A molecular toolkit for the green seaweed Ulva mutabilis

Abstract

The green seaweed Ulva mutabilis is an ecologically important marine primary producer as well as a promising cash crop cultivated for multiple uses. Despite its importance, several molecular tools are still needed to better understand seaweed biology. Here, we report the development of a flexible and modular molecular cloning toolkit for the green seaweed U. mutabilis based on a Golden Gate cloning system. The toolkit presently contains 125 entry vectors, 26 destination vectors, and 107 functionally validated expression vectors. We demonstrate the importance of endogenous regulatory sequences for transgene expression and characterize three endogenous promoters suitable to drive transgene expression. We describe two vector architectures to express transgenes via two expression cassettes or a bicistronic approach. The majority of selected transformants (50%–80%) consistently give clear visual transgene expression. Furthermore, we made different marker lines for intracellular compartments after evaluating 13 transit peptides and 11 tagged endogenous Ulva genes. Our molecular toolkit enables the study of Ulva gain-of-function lines and paves the way for gene characterization and large-scale functional genomics studies in a green seaweed.

Molecular cloning tools allow generating gain-of-function seaweed lines that will help to study seaweed biology.

Introduction

Important progress in molecular research has been made for several unicellular algae, including comparative genomics (Blaby-Haas and Merchant, 2019), genetic transformation (e.g. Doron et al., 2016; Faktorová et al., 2020), and large systems-biology studies (e.g. Mackinder et al., 2017; Strenkert et al., 2019). Multicellular algae, or seaweeds, are important organisms in different marine habitats and several species are cultivated for food, feed, biofuel, or pharmaceuticals (Bolton et al., 2016; Charrier et al., 2017; Wells et al., 2017). Notwithstanding their importance, molecular tools are needed to study seaweed biology and generate new strains with desirable characteristics (Loureiro et al., 2015). While genomic resources are increasing for brown, green, and red seaweeds (Cock et al., 2010; Collén et al., 2013; De Clerck et al., 2018; Wang et al., 2020), stable genomic integration and expression of exogenous DNA is still rare and limited to antibiotic resistance genes and/or reporter genes, such as ß-galactosidase (lacZ) or Green Fluorescent Protein (GFP; e.g. Jiang et al., 2003; Song et al., 2003; Son et al., 2012; Lim et al., 2013; Uji et al., 2014; Oertel et al., 2015; Lim et al., 2019). Recently, an attempt to overexpress a Pyropia yezoensis gene coding for a light-harvesting protein resulted in cosuppression (Zheng et al., 2020). To our knowledge, no transgenic macroalgal line has been described expressing an endogenous tagged gene. Developing molecular tools for macroalgae will allow the study of gene function, which is crucial to understand seaweed growth, reproduction, and dependence on symbiotic bacteria (Charrier et al., 2017).

The green seaweed Ulva is a multicellular marine algae model that has traditionally been studied from a morphological and physiological perspective (reviewed in Wichard et al., 2015), but is also a relevant commercial species (Bolton et al., 2016). Most research has been conducted on U. mutabilis, a species that has been studied and kept in culture since the 1950s (Føyn, 1958; Wichard et al., 2015). Advantages of this model organism are the short life cycle, ease of cultivation in well-defined media and the ability to generate stable transformants using plasmid vectors (Stratmann et al., 1996; Spoerner et al., 2012; Oertel et al., 2015). Importantly, the complete U. mutabilis genome is published and annotated (De Clerck et al., 2018). Transcriptomes, organelle genomes, or specific loci have been reported in other Ulva species (Melton III et al., 2015; Zhou et al., 2016a, 2016b; Cai et al., 2017; Yamazaki et al., 2017; Ichihara et al., 2019; Wang et al., 2019). Despite the availability of genomic resources, few molecular tools are described for investigating gene function in Ulva. So far, only four Ulva-specific promoter sequences have been isolated (Kakinuma et al., 2009; Oertel et al., 2015; Wu et al., 2017) and the expression of three transgenes has been reported (bleomycin resistance protein [ble], GFP, and lacZ; Huang et al., 1996; Kakinuma et al., 2009; Suzuki et al., 2014; Oertel et al., 2015). In a transient expression system of three Ulva species (Ulva prolifera, Ulva linza, and Ulva pertusa), the Cauliflower Mosaic Virus 35S promoter could only drive weak expression of a reporter gene (uidA encoding GUS) whereas the actin promoter of U. prolifera was relatively stronger (Huang et al., 1996; Wu et al., 2017). Stable U. mutabilis transformants can be generated through expression of a ble cassette containing the promoter/5′ UTR, first intron, and terminator of Rubisco SSU (denoted here as BleR; Oertel et al., 2015). Exogenous DNA is randomly integrated in the nuclear genome of Ulva, often in tandem vector repeat clusters, by an undefined recombination mechanism. Transgene expression is reported to be stable for three generations in the absence of selection (Oertel et al., 2015). Occasionally, some transformants display a mosaic phenotype, with green healthy blade cells and bleached, presumably nonphleomycin-resistant, cells. This has been tentatively attributed to transgene silencing (Oertel et al., 2015).

Efficient modular cloning systems are a prerequisite in the design of large-scale functional genomic studies and synthetic biology. Flexible molecular toolkits based on Golden Gate cloning allow efficient assembly of transcriptional units (Lampropoulos et al., 2013; Casini et al., 2015). Type IIS restriction enzymes, such as BsaI, BbsI, or BsmbI generate staggered cuts outside their recognition site, resulting in user-defined four nucleotide overhangs. During Golden Gate assembly, multiple DNA fragments are joined in parallel in a single reaction in a predetermined orientation based on the overhang sequences. Variations of Golden Gate, such as MoClo, GoldenBraid, or GreenGate reflect differences in the Type IIS enzyme used, the overhang site sequences or the number of possible fusion sites (reviewed in Casini et al., 2015; Patron et al., 2015). For example, a MoClo modular library containing 119 parts has been developed for the unicellular green algae Chlamydomonas reinhardtti, including 7 promoters, 11 signal and targeting peptides, 34 reporter genes, and 8 antibiotic resistance genes (Crozet et al., 2018). In general, new modules are easily made and their modularity allows the generation and testing of a multitude of vectors in a straightforward way. At present, no such system is available for seaweeds.

