A Simple Auxin Transcriptional Response System Regulates Multiple Morphogenetic Processes in the Liverwort Marchantia polymorpha

In land plants comparative genomics has revealed that members of basal lineages share a common set of transcription factors with the derived flowering plants, despite sharing few homologous structures. The plant hormone auxin has been implicated in many facets of development in both basal and derived lineages of land plants. We functionally characterized the auxin transcriptional response machinery in the liverwort Marchantia polymorpha, a member of the basal lineage of extant land plants. All components known from flowering plant systems are present in M. polymorpha, but they exist as single orthologs: a single MpTOPLESS (TPL) corepressor, a single MpTRANSPORT INHIBITOR RESPONSE 1 auxin receptor, single orthologs of each class of AUXIN RESPONSE FACTOR (ARF; MpARF1, MpARF2, MpARF3), and a single negative regulator AUXIN/INDOLE-3-ACETIC ACID (MpIAA). Phylogenetic analyses suggest this simple system is the ancestral condition for land plants. We experimentally demonstrate that these genes act in an auxin response pathway — chimeric fusions of the MpTPL corepressor with heterodimerization domains of MpARF1, MpARF2, or their negative regulator, MpIAA, generate auxin insensitive plants that lack the capacity to pattern and transition into mature stages of development. Our results indicate auxin mediated transcriptional regulation acts as a facilitator of branching, differentiation and growth, rather than acting to determine or specify tissues during the haploid stage of the M. polymorpha life cycle. We hypothesize that the ancestral role of auxin is to modulate a balance of differentiated and pluri- or totipotent cell states, whose fates are determined by interactions with combinations of unrelated transcription factors.


Introduction
Genetic and physiological studies have revealed that the phytohormone auxin, indole-3-acetic acid (IAA), controls central aspects of plant development in both haploid and diploid generations of plant species. Auxin triggers growth and tissue differentiation in places where it accumulates, perhaps acting as a morphogenetic trigger [1]. In the flowering plant Arabidopsis thaliana, a balance between auxin synthesis, transport, conjugation/de-conjugation and degradation regulates the spatial distribution of auxin, leading to activation of transcriptional responses that influence a variety of morphological and physiological processes [2,3,4,5,6,7,8].
The auxin transcriptional response in its simplest outline is as follows. In the absence of auxin, AUXIN/INDOLE-3-ACETIC ACID (AUX/IAA) repressors heterodimerize with AUXIN RESPONSE FACTOR (ARF) transcription factors (via two conserved domains, D34) and recruit TOPLESS (TPL) corepressors, thus preventing ARFs from regulating target genes [9,10,11,12,13]. In the presence of auxin, the AUX/IAA proteins are targeted for degradation by an SCF E3 ubiquitin ligase (TRANSPORT INHIBITOR RESPONSE1 [TIR1]/AUXIN SIG-NALING F-BOX [AFB]), via auxin dependent interactions between AUX/IAA and TIR1 proteins [14,15,16,17]. Phylogenetic analyses of ARF evolution in land plants identifies three ARF classes: class A (orthologues of Arabidopsis ARF5-8), class B (orthologues of Arabidopsis ARF1-4, ARF9) and class C (orthologues of Arabidopsis ARF10, ARF16, ARF17) [18,19,20]. However, not all classes of ARFs are equal, with class A acting primarily as activators, and classes B and C as repressors [19,20]. Since auxin application predominantly activates gene expression, it has been argued that class B and C ARFs may act to repress common target genes activated by class A ARFs since all classes can potentially bind the same auxin responsive element (AuxRE) in promoters of auxin responsive genes [19,20,21]. Among the direct targets of ARFs are genes encoding AUX/IAA proteins, establishing a negative feedback loop [22,23]. Studies in Arabidopsis have suggested that activator ARFs interact with AUX/IAAs and that AUX/IAAs may also homodimerize, but that repressor ARFs do not interact efficiently with the AUX/IAAs, leaving most class B and C repressors outside this protein network [10]. These results imply that many ARF repressors regulate gene expression independently of auxin, as TIR1/AFB mediated degradation of AUX/IAAs would not affect their activity. However, all ARF proteins potentially compete for the same targets as they recognize similar AuxREs [11,21]. Thus, a current view is that transcriptional activation by class A ARFs is negatively regulated in an auxin dependent manner by AUX/IAAs and an auxin independent manner by class B and C ARFs.
