An optimized transformation protocol for Anthoceros agrestis and three more hornwort species

SUMMARY Land plants comprise two large monophyletic lineages, the vascular plants and the bryophytes, which diverged from their most recent common ancestor approximately 480 million years ago. Of the three lineages of bryophytes, only the mosses and the liverworts are systematically investigated, while the hornworts are understudied. Despite their importance for understanding fundamental questions of land plant evolution, they only recently became amenable to experimental investigation, with Anthoceros agrestis being developed as a hornwort model system. Availability of a high‐quality genome assembly and a recently developed genetic transformation technique makes A. agrestis an attractive model species for hornworts. Here we describe an updated and optimized transformation protocol for A. agrestis, which can be successfully used to genetically modify one more strain of A. agrestis and three more hornwort species, Anthoceros punctatus, Leiosporoceros dussii, and Phaeoceros carolinianus. The new transformation method is less laborious, faster, and results in the generation of greatly increased numbers of transformants compared with the previous method. We have also developed a new selection marker for transformation. Finally, we report the development of a set of different cellular localization signal peptides for hornworts providing new tools to better understand the hornwort cell biology.


INTRODUCTION
Hornworts, together with the mosses and liverworts, belong to bryophytes, a monophyletic group sister to all other land plants (tracheophytes).Bryophytes are crucial for revealing the nature of the land plant common ancestor and improving our understanding of fundamental land plant evolutionary innovations (Puttick et al., 2018).There is mounting evidence that morphological, and developmental characters were independently lost and gained in each of the three lineages of bryophytes (Clark et al., 2022;Harris et al., 2022;Li et al., 2020).Such an example is stomata that are present in hornworts and mosses but are absent in liverworts (Frangedakis, Shimamura, et al., 2021;Harrison & Morris, 2018;Ligrone et al., 2012).Therefore, studying representative species from each of the three main bryophyte clades is necessary to understand the evolution and the likely ancestral state of a trait of interest.Hornworts can be instrumental in understanding the evolution of: (i) sporophyte; (ii) plant-microbe symbiosis; and (iii) plastids.Sporophytes of hornworts are morphologically and developmentally distinct from other bryophytes and land plants.In particular, in hornworts, the sporophyte grows from a multicellular basal meristem, whereas in mosses from an intercalary meristem (Frangedakis, Shimamura, et al., 2021;Harrison & Morris, 2018;Ligrone et al., 2012), and in the liverwort Marchantia polymorpha a distinct sporophytic multicellular proliferative region is absent.In contrast, in vascular plants the sporophyte grows from a shoot and root apical meristem.Hornworts establish symbiotic interactions with mycorrhizal fungi (Desiro et al., 2013;Frangedakis, Shimamura, et al., 2021;Read et al., 2000) but are also able to form symbiotic interactions with cyanobacteria, which occur only in a few, phylogenetically separated lineages of land plants (Adams, 2002;Chatterjee et al., 2022;Duggan et al., 2013).Finally, hornworts are unique among land plants as they have a single chloroplast per cell that contains a pyrenoid based alga-like carbon-concentrating mechanism (Li et al., 2017;Vaughn et al., 1992).Understanding the genetic pathways controlling the development of these unique traits in hornworts, and comparing them to mosses, liverworts as well as vascular plants, can greatly facilitate efforts towards a more comprehensive understanding of the mechanisms underpinning land plant evolution.
However, until recently, a tractable model system was lacking for hornworts.To further our understanding of hornwort biology, Anthoceros agrestis was recently proposed as an experimental model system (Frangedakis, Shimamura, et al., 2021;Gunadi et al., 2022;Li et al., 2020;Neubauer et al., 2022;Sz€ ov enyi et al., 2015) with two geographic isolates (Oxford and Bonn) available.Axenic culturing methods for both isolates have been established and their genomes have been sequenced.Importantly, an Agrobacteriummediated stable transformation method has been developed (Frangedakis, Waller, et al., 2021).Transgenic lines can be propagated vegetatively without losing the transgene.These technical advances provide us with the tools to unlock fundamental questions of hornwort biology.
While the published transformation method was applied successfully to recover stable transformants for the Oxford isolate of A. agrestis, it resulted only in a very low number of stable transformants for the Bonn isolate.The Bonn isolate (Figure 1a-d) is of particular interest for embryo and sporophyte development studies, as, in contrast to the Oxford isolate (Figure 1e,f), sporophytes can be induced en masse under axenic conditions (Figure S1).
Our goal was to optimize the previously published transformation protocol for the A. agrestis Bonn strain and potentially apply it to other hornwort species.To do so, we examined a range of approaches and parameters that were most likely to influence the infection efficiency of Agrobacterium and the recovery of stable transformants: (i) the effect of the pH of the co-cultivation media; (ii) the effect of the promoter driving the selection marker; (iii) the sucrose concentration of the co-cultivation solution; and (iv) the choice of the Agrobacterium strain used.We successfully developed a modification of the Frangedakis, Waller, et al. (2021) protocol, that allowed us to obtain up to 55 stable, independent, transformant lines from 200 mg of tissue for the A. agrestis Bonn isolate and 103 stable transformant lines from 200 mg of tissue for the Oxford isolate.
We then tested whether the optimized transformation method is applicable to three other hornwort species: Anthoceros punctatus, Leiosporoceros dussii, and Phaeoceros carolinianus.A. punctatus has long been used as the study system for hornwort associations with cyanobacteria (Nostoc punctiforme) (Chatterjee et al., 2022).L. dussii is the sister taxon to all hornworts (Duff et al., 2007).L. dussii chloroplasts lack a pyrenoid and one of its key morphological innovations include a unique symbiotic arrangement of endophytic cyanobacteria (Villarreal et al., 2018;Villarreal & Renzaglia, 2006).P. carolinianus represents a commonly found species that can be grown relatively easily under laboratory conditions.Finally, to facilitate future studies of hornworts at a cellular level, we developed a set of Loop cloning system/ OpenPlant kit (Sauret-G€ ueto et al., 2020) compatible tags for localization at the mitochondria, Golgi, peroxisomes, actin cytoskeleton, chloroplasts, and the endoplasmic reticulum (ER), and tested the utility of an additional fluorescent protein and chlorsulfuron as new selection marker for transformation.
In summary, we provide a streamlined transformation protocol, which can be used for a series of hornwort species and can potentially be applied to further species of hornworts.

