An Agrobacterium-mediated stable transformation technique for the hornwort model Anthoceros agrestis.

Despite their key phylogenetic position and their unique biology, hornworts have been widely overlooked. Until recently there was no hornwort model species amenable to systematic experimental investigation. Anthoceros agrestis has been proposed as the model species to study hornwort biology. We have developed an Agrobacterium-mediated method for the stable transformation of A. agrestis, a hornwort model species for which a genetic manipulation technique was not yet available. High transformation efficiency was achieved by using thallus tissue grown under low light conditions. We generated a total of 274 transgenic A. agrestis lines expressing the β-glucuronidase (GUS), cyan, green, and yellow fluorescent proteins under control of the CaMV 35S promoter and several endogenous promoters. Nuclear and plasma membrane localization with multiple color fluorescent proteins was also confirmed. The transformation technique described here should pave the way for detailed molecular and genetic studies of hornwort biology, providing much needed insight into the molecular mechanisms underlying symbiosis, carbon-concentrating mechanism, RNA editing and land plant evolution in general.

Introduction

The hornworts are one of the three lineages of bryophytes that diverged from the monophyletic group of liverworts and mosses over 460 million years ago (Morris et al., 2018; One Thousand Plant Transcriptomes Initiative, 2019; Li et al., 2020). Although having only around 220 extant species (Söderström et al., 2016), hornworts are key to address diverse questions about land plant evolution and terrestrialization. Hornworts display a unique combination of features (Frangedakis et al., 2020) such as a sporophyte that is produced by an indeterminate basal meristem, and has stomata similar to mosses and vascular plants (Renzaglia et al., 2017). In addition, it is the only extant land plant lineage (together with a few Selaginella species (Liu et al., 2020)), that has a single (or just a few) algal‐like chloroplast(s) per cell. The chloroplasts resemble those of algae in that they may contain pyrenoids, a carbon‐concentrating structure that is shared with many algal lineages (Villarreal & Renner, 2012; Li et al., 2017). Hornwort plastids are also unique by exhibiting the highest RNA editing rates amongst land plants (Yoshinaga, 1996, 1997; Kugita, 2003; Small et al., 2019). Finally, hornworts are among the very few plant lineages that can establish symbiotic relationships with both endophytic cyanobacteria (Renzaglia et al., 2009) and various glomeromycotina and mucoromycotina fungal partners (mycorrhiza) (Desirò et al., 2013).

Anthoceros agrestis has been established as an experimental model system for hornworts, and two isolates are currently available (Oxford and Bonn) (Szövényi et al., 2015). Anthoceros agrestis, like other bryophytes, has a haploid‐dominant life cycle through which the haploid gametophyte phase alternates with the diploid sporophyte phase. The life cycle of A. agrestis starts with the germination of the haploid spores which develop into an irregularly shaped thallus (Fig. 1a–c). Sexual reproduction occurs through fusion of the egg (produced in archegonia) and the motile sperm (produced in antheridia). The resulting embryo develops within the gametophyte and gives rise to the sporophyte (Fig. 1d). Anthoceros agrestis can be easily grown under laboratory conditions and its haploid‐dominant life cycle makes genetic analysis straightforward. The nuclear genome of A. agrestis was recently sequenced (Li et al., 2020; available at https://www.hornworts.uzh.ch/en/about.html) and is one of the smallest genomes amongst land plants (genome size: 116/123 Mb for the Bonn and Oxford isolates, respectively). The species is monoicous, with male and female reproductive organs produced by the same individual. The sexual life cycle of A. agrestis can be completed under laboratory conditions within c. 2–3 months (Szövényi et al., 2015). Efficient genetic manipulation methods have been published for the mosses Physcomitrium patens (Rensing et al., 2020), Ceratodon purpureus (Zeidler et al., 1999; Trouiller et al., 2007; Finiuk et al., 2014) and Scopelophila cataractae (Nomura et al., 2016), and the liverworts Marchantia polymorpha (Kohchi et al., 2021) and Riccia fluitans (Althoff & Zachgo, 2020). However, transformation of A. agrestis was not feasible until recently, posing a major obstacle to the analysis of gene function in hornworts, and more generally to land plant evo‐devo studies.

Fig. 1

Morphological features of Anthoceros agrestis and effect of light on growth. (a) Light micrograph (LM) of a germinating spore. Bar, 50 μm. (b) Surface view of the irregularly shaped thallus (gametophyte). Bar, 1 mm. (c) LM showing cells of mature gametophyte tissue with single chloroplasts. Bar, 10 μm. (d) Anthoceros agrestis Oxford gametophyte with sporophytes. Bar, 4 mm. (e) The conditions for the preparation of tissue used for transformation are critical. Plants must be propagated in axenic culture by transferring small thallus fragments (typically 1 × 1 mm) onto plates with fresh growth medium using sterile scalpels and then grown under low light conditions (3–5 μmol m−2 s−1) for 4 wk. (f) Anthoceros agrestis Oxford thallus tissue grown for 4 wk under low light intensity (3–5 μmol m−2 s−1). (g) Anthoceros agrestis Oxford thallus tissue grown for 4 wk under high light intensity (80 μmol m−2 s−1). Bars, 1 mm. Tissue morphology after: (h) the first homogenization before cocultivation, (i) 1 month after the first homogenization, (j) after the second homogenization before cocultivation. Bars, 1 mm.

Several approaches have been used for gene delivery in bryophytes, including polyethylene glycol (PEG)‐mediated uptake of DNA by protoplasts (Schaefer et al., 1991), particle bombardment (Cho et al., 1999; Chiyoda et al., 2008) and Agrobacterium tumefaciens (revised scientific name Rhizobium radiobacter (Young et al., 2001)) mediated transformation (Ishizaki et al., 2008; Kubota et al., 2013; Althoff & Zachgo, 2020). Agrobacterium‐mediated transformation is a commonly used method for various plant species (Gelvin, 2003), as it is relatively simple and does not require specialized or expensive equipment. In addition, Agrobacterium‐mediated transformation has several advantages over other transformation methods, such as the integration of a lower number of transgene copies into the plant genome and the ability to transfer relatively large DNA segments with intact transgene genome integration.

In this study we report the first successful stable genetic transformation method for hornworts. The method is based on Agrobacterium‐mediated transformation of A. agrestis thallus. We also show the successful expression and targeting of four different fluorescent proteins in two different cellular compartments, the plasma membrane and the nucleus. Finally, we characterize a number of native A. agrestis promoters for their potential to drive strong constitutive transgene expression that can be useful for future hornwort genetic studies.

Materials and Methods

Plant material and maintenance

In this study we used the A. agrestis Oxford and Bonn isolates (Szövényi et al., 2015). Anthoceros agrestis thallus tissue was propagated on KNOP medium (0.25 g l−1 KH2PO4, 0.25 g l−1 KCl, 0.25 g l−1 MgSO4.7H2O, 1 g l−1 Ca(NO3)2.4H2O and 12.5 mg l−1 FeSO4.7H2O). The medium was adjusted to pH 5.8 with KOH and solidified using 7.5 g l−1 Gelzan CM (G1910; Sigma) in 92 × 16 mm Petri dishes (82.1473.001; Sarstedt, Leicester, UK) with 25–30 ml per plate. Plants were routinely grown in a tissue culture room (21°C, 12 h : 12 h light : dark, 3–5 or 35 μmol m−2 s−1 light intensity, Philips TL‐D 58W (835)). To subculture the thallus tissue, a small part of it was cut using sterile disposable scalpels (0501; Swann Morton, Sheffield, UK) and placed on fresh media on a monthly basis (see Supporting Information Fig. S1).