That 23 ARF, 29 AUX/IAA, 6 TIR1/AFB, and 5 TPL paralogs are encoded in the Arabidopsis genome has prompted speculation that interactions differing in affinity among paralogs could influence auxin sensitivity in different cells [10,24]. Key features of the auxin signaling pathway are mechanistically similar in the moss Physcomitrella patens suggesting conservation of the auxin transcriptional response throughout land plants [25]. Given the complexity arising from multiple paralogs encoding each of the components of the auxin transcriptional response pathway in both flowering plants and P. patens [18], we sought to establish a model system where such complexity is reduced. It has been speculated that an increase in complexity of the auxin genetic toolkit (including genes for metabolism, synthesis, transport and signaling) is correlated with an increased complexity in land plant architecture, from bryophytes to flowering plants [26]. Although there is no universal agreement on the phylogenetic relationships of extant land plant clades, it is generally accepted that liverworts diverged earlier than other land plants based on morphology, fossil evidence, and some molecular analyses [27,28,29]. Liverworts in general have experienced a low rate of chromosomal evolution, with little evidence of widespread polyploidy, and molecular evolution within the Marchantiopsida is reported to be slow, compared with more derived liverwort lineages [30,31]. Given these evolutionary attributes, we reasoned that Marchantia polymorpha could possess a simplified version of the auxin response system. Furthermore, as M. polymorpha has been used as a model system for nearly two centuries to investigate both universal and specific biological processes, an extraordinary body of literature is available to inform our studies.
In M. polymorpha, the haploid spore produces a short protonemal filament only a few cells long [32,33,34]. The brief protonemal stage is followed by the formation of a globular sporeling that subsequently develops a marginal row of cells in its apical (distal) region, a central cell of which will form the apical cell of the gametophytic shoot meristem [33,34,35]. The marginal row of cells is conspicuous and forms the distal growing tip, and represents the transition from a sporeling to a 'prothallus'-a flattened unistratose heart shaped plant that grows by means of an apical cell with two alternate cutting faces [33,35]. Subsequently, a change in the wedgeshaped apical cell such that divisions occur along four planes, rather than two, transforms the prothallus into a thallus. Accompanying this transition is the establishment of dorsiventrality, with the production of specialized cell types both dorsally and ventrally [36]. M. polymorpha can also reproduce asexually via small, flattened lecticular gemmae [32,33,37,38]. Mature gemmae are dormant and lack dorsiventral polarity while inside dorsal gemmae cups, but acquire environmentally induced polarity when displaced from the gemmae cup. Individual gemmae are clonal and derived from a single cell of the parental thallus [32,33,37,39].
The effects of exogenous auxin on M. polymorpha gemmaling growth have been reported in previous studies. Some generalizations of the effects of exogenous auxin in young gemmalings can be made as follows: in low concentrations auxin increases the size of the thallus [40], breaks gemmae dormancy [41,42], produces formation of additional meristems in dormant gemmae [43], and confers apical dominance [44]. At higher concentrations auxin promotes the formation of numerous rhizoids on the dorsal side of the gemmae, a dorsiventral patterning defect, and further increases in concentration results in formation of a callus-like tissue with limited growth [40,41,42,43,44,45,46,47,48,49,50,51,52,53,54]. In transgenic M. polymorpha, an auxin-responsive regulatory sequence derived from soybean, the GRETCHEN HAGEN 3 (GH3) promoter, In the present work, we show that the liverwort M. polymorpha possesses the simplest auxin genetic toolkit so far found in any land plant. Dissection of the auxin response system of M. polymorpha, points to auxin being a facilitator rather than a determiner of cell fate, with auxin also being required for major developmental transitions during gametophyte development.

Results
Auxin influences most developmental stages of M. polymorpha gametophyte development To address the role of auxin in the M. polymorpha gametophyte, young undifferentiated gemmae and young thalli with established dorsiventral polarity were grown in the presence of the auxin analogues naphthalene-1-acetic acid (NAA) and 2,4-dichlorophenoxyactetic acid (2,4-D).
