolerance was analyzed using Coot 72 .
Biolistic transformation of P. tricornutum. The WT P. tricornutum strain UTEX 646 culture in mid-log phase was diluted with an equal volume of RSE medium and grown for another three days. On the day of transformation, cells were pelleted by centrifugation (2200 × g for 5 min), and 3 × 10 7 cells were spread on each RSE-agar plate. The plates were left to dry in the laminar hood for 1.5 h before particle bombardment. Tungsten-M17 particles (Bio-Rad) were coated with 5 µg of DNA, as per the manufacturer's instructions. Each DNA-coated micro-projectile preparation was spent in two shots. Biolistic gene delivery was performed using the PDS-1000/ Transformation optimization experiments were compared by t-test at α = 0.05 in Microsoft excel using Data AnalysisToolPak.
Southern hybridization of purified genomic DNA from transgenic lines was performed to confirm integration of pPtPDS1-M1 DNA in the P. tricornutum genome. A wet pellet weighing 3 g was lyophilized and subsequently used for high quality genomic DNA purification following the Dellaporta protocol 73 . KpnI-digested genomic DNA (9 µg) was resolved on a 1% agarose gel and neutral blotted to a HyBond-N + nylon membrane (GE Healthcare, Little Chalfont, UK) for Southern blotting using a DIG High Prime DNA Labelling and Detection Starter Kit I (Roche, Penzberg, Germany). Templates for DIG labeling were PCR amplified and gel-eluted. A template for the eGFP probe (606 bp) was amplified with eGFP-F and eGFP-R primers. Similarly, a template for the PtPDS1 probe (581 bp) was amplified with PtPDS1-F and PtPDS1-R primers, followed by the addition of XhoI restriction enzyme (New England BioLabs, Ipswich, USA) to the PCR reaction (incubated at 37 °C for 2 h), and gel-elution of a 581 bp fragment. DIG-labeled probes were hybridized at 54 °C overnight and stringency washed at 68 °C.
Western blot analysis. Expression of eGFP in transgenic P. tricornutum lines was validated by probing a western blot of 20 µg crude protein extract resolved on a 10% SDS-PAGE gel with an anti-TAG(CGY)FP antibody (Evrogen, Moscow, Russia) at a dilution of 1:5000 and a secondary antibody, goat anti-rabbit IgG conjugated to horseradish peroxidase (Bio-Rad, Hercules, USA) at a dilution of 1:3000. Chemiluminescence development of the western blot bound antibody was done using an EZ-ECL Kit (Biological Industries, Beit Haemek, Israel) according to the manufacturer's instructions and imaged with a MicroChemi detection system (DNR Bio-Imaging Systems Ltd., Jerusalem, Israel).

Fluorescence microscopy.
In vivo characterization of lines expressing eGFP was verified by fluorescence microscopy with filter sets 38 HE and 16 for visualization of eGFP and chlorophyll auto-fluorescence, respectively, using a Zeiss Imager A2 microscope (Carl Zeiss MicroImaging Inc., Germany) with a Zeiss AxioCamMRc mounted digital camera.