Blasticidin-S deaminase, a new selection marker for genetic transformation of the diatom Phaeodactylum tricornutum

Phaeodactylum tricornutum culture media

One liter of modified f/2 medium (Guillard, 1975) was prepared by dissolving 16.6 or 8.3 g of “Tropic-Marin CLASSIC” artificial sea salt (Dr. Biener GmbH, Wartenberg, Germany) in demineralized water (resulting in one half or one quarter of the salinity of natural sea water). After autoclaving the standard f/2 supplements (Guillard, 1975) and additionally one ml of 2M Tris–HCl pH 8.0 were added. For preparation of solid media, 12 g bacto agar (Becton; Dickinson and Company, Le Pont de Claix, France) per liter of salt water were added prior to autoclaving. After addition of the supplements, plates each containing 25 ml of solid medium were prepared in Petri dishes (633181; Greiner Bio-One, Kremsmünster, Austria).

Phaeodactylum tricornutum selection media

Zeocin (Thermo Fisher, Waltham, MA, USA): 750 μl of a 100 mg/ml solution of Zeocin was added to one liter of the f/2 medium to a final concentration of 75 μg/ml. Nourseothricin (clonNAT; WERNER BioAgents, Jena, Germany) was prepared as a stock solution of 100 mg/ml in water and added to the f/2 medium a final concentration of 150 μg/ml. Blasticidin-S (R21001; Thermo Fisher, Waltham, MA, USA) and tunicamycin (Abcam, Cambridge, UK) were prepared as stock solutions of 10 mg/ml in water. Zeocin, nourseothricin, blasticidin-S and tunicamycin stock solutions were added in the desired quantities to the media after autoclaving and prior to preparing the culture plates, or prior to the inoculation in case of liquid selection media. Voriconazole (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in dimethyl sulfoxide as stock solution of 20 mg/ml. Each voriconazole plate was prepared by plating the appropriate amount of the stock solution onto culture plates prepared as described above.

Transformation of P. tricornutum by particle bombardment

Genetic transformation of P. tricornutum Pt4 (single colony derived from UTEX646 which was previously used by Apt, Grossman & Kroth-Pancic (1996); UTEX culture collection of algae, Austin, TX, USA) was performed as previously described (Apt, Grossman & Kroth-Pancic, 1996; Kroth, 2007) with three single ballistic shots for one attempt. Selection on blasticidin-S plates was performed at 70 μE (Biolux L 36/965; Osram, Munich, Germany) continuous light conditions and at 22 °C for 3–5 weeks.

Construction of the vectors carrying the resistance genes

Amino acid sequences of the tunicamycin resistance protein (tmrB, GenBank P12921.4; Harada et al., 1988) and the bsr (GenBank P33967.1; Kobayashi et al., 1991) were retrieved from GenBank. The amino acid sequences were reverse translated with the online-tool “Reverse Translate” of the “Sequence Manipulation Suite” (SMS, http://www.bioinformatics.org/sms2/rev_trans.html) (Stothard, 2000). Recognition sites of popular commercially available restriction enzymes were then manually eliminated (the final gene sequences are shown in Data S1). We then checked the codon efficiency of the genes with a custom generated codon usage table for P. tricornutum (Table S1), prepared in the following way: The coding sequences (cds) of the P. tricornutum gene catalog Phatr3 (Bowler et al., 2008; Rastogi et al., 2018) were downloaded from ftp://ftp.ensemblgenomes.org/pub/protists/release-39/fasta/phaeodactylum_tricornutum/cds/ (file name Phaeodactylum_tricornutum.ASM15095v2.cds.all_.fa). The 12178 cds were then screened for sequence lengths that are divisible by three. If necessary, sequences were processed to a sequence length that is divisible by three with help of the custom perl script “divisible_by_three.pl” (https://bitbucket.org/grubio-/divisible_by_three). 12177 cds had lengths that were either divisible by three, or could be truncated to complete codons in reference to a start or a stop codon, resulting in a total of 18112998 base pairs (compared to 18113425 base pairs in the complete Phatr3 cds catalog). Sequences were then concatenated into a single sequence with the “Combine FASTA” tool of a locally installed version of SMS (Stothard, 2000), and the codon table was generated using the tool “Codon Usage” of SMS (Stothard, 2000).

The reverse translated sequences were synthesized and cloned into a standard plasmid by Eurofins Genomics (Ebersberg, Germany). The genes were amplified by PCR (primers 01 and 02 for bsr and 03 and 04 for tmrB, Table S2) using KapaHifi polymerase (Roche, Basel, Switzerland).