In this study, we present a modular vector toolkit for molecular research in U. mutabilis based on the GreenGate cloning system (Lampropoulos et al., 2013). The toolkit presently contains 125 entry vectors, 26 destination vectors, and 107 expression vectors. We demonstrate the importance of using endogenous promoter and intron regions for proper transgene expression and report stable overexpression of different transgenes using microscopy and molecular methods. The toolkit allows the rapid generation of transgenic Ulva lines expressing tagged transgenes, enabling the study of gain-of-function lines in green seaweeds.

Results

A universal cloning method and repository for Ulva vectors

An important prerequisite for a molecular toolkit is a universal and cost-effective cloning method that can be easily adopted by different laboratories. To this end, we generated a set of plasmids based on the GreenGate cloning system (Lampropoulos et al., 2013; Figure 1; Supplemental Table S1). This Golden Gate cloning system allows rapid and efficient assembly of up to six building blocks in a single-tube reaction. Every building block, or entry vector, is flanked by convergent BsaI restriction enzyme recognition sites. Upon BsaI cleavage, predefined 4-nt 5′ overhangs are generated on both ends of the DNA fragment that ensure the building blocks are assembled in the desired order and orientation in the destination vector (Lampropoulos et al., 2013). This modular assembly method is both fast (see “Materials and methods”) and inexpensive. The main disadvantage is the need to “domesticate” all sequences by removing internal BsaI sites. When building blocks are made, vector assembly, Escherichia coli transformation and plasmid quality checks with restriction digest and Sanger sequencing can be done in as little as 4 d. After assembly and quality control steps have been met, 5 µg of plasmid DNA is used to transform Ulva gametes (Oertel et al., 2015; Figure 1). Ulva gametes are haploid and develop parthenogenetically after settling. Transformants are selected using phleomycin, undergo normal development with the help of at least two bacterial symbionts (Roseobacter MS2 and Cythophaga MS6; Spoerner et al., 2012) and are fully grown after ∼4–5 weeks in Ulva Culture Medium (UCM; Figure 1). In summary, new vector combinations can be designed and tested in Ulva stable transformants in 5–6 weeks. All vectors generated in this study (Supplemental Table S1) are available through https://gatewayvectors.vib.be.

Figure 1.

Overview of cloning strategy and transformation of Ulva mutabilis. A, Summary of GreenGate cloning strategy (Lampropoulos et al., 2013). Up to six entry modules (depicted with A–B, B–C, C–D, D–E, E–F, and F–G) can be assembled in one destination vector in one reaction generating an expression vector. All entry modules have an ampicillin/carbenicillin (bla) resistance cassette for bacterial selection, the destination vectors contain a kanamycin (aph) resistance cassette. Although any sequence can be cloned in all entry modules, the most common sequences in every module are depicted (e.g. promoter sequence in A–B module). BleR is a module containing the ble gene, encoding Bleomycin resistance protein, the promoter/5′ UTR, first intron and terminator of Rubisco SSU (see Oertel et al., 2015). See Supplemental Figure S6 for a cloning scheme of custom destination vectors. B, Overview of Ulva transformation and selection. Gametogenesis of mature Ulva is induced by fragmentation and changing medium to wash out sporulation inhibitors. After 72 h, the cells have differentiated into gametes that are released from the tissue after one more medium changes. Gametes are isolated using their phototactic properties and subsequently DNA can be transfected using PEG-mediated transformation (Oertel et al., 2015). Transformed gametes develop parthenogenically into new thalli and can be selected using phleomycin. Representative images of transformants at different developmental stages.

Endogenous promoters ensure transgene expression in Ulva

The promoter/5′ UTR region of the small subunit of Rubisco (pRbcS2) is the only promoter functionally validated in Ulva by expression of BleR (Oertel et al., 2015). Therefore, we designed a promoter screen to expand the number of functional promoters for Ulva. We first identified 43 highly transcribed genes from previously generated transcriptome data (De Clerck et al., 2018) and selected genes that had a consistently high expression value >3,000 FPKM. We narrowed this list down to 14 genes (Figure 2; Supplemental Figure S1; Supplemental Table S1) by removing those with internal BsaI and BbsI restriction sites in the promoter/5′ UTR regions. These genes mainly code for ribosomal proteins or proteins involved in photosynthesis (Supplemental Figure S1). We confirmed the transcript expression levels of the selected genes in Ulva individuals using quantitative reverse transcription PCR (RT-qPCR; Supplemental Figure S2) and cloned the 770–2,081 bp region upstream of the predicted translation start site (TLSS), a region that includes both the promoter region and 5′ UTR (hereafter simply denoted as promoter). In addition, we generated one clone that contained the promoter region of the gene (UM020_0181) and its putative leader intron (Supplemental Figure S1). We assembled the promoter to the BleR resistance gene directly fused to GFP (Fuhrmann et al., 1999) and the terminator of RbcS2 (promoter-BleR-GFP-tRbcS; Figure 2; Supplemental Figure S1). Similarly, we generated pRbcS-BleR-GFP-tRBCS as a positive control (546 bp; Oertel et al., 2015) and generated expression vectors with the promoter region of an endogenous actin gene (pUmAct1; 2 kbp), the previously characterized U. prolifera actin1 promoter (pUpAct1; 1,941 bp; Wu et al., 2017), the Cauliflower Mosaic Virus 35S promoter (p35S; 864 bp; Odell et al., 1985) and the Arabidopsis UBIQUITIN10 promoter (pUBI; 633 bp; Norris et al., 1993). Ulva gametes were transformed with the expression vectors and the number of resistant individuals was counted for each promoter variant. To assess the relative strength of the promoters, we counted the number of resistant individuals and compared this to the number of resistant individuals generated using the positive control pRbcS (Oertel et al., 2015; Figure S1). Generally, around 100 total transformants were obtained for control pRbcS transformations. Fourteen promoters resulted in phleomycin-resistant individuals (Figure 2) whereas five promoter regions did not lead to any resistant individuals (pUM003_0274, pUM007_0106, pUmAct1, p35S, and pUBI). Two promoters (pUM140_0016 and pUM056_0044) resulted in approximately two-fold more resistant individuals compared with pRbcS, five promoters were more variable or similar to pRbcS (pUM010_0172, pUM020_0016, pUM064_0041, pUM050_0066, and pUM061_0014) and six promoters generated fewer events than pRbcS (pUpACT1, pUM120_0090, pUM021_0090, pUM182_0008, pUM060_0075, and pUM020_0181). Inclusion of a putative leader intron with pUM020_0181 did not increase the strength of this promoter (Figure 2).