For sporophyte induction, A. agrestis Bonn gametophyte tissue was subcultured as described above and grown for 2–3 months. Fertilization was facilitated by adding 5 ml sterile water per Petri dish when the first antheridia were observed. Several sporophytes appeared after 2 wk, at which point excess moisture was removed from the Petri dishes and the sporophytes were allowed to grow further and mature. Once the majority of sporophytes started to turn brown at their tips, the Petri dishes were removed from the humidity‐controlled light chamber and put into a dryer environment (e.g. lab bench) to facilitate sporophyte maturation. Once all the sporophytes had turned completely brown and looked dry, they were collected with tweezers under a laminar hood and stored for up to 8 months at 4°C (in a fridge) in 60 × 15 mm Petri dishes (82.1194.500; Sarstedt). Sporophytes were used for downstream applications at the earliest after 3 months of storage at 4°C to allow for complete desiccation.

Cocultivation medium

Cocultivation medium was liquid KNOP supplemented with 2% sucrose (0.25 g l−1 KH2PO4, 0.25 g l−1 KCl, 0.25 g l−1 MgSO4.7H2O, 1 g l−1 Ca(NO3)2.4H2O, 12.5 mg l−1 FeSO4.7H2O and 20 g l−1 sucrose, pH 5.8 adjusted with KOH).

Tissue preparation for transformation

Approximately 2 g of thallus tissue (15–20 Petri dishes) was divided into four parts, and each part (c. 0.5 g) was homogenized in 15 ml sterile water using a homogenizer (727407; IKA Ultra‐Turrax T25 S7 Homogenizer, Oxford, UK) and corresponding dispensing tools (10442743; IKA Dispersing Element), for 5 s, using the lowest speed of 8000 rpm. The homogenized tissue was washed with 50 ml sterile water using a 100 µm cell strainer (352360; Corning, Flintshire, UK), plated onto solid KNOP medium and placed at 21°C under 12 h : 12 h, light : dark at a light intensity of 3–5 μmol m−2 s−1. After 4 wk the tissue was rehomogenized in 15–20 ml sterile water and filtered using a 100 µm cell strainer. The rehomogenized tissue was transferred again onto four plates with solid KNOP medium and was allowed to grow for 2 d at 21°C under continuous light (35 μmol m−2 s−1; TL‐D58W/835; Philips).

For sporeling‐transformation, 20 mature and desiccated sporophytes were pooled in a 1.5 ml Eppendorf tube and the spores were surface sterilized as described by Sauret‐Güeto et al. (2020). The spores were incubated in the sterilizing solution for 15 min and then centrifuged at 10 000  for 2 min. The spore pellet was resuspended in 150 μl fresh sterilizing solution and plated onto solid KNOP medium with an additional 5 ml liquid KNOP medium added to the Petri dish. The Petri dishes were closed with micropore tape and kept at 21°C under 12 h : 12 h light : dark at a light intensity of 35 μmol m−2 s−1. The first germinating spores were observed 5–8 d after plating. Sporelings were left to grow for 7, 14 and 21 d (after observation of the first spore germination). For Agrobacterium cocultivation, the sporelings were scraped off from the surface of the solid KNOP medium with sterile scalpels and collected and washed using a 40 µm cell strainer (431750; Corning). Sporelings from a single Petri dish (20 sporophytes) were transferred into a well of a six‐well plate with Agrobacterium and transformation buffer (for cocultivation conditions see Agrobacterium culture preparation and cocultivation conditions).

Agrobacterium culture preparation

One to three Agrobacterium colonies (strain AGL1) were inoculated in 5 ml of LB medium supplemented with rifampicin 10 μg ml−1 (R0146; Duchefa, Haarlem, the Netherlands), carbenicillin 50 μg ml−1 (C0109; Melford, Chelsworth, UK) and the plasmid‐specific selection antibiotic spectinomycin 100 μg ml (SB0901; Bio Basic, Toronto, Canada). The preculture was incubated at 28°C for 2 d with shaking at 120 rpm. OD600 was c. 2.7 and was measured using an OD600 DiluPhotometer (Implen, München, Germany).

Cocultivation conditions

In total, 5 ml of a 2 d Agrobacterium culture was centrifuged for 7 min at 2000  . The supernatant was discarded, and the pellet was resuspended in 5 ml liquid KNOP supplemented with 2% (w/v) sucrose (S/8600/60; ThermoFisher, Loughborough, UK) and 100 μM 3′,5′‐dimethoxy‐4′‐hydroxyacetophenone (acetosyringone) (115540050; Acros Organics, dissolved in dimethyl sulfoxide (DMSO) (D8418; Sigma)). The culture was incubated with shaking (120 rpm) at 28°C for 5 h. The regenerating thallus tissue was transferred (one plate – 2 d after the second homogenization) into a well of a six‐well plate with 5 ml of liquid KNOP medium supplemented with 2% (w/v) sucrose. Then, 80 μl of Agrobacterium culture and acetosyringone at final concentration of 100 μM were added to the medium.

The tissue and Agrobacterium were cocultivated using a six‐well plate (140675; ThermoFisher) for 3 d with shaking at 110 rpm at 22°C on a shaker without any additional supplementary light (only ambient light from the room, 1–3 μmol m−2 s−1). After 3 d, the tissue was drained using a 100 µm cell strainer (352360; Corning) and moved onto solid KNOP plates (two Petri dishes from a single well) supplemented with 100 μg ml−1 cefotaxime (BIC0111; Apollo Scientific, Bredbury, UK) and 10 μg ml−1 hygromycin (10687010; Invitrogen). After 3–4 wk, plants were transferred to fresh solid KNOP plates supplemented with 100 μg ml−1 cefotaxime and 10 μg ml−1 hygromycin and grown at 22°C under 12 h : 12 h light : dark at a light intensity of 35 μmol m−2 s−1 (TL‐D58W/835; Philips).

GUS staining

Glucuronidase (GUS) staining was performed according to Plackett et al. (2014). GUS buffer stock solution was prepared by mixing 11.54 ml of 1 M Na2HPO4, 8.46 ml of 1 M NaH2PO4, 8 ml of 0.25 M EDTA (pH 7.0), 10 μl of 20% Tween20 (v/v) and topped up to 200 ml with dH2O. GUS staining solution (10 ml) was prepared by dissolving 10 mg 5‐bromo‐4‐chloro‐3‐indolyl‐β‐d‐glucuronic acid (X‐glcA) (B72200‐0.25; Melford) in 1 ml DMSO and added to 9 ml GUS buffer stock solution (final X‐glcA concentration 1 mg ml−1). The tissue was first fixed in cold 90% (v/v) acetone for 10 min. The acetone was removed, and the tissue was rinsed with cold dH2O. After that dH2O was removed and GUS staining solution was added. The samples in the solution were placed under vacuum for 4 min twice to enable efficient infiltration to the tissues. Then the sample dish was wrapped in tinfoil and incubated at 37°C overnight. Before visualization, Chl was removed (bleached) by a rising ethanol series (v/v): 15, 30, 50, 70 and 100%.

Genomic DNA extraction

A modified CTAB protocol from Porebski et al. (1997) was used for extraction of hornwort genomic DNA. Tissue (0.5 g) was harvested and frozen in liquid nitrogen. Tissue was ground to a fine powder using a chilled mortar and pestle and then added to 10 ml of DNA extraction buffer (100 mM Tris‐HCl pH 8, 1.4 M NaCl, 20 mM EDTA pH 8, 2% (w/v) CTAB, 0.3% (v/v) β‐mercaptoethanol and 100 mg polyvinylpyrrolidone (PVP)/g of tissue) that had been prewarmed at 60°C, 100 μl of RNAse A (100 mg ml−1) was added and the solution was mixed well. The mixture was incubated at 60°C for 30 min and then removed from heat and allowed to cool at room temperature for 4 min. Then, 12 ml of chloroform : isoamyl alcohol (24 : 1) was added, mixed well by inversion and then centrifuged at 12 000  for 10 min at room temperature. The upper aqueous phase was transferred to a new 50 ml centrifugation tube and 10 ml of chloroform : isoamyl alcohol (24 : 1) was added, mixed well by inversion and then centrifuged at 6000  for 10 min at room temperature to remove any remaining PVP in the aqueous phase. The upper aqueous phase was transferred to a new 50 ml centrifugation tube and ½ volume of 5 M NaCl was added. Two volumes of cold (−20°C) 95% (v/v) ethanol were also added and the contents of the tube were mixed well by inversion. The tube was spun at 20 000  for 6 min. The pellet was resuspended in 2 ml of TE buffer and the previous step was repeated. The pellet was washed with cold 70% (v/v) ethanol. The pellet was dried and dissolved in 500 μl of TE buffer and then stored at 4°C.