The plasmid pPha-T1 (Zaslavskaia et al., 2000) was used as a template for a deletion PCR (primers 05 and 06, Table S2) to remove the ShBle gene and to introduce recognition sites for the restriction enzymes MssI and EcoRI. The resulting PCR product was ligated and delivered to E. coli XL1 Blue by electroporation, followed by selection on ampicillin plates. Plasmid preparation was done according to the manufacturer’s protocol using “QIAprep Spin Miniprep Kit” (QIAGEN, Hilden, Germany). The plasmid DNA was treated with the restriction enzyme MssI (Thermo Fisher, Waltham, MA, USA) and ligated with the amplified resistance genes using T4 DNA ligase (Thermo Fisher, Waltham, MA, USA). These constructs were genetically transformed to E. coli XL1 Blue by electroporation. Resulting colonies were screened by PCR for correct orientation of the insert. Extracted plasmid DNAs of single colonies growing on ampicillin plates were sequenced (Microsynth Seqlab, Göttingen, Germany) and used for the experiments with P. tricornutum.

Vector sequences and maps

Vector maps were generated with Geneious version 9 (Kearse et al., 2012). The sequences of the reverse translated resistance genes and of the whole plasmids can be found in Supplemental Information S1. The sequence of the pPTbsr vector is additionally available at GenBank (accession number MH541819) and the vector will be available at Addgene with the deposit number 117696.

PCR test for presence of bsr gene

A small amount of blasticidin-S resistant colonies (transformed with pPTbsr) was scraped off the plates, dissolved in 100 μl of water and heated to 99 °C for 10 min to serve as template. The PCR mixture contained five μl of this template, 0.1 μl of Taq DNA Polymerase E (Genaxxon bioscience, Ulm, Germany), 2.5 μl of 10× buffer, 0.25 μl of each primer (01 and 02 for the whole gene, 07 and 08 for a shorter fragment; 10 mM, Table S2), 0.25 μl of dNTPs (10 mM), 17 μl of H₂O. The PCRs were performed in a PCR cycler (ThermoCycler; Eppendorf, Hamburg, Germany) as described by the manufacturer of the DNA polymerase with annealing temperature of 54 °C and an extension time of 40 s.

Cross activities of marker genes

Nine of the blasticidin-S-resistant P. tricornutum colonies, transformed with the plasmid “pPTbsr” and nine nourseothricin-resistant colonies created by transformation of wild-type P. tricornutum with the plasmid “pNat” (Zaslavskaia et al., 2000), as well as five of the Zeocin resistant “lhcf1” cell lines generated by Chu et al. (2016) as well as two TGS1 and two TGS2 cell lines generated by Huang, Río Bártulos & Kroth, (2016) were used for the detection of possible cross-resistances against the antibiotics conferred by one of the tested marker genes.

Determination of effective concentrations of the antibiotics

A classical way to determine the lethal concentration of an antibiotic is a dose-response experiment, in which cells are subjected to increasing amounts of an antibiotic to determine the minimal efficient antibiotic concentration that is needed to inhibit cell growth. Accordingly, we tested the efficiency of the antibiotics by cultivating P. tricornutum cells on plates containing different concentrations of antibiotics. According to previous publications, a lethal dose of blasticidin-S of about two μg/ml was reported for mammalian cells (Izumi et al., 1991), for tunicamycin the LD₅₀ was reported between 0.1 and five μg/ml (Elbein, 1981), while voriconazole at 0.35 μg/ml impaired plant growth (Rozhon et al., 2013). For our experiments, we used concentrations ranges from higher to lower than these values (Table 1). We plated 2.5 × 10⁷ wild-type cells, and the growth of cells in each plate was determined after 3 and 5 weeks, classifying the results into “no growth”, “scattered growth”, or “diatom lawns” (Fig. 1; Table 1). Some antibiotics like G418 and nourseothricin are inhibited by high salt concentrations in marine media (Zaslavskaia et al., 2000). Blasticidin-S has been described by the manufacturer to be affected by the presence of salts, we therefore additionally tested a lower salt concentration in the medium (50% and 25% sea salt concentration). P. tricornutum is known to occur in places with highly variable growth conditions (Martino et al., 2007) and to tolerate a wide range of salt concentrations (Kräbs & Büchel, 2011).

Voriconazole, an inhibitor of P450 in plants (Rozhon et al., 2013), did not show any effect on P. tricornutum growth even at 105 μg/ml, which is about 300-fold the concentration of effective doses for most land plants (Rozhon et al., 2013). Therefore, this antibiotic was not investigated further. Blasticidin-S is an inhibitor of protein synthesis binding to the P-site of the 60S ribosomal subunit (Garreau De Loubresse et al., 2014). Blasticidin-S inhibited the growth of P. tricornutum completely at a concentration of 3.5 μg/ml on plates with 25% salt and eight μg/ml on 50% salt plates (Table 1). In liquid 50% salt medium, blasticidin-S inhibited growth at a concentration of three μg/ml (Fig. S1). Tunicamycin, which acts as inhibitor of biosynthesis of peptidoglycan structures in prokaryotes and of N-glycosylation in eukaryotes (Takatsuki, Arima & Tamura, 1971; Wyszynski et al., 2012), inhibited the growth at concentrations of five μg/ml on 25% salt plates, and 10 μg/ml on 50% salt plates. However, in contrast to blasticidin-S, tunicamycin was not able to stop growth completely, as single colonies appeared after 5 weeks on the 50% salt plates.