Auxin application to gemmae and gemmalings grown up to four days old on unsupplemented media prior to auxin treatment triggered ectopic rhizoid formation on the dorsal surface of the thallus (Fig 1A and 1B). Normal dorsal structures including air chambers, air pores, and gemmae cups did not develop. However, plants were not entirely ventralized since scales, which are ventral structures, did not develop dorsally. Thalli also failed to develop a normal branching pattern and lacked lateral growth.
Application of auxin to gemmalings with established dorsiventral polarity, those grown for six and sixteen days on unsupplemented media prior to auxin treatment, resulted in two classes of aberrant phenotype. In the first class, occurring primarily in plants with auxin treatment after one week, the thallus developed excessive rhizoids, although they were spatially restricted to the original central zones of the gemmalings and to the ventral side of the mature thallus (S1A and S1B Fig)-locations where rhizoids arise in untreated plants. Thus, by day six of gemmaling development dorsal tissues have lost the capacity to produce rhizoids in response to auxin (S1B Fig). However, typical dorsal structures of the thallus were affected, with air pores often either partially formed or unopened (Fig 1C and 1D, S1B and S1C Fig), air chambers structure asymmetric and disorganized, and an undulating dorsal epidermal surface with conspicuously protruding air pores ( Fig 1D). Additionally, plants subject to this treatment were unable to form any gemmae cups (S1A-S1C Fig).
In the second, milder class, primarily occurring in plants with auxin treatment after two weeks, plants exhibited abnormal tubular elongation of gemmae cups that had been initiated prior to auxin treatment ( Fig 1E-1H and S1D Fig). The gemmae cups exhibited a positive gravitropic response in contrast to their normal negative graviotropism [55]. While previously formed gemmae cups continued to develop, no new cups were formed. Air chambers were asymmetric, spatially disorganized, and protruding upwards as in the first class (Fig 1F-1H). The growing thallus margins were bent downwards ( Fig 1F and S1D Fig), an epinastic growth pattern reminiscent of epinastic angiosperm leaf growth in response to high levels of auxin [56].

Lack of auxin affects gemmae cup formation and thallus branching
As a complementary approach we examined the phenotypic consequences of reducing endogenous auxin levels. We first tested the effects of the auxin biosynthesis inhibitor L-Kynurenine (L-Kyn) on gemmaling development [57]. In Arabidopsis L-Kyn acts as an alternative substrate and competitive inhibitor for the TRYPTOPHAN AMINOTRANSFERASE OF ARABIDOPSIS1 (TAA1) and TRYPTOPHAN AMINOTRANSFERASE RELATED1, 2 (TAR1, 2) enzymes, inhibiting the conversion of tryptophan (trp) to indole-3-pyruvic acid (IPA), a key step in auxin biosynthesis [4,57,58]. Arabidopsis plants grown in the presence of 100 μM L-Kyn phenocopy  [57,59].
Wild-type gemmae grown on 100 to 250 μM L-Kyn were unable to produce normal gemmae cups, with gemmae cup rims failing to fully develop (Fig 1I and 1J). Growth of gemmae was slower as compared to wild type (S1E and S1F Fig), with smaller propagules forming inside the cup (Fig 1I and 1J). Decreased rhizoid production was also observed but was not measured quantitatively (Fig 1K and 1L).
During vegetative growth, a gemmae cup is usually initiated following each branching event [36] and only a single cup is formed per branching event (S2 Fig). Immediately after branching a growth phase separates the two apical notches creating a mid-lobe prior to a new round of branching and gemmae cup formation. Application of L-Kyn to established seven-day-old thalli resulted in gemmae cups derived from successive branching events fusing into a single structure (Fig 1M and 1N). The extent of fusion varied, with the most extreme fusion resulting in single oval gemmae cups (e.g. Fig 1N and 1R). Therefore, inhibiting auxin synthesis reduces the rate of branching and the extent of growth separating branches.