Promoter identification and isolation

We used RNA‐sequencing (RNA‐seq) data to find genes showing constantly high levels of expression under various developmental stages and experimental conditions (‘constitutively expressed genes’). To do so, we estimated expression of genes under three developmental stages of the gametophyte and sporophyte phases and in symbiosis with cyanobacteria. We retrieved raw RNA‐seq data for these experiments from Li et al. (2020). We used trimmomatic to quality filter and trim the raw reads. Gene expression was estimated using salmon (Patro et al., 2017) and expressed as normalized expression counts. We identified candidates by selecting those showing the highest average expression level and the least gene expression variability across all conditions investigated. We then manually selected a subset of genes taking account their genomic location, exact expression pattern, and the length and sequence composition of their putative promoter sequences. We also assessed the suitability of our candidate promoters using the information available for M. polymorpha and P. patens (Table S1).

Putative promoter sequences were amplified from genomic DNA using KOD Hot start polymerase (71086‐5; Merck Millipore, Darmstadt, Germany) and cloned into pJET1.2 (K1231; ThermoFisher) before Sanger sequencing. Loop assembly‐compatible DNA parts were generated according to Sauret‐Güeto et al. (2020). Briefly, aliquots of the DNA parts (15 nM) and the pUAP4 acceptor vector (7.5 nM) were prepared. A type IIS assembly reaction was set up in a 0.2 ml tube (5 μl nuclease‐free H2O, 1 μl 10× CutSmart buffer (B7204S; NEB, Cambridge, UK), 0.5 μl 1 mg ml−1 BSA (B9000S; NEB), 1 μl 10 mM ATP (P0756S; NEB), 0.25 μl 400 U μl−1 T4 DNA ligase (M0202S; NEB), 0.25 μl 10 U μl−1 SapI (R0569S; NEB), 1 μl of L0 DNA part, and 1 μl of pUAP4) to clone the amplified DNA part into pUAP4. Samples were incubated in a thermocycler using the following programme: assembly: 26 cycles of 37°C for 3 min and 16°C for 4 min; termination and enzyme denaturation: 50°C for 5 min and 80°C for 10 min. Chemically competent Escherichia coli cells (30 μl) were transformed using 5 μl of the assembly reaction and then spread on LB agar plates containing 25 μg ml−1 chloramphenicol and 40 μg ml−1 X‐gal. Construct sequences are detailed in Table S2, and a list of primers in Table S3. Promoter sequences, their genomic localization and RNA‐seq coverage are also given in Notes S1.

Construct generation

Constructs were generated using the OpenPlant toolkit (Sauret‐Gueto et al., 2020). OpenPlant L0 parts used were: OP‐040 CTAG_linker‐N7, OP‐023 CDS12‐eGFP, OP‐020 CDS_hph, OP‐027 CDS12_mTurquoise2, OP‐029 CDS12_mVenus, OP‐037, CTAG_Lti6b, OP‐054 3TER_Nos‐35S and OP‐049 PROM_35S. For L1 construct generation: L0 plasmids with the DNA parts to be assembled were prepared at a concentration of 15 nM, and the acceptor pCk vector at a concentration of 7.5 nM. The loop assembly Level 1 reaction master mix contained 2 μl nuclease‐free H2O, 1 μl 10× CutSmart buffer (B7204S; NEB), 0.5 μl 1 mg ml–1 BSA (B9000S; NEB), 1 μl 10 mM ATP (P0756S; NEB), 0.25 μl 400 U μl−1 T4 DNA ligase (M0202S; NEB) and 0.25 μl 20 U μl−1 BsaI HF v2 (R3733S; NEB). Cycling conditions were 26 cycles of 37°C for 3 min and 16°C for 4 min. Termination and enzyme denaturation was 50°C for 5 min and 80°C for 10 min. Chemically competent E. coli cells (30 μl) were transformed using 5 μl of the Loop assembly reaction and spread on LB agar plates containing 50 μg ml–1 kanamycin and 40 μg ml–1 X‐Gal. For L2 construct generation: the protocol is the same as the L1 BsaI assembly protocol with the exception of the master mix composition: 2 μl nuclease‐free H2O, 2 μl nuclease‐free H2O, 1 μl 10× CutSmart buffer (B7204S; NEB), 0.5 μl 1 mg ml–1 BSA (B9000S; NEB), 1 μl 10 mM ATP (P0756S; NEB), 0.25 μl 400 U μl−1 T4 DNA ligase (M0202S; NEB) and 0.25 μl 10 U μl−1 SapI (R0569S; NEB), and spread on LB agar plates containing 100 μg ml−1 spectinomycin and 40 μg ml−1 X‐Gal. The donor plasmids were L1 constructs, and the acceptor plasmid was a pCsA vector (see Table S2 for full construct maps and sequences).

Western blotting

In total, 50 mg of A. agrestis thallus tissue was grown for 3 wk on KNOP medium at 21°C under continuous light (35 μmol m−2 s−1) and ground in liquid nitrogen. The tissue powder was resuspended in 500 μl 5× Laemmli loading buffer (0.2 M Tris‐HCl pH 6.8, 5% (w/v) sodium dodecyl sulfate, 25% (v/v) glycerol, 0.25 M dithiothreitol, 0.05% (w/v) bromophenol blue), supplemented with Roche cOmplete protease inhibitor (11836170001; Roche), heated at 95°C for 5 min and centrifuged at 10 000  for 10 min. The supernatant was transferred to a new tube. Equal amounts of proteins were separated by denaturing electrophoresis in NuPAGE gel (NP0322BOX; Invitrogen) and electrotransferred to nitrocellulose membranes using the iBlot2 Dry Blotting System (ThermoFisher). Enhanced green fluorescent protein (eGFP) was immunodetected with anti‐GFP antibody (1 : 4000 dilution) (JL‐8, 632380; Takara, Göteborg, Sweden) and anti‐mouse‐HRP (1 : 15 000 dilution) (A9044; Sigma) antibodies. Actin was immunodetected with antiactin (plant) (1 : 1500 dilution) (A0480; Sigma) and antimouse‐HRP (1 : 15 000 dilution) (A9044; Sigma) antibodies, using the iBind™ Western Starter Kit (SLF1000S; ThermoFisher). Western blots were visualized using the ECL™ Select Western Blotting Detection Reagent (GERPN2235; GE) following the manufacturer's instructions. Images were acquired using a G:BOX F3 Syngene Gel Documentation system.

Sample preparation for imaging

A gene frame (AB0576; ThermoFisher) was positioned on a glass slide and 30 μl of KNOP medium with 1.2% (w/v) Gelzan CM (G1910; Sigma) was placed within the gene frame. A thallus fragment was placed within the medium‐filled gene frame together with 30 μl milliQ water. The frame was then sealed with a cover slip. Plants were imaged immediately using a Leica SP8X spectral fluorescence confocal microscope.