Figure 2.

Functional validation of Ulva promoter and intron sequences. A, Twenty different regulatory promoter regions were cloned upstream of the resistance cassette (BleR-GFP; Oertel et al., 2015). To evaluate the relative strength of each promoter, the number of resistant individuals was counted relative to the number of resistant individuals for pRbcS (Oertel et al., 2015; Supplemental Figure S1). Visualization of BleR-GFP expression is included for constructs containing pRbcS, pUM140_0016 and pUM056_0044 (see Supplemental Figure S2 for other lines; scale bar: 50 µM). Each dot represents relative transformation efficiency for one independent experiment, the mean is indicated with a bar. n = 3–5. B, Promoter deletion experiment for three promoters (pRbcS, pUM140_0016 and pUM056_0044). The size of the promoter region is indicated for each independent line. Resistant individuals were counted and relative transformation efficiency was calculated as in (A). A control without promoter sequence was included. Each dot represents relative transformation efficiency for one independent experiment, the mean is indicated with a bar, error bars represent sd. n = 3. C, Effect of intron variants for transgene expression in Ulva. The Rbcs2 intron is originally an integral part of the BleR cassette (IBT; Oertel et al., 2015). The Rbcs2 intron was cloned as a separate module directly upstream of the TLSS (I_BT), incorporated in the BleR gene with a G4SGS protein linker (GSIBT) or incorporated in one of three promoter regions (pRbcS+I; pUM140_0016+I or pUM056_0044+I). The number of resistant individuals was counted and transformation efficiency was calculated relative to pRbcS_IBT. Each dot represents relative transformation efficiency for one independent experiment, the mean is indicated with a bar, error bars represent sd. n = 3.

To qualitatively evaluate transgene expression in mature transformants, we imaged the BleR-GFP protein using confocal microscopy. The BleR-GFP protein fusion is expected to localize in the nucleus where it prevents the formation of phleomycin-induced DNA breaks (Calmels et al., 1993; Fuhrmann et al., 1999; Oertel et al., 2015). We were able to visualize nuclear-localized GFP expression for each functional promoter-BleR-GFP-tRbcS combination in mature thalli of transformants (Figure 2; Supplemental Figure S2).

In addition to the promoter screen, we assessed the effect of successive deletions on the functionality of pRbcS and our two strong promoters pUM140_0016 and pUM056_0044. pRbcS is a short sequence of only 546 bp, so we successively deleted 100 bp. For pUM140_0016 and pUM056_0044 the size of the promoter was successively halved (1,000–500–250–125–63 bp; Figure 2). Similar to our promoter screen, we transformed Ulva gametes with all constructs, counted number of resistant individuals and represent the strength of the constructs relative to the control pRbcS (Oertel et al., 2015). In general, we observed that for all three promoters the successive deletion of fragments negatively affects the relative transformation efficiency. Promoter sequences ≥250 bp confer a comparable or higher relative transformation efficiency compared with pRbcS (Figure 2). When promoter size is ˂250 bp, the number of resistant individuals drops sharply, indicating that the main transcriptional enhancers are located in this region (Figure 2). One intriguing exception is the 63-bp promoter of pUM140_0016, which shows an average relative transformation efficiency of 0.62; whereas this value is 0.08 for the 125-bp promoter of this construct (Figure 2).

Intron placement affects transgene expression in Ulva

Heterologous genes introduced in green algal genomes are often poorly expressed. For example, only 13% of Chlamydomonas transformants with a simple VENUS-containing plasmid (pMO518) show fluorescence (Onishi and Pringle, 2016). One way to increase transgene expression is to incorporate endogenous regulatory sequences in the coding sequence (CDS; Lumbreras et al., 1998; Oertel et al., 2015; Jaeger et al., 2019). To evaluate the importance of the endogenous Ulva RbcS2 intron (RbcsI) sequence incorporated in the BleR CDS (Oertel et al., 2015), we generated intron variants of the resistance cassette. We incorporated the RbcsI sequence either as a stand-alone sequence before the TLSS of BleR, integrated in a sequence encoding a small G4SGS protein linker directly upstream of the CDS of BleR or incorporated in one of three promoter/5′ UTR sequences (pRbcS, pUM140_0016 or pUM056_0044) at 154 or 178 bp before the TLSS, respectively. For positive controls we used the BleR CDS containing the RbcsI eight bases downstream of the start codon (Oertel et al., 2015), controlled by either pRbcS, pUM140_0016 or pUM056_0044. Similar to the promoter screen, we transformed Ulva gametes with these constructs and counted the number of resistant individuals for every promoter–intron combination. We observed no, or a strongly reduced number of resistant individuals when RbcsI was incorporated in pRbcS, pUM140_0016, or pUM056_0044 (Figure 2). In contrast, combinations, where the RbcsI is incorporated as a stand-alone sequence or as part of a linker sequence, are functional and generate a similar number of resistant individuals compared with the positive controls (Figure 2).

Two approaches for transgene expression in Ulva

An important feature for a molecular toolkit is to allow overexpression of a transgene. In the green algae Chlamydomonas, expressing an unselected gene-of-interest (GOI) is challenging, but different strategies have been developed to increase efficiency (Rasala et al., 2012; Onishi and Pringle, 2016). We tested two approaches for Ulva: expressing a GOI and a selectable marker (BleR) from two transcripts or as a single, bicistronic transcriptional unit by utilizing ribosome-skipping 2A peptides (Liu et al., 2017).

