Simplified Transformation of Ostreococcus tauri Using Polyethylene Glycol

Ostreococcus tauri is an easily cultured representative of unicellular algae (class Mamiellophyceae) that abound in oceans worldwide. Eight complete 13–22 Mb genomes of phylogenetically divergent species within this class are available, and their DNA sequences are nearly always present in metagenomic data produced from marine samples. Here we describe a simplified and robust transformation protocol for the smallest of these algae (O. tauri). Polyethylene glycol (PEG) treatment was much more efficient than the previously described electroporation protocol. Short (2 min or less) incubation times in PEG gave >104 transformants per microgram DNA. The time of cell recovery after transformation could be reduced to a few hours, permitting the experiment to be done in a day rather than overnight as used in previous protocols. DNA was randomly inserted in the O. tauri genome. In our hands PEG was 20–40-fold more efficient than electroporation for the transformation of O. tauri, and this improvement will facilitate mutagenesis of all of the dispensable genes present in the tiny O. tauri genome.


Introduction
Marine microalgae in the class Mamiellophyceae are emerging as useful models for integrative biology for several reasons. They can be easily cultured in the laboratory, witnessed by more than 300 strains maintained in the Roscoff Culture Collection (RCC) in France [1], and they diverged at the base of the green lineage that includes land plants, thus providing a starting point for phylogenetic analyses [2]. Several of their streamlined haploid genomes have been completely sequenced and analysed [3][4][5][6][7][8], and they harbour a minimal number of genes for cellular processes, as reflected in the size of their nuclear genomes. Species in the genus Ostreococcus, for example, which so far includes four phylogenetically divergent [9] species with fully sequenced genomes (see [3,4,8] and https://genome.jgi.doe.gov/OstRCC809_2/OstRCC809_2.home.html), have only 12-14 Mb of nuclear DNA, comparable to that of yeast (12.1 Mb) [10]. This is much smaller than those of other model species in the green lineage such as mosses (Physcomitrella patens, 476 Mb, [11]) Arabidopsis thaliana (135 Mb [12]) or maize (2.4 Gb [13]). Ostreococcus spp. thus often possess only one copy of genes that are multigene families in other photosynthetic organisms, so alteration of individual genes by directed mutagenesis immediately gives rise to new phenotypes [14,15]. Various kinds of biological analyses involving these algae are being undertaken, for example, the functional analysis of casein kinase 1 in cell cycle [16], nitrogen status sensing by calcium-dependent protein kinases (CDPKs) [17],

Transformation
Polyethylene glycol, PEG MW 4000, 6000 or 8000 (reference 81240 Sigma-Aldrich) solution was prepared by diluting in MilliQ water to obtain a concentration of 60% (w/v) before filter-sterilisation (0.22 µm pore size). For each transformation, 50 mL of cells at density of 20-30 × 10 6 mL −1 were used. In our culture conditions this cell density occurred during the late part of the exponential growth, about one week after subculturing.
We then centrifuged the culture for 10 min at 6000× g at 20 • C and the pellet was resuspended in a microtube (1.5 mL) with 500 µL of L1 medium to reach a final cell concentration of about 10 9 cells·mL −1 . DNA/tRNA was added, then gently mixed in 500 µL PEG (60%), giving a final PEG concentration of 30% (w/v), and incubated on the laboratory bench (~10 µE m −2 s −1 ) at 20 • C for 2 min. Cells were then diluted into 40 mL of fresh L1 medium, thereby diluting the concentration of PEG to 1.5% in the recovery culture, and transferred to a growth chamber for 2-6 h before plating out.

Selection on Plates in Semi-Solid Medium
One millilitre aliquots of a solution of 2.1% (w/v) low-melting-point agarose in MilliQ water were maintained at 60 • C in a water bath. For each plate (Petri dish, diameter 55 mm), 8 mL of L1 medium with the required concentration of antibiotic (G418 disulphate salt (Sigma-Aldrich, A1720) at a final concentration of 2 mg·mL −1 ) were added to the tube containing the warm agarose, and quickly mixed. One millilitre of freshly transformed cells in L1 + G418 were then added and poured. Plates were dried by leaving the lids halfway open in a sterile laminar air flow cabinet for 20 min before returning them to the culture chamber (100% humidity) at 20 • C.