Effect of 4-morpholineethanesulfonic acid concentration on transformation efficiency
It has been reported that a stable pH of 5.5 during cocultivation leads to significantly increased Agrobacterium infection of Arabidopsis thaliana cells, likely due to inhibited calcium-mediated defence signalling (Wang et al., 2018).As this innate immune response is possibly conserved among land plants, the pH during co-cultivation may be a crucial factor for increased Agrobacterium infection rates of hornwort cells.While the published A. agrestis transformation method is employing phosphate-buffered KNOP medium with pH 5.8 for co-cultivation, monitoring pH during co-cultivation of A. agrestis with Agrobacterium, revealed a pH value increase from an initial value of 5.8 to about 7-8, likely due to Agrobacterium growth.A test trial with the p-AaEf1a::hphp-AaTip1;1::eGFP-Lti6b construct (AaEf1a: A. agrestis Elongation Factor 1a, hph: hygromycin B resistance phosphotransferase, AaTip1;1: A. agrestis Gamma Tonoplast Intrinsic Protein 1;1, full list of all construct maps used in this study in Table S1), the initial pH set to 5.5 and the co-cultivation medium supplemented with 10 mM 4-morpholineethanesulfonic acid (MES) for improved buffering, increased the number of recovered stable transformants for A. agrestis Bonn.In this pre-trial we also replaced the use of a homogenizer to fragment the thallus tissue (Figure S2, Video S1) before the cocultivation with a razor blade, to eliminate the need for special equipment.
To further test the influence of the pH range and stability during co-cultivation on transformation efficiency, we added MES in a range of concentrations, 0-50 mM, to the co-cultivation medium and evaluated the effect of different pH values on transformation efficiency.Increasing the concentration of MES in the co-cultivation medium led to a slower rise of pH during co-cultivation and greatly increased the number of transformants (Figure 1g).However, MES at concentrations higher than 40 mM had an apparent toxic effect on A. agrestis gametophyte tissue (Figure S3), as indicated by chloroplasts turning brown or grey, plant fragments immediately after co-cultivation looked generally less healthy compared with plants co-cultivated using <40 mM MES concentrations.When applying an MES concentration of 50 mM, most plant fragments appeared to be dead within 4-5 days after co-cultivation, leading to a decrease in the total number of recovered transformants.The optimal MES concentration was therefore determined to be 40 mM MES.

Effect of initial pH of co-cultivation media on transformation efficiency
We also tested the effect of the initial pH of the cocultivation medium on transformation efficiency.Comparison of the different initial pH from 5 to 5.5 showed no significant difference in transformation efficiency, but the number of transformants decreased when the starting pH was set to ≥5.75 (Figure 1h-j).In conclusion, the highest number of stable transformants could be recovered with an initial pH between 5 and 5.5 and 40 mM MES for improved buffering.It should be noted that this is only valid for the specific tissue culture conditions used in our trials, and using thallus fragments that have been propagated for up to 6 weeks after subculturing (Figure S3).Non-empirical observations suggest that thallus tissue from older cultures (2-3 months after subculturing) show a higher sensitivity to MES, with 20 mM MES frequently killing many cells.In this case, higher numbers of recovered stable transformants were usually obtained with only 10 mM MES.

Effect of promoter on selection and recovery of transformant lines
The promoter used to drive expression of the selection gene hygromycin B resistance-encoding hph may influence the number of recovered stable transformants.The cauliflower-mosaic virus (CaMV) 35S promoter previously used for this purpose shows stronger expression in differentiated cells of the thallus tissue and weaker expression in new grown parts of the thalli (Frangedakis, Waller, et al., 2021).In contrast, the endogenous AaEf1a promoter is more strongly expressed in new grown tissue of the Oxford isolate (Frangedakis, Waller, et al., 2021), which we hypothesize to be more susceptible to Agrobacterium infection.Interestingly, the AaEf1a promoter shows high activity in protoplasts of A. agrestis Bonn, while CaMV 35S activity could not be detected (Neubauer et al., 2022).Thus, we reasoned that driving expression of the hph under the AaEf1a promoter might improve transformation efficiency.To test the effect on transformation efficiency when driving the expression of the hygromycin resistance (hph) gene with either the AaEf1a or the CaMV 35S promoter, we transformed equal amounts of A. agrestis Bonn tissue with two different vectors.One contained a p-35S::hph transcription unit and the other a p-AaEf1a::hph transcription unit.Both vectors further contained a p-AaTip1;1::eGFP-Lti6b transcription unit to allow the use of fluorescence as an additional marker for successful transformation.We then compared the number of isolated enhanced green fluorescent protein (eGFP) expressing transformants and their growth after a total of 7 weeks on selective media.For both constructs, the number of primary transformants (thalli) expressing eGFP in at least some cells of the thallus was comparable (Figure 1k).However, the majority of plants carrying the hph gene driven by the CaMV 35S promoter showed retarded growth compared with plants carrying the hph gene driven by the AaEf1a promoter.In particular, after 7 weeks of growth, the surface area of thalli of transformants carrying the p-AaEf1a::hph transcription unit, was on average 10 times that of plants carrying the p-35S::hph transcription unit (Figure 1l,m) (based on three independent experimental replicates).Therefore, while the number of initially isolated transformant lines is not affected by the choice of the promoter driving hph, it does have a large effect on the recovery and growth rate of transformed thalli.Different 'enhanced' versions of the CaMV 35S promoter might be more effective in driving selection markers in the Bonn strain.