To express two transcripts from two promoters, we made expression vectors containing both the selection marker (pRbcS-BleR-tRbcS; Oertel et al., 2015) and pUM140_0016-GOI-NOST, where NOST denotes the NOS terminator (Bevan et al., 1983). For each GOI (mTagBFP2, YFP, or mCherry), a second expression vector was created that also included RbcsI upstream of TLSS (Figure 3; Supplemental Figures S3 and S4). A second approach for transgene expression is using a bicistronic mRNA, where BleR is transcriptionally coupled to the transgene of interest via a sequence encoding a ribosome-skipping 2A peptide (Rasala et al., 2012; Liu et al., 2017). In such a vector composition, two or more genes are transcribed as one mRNA, but the 2A peptides mediate ribosome skipping during translation (Liu et al., 2017). To investigate this approach in Ulva, we fused BleR to mTagBFP2, YFP, or mCherry via four different 2A peptides: equine rhinitis A virus (E2A), foot and mouth disease virus 2A (F2A), porcine teschovirus-1 2A (P2A), or toshea asigna virus 2A (T2A; Figure 3; Supplemental Figures S4 and S5; Liu et al., 2017). In addition, we used one of two promoters (pRbcS or pUM140_0016) to control expression of the bicistronic transcript (Figure 3; Supplemental Figure S4 and S5). We transformed Ulva gametes with these vector combinations and selected transformants using phleomycin (Oertel et al., 2015).

Figure 3.

Stable expression of transgenes in Ulva. A, Confocal imaging visualizing the expression YFP in transgenic Ulva individuals and untransformed control (WT). Ulva gametes were transformed with expression vectors containing two promoters (pRbcS-BleR-tRbcs and pUM140_0016-YFP-tNOS with RbcsI or not [−]), one promoter (pUM140_0016-BleR-2A-YFP-tRbcS and pRbcS-BleR-2A-YFP-tRbcS) or a direct fusion to BleR (pRbcS-BleR-YFP-tRbcS; -2A). The 2A sequence is E2A, F2A, P2A, or T2A. Eight primary transformants per construct are shown. See Supplemental Figure S4 for constructs containing mCherry or mTagBFP2. Scale bar: 5 mm. B, RT-qPCR analysis of relative YFP, mCherry, and mTagBFP2 expression in Ulva transgenic lines transformed with expression vectors containing two promoters (pRbcS-BleR-tRbcs and pUM140_0016-GOI-tNOS) or one promoter (pRbcS-BleR-2A-GOI-tRbcS) or a direct fusion to BleR (pRbcS-BleR-GOI-tRbcS; -2A) and untransformed control (WT). For each sample, at least 10 independent primary transformants were pooled. Relative expression values are calculated by normalization against two reference genes (EFLa and UBQ10) and are the mean of three technical replicates ± se. C, Immunoblot analysis on Ulva transgenic lines transformed with expression vectors containing two promoters (pRbcS-BleR-tRbcs and pUM140_0016-mCherry-tNOS) or one promoter (pRbcS-BleR-2A-mCherry-tRbcS), a direct fusion of BleR-mCherry (-2A), vector control (Neg.) and untransformed control (WT). For each sample, at least 10 independent primary transformants were pooled. The expected size of the fusion protein (BleR-mCherry) and the cleaved mCherry is indicated with blue and yellow arrows, respectively. Loading control for each line is represented with ponceau staining (P) and stain-free gel imaging (G). N.D., not determined. See Supplemental Figure S4 for constructs containing YFP or mTagBFP2.

Expression of transgenes was confirmed for all constructs by confocal microscopy, RT-qPCR, and immunoblot analyses (Figure 3; Supplemental Figures S3–S5). For individuals expressing YFP using a two-promoter approach, we visualized expression in 67% and 66% of transformants in the presence or absence of RbcsI before the TLSS, respectively (Supplemental Figure S3). Although there is variability in transgene expression between individuals, we see high and constitutive expression in ∼25% of individuals. YFP signal was lower or confined to a subset of cells in other individuals (Supplemental Figure S3). Furthermore, we did not observe major differences in transgene expression when the vectors were delivered as linear or circular molecules (Supplemental Figure S3). The immunoblot analysis demonstrated that inclusion of E2A, F2A, and P2A resulted in efficient separation of the GOI from BleR, although the fusion protein was detected as well for all lines (Figure 3; Supplemental Figure S4). Although no clear differences in RNA or protein accumulation are detected using RT-qPCR and immunoblot where pooled individuals were investigated; a bicistronic expression cassette under control of pRbcS appears to result in lower accumulation of fluorescent proteins compared with pUM140_0016 (Figure 3; Supplemental Figures S4 and S5). The transgene is stably integrated in the Ulva genome and inherited as expression was observed in the following generation (T1; Supplemental Figure S5). Building on these results, we made expression vectors containing BleR fused to F2A and one of three other genes encoding fluorescent proteins (mCerulean, GFP, and tdTomato; Rasala et al., 2013). We demonstrated expression of the transgenes using confocal imaging (Supplemental Figure S5). In conclusion, high expression of transgenes can be achieved in Ulva with one or two transcriptional units.

Creating marker lines using transit peptides and transgenes

To facilitate the assembly of expression vectors containing endogenous or heterologous genes, we designed standard destination vectors where only the GOI and tag need to be inserted (Supplemental Figure S6; “Materials and methods”). When two expression cassettes are preferred, we designed pGG-pUM140_0016-BE-NOST-PIBT3 and pGG-pUM140_0016-BE-NOST-PIBGT, PIBGT, and PIBT3 denote BleR fused to GFP, or not, and BE the Golden Gate overhang sites (Lampropoulos et al., 2013; Supplemental Figure S6). Similarly, we designed pGG-pUM140_0016-BleR-2A-CE-UmtRbcS and pGG-UmpRbcS-BleR-2A-CE-UmtRbcS with the four 2A sequences (E2A, F2A, P2A, and T2A). In addition, we created destination vectors that allow straightforward assembly of more complex or multicistronic constructs (pGG-pUM140_0016-BleR-2A-BG-UmtRbcS and pGG-UmpRbcS-BleR-2A-BG-UmtRbcS) together with the necessary 2A entry vectors (Supplemental Table S1; Supplemental Figure S6). The GOI with an N- and/or C-terminal tag will replace the ccdB/CmR counterselection module.