In order to further deplete free auxin levels in the thallus, we created transgenic plants overexpressing a heterologous bacterial auxin conjugation enzyme iaaL [60,61,62] under the control of the EF1-α promoter (proEF1), which drives high levels of expression in the meristematic regions and throughout the thallus [63]. pro EF1:iaaL plants exhibited disruptions to patterning and growth, with thallus growth severely stunted, despite evidence of continuing cell division (Fig 1O  and 1P). Except for a few cells developing as rhizoids, most cells in pro EF1:iaaL plants do not acquire a particular fate, but instead resemble cells present in early stages of sporeling development.
To manipulate free auxin levels specifically in the meristematic region we utilized a 3.2kb promoter fragment of the single M. polymorpha orthologue of the SHORT INTERNODES/ STYLISH (SHI/STY) gene family [64,65,66]. Under normal growth conditions, pro MpSHI:GUS activity was detected mainly in the apical notch (meristematic region) during all stages of development ( Fig 1Q). Most pro MpSHI:iaaL transformants exhibited weaker phenotypes than pro EF1:iaaL and were able to form a narrow thallus with compromised lamina expansion and a reduced rate of branching (Fig 1S, S1G and S1H Fig). Gemmae cups sometimes developed to the side of the midline, and/or close together, suggesting an irregular zig-zag formation and a degree of fusion (Fig 1S), resembling L-Kyn treatments ( Fig 1R). We thus conclude that auxin influences the growth and differentiation of multiple tissues and processes in different developmental stages of the M. polymorpha gametophyte.
Auxin transcriptional response genes are conserved in M. polymorpha Analysis of the MpTPL sequence revealed highly conserved LisH, CTLH and WD40 repeat domains, characteristic of this gene family (amino acid sequence identity between MpTPL and Arabidopsis TPL is 68%). MpIAA harbors a conserved EAR domain (D1), as well as a proteolysis domain (D2); D1 allows AUX/IAA proteins to interact with the TPL CTLH domain, and D2 domain has been shown to facilitate interactions between AUX/IAA proteins and TIR1/ AFB [15]. Both MpIAA and all three MpARFs possess conserved protein-protein interaction domains (D34; S3 Fig and S1 Table). In an accompanying paper, Kato et al [67] describe the extent of MpARF-MpIAA interactions in M. polymorpha using yeast two-hybrid (Y2H) and biomolecular fluorescence complementation (BiFC).
Bayesian phylogenetic analyses were performed to elucidate the evolution of the TPL, AUX/IAA, and ARF gene families using sequences from M. polymorpha, P. patens, Selaginella moellendorffii, Pseudotsuga menziesii, Amborella tricopoda, Solanum lycopersicum and Arabidopsis thaliana. The TPL phylogeny suggests that the common ancestor of land plants had a single TPL gene, with the M. polymorpha genome retaining this ancestral feature. Our analysis suggests a gene duplication in the lineage leading to the origin of tracheophytes (maximum support), creating two classes of TPL genes (Class A and B). This topology implies a loss of class A genes in S. moellendorffii (S4 Fig) and further sequence sampling may clarify if this is a trend in other lycophytes.
Phylogenetic analysis of the AUX/IAA family also suggests a single gene in the embryophyte ancestor, rendering M. polymorpha ancestral for AUX/IAA gene copy number. Our phylogenetic tree indicates that angiosperms and gymnosperms also have two types of AUX/IAAs, class A (similar to the Arabidopsis IAA1-7, IAA9, IAA14-17, IAA19 and IAA27) and class B AUX/IAAs (including IAA10-13, IAA18, IAA26, IAA28 and IAA29), suggesting a gene duplication in the common ancestor of seed plants (S5 Fig).