Luciferase Assays
We prepared a suspension of algae at 2 × 10 7 cells·mL −1 , adding aliquots of 200 µL to each well of a 96-well white plate (Greiner Bio-One™ LUMITRAC, SAS, Les Ulis, France), then added D-Luciferin (Pierce, Waltham, MA, USA, ref 88293), final concentration 10-100 µM and placed the plate at 20 • C for 10 min in the dark before measuring luminescence with a Victor3 Multilabel Plate Counter Spectrofluorimeter (Perkin Elmer, Waltham, MA, USA) by digital photon counting.

Pulsed-Field Gel Electrophoresis
Pulsed-field gel electrophoresis (PFGE) and in-gel hybridisation using dried gels [50,51] was conducted as previously described [52,53], with the following modifications. The cell cultures were grown to mid-exponential phase (~1.6 × 10 7 cells·mL −1 ), 8.7 × 10 7 cells were harvested by centrifugation (8000× g for 20 min) and resuspended in 150 µL of TE buffer (10 mM Tris-HCl, 125 mM EDTA, pH 8). Cells were embedded into plugs by mixing with an equal volume of molten low-melting-point agarose (1% in TE buffer precooled to 45 • C), to a final volume of 300 µL before adding to moulds (Bio-Rad, Hercules, CA, USA). After setting, cells were then lysed in the plugs using proteinase K buffer (10 mM Tris HCl pH 8, 0.5 M EDTA pH 8, 1% lauryl sarcosinate and 1 mg·mL −1 proteinase K (final concentration), 37 • C, 24 h under constant agitation), before washing three times in 0.5 M EDTA pH 8 at 37 • C for 2 h under constant agitation and storage at 4 • C. Electrophoresis was performed subsequently in 0.8% agarose gels in 0.5 × TBE buffer (final 44.5 mM Tris, 44.5 mM boric acid, 1 mM EDTA at pH 8) using a CHEF-DR III (Bio-Rad) system. For each sample, 2 mm of plug was loaded into the wells. Electrophoresis was run at 14 • C, 6 V cm −1 with 120 • pulse angle for 15 h with a switch time of 60 s and followed 9 h at a switch time of 90 s. After PFGE, the gel was moved to flat-bottomed dish and enough 0.4 M NaOH was added to cover the gel for chemical denaturation of the DNA by incubation for 30 min at room temperature with agitation. Then the gel was washed three times for 10 min in a 6 × SSC solution (20 × SSC, 3 M NaCl, 0.3 M Na 3 citrate, pH 7). Finally, the gel was dehydrated under vacuum (Hoefer inc. GD2000, Richmond, CA, USA).
PCR was used to amplify a 630 bp region of the G418 gene selectable marker gene with the primer pair Kan-FW CCTGTACGGGTACAAGTGGG and Kan-RV CAGGTGGAACTGGAGCTTGT and the product was cleaned (Wizard ® , Promega, SV Gel and PCR Clean-Up System). The amplicon was randomly labelled with [α-32 P]CTP (Perkin-Elmer ref BLU008H250UC) according to the manufacturer's instructions (Prime-a-Gene kit, Promega ref U1100) for use as a DNA probe. Dried gels were equilibrated in hybridisation buffer (6 × SSC, 5 × Denhardt's solution, 0.1% (v) sodium dodecyl sulphate, 10 µg·mL −1 tRNA), radiolabelled probe was added, hybridised overnight at 65 • C, and the gel exposed to radiographic film.
To synthesize a suitable vector, the coding sequences (CDSs) of genetic marker genes were first chosen: (i) the aminoglycoside 3 -phosphotransferase type I (APHI) coding gene of the bacterial transposon Tn903 for G418 resistance [58] and (ii) the firefly luciferase CDS [59] was chosen as a reporter for gene expression. The predicted amino acid sequences of all of these genes were then back-translated using the EMBOSS [60] function backtranseq together with a codon usage table for O. tauri produced by EMBOSS cusp, which was built from the characterised O. tauri CDS available from the ORCAE website [61]. These sequences were then edited by replacing codons for glycine, leucine, serine and arginine with the optimum codon sequences for highest expression [62] and, when possible, by introducing a synonymous base substitution in one of the codons in undesirable restriction enzyme sites (SmaI, SacI, XmaI, XhoI from the APHI CDS and ScaI, NcoI, KpnI, BglI, BamHI from the luciferase CDS). Next, we chose promoter and terminator regions from highly expressed constitutive genes, firstly α-tubulin because it was shown to have constitutive expression in algae [63][64][65] then histone and thioredoxin genes because they were expressed constitutively in healthy and virus-infected cells [66]. The length of DNA chosen for promoters was rather subjective, in general being about half of the individual intergenic sequences concerned. A "strong stop" (SS1, Figure 1) sequence was used to avoid translational or transcriptional read-through from adjacent sequences. This was generated by using the sequence of a short intergenic region from between two convergently strongly expressed genes (ostta05g02170 and ostta05g02170, O. tauri genome version 2 [61]) and adding three out-of-frame translational stops to both ends of the intergenic region. Some commonly used restriction enzyme sites were then added in-between or around the genes to facilitate subsequent cloning steps ( Figure 1). The two genes with suitable promoters were then synthesized commercially (Genscript, USA) and ligated to a standard vector pUC57 to produce pOLK1 ( Figure 1). Subsequently, three other constructions were produced from the pOLK1 vector by replacing the BglII-NcoI α-tubulin promoter sequence upstream of the luciferase gene with native O. tauri promoter sequences for histone 2A (pOLK2), histone 3 (pOLK3), ubiquitin (pOLK4) and thioredoxin (pOLK5).  Table S1).