In summary, our molecular toolkit allows one to efficiently generate a diverse set of vectors for Ulva stable transformation. In total, 15 out of 20 promoter-BleR-GFP-tRbcS combinations were functional in Ulva and we identified two promoters that consistently result in more resistant individuals compared with the previously described pRbcS. A promoter deletion study indicated that the main transcriptional enhancers are located in the first 250 bp of pRbcS, pUM140_0016, and pUM056_0044. Furthermore, our intron-variant experiment illustrates that RbcsI is functional in different configurations. Using two expression systems, we can efficiently generate transgenic Ulva lines.

With the development of our molecular toolkit in hand, we tested the possibility of generating tagged Ulva lines. We selected transit peptides (TPs) functionally characterized in Chlamydomonas and endogenous Ulva genes for overexpression, as a bicistronic construct fused to BleR-F2A and as a second expression cassette, respectively. Thirteen TPs were cloned, six C-terminal peptides including three different nuclear localization signals (SV40, 2xSV40, and N7) and three markers for microbodies (Pumpkin Malate Synthase [PMS], Chlamydomonas Malate Synthase [CrMS], and Random Peroxisome Targeting Sequence [RMS]), five N-terminal tags including two peptides for mitochondrial targeting (HSP70C and atpA), two chloroplast targeting signals (psaD and USPA), one signal for excretion to the extracellular matrix (cCA), and two N-terminal endoplasmic reticulum (ER) targeting signals combined with a C-terminal HDEL ER-retention signal (ars1 and BIP; Supplemental Tables S1 and S2; Lampropoulos et al., 2013; Lauersen et al., 2015; Crozet et al., 2018). We cloned the TPs in the appropriate entry clones and assembled them with YFP in pGG-pUM140_0016-BleR-F2A-BG-UmtRbcS-KmR. We observed YFP expression in 75%–100% selected transformants for 11 out of 14 constructs (Figure 4; Supplemental Figure S7). All constructs where no YFP signal was observed (atpA, BIP, and USPA) still contain their endogenous Chlamydomonas intron, possibly suggesting that correct splicing is impaired in the transgene. HSP70c is the only TP containing the Chlamydomonas intron that is functional. For atpA we observed YFP signal with anticipated localization when using the CDS (Figure 4; Supplemental Fig. S7). For two TPs, cCA, and psaD, we only observed free YFP signal, indicating that targeting to the extracellular matrix or chloroplast was impaired (Figure 4; Supplemental Figure S7). In contrast, we observed YFP expression consistent with localization in mitochondria (atpA [CDS] and HSP70c), microbodies (CMS, PMS, and RMS), the nucleus (SV40, CrNLS, and N7), and ER (ars1). Visual inspection of fluorescent signal in the lines with mitochondrial, microbody, or ER localization reveals no obvious nuclear localization as visualized by direct fusion of BleR with a fluorescent protein (Supplemental Figures S2 and S5; Oertel et al., 2015), suggesting that ribosome-skipping is efficient and those fusion protein products are present in amounts below the detection limit of our imaging setup.

Figure 4.

Tagged Ulva transgenic lines. A, Visualization of Ulva transgenic lines expressing YFP targeted to different intracellular locations using TPs: mitochondria (AtpA and HSP70c), microbodies (PMS, CrMS, and RMS) and nucleus (SV40; 2xSV40 and N7). One transgenic line (psaD) shows free YFP signal. For each line a representative individual is shown, green indicates the YFP signal and red represents chlorophyll autofluorescence. Scale bar: 50 µM. See also Supplemental Figure S7. B, Visualization of Ulva transgenic lines expressing endogenous Ulva genes tagged with YFP targeted to different intracellular locations: chloroplast (UM120_0017.1), nucleus (UM003_0201.1 and UM001_0379.1), microbodies (UM079_0010.1), and vacuole (UM110_0052.1). One transgenic line (UM025_0020.1) shows YFP signal at the outer chloroplast membrane and cytosol. For each line a representative individual is shown, green indicates the YFP signal and red represents chlorophyll autofluorescence. Scale bar: 50 µM. See also Supplemental Figure S8.

Besides small TPs, we identified twenty Ulva orthologs of Chlamydomonas marker genes for different intracellular locations (nucleus, mitochondria, chloroplast, ER, and peroxisomes; Mackinder et al., 2017). We cloned 11 genes from genomic DNA in the appropriate entry vector and cured BsaI sites where necessary (Supplemental Table S3). Genomic sequences containing endogenous introns were chosen based on our and other data that these introns could facilitate transgene expression (Figure 2; Lumbreras et al., 1998). We assembled the gene sequences in destination vector pGG-pUM140_0016-BE-NOST-PIBT3-KmR and included a C-terminal YFP. Ulva gametes were transformed and YFP expression was visualized in blade cells. We observed expected nuclear localization for two tagged histone H3 proteins (encoded by UM001_0379 and UM003_0201), peroxisomal localization for the tagged protein encoded by an ortholog (UM079_0010) of the peroxisomal marker encoded by the Chlamydomonas gene CR10G07890, UM119_0052-YFP was localized in vacuole membranes and the tagged Ferredoxin ortholog (encoded by UM120_0017) localizes in Ulva chloroplasts (Figure 4; Supplemental Figure S8). Furthermore, we were able to generate tagged lines for Ulva orthologs coding for α-tubulin (TUA2; UM021_0090), EF-hand-containing Ca2+/calmodulin-dependent protein kinase (EF; UM005_0221), 20-kDa translocon at the inner membrane of chloroplasts (TIC20; UM014_0107) and Translocase of chloroplast 34 (TOC34; UM025_0020; Figure 4; Supplemental Figure S8). No or unclear localization was observed for the remaining two tagged genes.

In conclusion, we illustrate that a variety of Ulva marker lines can be generated by fusing TPs or endogenous proteins to a fluorescent marker.