Effect of sucrose concentration of the co-cultivation medium on transformation efficiency
We compared the effect on transformation efficiency of two different concentrations of sucrose, 1% (w/v) and 2% (w/v).We reasoned that lowering the concentration of sucrose in the co-cultivation medium might reduce the chances of Agrobacterium overgrowth during cocultivation that can potentially affect the pH of the medium and consequently transformation efficiency.We infected equal amounts of A. agrestis Bonn tissue using cocultivation media with 20 mM MES, the AGL1 Agrobacterium harbouring the p-AaEf1a::hphp-35S_s::eGFP-Liti6b construct and either 1% (w/v) or 2% (w/v) sucrose concentration.Our results suggest that the transformation efficiency is comparable (Figure 1n).

Effect of Agrobacterium strain used
Previously, only a very small number of transformants were obtained when the GV3101 Agrobacterium strain was used.We tested again the GV3101 Agrobacterium strain for its ability to infect and transform A. agrestis thallus compared with the AGL1 strain.We infected equal amounts of A. agrestis Bonn and Oxford tissue using co-cultivation media with 20 mM MES with GV3101 or AGL1 Agrobacterium harbouring the p-AaEf1a::hphp-AaEf1a:: eGFP-Liti6b construct.We repeated the same experiment using co-cultivation media with 20 and 30 mM MES concentrations.Our results revealed that when the GV3101 strain was used the number of transformants was greater (Figure 1o).This contradicts our previous findings (Frangedakis, Waller, et al., 2021), which could be explained by differences in the quality of GV3101 Agrobacterium batches, and/or the difference in pH stability during co-cultivation.

Comparison of homogenized versus blade fragmented tissue
To test the effect of the tissue fragmentation method on transformation efficiency we infected equal amounts of A. agrestis Bonn tissue fragmented using a razor blade or a homogenizer and using co-cultivation media with 30 mM MES and the AGL1 or GV3101 Agrobacterium harbouring the p-AaEf1a::hphp-AaEf1a::eGFP-Liti6b construct.Our results show that using a razor blade instead of a homogenizer does not significantly affect overall transformation efficiency.This indicates that fragmentation can be carried out without the need of specialized equipment (Figure 1p).

Comparison of transformation efficiency between the two Anthoceros agrestis isolates and Anthoceros punctatus
We then confirmed that the protocol can successfully be used to recover stable transformants for the A. agrestis Oxford isolate (Figure 2a,b).
To estimate the transformation efficiency of the protocol for the two A. agrestis isolates (Bonn and Oxford), we performed transformation trials using 200 mg of tissue per trial, co-cultivation media with 20, 30, or 40 mM MES, AGL1 or GV3101 Agrobacterium harbouring the p-AaEf1a::hphp-AaEf1a::eGFP-Lti6b construct.The number of successful transformants (plant thalli) per experiment is summarized in Figure 2c,e.We could recover up to three times more transformants for the Oxford isolate compared with the Bonn isolate.