Optimisation of the Transformation Procedure
To develop the transformation procedure, we performed a series of test experiments (Figure 2  and Table S2), each conducted over a period of about two months. Usually, four to eight tests were regrouped to be done simultaneously, to allow enough time for handling cells in a day and to make better within-group treatment comparisons. After the PEG treatment, the cells were allowed to recover in culture for 24 h before plating them out unless otherwise stated, to remain consistent with previous studies. The time from transformation to picking off individual colonies on selective plates was usually about 21 days.
In first a series of experiments we compared electroporation [27] with PEG, using PEG conditions similar to those optimised for C. merolae [45], and found that in our hands PEG gave about 20-40 times as many transformants than electroporation ( Figure 2). We then tested the effects of several other conditions ( Figure 2 and Table S2), including the time length of the PEG treatment, the concentration of PEG, the PEG molecular weight, linear or circular transforming DNA, the concentration of transforming DNA, the presence or absence of carrier DNA or tRNA, and the use of different transforming DNAs. Short incubation periods with PEG were found to be optimal ( Figure  2), and we retained two minutes as the standard for further tests.    Table S1).

Optimisation of the Transformation Procedure
To develop the transformation procedure, we performed a series of test experiments (Figure 2  and Table S2), each conducted over a period of about two months. Usually, four to eight tests were regrouped to be done simultaneously, to allow enough time for handling cells in a day and to make better within-group treatment comparisons. After the PEG treatment, the cells were allowed to recover in culture for 24 h before plating them out unless otherwise stated, to remain consistent with previous studies. The time from transformation to picking off individual colonies on selective plates was usually about 21 days.
In first a series of experiments we compared electroporation [27] with PEG, using PEG conditions similar to those optimised for C. merolae [45], and found that in our hands PEG gave about 20-40 times as many transformants than electroporation ( Figure 2). We then tested the effects of several other conditions ( Figure 2 and Table S2), including the time length of the PEG treatment, the concentration of PEG, the PEG molecular weight, linear or circular transforming DNA, the concentration of transforming DNA, the presence or absence of carrier DNA or tRNA, and the use of different transforming DNAs. Short incubation periods with PEG were found to be optimal (Figure 2), and we retained two minutes as the standard for further tests.  Flow cytometry measurements confirmed that longer times in the PEG treatment or use of higher-molecular-weight PEG (MW 4000, 6000 and 8000 were tested) were indeed more toxic ( Figure S1A,B). Linearized DNA gave 2-to-9-fold more transformants than circular DNA in four comparisons ( Figure 2). Increasing the concentration of PEG MW 4000 from 30% to 40% reduced the frequency of transformation (Figure 2). Carrier tRNA increased the transformation frequency ( Figure 2D). At lower concentrations of transforming DNA (2 µg) the transformation was more likely to fail completely if there was no carrier (overall 3/11 tests failed with no carrier, data not shown). Overall, comparing between transformations where 2 µg of linear transforming DNA was used, absence of carrier yielded 6.35 × 10 3 ± 8.44 × 10 3 transformants per microgram of vector DNA (n = 11), whereas salmon sperm carrier DNA gave 1.59 × 10 4 ± 7.45 × 10 3 transformants per microgram of vector DNA (n = 5) and yeast tRNA carrier yielded 2.11 × 10 4 ± 4.83 × 10 3 transformants per microgram of vector DNA (n = 4). Since all of the 11 independently transformed lines that we tested by Southern blot carried integrated vector DNA (see Figure 4 in Section 3.3 below), all of these lines were transformed. Negative controls (no vector DNA) were done in all tests (data not shown) but never gave rise to G418-resistant lines.