For the regeneration test experiment, five thallus fragments were placed into a KNOP medium‐filled gene frame as described above (three slides and 15 plants in total). Images were acquired on a daily basis, for a total duration of 1 wk, using a Leica SP8X spectral fluorescence confocal microscope and a 10× air objective (HC PL APO 10×/0.40 CS2).

Imaging with confocal microscopy

Images were acquired on a Leica SP8X spectral confocal microscope. Imaging was conducted using either a 10× air objective (HC PL APO 10×/0.40 CS2) or a 20× air objective (HC PL APO 20×/0.75 CS2). Excitation laser wavelength and captured emitted fluorescence wavelength window were as follows: for monomeric Turquoise 2 fluorescent protein (mTurquoise2) (442 nm, 460−485 nm), for eGFP (488 nm, 498−516 nm), for mVenus and enhanced yellow fluorescent protein (eYFP) (515 nm, 522−540 nm), and for Chl autofluorescence (488 or 515, 670−700 nm). When observing lines expressing both eGFP and mTurquoise2, the sequential scanning mode was used.

Light microscopy

Images were captured using a Keyence VHX‐S550E microscope (VHX‐J20T lens) or a Leica M205 FA stereomicroscope (with GFP longpass (LP) filter).

Sequencing transformant line genomes

Transformant lines were grown on solid KNOP medium containing 10 μg ml−1 hygromycin and 100 μg ml−1 cefotaxime. Genomic DNA was extracted from 600 mg fresh tissue per line using either the DNeasy Plant Pro kit (69204; Qiagen) (Cam‐1 and Cam‐2 lines) or the procedure from Li et al. (2020) (BTI1‐3 lines) to reach a total yield of at least 200 ng per line. Illumina libraries were prepared using the TruSeq DNA nano kit (20015964; Illumina, Zurich, Switzerland) and were sequenced on an Illumina Novaseq 6000 platform with an expected sequencing depth of 80–150× for all five lines in paired‐end mode (read length: 151 bp).

After sequencing, we quality filtered and trimmed reads using trimmomatic (command line: ALL_TruSeq‐PE.fa:2:30:10:2:keepBothReads LEADING:3 TRAILING:3 SLIDINGWINDOW:4:20 MINLEN:36) and assembled the reads with spades3.14.1 using the ‐‐isolate ‐‐cov‐cutoff auto ‐‐only‐assembler options as recommended (Nurk et al., 2013). To localize the insertion and its copy number, we used the insert sequence as a query in a blastn search (Altschul et al., 1990) against the database containing the assembly (Altschul et al., 1990) with an e‐value threshold of 10−4. As evidence of genomic integration, we only accepted hits covering the full length of the insert sequence with one or no mismatches. We manually inspected blast hits to eliminate false positives. Finally, we used the A. agrestis Oxford genome sequence (Li et al., 2020) to localize the hits on the genomic scaffolds.

PCR confirmation of transformant insertion sites

Based on the genome assemblies of the transformant lines, primers were designed to amplify regions of 0.6–1.3 kb spanning the 5′‐ and 3′‐ends of the T‐DNA inserts and their respective adjacent genomic regions. Sequences were amplified from genomic DNA with Phusion High‐Fidelity DNA Polymerase (F‐530S; ThermoFisher). For each amplified region, a second set of nested primers were designed to amplify a shorter amplicon, using a 1 : 100 dilution of the previous PCR product as a template. Resulting nested PCR products were either purified and Sanger sequenced (Eurofins) from both ends, or cloned into pJET1.2 (K1231; ThermoFisher) before Sanger sequencing from both ends using pjet1.2 sequencing primers.

Results

We tested the potential of the Agrobacterium‐mediated gene delivery method to recover stable A. agrestis transgenic lines. There are several critical factors that determine the efficiency of Agrobacterium‐mediated gene delivery: the selection of appropriate plant tissue and its preparation for infection; the type and concentration of antibiotics applied to select for transgenic lines; the choice of transformation vectors; the choice of Agrobacterium strain; and the optimal conditions for cocultivation. Each of these factors is examined in this study.

Selection of tissue and optimal growth conditions

Anthoceros agrestis gametophyte thallus was chosen as the appropriate tissue for transformation because it is easily accessible, has a remarkable regenerative capacity and is haploid. Anthoceros agrestis thallus tissue cultures can easily be propagated and maintained by transfer of small thallus fragments (c. 1 × 1 mm) onto fresh growth medium on a monthly basis (Fig. S1). Anthoceros agrestis is similar to several other bryophytes in that an entire plant can regenerate from a small thallus fragment. This is in striking contrast to vascular plants, where usually the transition to an undifferentiated tissue state (callus), after treatment with extrinsic hormones such as auxin and cytokinin, is necessary for the regeneration of new plant tissue (Ikeuchi et al., 2013). The use of thallus has the additional advantage that the resulting transformants have a uniform genetic background.

We reasoned that similar to the liverwort M. polymorpha (Kubota et al., 2013), fragmented tissue will be susceptible to Agrobacterium infection. In the case of M. polymorpha, the apical part of the thallus is removed to induce regeneration, followed by cocultivation with Agrobacterium to generate transformed plants. Anthoceros agrestis thallus regeneration is similarly induced by fragmentation and presumably removal of the apical parts of the thallus. Consequently, a transformation approach similar to the one used for M. polymorpha was utilized and adapted to A. agrestis. However, unlike in M. polymorpha, the A. agrestis notch area and apical cells are not easily distinguishable, and thus determining which part of the thallus should be removed is not obvious (Frangedakis et al., 2020). Therefore, we tested whether homogenization using dispensing tools is a suitable method for thallus tissue fragmentation. Different speeds and duration of homogenization were examined. We found that homogenization of 0.5 g of thallus tissue in 15 ml of sterile water for 5 s (see Materials and methods) is sufficient to fragment the tissue and results in rapid regeneration.

In addition, we found that the light intensity is a critical factor during growth of the tissue used for transformation (Fig. 1e–g). Tissue that was grown under low light conditions, even though smaller in size, had a more regular and flattened shape and was optimal for transformation. (Fig. 1f,g and Fig. S1). When tissue was grown under high light intensity (>40 μmol m−2 s−1), no transformants were obtained (for detailed explanation see ‘Transformation efficiency and optimization’ section).

Besides fragmented thallus tissue, immature thalli grown from spores can also be used to achieve high numbers of stable transformants in M. polymorpha (Ishizaki et al., 2008). Therefore, we also tested the transformation of A. agrestis Bonn thalli grown from spores after 1, 2 and 3 wk of germination. However, no stable transformants could be obtained using these tissues (see details in the Materials and methods section).

Selection of appropriate antibiotics

A selectable marker gene, most commonly one that confers antibiotic resistance, is necessary for efficient recovery of stable transgenic lines following cocultivation with Agrobacterium. To identify antibiotics and their appropriate concentration with cytotoxic effect on A. agrestis, untransformed A. agrestis (Oxford and Bonn) thallus fragments were subjected to different concentrations of hygromycin and geneticin/G418 (an analogue of neomycin and kanamycin, respectively). The tested concentrations ranged from 0 to 20 μg ml−1 for hygromycin and 0 to 150 μg ml−1 for geneticin/G418. We found that a 3 wk incubation period with 10 μg ml−1 hygromycin was sufficient to inhibit growth of untransformed thallus tissue for both the Oxford and the Bonn isolates (Figs S2, S3), whereas thallus tissue was not susceptible to geneticin/G418 even when supplied in high concentration (150 μg ml) (Fig. S4). Thus, we selected hygromycin as an appropriate selection agent for A. agrestis transformation.

Preliminary tests with the GUS reporter

Preliminary transformation experiments were performed using the Oxford isolate and the pCAMBIA1305.2 plasmid containing the hygromycin B phosphotransferase gene (hph, conferring hygromycin resistance) driven by the cauliflower mosaic virus 35S (CaMV 35S) promoter, terminated with a 35S polyadenylation signal, and a p‐35S_s::GUSPlus (β‐glucuronidase) transcription unit. GUSPlus contains a catalase intron to ensure that the observed GUS expression is not due to the Agrobacterium.