Characterization of the activity of the AaTip1;1 promoter
Previously, a 1368 bp putative promoter fragment of the A. thaliana Tip1;1 gene homologue in A. agrestis was selected as a candidate for a constitutive hornwort promoter (Frangedakis, Waller, et al., 2021).Here we further characterize the AaTip1;1 promoter and compared it with the two other constitutive, AaEf1a and the CaMV 35S, promoters used for A. agrestis.
Expression of eGFP driven by the AaTip1;1 promoter is equally strong across the thallus (Fig 2f-n).As reported earlier (Frangedakis, Waller, et al., 2021), expression of eGFP driven by the CaMV 35S promoter seems to be weaker in meristematic areas of the thallus (Figure 2f-n Plant Journal, (2023), 114, 699-718 unlike expression of eGFP driven by the AaEf1a promoter, which is stronger in meristematic areas of the thallus.It must be noted that the eGFP expression in the stable transformants can be categorized into four groups: expression throughout the thallus, expression in the rhizoids, expression in patches, or no expression (Figure S4).Frequency of each of those four events is summarized in Table 1.
We also compared the amount of eGFP protein accumulated in plants expressing eGFP under the AaTip1;1, the CaMV 35S or the AaEf1a promoter.Western blot analysis showed that the AaTip1;1 and the AaEf1a promoters lead to similar abundances of fluorescent proteins at the whole thallus level.By contrast, the CaMV 35S promoter was weaker (Figure 2o).
We also tested the expression of the three constitutive promoters at the sporophyte stage of A. agrestis Bonn.When eGFP was driven by the CaMV 35S promoter no eGFP fluorescence could be observed in the sporophyte (Figure 3a,d,g).In contrast, when eGFP expression was driven by the AaEf1a or the AaTip1;1 promoter, eGFP fluorescence was detectable in most sporophyte tissues (Figure 3b,c,e,f,h,i,j).
A. punctatus, as A. agrestis, can routinely be propagated vegetatively in lab conditions, with monthly subculturing.Its genome is sequenced and has a size of approximately 132 Mbp.However, sexual reproduction under lab conditions has been challenging.Spores are punctuated (Figure 4b).The A. punctatus thallus is irregularly shaped (Figure 4c), lacks specialized internal differentiation and is composed of mucilage chambers and parenchyma cells (Figure 4d).The male (Figure 4e) and female reproductive organs are embedded in the thallus.Sporophytes grow on the gametophyte (Figure 4f) and contain columella, spores, pseudoelaters, assimilative tissue, epidermis, and stomata (Figure 4g,h).A. punctatus chloroplasts have pyrenoids (Figure 4i) and its thallus has ellipsoidal cavities colonized by cyanobacteria (Figure 4j).
L. dussii typically has a solid thallus with schizogenous cavities (Figure 5a).Antheridia are numerous  S1. (o) Western blot analysis of eGFP accumulation in transgenic lines.Total cellular proteins were separated by polyacrylamide gel electrophoresis, blotted, and the probed with anti-GFP and anti-actin antibodies.Numbering above the blot images corresponds to the identifiers of independent transformed lines.(Figure 5b,c) and sporophytes (Figure 5d) develop on the gametophyte.Unlike other hornworts, spores are monolete not trilete and smooth (Figure 5e).The sporophyte (Figure 5f) is consisting of the columella, the sporogenous tissue, the assimilative layer, epidermis, and stomata.L. dussii has a single chloroplast per cell that lacks a pyrenoid (Figure 5g,h).Chloroplasts have numerous channel thylakoids and extensive grana stacks (Figure 5i,j).The thallus of L. dussii is colonized with Nostoc cyanobacteria (Figure 5k), which are longitudinally oriented strands in mucilage-filled schizogenous canals.The P. carolinianus thallus is irregularly shaped (Figure 5l) and lacks specialized internal differentiation (Figure 5m).P. carolinianus is monoicous and produces both antheridia (Figure 5n) and archegonia on the dorsal side of the thallus.The sporophyte (Figure 5o) consists of the columella, sporogenous tissue, assimilative layer, epidermis, and stomata.P. carolinianus has a single chloroplast per cell with pyrenoids (Figure 5p) and its thallus is colonized by cyanobacteria (Figure 5q).
We performed preliminary trials using as starting material 200 mg of tissue per trial, co-cultivation media with 30 or 40 mM MES, AGL1 Agrobacterium harbouring the p-AaEf1a::hphp-AaEf1a::eGFP-Lti6b plasmid.We successfully recovered stable transformants for all three species (Figure 6a-f).We also performed trials for the three species using 200 mg of tissue per trial, co-cultivation media with 40 mM MES and the GV3101 Agrobacterium harbouring the p-AaEf1a::hphp-35S_s::eGFP-Lti6b plasmid; however, we only recovered transformants for A. punctatus.The number of successful transformants (plant thalli) per experiment are summarized in Figure 6g,h   Optimised transformation protocol for hornworts 709 punctatus has the highest transformation efficiency and L. dussii has the lowest efficiency with only three stable lines recovered in total.

Simplified new transformation method
In summary the steps of the new optimized protocols are as follows (Figure 7, detailed description is provided in Methods and Figures S5-S8): fragmented regenerating thallus tissue (grown for at least 4 weeks) was co-cultivated with Agrobacterium in liquid KNOP supplemented with 1% (w/v) sucrose medium.Liquid media were supplemented with 40 mM MES and 3,5-dimethoxy-4hydroxyacetophenone (acetosyringone) at a final concentration of 100 lM.The pH was adjusted to 5.5.Co-cultivation duration was 3 days with shaking at 21°C on a shaker without any light supplementation (only ambient light from the The Plant Journal, (2023), 114, 699-718 room).After co-cultivation the tissue was plated on solid KNOP plates supplemented with 100 lg ml À1 cefotaxime and 10 lg ml À1 hygromycin.A month later successful transformants based on rhizoid production were visible on the plate.After 4 weeks it is recommended to transfer the tissue on fresh selective media plates.As we have demonstrated previously, the transgene is stably integrated into the genome (Frangedakis, Waller, et al., 2021) and lines can be propagated for years without losing the transgene.