In a second series of experiments, the length of time the cells spent recovering in liquid medium in the growth chamber before plating them out on selective medium was tested. In the 12 h light/12 h dark cycle that was used, cell division occurred mainly in the night. Although overnight incubation in culture chambers (up to 24 h) produced about three times as many colonies as 2 h incubation, it is likely that a large proportion of the surviving cells divided once, potentially producing duplicate transformants. The density of cells in control untreated cultures rose only 2% over 2 h (at t = 2 above the t = 0 level, 11:30 a.m.) and 9% at t = 6 h, but at t = 24 h it had risen to 125%, confirming that nearly all cell division occurred in the night (Figure 2 and [67]). In contrast, the number of transformants increased to 177% at t = 6 h (when t = 2 h was arbitrarily considered to be 100%) and to 290% at t = 24 h (Figure 2A). This step was thus shortened to 2-6 h in the optimised protocol presented. Two hours is sufficient to obtain about 20 transformants per million cells input, and 6 h in recovery (before the cells divide) permits about twice this frequency, with only a small risk of finding duplicate transformants if sufficient time is available on the same day to do the experiment.
A schematic presentation of this optimised procedure is shown in Figure 3 and described in the Methods section. Correcting for the shorter recovery period in culture of 2-6 h, this procedure gave 10 4 transformants per microgram of transforming DNA, or about 4 × 10 5 transformants from each 50 mL flask of cultured cells used in the experiment.

Integration of Transforming DNA is Random
Radioactive probing of O. tauri chromosomes in gels separated by pulsed field gel electrophoresis (PFGE [52] was used to verify that the putative transformants had stably integrated plasmid DNA. The karyotypes of 11 individual colonies arising from transformations using pOLK1 were visualized by PFGE [68]. The gels were then probed with the 0.63 kb long fragment prepared from the APHI gene present in the vector. Usually insertions occurred on a single chromosome, visualized as a dark band on the gel track, migrating at the same position as a chromosomal band (Figure 4).

Integration of Transforming DNA is Random
Radioactive probing of O. tauri chromosomes in gels separated by pulsed field gel electrophoresis (PFGE [52] was used to verify that the putative transformants had stably integrated plasmid DNA. The karyotypes of 11 individual colonies arising from transformations using pOLK1 were visualized by PFGE [68]. The gels were then probed with the 0.63 kb long fragment prepared from the APHI gene present in the vector. Usually insertions occurred on a single chromosome, visualized as a dark band on the gel track, migrating at the same position as a chromosomal band (Figure 4).

Transgene Expression from Integrated DNA
In order to test the utility of our vector, and to test several promoters for future work, we replaced the BglII-NcoI fragment of the vector, immediately upstream of the luciferase gene ( Figure  1) with four other native O. tauri promoter regions (Materials and Methods, and Table S1). Expression from these promoters was assayed in transgenic lines by comparing their bioluminescence. Sixteen independent transformed lines from pOLK1, pOLK2, pOLK3 and pOLK4 were chosen randomly and assayed for expression of luciferase (64 independent transformants in total). All of the independent transformed lines expressed the luciferase reporter gene.
The majority of these transformants showed high levels of expression, but the level of expression of individual lines was very variable ( Figure 5 and Table S3). In relative units, P_Ub and P_H2A gave the highest levels of expression (pOLK2 and pOLK4, overall about one-hundred-fold higher than the control), followed by pOLK4 (P_Ub, 97-fold), pOLK1 (P_aTub, 77-fold) and pOLK3 (P_H3, 23-fold).