Thallus grown under low light conditions (3–5 μmol m−2 s−1) (similar morphology with the tissue in Fig. 1f) was homogenized, grown for 1 month under low light conditions (3–5 μmol m−2 s−1) and then homogenized a second time (Fig. 1g–i). Two days after the second homogenization, the regenerating thallus tissue was cocultivated with Agrobacterium AGL1 strain containing the pCAMBIA1305.2 plasmid, as well as Agrobacterium without a transformation vector as a negative control, in liquid KNOP media supplemented with sucrose. Media were also supplemented with acetosyringone because phenolic compounds such as acetosyringone have been shown to be important for activation of Agrobacterium virulence genes (Stachel et al., 1985). Cocultivation duration was 3 d at 22°C on a shaker without any light supplementation (only ambient light from the room). After cocultivation, the tissue was spread on solid KNOP plates supplemented with cefotaxime and hygromycin. After 3–4 wk the surviving, regenerating and putatively transformed tissue was transferred on fresh selective media (tissue morphology used for transformation is shown in Fig. 1g–i, transformation outline in Fig. 2; see Fig. S5 for detailed step‐by‐step protocol description). Emergence of rhizoids on surviving tissue fragments is a reliable indicator of successful transformation events (Fig. 3a,b). One to 2 months later successful transformants (plant fragments producing rhizoids) were visible. Finally, surviving plants were subjected to a third round of antibiotic selection to ensure false positives were eliminated. Thallus surviving selection on hygromycin (Fig. 3c) exhibited GUS expression (Figs 3d, S6). No surviving plants were observed for those trials using the Agrobacterium without the transformation vector. These results indicated that A. agrestis Oxford thallus tissue is susceptible to Agrobacterium infection and that the CaMV 35S promoter driving the hph gene is sufficient for selection of transformants.

Fig. 2

Outline of Anthoceros agrestis transformation method. (1a–b), tissue grown under low light is homogenized, transferred to growth medium and placed again under low light conditions; (2a–c), after 4 wk, the tissue is homogenized again and grown for two additional days (the purpose of the first homogenization is tissue amplification); (3), the tissue is cocultivated with Agrobacterium for 3 d (under low light) and then (4a–b); spread on appropriate antibiotic‐containing growth medium; (5), after 3 wk, the tissue is transferred again onto freshly prepared antibiotic‐containing growth medium for a second round of selection; (6), after c. 4–8 wk, putative transformants are visible; (7), a final round of selection is recommended to eliminate false‐positive transformants.

Fig. 3

Schematic representation of transformation constructs and transgenic Anthoceros agrestis expressing different fluorescent proteins. (a, b) The emergence of rhizoids (shown with (a) white and (b) purple arrows) is a reliable indicator of successfully transformed plant fragments. In (a) the red arrowheads show false positives regenerating (green) tissue fragments that lack rhizoids. Bars, 2 mm. (c) Example of transgenic A. agrestis plants (gametophyte thallus). Petri dish dimensions: 92 × 16 mm. (d) GUS activity detected as blue staining in thallus tissue fragments from a plant transformed with the pCAMBIA1305.2 plasmid. Bar, 200 μm. (e) Loop assembly Type IIS cloning system outline: Level 0 DNA parts (for simplicity only promoter, coding sequence and terminator genetic DNA parts are indicated) are assembled in Level 1 transcription units (TUs) into one of the four pCk vectors, depicted with numbered circles, by BsaI‐mediated Type IIS assembly (sequential restriction enzyme digestion and ligation reactions). Level 1 TUs are assembled to Level 2 multi‐TUs into one the four pCs vectors, depicted with lettered circles, by SapI‐mediated Type IIS assembly. This workflow is then repeated for higher level assemblies. (f–i top) Schematic representation of constructs for the expression of two TUs: one TU for the expression of the hygromycin B phosphotransferase (hph) gene under control of the cauliflower mosaic virus (CaMV) 35S promoter and one TU for the expression of p‐35S_s::eGFP‐Lti6b (f), p‐35S_s::mTurquoise2‐N7 (g), p‐35S_s::myr‐eYFP (h) and p‐35S:mVenus‐N7 (i). hph, hygromycin B phosphotransferase; 35S, CaMV 35S promoter; eGFP, enhanced green fluorescent protein; mTurquoise2, monomeric turquoise 2 fluorescent protein; eYFP, enhanced yellow fluorescent protein; Lti6b, low temperature induced protein 6B signal for membrane localization; N7, Arabidopsis At4g19150/N7 nuclear localization signal; myr, myristoylation signal for membrane localization; nosT, 3′ signal of nopaline synthase. (f–i middle and bottom) Images of A. agrestis Oxford thallus tissue expressing different combinations of CaMV 35S promoter – fluorescent protein – localization signal. (f) p‐35S_s::eGFP‐Lti6b for plasma membrane localization (white arrow). Bars: (top) 50 μm; (bottom) 20 μm. (g) p‐35S_s::mTurquoise2‐N7 for nuclear localization (white arrow). Bars: (top) 20 μm; (bottom) 10 μm. (h) p‐35S_s::myr‐eYFP for plasma membrane localization (white arrow). Bars: (top) 50 μm; (bottom) 20 μm. (i) p‐35S::mVenus‐N7 for nuclear localization (white arrow). Bars: (top) 50 μm; (bottom) 25 μm. The bottom image is a magnification of the image in the middle. Red, Chl autofluorescence. (j) Light micrograph image: surface view of thallus (gametophyte), similar to the area imaged in (f) (and also g–i). Bar, 100 μm. (k) Images of A. agrestis Bonn gametophyte tissue expressing the p‐35S_s::eGFP‐Lti6b TU for eGFP plasma membrane localization. Bar, 50 μm. (l) Anthoceros agrestis Bonn with mature sporophytes indicated with white arrows. Bars, 2 mm. (m) Anthoceros agrestis Bonn transgenic spores expressing the p‐35S_s::eGFP‐Lti6b TU (white arrow indicates rhizoid and yellow arrow indicates young thallus). Top left: light microscopy of A. agrestis Bonn wild type germinating spore. Bar, 20 μm.

Tests with eGFP as a reporter

Subsequent experiments were carried out using the A. agrestis Oxford isolate and a construct containing the eGFP reporter gene (Cormack et al., 1996). eGFP makes the identification of successful transformation events easier without the need of laborious GUS staining. For construction of the eGFP transformation vector, we used the OpenPlant toolkit (Sauret‐Gueto et al., 2020), which is based on the Loop assembly Type IIS cloning system (Pollak et al., 2019).

Loop assembly is a Type IIS assembly system that alternates between two enzymes and, unlike other systems, requires only two sets of vectors (Fig. 3e). The restriction enzymes are BsaI, which generates four base 5′ overhangs, and SapI, which generates three base 5′ overhangs. Alternate use of BsaI and SapI and corresponding vector sets allows efficient and parallel assembly of large DNA circuits. In a single step reaction, the standardized DNA parts can be assembled into a transcriptional unit (TU) and TUs can then be combined into multi‐TU constructs. More specifically, level 2 parts can be digested with BsaI to combine 16 TUs in a level 3 part, which could be digested again with SapI to create a level 4 part with 64 TUs and so on.