Discussion

Ulva mutabilis is a model organism for multicellular green seaweeds. With the completion of the Ulva genome project (De Clerck et al., 2018), the development of a flexible molecular toolkit is required for continued functional genomic research in this species. Prior to this report, stable transgene expression in Ulva had only been demonstrated for the BleR cassette, allowing selection of transformants (Oertel et al., 2015). The original vectors contain multiple cloning sites that permit the creation of larger expression vectors containing more modules (Oertel et al., 2015). However, to allow more uniform and modular assembly of expression vectors, we opted to base our toolkit on the GreenGate cloning system (Lampropoulos et al., 2013). Our toolkit differs from other MoClo-based modular kits generated for Chlamydomonas and Cyanobacteria (Crozet et al., 2018; Vasudevan et al., 2019) that allow fusion of 11 sites. The GreenGate strategy contains six modules (Lampropoulos et al., 2013), which implies a more straightforward design for assembly of common expression vectors, but still permits to generate more complex constructs such as gene stacking (Lampropoulos et al., 2013). Practically, the Ulva molecular toolkit is largely compatible with the other MoClo-based systems following PCR to add the necessary overhangs and incorporation into the appropriate entry (or level 0) plasmids. Since BsaI is also used for assembly of level 1 plasmids in MoClo toolkits (Patron et al., 2015; Crozet et al., 2018; Vasudevan et al., 2019), only BpiI/BbsI or BsmBI recognition sites need to be cured for 24 or 16 of the 105 entry vectors described here, respectively, when level 2 assemblies are intended using other Golden Gate standards. In this report, we demonstrate the efficient generation of transgenic Ulva lines expressing transgenes, a feature that has not been achieved before in other multicellular algae. Ulva, therefore, has the potential to complement unicellular model algae, such as Chlamydomonas as a workhorse for functional genetic research and as a chassis for synthetic biology.

Although it remains unclear what sequence characteristics are indispensable for a functional promoter in (green) algae, we report 15 functional constitutive promoters for Ulva with apparent varying degrees of activity. Interestingly, promoter regions of homologous genes between Ulva and Chlamydomonas are functional. Not only the endogenous promoter of the small subunit of rubisco gene are functional both in Ulva and Chlamydomonas, but also promoter regions from PSAD and two genes encoding chlorophyll a/b binding proteins (Blankenship and Kindle, 1992; Fischer and Rochaix, 2001). Additional promoter screens to identify conditional promoters that are only active upon addition of a stimulus (e.g. light or heat shock), in specific cell types or during specific developmental phases would be a useful addition to the collection we report here.

Our study illustrates that an endogenous intron sequence in the BleR resistance cassette is functional in different configurations (Lumbreras et al., 1998; Oertel et al., 2015). Importantly, RbcsI does not necessarily need to be introduced in BleR to promote transgene expression but can be placed immediately before the TLSS. This implies that transcriptional enhancers do not necessarily need to be built into a coding sequence to increase transgene expression. However, in contrast to Chlamydomonas (Lumbreras et al., 1998), we did not observe an increase in transformation efficiency when RbscI was included more upstream in a promoter sequence. The endogenous promoter pUM140_0016 promotes transgene expression in a vector system containing two expression cassettes without the addition of RbscI, suggesting that all necessary regulatory and enhancer elements are present.

We developed two vector systems for transgene expression in Ulva, one containing two expression cassettes and one for bi- or multicistronic cassettes. Our vector containing two expression cassettes generates more transgenic individuals with detectable transgene expression compared with the model Chlamydomonas (67% versus 13% Onishi and Pringle, 2016). In a bicistronic context, the transgene is forced to be transcribed together with BleR. For the creation of bicistronic expression cassettes, we tested four different 2A ribosome-skipping peptides but alternative approaches exist, such as the addition of internal ribosome entry site (IRES) or even short linkers of unstructured sequences between the selectable marker and the GOI (Onishi and Pringle, 2016). Here, we chose to characterize 2A peptides as this was reported to be compatible with BleR expression in Chlamydomonas, whereas IRES or short linkers were not (Rasala et al., 2012; Onishi and Pringle, 2016). Although both vector systems allow high expression of transgenes in Ulva, immunoblot analysis suggests that while ribosomal skipping is efficient, it is not perfect. As a consequence, we advise to first attempt overexpression using two expression cassettes.

We successfully developed 18 Ulva marker lines for six intracellular locations (chloroplast, ER, microbodies, mitochondria, nucleus, and vacuole). These are important references to investigate the intracellular localization of tagged genes. We generated multiple independent lines targeting three intracellular compartments. Our results suggest that N7, AtpA (CDS), and PMS result in highly specific localization in nucleus, mitochondria, and microbodies, respectively, whereas more aspecific YFP signal was observed in transgenic lines expressing other TPs. Importantly, we demonstrate that eight TPs tested are functional both in Chlamydomonas and Ulva. Furthermore, we illustrate Ulva genes orthologous to known Chlamydomonas marker genes show intracellular localization consistent with their proposed function, e.g. nuclear localization for two proteins encoding a histone H3 protein. With the generation of these lines, we report the first tagged endogenous genes described in Ulva. Generating transgenic lines with tagged transgenes allows the inspection of intracellular localization, but also enables the development of techniques, such as chromatin immunoprecipitation or affinity purification–mass spectrometry.

We are currently able to generate gain-of-function mutants in Ulva. Although the generation of T-DNA insertion mutants has been described (Oertel et al., 2015), the development of RNAi techniques or programmable genome editing techniques, such as CRISPR/Cas9 would be an important addition to the molecular toolkit of Ulva. The overall low efficiency of CRISPR/Cas9 reported in algae and the single functional selection marker (BleR) in Ulva make the development of this technique challenging at this time. Although Ulva is resistant to many antibiotics, or the symbiotic bacteria are sensitive to antibiotics hampering proper development (Oertel et al., 2015), identifying novel resistance markers is a subject of ongoing research.