Functionality of the respective resistance genes

In order to confer resistance against tunicamycin and blasticidin-S in P. tricornutum, we generated DNA fragments encoding the tunicamycin resistance protein (tmrB) (Harada et al., 1988) and the bsr (Endo et al., 1987), with codons suitable for P. tricornutum and manually eliminated recognition sites of popular commercially available restriction enzymes. We inserted the synthetic genes into the plasmid pPha-T1 (Zaslavskaia et al., 2000), replacing the Zeocin resistance cassette ShBle between the fcpB-promoter and the fcpA-terminator (Fig. 2, example for the bsr gene). P. tricornutum was genetically transformed with the constructs by particle bombardment. To test the antibiotic resistance, we plated the cells on selection plates containing four μg/ml (25% salt) or 10 μg/ml (50% salt) of blasticidin-S, and 10 μg/ml (50% salt) tunicamycin, respectively. For unknown reasons, we only obtained blasticidin-S-resistant colonies for the cells transformed with the bsr gene, while no tunicamycin-resistant colonies were obtained when we tested the tmrB gene using two independent attempts.

In total, we obtained 78 blasticidin-S-resistant colonies on four μg/ml, 25% salt medium in two independent transformations, plus 22 colonies on 10 μg/ml, 50% salt medium in one transformation approach. We re-plated each of the obtained colonies on fresh blasticidin-S plates, on which all of them continued to grow stably and were able to survive in liquid medium under selection pressure (Fig. S1). In more than one year that has passed since we obtained the first blasticidin-S cell lines, we did not observe a loss of resistance in any of the colonies. For further experiments we focused only on colonies that were selected on the 25% salt plates. These 78 colonies were checked by PCR for the presence of the bsr gene. We could amplify a PCR product in the expected size range from 75 of the 78 analyzed templates derived from the colonies, indicating the presence of the resistance gene (Fig. S2).

Cross reactivity test

In order to perform successive transformations with different selection markers, we tested whether blasticidin-S might interfere with the usage of Zeocin- or nourseothricin-based screening protocols. We therefore plated nine of the blasticidin-S-resistant colonies on agar plates containing either Zeocin or nourseothricin (Fig. 3). Conversely, nine nourseothricin- or Zeocin-resistant colonies were plated on plates with four μg/ml blasticidin-S. As a control, the same colonies were inoculated on culture plates containing the antibiotic to which they show resistance. After 11 days of growth, colonies were detected only on plates on which cells were plated that had the respective antibiotic resistance (Fig. 3). This indicates that blasticidin-S selection can be used in combination with Zeocin- or nourseothricin-resistant cells. Such double resistant strains were already reported for the combination of Zeocin and nourseothricin (Serif et al., 2017). Especially the combination of blasticidin-S- and nourseothricin might be important for future applications, for instance to replace Zeocin in the TALEN vectors or for complementation experiments following gene knockouts. We co-transformed the plasmids pPTbsr and pPha-NR (without BsaI) (Serif et al., 2017), which mediates nourseothricin resistance, leading to 29 resistant colonies on the 50% salt selection plates containing 10 μg/ml blasticidin-S and 150 μg/ml nourseothricin.

Transferability of the results to other strains and diatom species

In this study we used P. tricornutum strain UTEX 646 (Pt4 in the nomenclature by Martino et al. (2007)) in all our experiments. Experience with other antibiotics like Zeocin/phleomycin shows that there is apparently no difference between different strains regarding their sensitivity (Apt, Grossman & Kroth-Pancic, 1996; Falciatore et al., 1999; Siaut et al., 2007; Tanaka et al., 2005). Furthermore, Martino et al. (2007) used the ShBle marker gene for ten different strains of P. tricornutum. We therefore expect that our results regarding blasticidin-S are directly transferable to other P. tricornutum strains.

In contrast to this, the sensitivity of different diatom species to the currently used antibiotics is quite variable (Huang & Daboussi, 2017) and although some antibiotics, after dose adjustment, can be used in several species of diatoms, marker genes are not generally transferrable between diatom species (Huang & Daboussi, 2017). It therefore remains to be shown if blasticidin-S is suitable for the selection of genetically transformed cell lines of other diatom species. For this, most likely adjustments of the promoters and antibiotic doses would be needed.