All the DNA parts described here are generated following the common syntax (Patron et al., 2015) and are compatible with Type IIS cloning systems, such as GoldenGate and Loop assembly, facilitating the exchange of DNA parts between different laboratories. The transformation vector contained the hph gene driven by the CaMV 35S promoter and terminated with a CaMV 35S polyadenylation signal. It also contained a p‐35S_s::eGFP‐Lti6b transcription unit (same CaMV 35S promoter with the one driving GUSPlus in pCAMBIA1305.2 plasmid) terminated by the double nopaline synthase (Nos)–35S polyadenylation signal (Sauret‐Güeto et al., 2020) (Fig. 3f), which was fused to the Low Temperature Induced Protein 6b (Lti6b) signal for membrane localization from Arabidopsis thaliana (Arabidopsis) (Cutler et al., 2000) (see Notes S1 section ‘Map of the transformation vector used for the generation of the p‐35S_s::eGFP‐Lti6b expressing lines’ and Table S2 for full map and sequence). The vector was transformed into A. agrestis using the method described above and eGFP was successfully expressed in A. agrestis with the expected localization in the plasma membrane (Fig. 3f,j). During the course of this study we generated 215 stable A. agrestis transgenic lines (in a total of 23 transformation experiments) expressing the p‐35S_s::eGFP‐Lti6b (Table 1). There was variability in eGFP expression patterns between different transgenic lines (Fig. S7) presumably due to differences in the transgene copy number or genome location of transgene insertion. eGFP expression remained mosaic during the first, second and third round of selection. During the initial round of selection, the first transgene expression (based on fluorescence) was observed in regenerating thallus fragments that consist of only a few cells (about five). Thus, we speculate that the mosaic expression observed relates to the preferential expression of the promoter driving eGFP, in some parts of the thallus (see ‘Comparison of the CaMV 35S and AaEF1a promoters’) rather than to chimeric plants that consist of transformed and untransformed sections/cells. A small fraction of hygromycin resistant lines (four out of 157) do not show visible eGFP fluorescence, which could be attributed to potential silencing events or truncation of the inserted T‐DNA. These plants have been through at least five rounds of hygromycin antibiotic selection, so it is unlikely they are false positives. Fifteen lines have been propagated vegetatively for more than 2.5 yr without abolishing transgene expression (antibiotic resistance or eGFP).

Table 1

List of Anthoceros agrestis lines generated.

Construct	Total number of lines	Number of lines exhibiting fluorescence/GUS staining	Number of lines lacking fluorescence/GUS staining
pCAMBIA1305.1	10	10	–
p‐35S::hph ‐p‐35S_s::eGFP‐Lti6b	215	210	5
p‐35S::hph ‐p‐35S_s::eGFP‐Lti6b (Bonn)	2	2	–
p‐35S::hph ‐p‐35S_s::mTurquoise2‐N7	6	5	1
p‐35S::hph ‐p‐35S_s::myr‐eYFP	5	5	–
p‐35S::hph ‐p‐35S::mVenus‐N7	3	3	–
p‐35S::hph ‐p‐AaEF1a::eGFP‐Lti6b	8	6	2
p‐35S::hph ‐p‐AaUbi::eGFP‐Lti6b	3	3	–
p‐35S::hph ‐p‐AaTip1;1::mTurquoise2‐N7	3	3	–
p‐EF1a::hph‐ p‐AaTip1;1::eGFP‐Lti6b	1	1	–
p‐EF1a::hph‐p‐35S_s::eGFP‐Lti6b ‐p‐EF1a:mTurquoise2‐N7	3	2	1
p‐EF1a::hph‐ p‐35S_s::eGFP‐Lti6b	3	3	–
p‐35S::hph ‐p‐AaActin_1::eGFP‐Lti6b	3	–	3
p‐35S::hph ‐p‐AaActin_2::eGFP‐Lti6b	9	–	9

See Supporting Information and Table S2 for promoter sequences and full plasmid maps.

Testing additional fluorescent proteins, Bonn isolate and transgene inheritance

Fluorescent proteins have been shown to be a powerful tool for plant cell biology studies, permitting temporal and spatial monitoring of gene expression patterns at a cellular and subcellular level (Berg & Beachy, 2008). To expand the palette of fluorescent proteins that can be used in A. agrestis, we tested the expression of the monomeric turquoise 2 (mTurquoise2) fluorescent protein (Kremers et al., 2006; Goedhart et al., 2012), the enhanced yellow fluorescent protein (eYFP) (Orm et al., 1996) and the monomeric Venus (mVenus) fluorescent protein (Kremers et al., 2006). We used a construct similar to the one for the expression of eGFP protein, but with different subcellular localization signals. mTurquoise2 and mVenus were fused to the nuclear‐localization peptide sequence of At4g19150/N7 (Cutler et al., 2000) with a linker to the amino (C)‐terminus (Cutler et al., 2000), and eYFP was fused to the membrane‐targeting myristoylation (myr) signal to the amino (N)‐terminus (Resh, 1999) (see Table S2 for plasmid sequences). mTurquoise2 (Fig. 3g), eYFP (Fig. 3h) and mVenus (Fig. 3i) were successfully expressed in A. agrestis and were targeted to the predicted cellular compartments (for further information on the number of lines generated see Table 1).

We then tested whether the protocol developed for the Oxford isolate can be used successfully for the Bonn isolate (Fig. S8). Four trials resulted in two successful transformants, which is considerably less than the average number of transformants obtained for the Oxford isolate (Fig. 3k).

We finally tested whether the transgene can be stably inherited through the sexual life cycle. Two eGFP‐expressing transgenic lines for both Oxford and Bonn isolates were brought to sexual reproduction. Sporophytes were produced and young gametophytes germinating from the spores (sporelings) were expressing eGFP, indicating that the transgene and its expression was successfully passed on to the next generation (Fig. 3l,m).

Identification and selection of A. agrestis endogenous gene promoters

It is important to identify promoters that can be used to drive constitutive transgene expression (i.e. high‐level expression across almost all tissues and development stages). Commonly used constitutive promoters in other bryophyte model species include the M. polymorpha ELONGATION FACTOR 1 ALPHA (EF1a) promoter (Althoff et al., 2014), the rice Actin1 and the M. polymorpha ubiquitin‐conjugating enzyme E2 promoter (Sauret‐Güeto et al., 2020). Using the genomic sequence (Li et al., 2020) and RNA‐seq data (Fig. 4a,b), we identified a series of candidate promoter regions as constitutive A. agrestis promoters. In particular, we selected the promoter regions of the putative A. agrestis homologs of EF1a, Ubiquitin, Actin and the Arabidopsis GAMMA TONOPLAST INTRINSIC PROTEIN (Tip1;1) genes (Fig. 4a,b; Table S1). We amplified a 1532 bp long stretch of the 5′ flanking region including the 5′ untranslated region (UTR) of the EF1a gene, a 933 bp segment for the Ubiquitin (Ubi) gene, two fragments (1729 and 1516 bp) for the Actin gene (that correspond to two different predicted translational start sites), and a 1368 bp putative promoter for the Tip1;1 gene (sequence cloned, gene models, position of promoters and the corresponding RNA‐seq coverage tracks are shown in Fig. S9).

Fig. 4

Identification of constitutive promoters for Anthoceros agrestis. (a) Analysis of expression levels from RNA‐seq experiments on A. agrestis Oxford using datasets from Li et al. (2020). To generate this dataset, gametophytes were grown under varying carbon sources in the growth medium (indicated as C depleted, C enriched and C normal), as well as two different nitrogen‐depleted conditions (N‐ with cyanobacteria symbiosis and N‐ without cyanobacteria symbiosis). Error bars indicate standard error based on three independent experimental replicates. (Included for comparison: ‘CN‐Max 1, 3 and 5’, highest expressing genes based on N conditions; and ‘CN‐Max 2 and 4’, highest expressing genes based on C normal conditions (for gene ID, see Supporting Information Table S1). (b) Analysis of expression levels from RNA‐seq experiments on A. agrestis Bonn using data sets from Li et al. (2020). Note: normalized expression level of A. agrestis Bonn genes selected to represent strong and constitutive expression across various developmental stages of the gametophyte and the sporophyte phases. (Included for comparison: ‘T‐Max 1 to 5’, highest expressing genes under all conditions (for gene ID, see Table S1).) Error bars indicate standard error based on two independent experimental replicates.