In summary, the toolkit described here allows the generation of Ulva transgenic lines stably overexpressing transgenes. We illustrate this with the expression of fluorescent proteins, marker lines, and tagged endogenous proteins. To our knowledge, Ulva is the first seaweed that is amendable for these experiments at this time. Besides enabling molecular studies in relevant processes for Ulva biology (De Clerck et al., 2018), the homologs of genes from model microalgae, such as Chlamydomonas can now be studied in a multicellular context.

Materials and methods

Algae growth and cultivation

Ulva mutabilis Føyn “slender” mutant strain is a direct descendant of the original isolates of B. Føyn from the South Atlantic coast of Portugal near Ohlão and Faro (Føyn, 1958). The untransformed individuals were kindly provided by Thomas Wichard (University of Jena, Germany). Individuals were cultivated in tissue culture chamber under long-day conditions (16-h light:8-h dark; 21°C; 75 µM; Spectralux Plus NL-T8 36W/840/G13 fluorescent lamp) in standard Petri dishes containing synthetic UCM (Stratmann et al., 1996; Boesger et al., 2018). Ulva was grown and parthenogenetically propagated as described previously (Wichard and Oertel, 2010).

Ulva transformation

Ulva gametes were transformed using the PEG-mediated protocol described in (Oertel et al., 2015; Boesger et al., 2018). For each transformation, 3–5 µg of plasmid DNA was used per construct. Plasmids were purified using a GeneJet Plasmid Miniprep kit (Thermo Scientific, Waltham, MA, USA ). About 50-µg/mL Phleomycin (Invivogen, San Diego, CA) was added 48 h after transformation, resistant germlings develop parthenogenetically from the transformed gametes and can be detected after 2–3 weeks of cultivation. For each transformation experiment, nonvector controls were included to confirm proper selection. For experiments where relative transformation efficiency was scored, all constructs and controls were transformed in the same transformation experiment to reduce variability in transformation efficiency between different batches of gametes and to ensure the same input DNA was delivered in the same number of input gametes. Resistant individuals were counted and removed from the transformation plates; additional selection medium was added to allow detection of slower growing germlings. Mutant cultures are preserved and are available upon request.

Vector assembly

Vector assembly is based on the GreenGate cloning system (Lampropoulos et al., 2013). Sequences for entry modules were obtained via gene synthesis (BioXP3200, Codex DNA) or PCR-amplification of the target DNA sequence using primers that contain a ∼20-bp overlap with the entry module using high-fidelity polymerase Q5 (NEB, Ipswich, MA, USA) or GXL (Takara Bio, Shiga, Japan). After gel electrophoresis, the amplicon is purified (Zymoclean Gel DNA Recovery Kit; Zymo Research, Irvine, CA) and mixed with the predigested (BsaI; Thermo Scientific) entry clone and NEBuilder (NEB). After assembly (15–60 min at 50°C), the mix is transformed into chemically competent DH5α Escherichia coli cells and plated out on LB + 100 µg/mL Carbenicillin. PCR-mediated silent mutations were introduced to remove internal BsaI recognition sites. All template plasmids containing genes encoding fluorescent proteins or TPs were ordered via Chlamydomonas Research Center (https://www.chlamycollection.org; Supplemental Tables S1 and S2). All Ulva gene and promoter sequences were cloned from genomic DNA, isolated via CTAB (Clarke, 2009) method or Omniprep for plant (G-Biosciences, St. Louis, MO).

For a Golden Gate reaction, 100 ng of each entry and destination vector are assembled in one reaction mix containing 10 U BsaI-HF-v2 (NEB), 200 U T4 Ligase (NEB), 1 mM ATP (Thermo Scientific), and 1× Cutsmart buffer (NEB). The reaction mix is incubated at 30× (18°C [3 min] and 37°C [3 min]), followed by 5 min 50°C and 5 min 80°C. After assembly, the mix is transformed in DH5α E. coli cells and plated out on LB + 25 µg/mL Kanamycin. Vectors were validated using Sanger sequencing (Mix2seq; Eurofins Luxembourg, UK) and restriction digest, typically using NcoI (Promega, Madison, WI, USA), PstI (Promega), or XhoI (NEB).

All entry clones in this study are designed as described in (Lampropoulos et al., 2013). The destination vector used is a high-copy number pGG-AG-KmR. The ccdB/CmR cassette from pEN-L1-AG-L2 (Houbaert et al., 2018) was amplified using pGG-AG-KmR primers (Supplemental Table S4). The PCR product was purified and DNA was digested by ApaI (Promega), then the digest was purified. The pEN-L1-AG-L2 vector was digested by EcoRV and ApaI (Promega) and purified from gel. The vector fragment was ligated to the ApaI digested PCR product to generate pGG-AG-KmR. For vectors with two expression cassettes, the BleR resistance cassette was inserted proximal to the G BsaI recognition site. The pGG-AG-KmR was digested using EcoRV and NotI (Promega) and purified from gel, the PIBT3 or PIBGT cassettes were amplified from pPIBT3 or pPIBGT8 (Oertel et al., 2015) using pGG-AG-PIBT-KmR primers (Supplemental Table S4) and gel purified. The amplicons were inserted in the digested pGG-AG-KmR vector via Gibson Assembly to generate pGG-AG-PIBT3-KmR or pGG-AG-PIBGT-KmR, where PIBGT contains a BleR-GFP fusion.

For the construction of modified destination vectors, first a Golden Gate reaction was performed containing all desired fragments and two or three variable linkers containing AarI sites and the SacB counterselectable marker (Decaestecker et al., 2019). After sequence verification, the vectors were digested with AarI and purified from gel. The ccdB/CmR cassette from pEN-L1-AG-L2 (Houbaert et al., 2018) was amplified using B-ccdB/CmR-G or C-ccdB/CmR-E primers (Supplemental Table S4). The amplicons were inserted in the digested vectors vector via Gibson Assembly to generate modified destination vectors for two expression cassettes (pGG-pUM140_0016-BE-NOST-PIBT3 or pGG-pUM140_0016-BE-NOST-PIBGT) or one expression cassette (pRbcS-BleR-2A-CE-tRbcS, pRbcS-BleR-2A-BG-tRbcS, pUM140_0016-BleR-2A-CE-tRbcS, or pUM140_0016-BleR-2A-BG-tRbcS containing E2A, F2A, P2A, or T2A; Supplemental Figure S7). All destination vectors are transformed in One Shot ccdB Survival 2 T1R Chemically Competent Cells (ThermoFisher).