New tools to label and target specific compartments of the hornwort cell
Hornwort cell biology is poorly understood.While other bryophytes such as the moss Physcomitrium patens and the liverwort M. polymorpha are being used increasingly to study cellular processes, to what extent these findings can be extended to hornworts is unclear due to their independent evolutionary history spanning over 400 million years (Naramoto et al., 2022;Pfeifer et al., 2022).Consequently, developing tools for hornwort specific cell biology studies is necessary.Hornworts also have multiple unique cellular features absent in mosses, liverworts, or vascular plants awaiting to be investigated.For instance, monoplastidic cells, putative plastid stromules, and the presence/ absence of a carbon concentrating structure in the chloroplast known as the pyrenoid.More generally, hornworts can be useful experimental systems to understand fundamental aspects of plant cell biology such as cell polarity, plasmodesma-related processes, and cell division.This is thanks to the simple morphology and the relatively flat shape of hornwort thallus that makes imaging at the cellular level easier compared with other systems.Thus, to develop tools facilitating the study of hornworts at a cellular level, we tested a series of potential signal peptides for subcellular localization at the mitochondria, Golgi, peroxisome, actin cytoskeleton, chloroplast, and the ER (Figure 8a; Figure S9).We further tested the applicability in hornworts, of new fluorescent proteins, a different reporter, and a new selection marker.Mitochondria in the liverwort, M. polymorpha, have been visualized using a targeting sequence derived from the A. thaliana Segregation Distortion 3 (SD3) gene, which encodes a protein with high similarity to the yeast translocase of the inner mitochondrial membrane 21 (TIM21) (Ogasawara et al., 2013).A mitochondrial targeting sequence from the Saccharomyces cerevisiae COX4 (ScCOX4) gene (Hamasaki et al., 2012) can also be used successfully in M. polymorpha (Figure S10).However, mitochondrial localization was not achieved when using either the A. thaliana SD3 or the ScCOX4 targeting sequences in A. agrestis Oxford.We thus tested the N-terminal sequence of a predicted SD3 A. agrestis gene (Sc2ySwM_117.1379.1)(Figure S9), for its ability to direct fluorescent localization in mitochondria.Using this sequence (p-AaEf1a::hph -AaEf1a::mVenus-AaSD3 construct), we observed fluorescence in structures in the cytosol that resemble mitochondria (Figure 8b).To confirm that these structures are indeed mitochondria, we stained thallus fragments with MitoTracker.The signal from mVenus overlapped with that from MitoTracker (Figure S11a-c), demonstrating that the N-terminal sequence of AaSD3 is sufficient to visualize mitochondria.We note that the observed signal has low fluorescence intensity and the use of a confocal microscope at high magnification is necessary for observation.
In M. polymorpha the Golgi has been visualized using the transmembrane domain of the rat sialyltransferase (Figure S10) as a targeting sequence (Kanazawa et al., 2016).Another Golgi localization sequence, which can be used in M. polymorpha, is a targeting sequence derived from the soybean (Glycine max) a-1,2 mannosidase 1 (GmMan1) gene (Luo & Nakata, 2012) (Figure S9).We tested whether the sialyltransferase or the GmMan1 peptides can be successfully used in A. agrestis Oxford for Golgi targeting.Only the GmMan1 peptide (p-AaEf1a::hphp-35Sx2:: mTurquoise2-GmMan1 construct) led to fluorescence in structures in the cytosol that resemble Golgi (Figure 8c).
The C-terminus of mouse talin (mTalin) has been used for actin labelling in A. thaliana and M. polymorpha (Kimura & Kodama, 2016;Kost et al., 1998).We generated A. agrestis Oxford plants stably transformed with a p-35S:: hph -p35S::mVenus-mTalin construct and we observed fluorescence localization in the filamentous structures in the cytosol (Figure 8e).To confirm that the observed structures are indeed actin filaments we treated thallus fragments with Latrunculin A (LatA), a reagent that depolymerizes actin.Actin filaments become disassembled after treatment with LatA indicating that mTalin can be used successfully in A. agrestis to target actin (Figure S11d,e).
In M. polymorpha the N-terminal targeting sequence from a predicted chitinase (Mp2g24440) in combination with the C-terminal HDEL ER retention peptide, has been successfully used for ER localization (Sauret-G€ ueto et al., 2020).We tested the N-terminal targeting sequence of a predicted chitinase in A. agrestis (Sc2ySwM_228.5627.1)(Figure S9), for its ability to direct fluorescent localization in ER (p-AaEf1a::hph -AaEf1a::mVenus-AaChit construct).We observed fluorescence in reticulate structures in the cytosol around the nuclei and in the plasma membrane (Figure 8g).Finally, we tested the targeting sequence derived from the A. thaliana Calcineurin B-like 3 (AtCBL3) gene for tonoplast localization, which is functional in M. polymorpha (Figure S10).However, we were not able to detect any fluorescent signal in A. agrestis Oxford.
We also confirmed that the tags targeting mitochondria, Golgi, chloroplast, and ER can be also used for A. punctatus; however, the signal is not as strong as in A. agrestis (Figure S12).The number of lines obtained, and the constructs used for transformation for both A. agrestis and A. punctatus are given in Table 2.

Applicability of the fluorescent protein mScarlet in hornworts
To expand the palette of fluorescent proteins (FP) that can be used in A. agrestis, we tested the expression of the synthetic monomeric Scarlet (mScarlet) 'red' fluorescent protein (Bindels et al., 2016).mScarlet is the brightest among the monomeric red FP with, for example, >3.5 times increase in brightness compared with mCherry.Given that its emission maximum is 594 nm, mScarlet can be combined with other FP such as eGFP, mVenus, or mTurquoise2 with minimal spectral overlap.We successfully generated A. agrestis Bonn lines expressing mScarlet (Figure 8h).The number of lines obtained are given in Table 2.
Applicability of the RUBY reporter and 2A peptides in hornworts RUBY is a new reporter (He et al., 2020) that converts tyrosine to the red coloured betalain that is visible without the need for extra processing steps.Three betalain biosynthetic genes (CYP76AD1, DODA, and Glucoysl transferase) are fused into a single transcription unit using a single promoter the P2A self-cleavage peptides and a terminator.We generated a construct where we fused the AaTip1;1 promoter with the RUBY cassette and the double NOS-35S terminator.However, when the plants transformed with the construct were examined, we did not observe any red colour (Figure S13a).
2A self-peptides are known to have different cleavage efficiencies (Liu et al., 2017).To test the efficiency of 2A  Optimised transformation protocol for hornworts 713 self-cleavage peptides we generated two constructs where the mVenus FP fused to a nucleus targeting sequence and the mTurquoise2 FP fused to a plasma membrane targeting sequence were combined into a single transcription unit separated by either the P2A or the E2A peptide (Figure S13b-i).When used in M. polymorpha the expected FP localization was observed (Figure S13j,k).However, in the case of A. agrestis Oxford mainly mVenus localization into the nucleus was observed suggesting that cleavage efficiency is very low.The number of lines obtained are given in Table S2.