The candidate promoter regions were cloned (and if necessary domesticated to generate a Loop assembly cloning system‐compatible DNA part), fused with the eGFP or the mTurquoise2 reporter genes, and terminated with the double Nos–35S terminator (Sauret‐Güeto et al., 2020). The AaEF1a (Fig. 5a) promoter region was sufficient to drive expression of eGFP throughout the thallus. Similarly, the AaTip1;1 promoter region was sufficient to drive expression of eGFP and mTurquoise2 (Fig. 5b). However, only three independent lines were obtained for the p‐AaTip1;1::mTurquoise2‐N7 construct (with one showing growth retardation probably due to the insertion site of the T‐DNA) and one for the p‐AaTip1;1:: eGFP construct (Fig. S10). Thus, further characterization of the AaTip1;1 promoter is needed. The AaUbi (Fig. 5c) promoter gave less uniform expression patterns throughout the thallus, and the two AaActin promoters produced no detectable signal (Fig. 5d) (a summary of the number of lines generated is shown in Table 1). Our data thus indicate that out of the five candidates, AaEF1a is the best promoter for driving relatively strong expression across cells of the gametophyte thallus. We generated a total of nine p‐AaEF1a::eGFP lines, four of which are shown in Fig. S10. Of the nine hygromycin‐resistant lines, two do not express eGFP, which could be due to transgene silencing or truncation of the inserted T‐DNA. In addition, we showed that the AaEF1a promoter can drive adequate hph expression (Fig. 5e). Finally, we were also able to successfully express simultaneously three different transcription units, p‐AaEF1a::mTurquoise2‐N7, p‐35S_s::eGFP‐Lti6b and p‐35S::hph (Fig. 5f), which was the largest construct (c. 7.4 kb) we successfully introduced into the A. agrestis genome. Three lines were obtained, one of which did not show eGFP expression (Table 1). The three lines have been propagated vegetatively for more than 1.5 yr without abolishing transgene expression (antibiotic resistance or fluorescence protein).

Fig. 5

Identification of constitutive promoters for Anthoceros agrestis. (a–f top) Schematic representation of constructs for the expression of two to three transcription units (TU): one TU for the expression of the hph gene under control of the CaMV 35S or the p‐AaEF1a promoter and one TU for the expression of plasma membrane‐localized eGFP and/or nucleus‐localized mTurquoise2 under the control of different native A. agrestis promoters: (a) p‐AaEF1a, (b) p‐AaTip1;1, (c) p‐AaUbi and (d) p‐AaActin_1. (e) p‐AaEF1a driving hph and (f) p‐AaEF1a driving mTurquoise2‐N7. (a–f middle and bottom) Images of A. agrestis Oxford gametophyte tissue expressing different combinations of A. agrestis native promoter – fluorescent protein – localization signal. (a) p‐AaEF1a:eGFP‐Lti6b for plasma membrane localization. Bars: (top) 50 μm; (bottom) 10 μm. (b) p‐AaTip1;1::mTurquoise2‐N7 for nuclear localization. Bars; (top) 20 μm; (bottom) 10 μm. (c) p‐AaUbi::eGFP‐Lti6b for plasma membrane localization. Bars, top: 20 μm, bottom: 10 μm. The bottom images are a magnification of the images in the top. (d) p‐AaActin_1::eGFP‐Lti6b. Bars, 50 μm. (e) p‐AaEF1a::hph ‐ p‐35S_s::eGFP‐Lti6b. Bar, 50 μm. (f) Image of A. agrestis Oxford gametophyte tissue expressing both p‐AaEF1a::mTurquoise2‐N7 for nuclear localization and p‐35S_s::eGFP‐Lti6b for plasma membrane localization. Bar, 50 μm. Red, Chl autofluorescence.

Comparison of the CaMV 35S and AaEF1a promoters

Expression of eGFP driven by the CaMV 35S promoter seems to be weaker in newly grown parts of the thallus (Figs 6a, S7). This is similar to the expression patterns of transgenes driven by the CaMV 35S promoter in M. polymorpha, which has a strong activity in all parts of the thallus except the notch area (Althoff et al., 2014). Expression of eGFP driven by the AaEF1a promoter seems to be stronger in the putatively younger parts of the thallus (Figs 6b, S10). This is again similar to the expression patterns of transgenes driven by the EF1a promoter in M. polymorpha, showing a strong activity in all parts of the thallus, particularly the notch area (Althoff et al., 2014). We also assessed the amount of eGFP protein accumulated in plants containing either the CaMV 35S or the AaEF1a promoter. Our western blot analysis showed that a similar amount of eGFP is accumulated in both types of plants (Fig. 6c). Therefore, our experiments imply that the CaMV 35S and the AaEF1a promoters have similar overall activity in the gametophyte thallus of A. agrestis. Nevertheless, CaMV 35S drives the transgene more weakly in newly grown parts of the thallus, which needs to be taken into account when designing experiments.

Fig. 6

Comparison of the CaMV 35S with the AaEF1a promoter and factors affecting the efficiency of Agrobacterium‐mediated transformation of Anthoceros agrestis. (a) Expression of eGFP driven by the CaMV 35S promoter. Younger part of the thallus indicated with a white arrow. Bar, 50 μm. (b) Expression of eGFP driven by the AaEF1a promoter. Younger part of the thallus is indicated with white arrows. Bar, 100 μm. (c) Western blot analysis of eGFP accumulation in transgenic lines. Total cellular proteins were separated by polyacrylamide gel electrophoresis, blotted and probed with anti‐GFP and anti‐actin antibodies. Numbering above blot images corresponds to the identifier of independent lines. Information on the genomic location and copy number of T‐DNA insertion for lines cam1 and cam2 is provided in Fig. 7 and in the Supporting Information in Notes S1 section ‘Detailed description of insertion localization in the sequenced transformant lines’. Images of CaMV 35S lines 5, 7 and 9 can be found in Fig. S7. Images of the four p‐AaEF1a lines can be found in Fig. S10. Lower band (marked with red asterisk) in ‘CaMV 35S line 1’ potentially corresponds to degraded eGFP protein. (d) Confocal microscopy images of fragmented A. agrestis thallus tissue taken on seven consecutive days after homogenization. Five days after homogenization plants start to regenerate. Yellow arrowhead indicates the fragmented thallus part. White arrows indicate new cell divisions. Bars, 50 μm. (e) Number of transgenic lines obtained using A. agrestis thallus tissue 2, 5, 7, 14 and 21 d after homogenization. Values of three independent experimental replicates are shown. (f) Effect of light intensity used to grow plant tissue for transformation after the first homogenization on the number of successful transformants. Cocultivation started 2 d after the second homogenization. Values of three independent experimental replicates are shown. Bars, 1 mm. (g) Effect of Agrobacterium strain on the number of transformants obtained. Dots represent values of individual experimental replicates. Cocultivation started 2 d after the second homogenization. Values of three independent experimental replicates are shown.

Fig. 7

Stable incorporation of transgene into Anthoceros agrestis genome. (a–f left) Light micrograph (LM) of transgenic thallus of A. agrestis plants. Bar, 500 μm. Right: confocal fluorescence microscopy images of thallus expressing eGFP in the plasma membrane, driven by the CaMV 35S promoter. Bars: (a–e) 100 μm; (f) 50 μm. Bottom, location of transgene insertion in the genome (see details in the Supporting Information in Notes S1 section ‘Detailed description of insertion localization in the sequenced transformant lines’). Black arrows indicate directionality of T‐DNA insert. (g) PCR analysis of genomic DNA from transgenic plants. L, fragment amplified from sequences spanning the 5′‐end of the T‐DNA inserts and their respective adjacent genomic regions. R, fragment amplified from sequences spanning the 3′‐end of the T‐DNA inserts and their respective adjacent genomic regions. Note: LM images (a), (e) and (f) were acquired using a KEYENCE VHX‐S550E microscope (VHX‐J20T lens) and confocal fluorescence images with a Leica SP8X microscope, in Cambridge University. (d) LM and fluorescence images were acquired using a Leica M205 FA stereomicroscope with GFP longpass (LP) filter, in BT Institute.