For the annotation of 5′ UTR regions, the gene sequences were searched against the full-length transcripts database, assembled with in-house RNAseq reads generated by Trueseq and Nanopore sequence technology, using BLAST. The CDS sequence and the full-length transcripts were aligned by the online server MAFFT (https://mafft.cbrc.jp/alignment/server/). The sequence of the full-length transcripts located before the start codon is identified as the 5′ UTR of the gene.

All vectors generated in this study (Supplemental Table S1) are available at https://gatewayvectors.vib.be.

Confocal imaging

Individual thalli of 1-month-old resistant individuals were selected for imaging. Entire individuals were imaged using an Opera Phenix High-Content Screening System (Perkin Elmer, Waltham, MA) and analyzed using Columbus software. Detailed imaging of tissues and cells was performed using Olympus Fluoview (FV1000) confocal microscope with FluoView software. The following excitation and emission settings were used: YFP, 515 excitations with 530–545 emission; mCherry, 559 excitations with 564–644 emission; mTagBFP2 405 excitation with 446–546 emission and chlorophyll, 559 excitations with 650–750 emission. Laser intensity was adapted in the ∼20%–50% range, depending on expression variety between individuals. A single laser intensity was chosen when such variation was relevant. For each transgenic line, 4–10 independent individuals were screened, a representative image is selected for all lines with a functional construct.

RT-qPCR

One-month-old resistant individuals were harvested, flash frozen and RNA was extracted using the ReliaPrep RNA Tissue Miniprep System (Promega) according to the manufacturer’s specifications. RT-qPCR experiments were performed in a Light-Cycler480 Real-Time SYBRgreen PCR system (Roche, Basel, Switzerland). About 500 ng of RNA was reverse-transcribed with the qScript cDNA Supermix kit (QuantaBio, Beverly, MA) according to the manufacturer’s specifications. RT-qPCR results were normalized against two reference genes (EFLa; UM025_0053 and UBQ10; UM008_0183). For each experiment, three technical replicates were performed on the same biological replicate. Each biological replicate consists of at least 10 pooled independent primary transformants grown under the same conditions; harvesting and further sample handling were done in separate tubes for the indicated transgenic lines. All primers used in this study are listed in Supplemental Table S4.

Protein extraction and immunoblot

Protein extracts were extracted from different transgenic lines using in a 2/3 ratio (v/w) of extraction buffer (25 mM Tris–HCl, pH 7.6, 15 mM MgCl₂, 150 mM NaCl, 15 mM p-nitrophenyl phosphate, 60 mM β-glycerophosphate, 0.1% (v/v) Nonidet P-40, 0.1 mM sodium vanadate, 1 mM NaF, 1 mM PMSF, 1 μM trans-epoxysuccinyl-l-leucylamido-(4-guanidino)butane (E64), 5% (v/v) ethylene glycol; EDTA-free Ultra Complete Tablet; Sigma, St Louis, MO, USA). After two freeze-thaw cycles in extraction buffer, the soluble protein fraction was obtained by two centrifugation steps at 14,000 rpm (Eppendorf 5424) for 15 min 4°C. Concentration was determined via Bradford assay (Bio-rad, Hercules, CA). About 15–20 µg protein was loaded on a 4%–15% Mini-PROTEAN TGX Precast gel (w/v; Bio-Rad). After SDS-PAGE, proteins were transferred to PVDF membrane using Trans-Blot Turbo Transfer System (Bio-Rad). Correct transfer was visualized using Ponceau Staining. After destaining, an excess of 3% SM-milk (w/v) in TBT was applied to block the membrane (RT; 20 min). The blot was incubated overnight on a shaker (4°C) with the primary antibody (anti-GFP 3H9, anti-RFP 6G6; Chromo Tek, Munich, Germany or anti-tRFP; Evrogen, Moscow, Russia). After three washes with TBT buffer, the secondary HRP-conjugated antibody was added (anti-mouse A9917 [Sigma] or anti-rabbit GENA934 [Sigma]; RT, 1 h). Detection was performed using Clarity Western ECL Substrate (Bio-Rad) and a ChemiDoc MP System (Bio-Rad). Each biological replicate consists of at least 10 independent primary transformants grown under the same conditions; harvesting and further sample handling were done in separate tubes for the indicated transgenic lines.

Large datasets

All generated vectors of the toolkit (Supplemental Table S1) are available on https://gatewayvectors.vib.be. Transgenic lines are maintained and are available from the corresponding author upon reasonable request.

Accession numbers

All accession numbers are indicated in Supplemental Tables S1–S3.

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Promoter screen gene identifiers and cloning strategy.

Supplemental Figure S2. Promoter screen expression analysis.

Supplemental Figure S3. Visualization of YFP expression in Ulva transgenic lines expressing pUM140_0016-RbcsI-YFP-NOST or pUM140_0016-YFP-NOST.

Supplemental Figure S4. Expression analysis of transgenic Ulva lines.

Supplemental Figure S5. Expression analysis of transgenic Ulva lines using bicistronic expression vectors.

Supplemental Figure S6. Custom destination vectors cloning scheme.

Supplemental Figure S7. Expression analysis of transgenic Ulva marker lines using TPs.

Supplemental Figure S8. Expression analysis of transgenic Ulva marker lines using endogenous genes.

Supplemental Table S1. Complete list of generated vectors (entry, destination, and expression).

Supplemental Table S2. Selection of TPs.

Supplemental Table S3. Selection of Ulva marker genes.

Supplemental Table S4. List of primers used in this study.

 

J.B., T.J., and O.D.C. designed the study. J.B. performed all experiments and analyzed the data. X.L. maintained Ulva cultures for experiments and annotated 5′ UTR sequences. J.B. wrote the manuscript with contributions from all authors. O.D.C. agrees to serve as the author responsible for contact and ensures communication.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/general-instructions) is: Olivier De Clerck (olivier.declerck@ugent.be).