New selection marker
The acetolactate synthase (ALS) is one of the enzymes that catalyses the biosynthesis of three essential amino acids, leucine, isoleucine, and valine.The herbicide chlorsulfuron is an ALS inhibitor and has been used in combination with mutated ALS genes as a selection marker in different plant species (Ogawa et al. 2008).We first tested whether A. agrestis is susceptible to chlorsulfuron and found that a 3-week incubation with 0.5 lM chlorsulfuron was sufficient to inhibit growth of untransformed thallus tissue (Figure S14).We then performed trials where we used a M. polymorpha mutated ALS gene (Ishizaki et al., 2015) driven by the AaEf1a promoter as a selectable marker, the GV3101 Agrobacterium strain and 30 or 40 mM MES.Up to 18 chlorsulfuron resistant plants were successfully recovered from 200 mg of tissue (Figure 8i-k).It must be noted that growth of transgenic plants on chlorsulfuron selection media is slower compared with hygromycin.It takes approximately an additional 2 weeks until primary transformants are visible by naked eye.

DISCUSSION
After testing several different approaches to increase transformation efficiency in the hornwort A. agrestis, it was pH control during co-cultivation that proved most successful.
Our results suggest that a pH of the co-cultivation medium about 5.5 is a key parameter to increase Agrobacterium infection rates.A Similar observation has been for A. thaliana Agrobacterium-mediated transformation (Wang et al., 2018).Consequently, our results support the notion of conserved innate immune responses via calcium signalling in land plants.These results may help to establish Agrobacterium-mediated transformation methods for plants that currently lack an efficient transformation method (e.g., lycophytes or various streptophyte algal lineages).

Plant material and maintenance
In this study we used the A. agrestis Oxford and Bonn isolates (Sz€ ov enyi et al., 2015) and Anthoceros punctatus (Li et al., 2014).

Tissue preparation for transformation
For the preparation of tissue used for transformation (fragmentation approach using razor blades), small pieces of thallus, approximately 2 mm 9 2 mm, were cut using sterile disposable scalpels (no.0501; Swann Morton) and placed on plates containing fresh growth medium (15-20 fragments per plate).The plates were grown for 4-6 weeks, at 21°C, 12 h of light, and 12 h of dark, 3-5 lmol m À2 sec À1 light intensity (Figure S2).
After 4-6 weeks, 1 g of thallus tissue (approximately 10 petri dishes) was transferred into an empty Petri dish, 2-10 ml of water was added and then the thallus fragmented using a razor blade (no.11904325; Fisher Scientific) for approximately 5 min (Video S1).The fragmented tissue was washed with 100 ml of sterile water using a 100-lm cell strainer (352 360; Corning) or until the flow through was clear (see Figures S5, S6, and S8).
When tissue was fragmented using a homogenizer, approximately 1 g of thallus tissue was homogenized in 20 ml sterile water using an Ultra-Turrax T25 S7 Homogenizer (727407; IKA) and corresponding dispensing tools (10442743; IKA Dispersing Element), for 5 sec, using the lowest speed of 8000 rpm.The homogenized tissue was washed with 100 ml of sterile water using a 100-lm cell strainer (352360; Corning) or until the flow through was clear.
The tissue and Agrobacterium were co-cultivated for 3 days with an orbital shaker at 110 rpm at 22°C without any additional supplementary light (only ambient light from the room, 1-3 lmol m À2 sec À1 ).After 3 days, the tissue was drained using a 100-lm cell strainer (352360; Corning) and moved on to solid KNOP plates (two Petri dishes from a single well) supplemented with 100 lg ml À1 cefotaxime (BIC0111; Apollo Scientific, Bredbury, UK) and 10 lg ml À1 hygromycin (10687010; Invitrogen) or 0.5 lM chlorsulfuron (Sigma 64902-72-3; to prepare 10009 stock chlorsulfuron solution, 100 mM chlorsulfuron super-stock solution was prepared in DMSO first, and then a 500 lM stock solution in dH 2 O).
If necessary, after 4 weeks, plants were transferred to fresh solid KNOP plates supplemented with 100 lg ml À1 cefotaxime and 10 lg ml À1 hygromycin and grown at 22°C under 12 h light: 12 h dark at a light intensity of 35 lmol m À2 sec À1 (TL-D58W/835; Philips).