Transgene Expression from Integrated DNA
In order to test the utility of our vector, and to test several promoters for future work, we replaced the BglII-NcoI fragment of the vector, immediately upstream of the luciferase gene ( Figure 1) with four other native O. tauri promoter regions (Materials and Methods, and Table S1). Expression from these promoters was assayed in transgenic lines by comparing their bioluminescence. Sixteen independent transformed lines from pOLK1, pOLK2, pOLK3 and pOLK4 were chosen randomly and assayed for expression of luciferase (64 independent transformants in total). All of the independent transformed lines expressed the luciferase reporter gene.
The majority of these transformants showed high levels of expression, but the level of expression of individual lines was very variable ( Figure 5 and Table S3). In relative units, P_Ub and P_H2A gave the highest levels of expression (pOLK2 and pOLK4, overall about one-hundred-fold higher than the control), followed by pOLK4 (P_Ub, 97-fold), pOLK1 (P_aTub, 77-fold) and pOLK3 (P_H3, 23-fold).  Table S3. To compare between constructs, we preferred to plot the relative bioluminescence units (ordinate) from each transformant as jittered points [69] ("geom_jitter" ggplot2 function in R). The means and standard errors for each construct are shown as large red dots and vertical bars, respectively.

Discussion
One inconvenience of using electroporation as a means for transformation is that the cells being used must first be washed and resuspended in a non-conductive solution of sufficient osmolarity to prevent them from bursting. Any traces of salt left in the mixture will cause electrical arcing and loss of the cells, so a second washing step is necessary [27]. In contrast, PEG transformation allows cells to be centrifuged and resuspended directly in the same culture medium, and no electroporation step is required.
Once we optimised the use of PEG, we went on to further refine the technique by shortening the delay between transformation treatment and plating out for individual cell lines. Previous protocols [34,37] allowed one day of recovery before the plating out of cells. While this gives rise to more antibiotic-resistant cell lines, an unknown proportion of these lines might arise by cell division, probably allowing duplication of some of the individual transformation events. In control cultures not exposed to PEG, the population density of cells increased in the night (Figure 2 and [67]). Although there was no apparent increase in the number of viable cells after PEG, we do not know whether PEG stopped cell division or whether there was a mixture of some cell division and some cell death. It was not necessary to allow cells to recover overnight, and the cells obtained could be plated out on the same day after the transformation treatment. This modification not only allows completion of the experiment in a single day, but also assures that all of the transformed colonies of  Table S3. To compare between constructs, we preferred to plot the relative bioluminescence units (ordinate) from each transformant as jittered points [69] ("geom_jitter" ggplot2 function in R). The means and standard errors for each construct are shown as large red dots and vertical bars, respectively.