Transformation efficiency and optimization

To estimate the transformation efficiency of the protocol, we performed 10 independent trials using as starting material c. 2 g of tissue per trial and the p‐35S::hph ‐ p‐35S_s::eGFP‐Lti6b construct. The number of successful transformation events per experiment varied from three to 23 (Table 2).

Table 2

Number of p‐35S::eGFP‐Lti6b transgenic lines of Anthoceros agrestis obtained from 10 different experiments to estimate transformation efficiency.

Experiment	Amount of tissue (g) used in first homogenization	Number of transformants
1	2.7	21
2	3.1	12
3	2.9	23
4	2.5	5
5	1.7	3
6	2.7	11
7	2.6	19
8	1.9	4
9	2.23	12
10	2.1	6

We then carried out further experiments to optimize transformation efficiency. We reasoned that tissue susceptibility to Agrobacterium infection may differ during different stages of regeneration after homogenization, thereby affecting transformation efficiency. To estimate when plant regeneration is initiated, we set up a microscopy time course using homogenized thallus fragments. The first cell division was observed 5 d after homogenization (Fig. 6d). Based on this result, we carried out an optimization experiment starting cocultivation at 2, 5, 7, 14 and 21 d after homogenization. We found that the number of stably transformed lines decreased when using tissue that was recovered for more than 5 d after homogenization (Fig. 6e). The highest number of transformants could be obtained when using tissue 2 d after homogenization.

We also tested the effect of tissue growth (after the first homogenization) under three different light intensities, 7, 35 and 80 μmol m−2 s−1, on transformation efficiency. Transformants were obtained only when using the lowest light intensity of 7 μmol m−2 s−1(Fig. 6f).

Finally, we tested another Agrobacterium strain (GV3101) for its ability to infect A. agrestis thallus. However, only up to two successful transformation events were obtained when the GV3101 strain harboring the p‐35S::hph ‐ p‐35S::eGFP‐Lti6b construct was used (Fig. 6g).

Verification of transgene incorporation into the A. agrestis genome

To confirm the genomic integration of the transgene, we sequenced and assembled the genomes of five stable transformant lines (Fig. 7a–f). For all five lines, we found a single integration site, with one line showing a single full‐length insertion of the T‐DNA. The other four lines additionally showed one or multiple partial insertions in inverted and/or tandem directions (see in Notes S1 section ‘Detailed description of insertion localization in the sequenced transformant lines’). To confirm the transgene integration site in the five lines, fragments overlapping the 5′‐ and 3′‐ends of the inserts, and their adjacent genomic regions were amplified by nested PCR and Sanger‐sequenced. The resulting sequences confirmed the integration sites identified by the genome assemblies (Fig. 7g, and see in Notes S1 section ‘Detailed description of insertion localization in the sequenced transformant lines’). We conclude that the transformation method described here results in the stable integration of one or more targeted transcriptional units into the A. agrestis nuclear genome.

Discussion

The protocol described in this study successfully generated stable transformants in A. agrestis Oxford and Bonn isolates and may be applicable to other hornwort species. We generated a total of 274 stable lines in 38 transformation experiments. We showed that transgenic lines can be propagated for more than 2 yr without abolishing transgene expression. Additionally, we confirmed that the transgene is integrated into the genome of A. agrestis and can be successfully inherited.

The genome sequencing of transgenic lines showed that the integration occurs in a single locus with one or more copies, which is similar to the reports in other organisms based on DNA gel blot analysis (Feldmann & Marks, 1987; Ishizaki et al., 2008; Plackett et al., 2014). The utilization of recent high‐throughput sequencing technologies combined with the genome size of A. agrestis allows precise determination of the transgene insertion site. Thus, it should be relatively simple to perform enhancer‐trap or T‐DNA‐based mutagenesis experiments in A. agrestis.

The light conditions under which the plant tissue was grown significantly affected the thallus morphology and was critical for successful transformation. It is likely that high light intensity triggers the accumulation of secondary metabolites (such as mucilage) and/or affects the composition of the cell wall, thereby significantly reducing transformation efficiency. Light intensity might affect the expression of genes such as the cytochrome P450 (CYP) gene family for primary and secondary metabolism, an expansion of which family has been reported in the genome of Anthoceros angustus (Zhang et al., 2020). Multiple photoreceptors are also present in the A. agrestis genome (Li et al., 2014, 2015a,b). Identifying which receptors determine the response to high light intensity could help to further improve the transformation efficiency.

We have tested different methods for tissue fragmentation before cocultivation with Agrobacterium, such as vortexing, and use of scalpels or razor blades as an alternative to tissue homogenizers. The use of scalpels or razor blades resulted in successful transformation events for both Oxford and Bonn strains, but this method is still under optimization.

We are currently developing genome editing tools for A. agrestis using CRISPR/Cas9 (Jinek et al., 2012). More selection markers are needed, and preliminary data suggest that blasticidin (Tamura et al., 1995; Ishikawa et al., 2011) is a promising candidate. We are also testing whether inducible gene expression systems such as the glucocorticoid receptor (Schena et al., 1991) or estrogen receptor (Zuo et al., 2000) can be applied successfully in hornworts. Finally, we are testing alternative gene delivery methods for both A. agrestis Oxford and Bonn isolates, such as particle bombardment.

The development of a hornwort transformation method, in combination with the recently published genome, will greatly facilitate more comprehensive studies of the mechanisms underpinning land plant evolution. It can also help with engineering hornwort traits into plants with agronomic value. For example, engineering pyrenoids in crops has the potential to increase carbon fixation and therefore increase crop yield (Li et al., 2017).

Author contributions

KS, EF, HT, MW and PS conceived and designed the experiments. TN and MW identified gene promoter regions. EF and MW performed cloning. EF generated and characterized the transgenic lines and performed imaging. XX, MT and JVC generated the BTI lines and AG imaged the lines. MW, PS, F‐WL and AG analyzed the transcriptomic data. EF performed qPCR analysis. MW performed next‐generation sequencing of transgenic lines. PS and MW assembled the genomes. AG and MW confirmed the insert locations. YY provided technical assistance. KS, EF, TN, MW and PS wrote the article with contributions from all the authors.

Fig. S1 Effect of light on Anthoceros agrestis growth and tissue culture.

Fig. S2 Hygromycin sensitivity of Anthoceros agrestis gametophytes (Oxford).

Fig. S3 Hygromycin sensitivity of Anthoceros agrestis gametophytes (Bonn).

Fig. S4 G418 sensitivity of Anthoceros agrestis gametophytes (Oxford).

Fig. S5 Anthoceros agrestis Oxford transformation protocol.

Fig. S6 GUS activity detected in transformed Anthoceros agrestis plants.

Fig. S7 Examples of Anthoceros agrestis Oxford transgenic lines I.

Fig. S8 Anthoceros agrestis Bonn transformation protocol.

Fig. S9 Promoters and their localization on the Anthoceros agrestis genome.

Fig. S10 Examples of Anthoceros agrestis Oxford transgenic lines II.

Notes S1 Native promoters, vector maps and insertion localization.

Table S1 Highly and/or constitutively expressed Anthoceros agrestis genes.

Table S2 List of construct sequences.

Table S3 List of primers used in this study.

Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Click here for additional data file.