Optimization trials
To test the effect of MES concentration on the transformation efficiency 3 g of A. agrestis Bonn thallus tissue was harvested 4 weeks after subculturing, fragmented with a razor blade, washed with 100 ml water, and distributed into six equal parts of 500 mg.Each part was then placed into a well of a six-well plate with 5 ml co-cultivation medium with different MES concentrations.The MES concentration of the co-cultivation medium in each of the six wells was: 0, 10, 20, 30, 40, or 50 mM.One hundred microlitre of the same Agrobacterium pre-culture, containing a p-AaEf1a::hphp-AaTip1;1-eGFP-Lti6b construct, was added to each well.Three replicates of this trial were performed.
To test the effect of co-cultivation medium pH value on the transformation efficiency, 2.5 g of A. agrestis Bonn thallus tissue was harvested 4 weeks after the subculture was made, fragmented with a razor blade, washed with 100 ml sterile distilled water, and distributed into five equal parts of 500 mg.Each part was then placed into a well of a six-well plate with 5 ml co-cultivation medium containing 40 mM MES, adjusted to different pH values.The pH value of the co-cultivation medium in each of the five wells was adjusted with 1 M KOH to: 5, 5.25, 5.5, 5.75, or 6.One hundred microlitre of the same Agrobacterium pre-culture, containing a p-AaEf1a::hphp-AaTip1;1-eGFP-Lti6b construct, was added to each well.Three replicates of this trial were performed.
To test the effect of co-cultivation sucrose concentration on the transformation efficiency, approximately 1.2 g of A. agrestis Bonn thallus tissue was harvested 4 weeks after subculturing, fragmented with a razor blade, washed with 100 ml water, distributed into six equal parts of 200 mg, and each part was then placed into a well of a six-well plate.In each plate, the sucrose concentration in three wells was 1% (w/v) and in the remaining three was 2% (w/v).Eighty microlitre of the same Agrobacterium (transformed with the p-AaEf1a::hph -35S_s::eGFP-Lti6b construct) preculture was added to each well.The experimental set-up was similar for testing the effect of the Agrobacterium strain (AGL1 or GV3101) on the transformation efficiency as well as the A. agrestis Oxford optimization trials with the only difference that the Agrobacterium was transformed with the p-AaEf1a::hphp-AaEf1a:: eGFP-Lti6b construct.
Co-cultivation and selection were carried out as described in the previous section.After 3-4 weeks on a first round of selection, surviving and growing thallus fragments were transferred to fresh selection plates.After an additional 2 months on second selection, growing thallus pieces that showed GFP expression (observed with a Leica M205 FA Stereomicroscope, GFP longpass filter) were counted as stable transformants.

Sample preparation for imaging
A gene frame (no.AB0576; ThermoFisher) was positioned on a glass slide.A thallus fragment was placed within the mediumfilled gene frame together with 30 ll of MilliQ water.The frame was then sealed with a coverslip.Plants were imaged immediately using a Leica SP8X spectral fluorescent confocal microscope.
Sporophytes of promoter-eGFP reporter lines were dissected from the gametophyte, and the gametophyte tissue around the base carefully removed with scalpels under a dissecting scope without disturbing the sporophyte base.The sporophytes were then transferred into a Lab-Tek chambered coverglass (no.155361; Lab-Tek) and overlaid with 5 mm solid KNOP medium (solidified with Gelrite) and water to keep them in place and prevent desiccation.eGFP expression within the sporophyte base of the living sporophytes was then visualized with a Leica TCS SP8 MP, excitation with a multiphoton laser at 976 nm, eGFP fluorescence detected with a bandpass filter (HyD-RLD 2ch FITC filter 525/ 50), and chloroplast autofluorescence with HyD-RLD 675/55.

MitoTracker staining
A 1 mM MitoTracker Red CMXRos (no.9082; Cell Signaling Technology, UK) stock solution was prepared in DMSO.Stock solution was diluted 100 times in distilled water and added to small A. agrestis thallus fragments (approximately 2 9 2 mm) and then incubated at room temperature for 1 h.Plant fragments (n = 5) were washed with distilled water four times and then imaged immediately using a confocal microscope.Excitation laser wavelength and captured emitted fluorescence wavelength window were as follows: 579 nm, 590-610 nm.

Latrunculin a treatment
A stock solution of 0.1 mM Latrunculin A (no. 9082; Abcam) in DMSO was prepared.For the LatA treatment, the stock solution was diluted 10 times in distilled water.For the control, DMSO was diluted 10 times in distilled water.Two gene frames (no.AB0576; ThermoFisher) were positioned on top of each other on a glass slide.A thallus fragment was placed within the gene frames together with approximately 60 ll of diluted LatA or diluted DMSO.Plants were imaged immediately using a Leica SP8X spectral fluorescent confocal microscope.Three independent experiments were performed.

AUTHOR CONTRIBUTIONS
MW, EF, and PS conceived and designed the experiments.EF and MW performed optimizations, and generated and characterized the hornwort transgenic lines.EF, MW, and AOM performed imaging.KSR collected, processed and imaged hornwort specimens in the light and transmission electron microscope.JR and SSG generated and characterized the transgenic M. polymorpha lines.JH and JMH provided resources.CRRS performed A. punctatus optimizations.MW, EF, and PS wrote the article with contributions from all the authors.

Figure 1 .
Figure 1.Characteristics of the Anthoceros agrestis Bonn strain and results of the optimization experiments to improve transformation efficiency.(a) A. agrestis Bonn plant bearing sporophytes.Scale bar: 3 mm.(b) Longitudinal section of the foot surrounded by the involucre showing the basal meristem (arrow) and differentiating sporogenous tissue.(c) Light microscopy image of the sporophyte epidermis showing stomata (see arrows).Scale bar: 10 lm.(d) Image of A. agrestis Bonn thallus fragment (before razor blade or ultra-turrax aided fragmentation) used as starting material for the co-cultivation with Agrobacterium.Scale bar: 2 mm.(e) A. agrestis Oxford plant bearing a young sporophyte.Scale bar: 3 mm.(f) Image of A. agrestis Oxford thallus fragment grown under the same conditions as the Bonn thallus in (d) (used as starting material for the co-cultivation with Agrobacterium).Scale bar: 2 mm.(g, h) Comparison of the number of transformants (per 500 mg starting tissue) and final pH of the co-cultivation media (after 3 days co-cultivation) for (g) different MES concentrations in the co-cultivation media (initial pH set to 5.5), and (h) different initial pH values of the co-cultivation media with 40 mM MES. Graphs show values of triplicate experiments (symbols) and their average (bars).Error bars on pH values depicting SDs.(i) Transgenic A. agrestis Bonn transformed with the p-AaEf1a::hphp-AaTip1;1::eGFP-Lti6b construct.Scale bar: 1 mm.(j) Confocal microscopy image of A. agrestis Bonn gametophyte transformed with the p-AaEf1a::hphp-AaTip1;1::eGFP-Lti6b construct.Scale bar: 50 lm.Construct maps at Table S1.(k-m) Effect of the promoter (AaEf1a versus the CaMV 35S) driving the hygromycin selection cassette.(k) Histogram showing numbers of recovered independent transformant lines expressing GFP.(l) Boxplot showing distribution of thallus size between the two constructs (box depicts quartile and median), with whiskers showing maximum and minimum values inside 1.59 IQR, outliers outside that range depicted as dots.(m) Top: transgenic A. agrestis Bonn transformed with the p-35S::hphp-35S_s::eGFP-Lti6b construct.Bottom: transgenic A. agrestis Bonn transformed with the p-AaEf1a::hphp-AaTip1;1::eGFP-Lti6b construct.Scale bar: 1 mm.(n-p) Comparison of the number of transformants (per 200 mg starting tissue) (after 3 days co-cultivation) for (n) different sucrose concentrations in the transformation buffer (initial pH set to 5.5, 20 mM MES), (o) different Agrobacterium strains used for the transformation (tested with A. agrestis Bonn), and (p) different fragmentation methods (tested with A. agrestis Bonn).Ó 2023 The Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Plant Journal, (2023), 114, 699-718 ), Ó 2023 The Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Plant Journal, (2023), 114, 699-718