Discussion
One inconvenience of using electroporation as a means for transformation is that the cells being used must first be washed and resuspended in a non-conductive solution of sufficient osmolarity to prevent them from bursting. Any traces of salt left in the mixture will cause electrical arcing and loss of the cells, so a second washing step is necessary [27]. In contrast, PEG transformation allows cells to be centrifuged and resuspended directly in the same culture medium, and no electroporation step is required.
Once we optimised the use of PEG, we went on to further refine the technique by shortening the delay between transformation treatment and plating out for individual cell lines. Previous protocols [34,37] allowed one day of recovery before the plating out of cells. While this gives rise to more antibiotic-resistant cell lines, an unknown proportion of these lines might arise by cell division, probably allowing duplication of some of the individual transformation events. In control cultures not exposed to PEG, the population density of cells increased in the night (Figure 2 and [67]). Although there was no apparent increase in the number of viable cells after PEG, we do not know whether PEG stopped cell division or whether there was a mixture of some cell division and some cell death. It was not necessary to allow cells to recover overnight, and the cells obtained could be plated out on the same day after the transformation treatment. This modification not only allows completion of the experiment in a single day, but also assures that all of the transformed colonies of cells after the transformation are likely to arise independently. Under the 12 h/12 h day/night growth regime used to grow the cells, the cells normally divide in the early part of the night [67], in contrast to cells grown under continuous illumination [34], whose asynchronous division is about 1.8 per day. This consideration is important for determining the transformation frequency (e.g., when calculating the number of transformations required to mutate all dispensable genes). Reducing the recovery time in culture after PEG treatment to 2 h gave about half as many G418-resistant as at 24 h, and reducing it to 6 h (also possible in one working day) gave about two-thirds the number ( Figure S1B). The integration events generally occurred on a single chromosome in each transformed line. According to the variation in band intensity, multiple copies of introduced genes would be integrated into a certain chromosome (Figure 4).
The expression of bioluminescence has been used as a visible reporter successfully in many previous studies in O. tauri. As examples, to mention but a few, Corellou et al. [27] used the luciferase expression in transformed lines to investigate circadian gene regulation in an article that also first reported the use of electroporation for O. tauri transformation. Luminescence reporter constructs have also been used to investigate the levels of expression of an inducible phosphate transporter [30] and to monitor the involvement of magnesium flux in cellular clocks [70]. We thus chose luciferase as a marker in PEG-mediated transformation based on the strength of its use in previous work. However, while all of these studies used purpose-made promoter-luciferase constructions to monitor expression by emission of bioluminescence, it was not their purpose to present more details about the transformation procedure by electroporation, which were reported elsewhere [34]. However, the levels of expression of the promoters we tested in our transformed lines were very variable, as we showed by testing the expression from 16 independent transformants for each construction (Figure 4). The silencing of transgenes has been extensively documented in the past (see [71,72] for reviews), and copy number can be positively or negatively correlated with gene expression [73,74]. The possible reasons for this variability include insert copy number, position effect of the site of integration, non-coding-RNA-mediated silencing, methylation of transforming DNA and chromatin conformation at the integration site. Since our main objectives here were to improve transformation frequency and to identify promoters that would be of use for future use in host-virus interaction studies, we did not pursue a more extensive analysis about this variability, which could be a subject of further research, particularly since certain key genes involved in classical gene silencing by siRNA have not yet been identified in O. tauri [75]. The question of the stability of the expression of transformed lines is also a subject that merits future investigation. We did keep some of the transformed lines expressing luciferase from either histone 2A or from α-tubulin promoters in culture, and recently remeasured their bioluminescence after about 400 cell divisions in the absence of genetic selection (no G418, >13 months in culture). This confirmed that all lines retained were expressing luciferase (five lines from each of the constructions) with a relative fluorescence for α-tubulin of 2.9 × 10 3 ± 2.9 × 10 3 (26-fold control level) and for histone 2A, 7.8 × 10 3 ± 3.7 × 10 3 (75-fold control level). We have no reason to believe that the stabilities of PEG-derived transformants would differ from those of electroporation-derived transformants.
Whereas the stress of prasinovirus infection almost always induces rearrangements in the size of chromosome 19 in O. tauri [76,77], no changes in the karyotype were visible following PEG-mediated transformation (Figure 4), indicating that the stress of the transformation treatment did not result in chromosomal rearrangements. Our protocol is thus suitable for investigating genes involved in host-virus interactions independently of this response, and will facilitate approaches that require an efficient delivery of transforming DNA, such as homologous (targeted) recombination [15].

Conclusions
Several different aspects of the O. tauri transformation procedure have been improved in this work: (i) the frequency of transformation was increased by 20-40-fold; (ii) the procedure was simplified by the use of PEG, which renders centrifugation and resuspension of cells in a non-ionic solution for electroporation unnecessary; (iii) the procedure was shortened, so that it could be completed in one instead of two days; and (iv) accuracy in determining the transformation frequency was improved. These ameliorations will widen the accessibility of the technique to more plant cell culture laboratories and facilitate its use for applications requiring high frequencies of independent transformants, such as the mutagenesis of dispensable genes involved in responses to viral infection [76,77].
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4425/10/5/399/s1, Table S1: Nucleotide sequences of the vector component parts used for transformations; Table S2: Optimization of PEG transformation conditions; Table S3: Raw bioluminescence data; Figure S1: Flow cytometric estimation of viable cells during treatment with PEG, estimated by autofluorescence of chlorophyll.