Figure 4 .
Figure 4. Hornwort phylogeny and the morphological features of Anthoceros punctatus.(a)Hornwort phylogeny(Villarreal & Renner, 2012).Genera whose species could be successfully transformed using our improved protocol are marked with red arrowheads.(b-j) Morphology of A. punctatus.(b) Germinating spore that develops into a thallus.Scale bar: 4 lm.(c) The thallus is irregularly shaped, lacks specialized internal differentiation and is composed of mucilage chambers and parenchyma cells.Scale bar: 2 mm.(d) Scanning electron microscopy of thallus.Scale bar: 100 lm.The male (antheridia are shown with a yellow arrow) (e) and female (archegonia) reproductive organs are embedded in the thallus.Scale bar: 0.5 mm.(f) Thallus with sporophytes (see white arrow).(g) Longitudinal section of the foot (see yellow arrow) surrounded by the involucre showing the basal meristem and differentiating sporogenous tissue.Scale bar: 100 lm.(h) Transverse section of the sporophyte showing its morphology.From the centre to outside: columella, spores, pseudoelaters, assimilative tissue, epidermis, and stomata with substomatal cavities.Scale bar: 100 lm.(i) Transmission electron microscopy of the chloroplast (yellow arrow: pyrenoid).Scale bar: 1 lm.(j) Hand section of A. punctatus thallus showing ellipsoidal cavity colonized by cyanobacteria.Scale bar: 100 lm.Ó 2023 The Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Plant Journal, (2023), 114, 699-718 Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Plant Journal, (2023), 114, 699-718 . A.

Figure 7 .
Figure 7. Schematic representation of the optimized transformation protocol.(a) Thallus tissue is routinely propagated on a monthly basis under low light.Four-to 6-week-old tissue is fragmented with the aid of a razor blade, transferred to a cell strainer, and washed thoroughly with sterile water.(b) The tissue is then co-cultivated with Agrobacterium for 3 days (under low light) and (c) spread on antibiotic-containing growth medium.After approximately 4-6 weeks, putative transformants are visible.A final round of selection is used to eliminate false-positive transformants.Ó 2023 The Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Plant Journal, (2023), 114, 699-718

Ó
2023 The Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Plant Journal, (2023), 114, 699-718 Optimised transformation protocol for hornworts 711 Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Plant Journal, (2023), 114, 699-718

Figure S7 .
Figure S7.Workflow showing steps of the protocol used to transform A. punctatus.Figure S8.Workflow showing steps of the protocol used to transform L. dussii.Figure S9.Summary of sequences used as transit peptides in this study to tailor specific localization of fluorescent proteins.Figure S10.Localization of fluorescent proteins tagged with various transit peptides in the liverwort Marchantia polymorpha.Figure S11.MitoTracker and LatA treatment.

Figure S10 .
Figure S7.Workflow showing steps of the protocol used to transform A. punctatus.Figure S8.Workflow showing steps of the protocol used to transform L. dussii.Figure S9.Summary of sequences used as transit peptides in this study to tailor specific localization of fluorescent proteins.Figure S10.Localization of fluorescent proteins tagged with various transit peptides in the liverwort Marchantia polymorpha.Figure S11.MitoTracker and LatA treatment.

Figure S11 .
Figure S7.Workflow showing steps of the protocol used to transform A. punctatus.Figure S8.Workflow showing steps of the protocol used to transform L. dussii.Figure S9.Summary of sequences used as transit peptides in this study to tailor specific localization of fluorescent proteins.Figure S10.Localization of fluorescent proteins tagged with various transit peptides in the liverwort Marchantia polymorpha.Figure S11.MitoTracker and LatA treatment.

Figure S12 .
Figure S12.Targeting fluorescent proteins to various subcellular compartments in the hornwort A. punctatus.Figure S13.Testing utility of the RUBY reporter and the 2A self cleavage peptides in A. agrestis.Figure S14.Chlorsulfuron sensitivity of A. agrestis gametophytes.
Ó 2023 The Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Plant Journal, (2023), 114, 699-718

Table 1
eGFP expression patterns in the stable transformants Ó 2023 The Authors.The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd., The Plant Journal, (2023), 114, 699-718 Optimised transformation protocol for hornworts 705 p-AaEf1::eGFP-Lti6b

Table S2 .
Number and expression of RUBY constructs.