We performed Bayesian analysis of land plant ARF genes, and as noted in previous analyses, our phylogenetic analysis results in a tree with three distinct and well-supported ARF clades

Loss of MpIAA results in auxin hypersensitivity
To test whether the components identified by homology act in an auxin signaling pathway, we created loss-and gain-of-function alleles of the genes. Two independent artificial microRNAs (amiRs) were designed to target MpIAA (amiRMpIAA 7 and amiRMpIAA 9 ). For amiR design in M. polymorpha, we used the backbone of the endogenous MpMIR160 precursor (S7 Fig), replacing the miR and miR Ã sequences with sequences to target the gene of interest, and cleavage was corroborated using RLM-RACE (Fig 2A; [68]). Both amiRs were constitutively expressed under the EF1-α promoter (proEF1) and the reduction of target MpIAA transcripts was corroborated (S8 Fig). While most pro EF1:amiRMpIAA 7 and pro EF1:amiRMpIAA 9 lines had nearly wild-type phenotypes, some lines exhibited a curly and convoluted thallus with long rhizoids, and miRMpIAA transcript levels are largely, albeit not perfectly, correlated with the severity of the phenotype (S8 Fig). Consistent with MpIAA having a role in auxin signaling, a majority of amiRMpIAA 7/9 lines showed severe growth defects in exogenous auxin compared to wild-type controls (Fig 2). When grown in 2.5 and 5μM 2,4-D auxin media from the sporeling stage, pro EF1:amiRMpIAA 7/9 plants were dramatically smaller compared to wild-type controls, suggesting that exogenous auxin causes lethality when transcript levels of MpIAA are reduced ( Fig  2B and S8D Fig). When mature pro EF1:amiRMpIAA 9 thalli were grown in mock media for 15 days and then transferred to 7.5μM 2,4-D plates for 20 days, loss of MpIAA resulted in profuse rhizoid production and a decrease in growth compared to controls ( Fig 2B). In summary, knock-down MpIAA lines were auxin hypersensitive, indicating that MpIAA represses the auxin signaling pathway in M. polymorpha.

MpARF1 is a modulator of the auxin signaling pathway
To elucidate whether auxin hypersensitivity observed in MpIAA knockdown lines was due to misregulation of ARF genes, we explored the functional role of MpARF1. We focused on MpARF1 since MpARF1 transcripts are readily detectable in the thallus, whereas we did not detect transcripts of  . When MpARF1 and MpARF1 ΔD34 genes were driven by pro EF1-α, plants displayed auxin hypersensitivity, exemplified by formation of ectopic rhizoids in adults exposed to 5μM 2,4-D ( Fig 3A). Even in the absence of auxin, proEF1:MpARF1 ΔD34 plants exhibited excessive rhizoid production with rhizoids growing to greater lengths compared to a Hyg Res control, visible when plants are grown on vertically oriented plates such that gravity influenced rhizoid growth is evident (Fig 3B). proEF1:MpARF1 and proEF1:MpARF1 ΔD34 plants exhibited higher rates of branching, seen as a significant increase in apical notches compared to controls (Fig 3A and 3C; S3 Table), and proEF1:MpARF1 ΔD34 lines had increased gemma cup production compared to proEF1:MpARF1 lines (Fig 3C). The extent of phenotypic aberration was correlated with MpARF1 transgene expression ( Fig 3D). Additionally, the levels of MpIAA expression appear to be inversely correlated with levels of MpARF1 expression, supporting the idea that MpARF1 inhibits transcription of MpIAA MpTPL is involved in patterning multiple tissues in M. polymorpha development The lack of strong phenotypic effects in knock-down MpIAA lines prompted us to examine the single member of the TOPLESS corepressor family in M. polymorpha. Expression patterns of a pro MpTPL:3xVENUS fusion, including 3.6kb upstream of MpTPL transcript up to the nearest upstream transcript, exhibited intense fluorescent signal in the meristematic regions during most developmental stages, including mature gemmae (Fig 4A), five-day-old gemmalings ( Fig  4B), and mature 25-day-old thalli ( Fig 4C) compared to autofluroesence in the wild type (S10A Fig). Expression was also detected in gemmae cups ( Fig 4D). Transfer of mature pro MpTPL:3x-VENUS plants to media containing 10 μM 2,4-D resulted in a decreased level of expression of the reporter gene, suggesting regulatory feedback between auxin and MpTPL ( Fig 4E). Expression of pro MpTPL:3xVENUS was observed throughout the thallus, although the signal was weaker compared to pro EF1:3xVENUS lines (Fig 4F).
The wild-type dorsiventral M. polymorpha thallus is several cell layers thick. The single apical cell in the shoot mersitem has four cleavage planes: one dorsal, one ventral, and two lateral [70,71]. Dorsal and lateral derivatives contribute to the formation of dorsal photosynthetic air chambers and gemmae cups (Fig 4G). Notably both dorsal gemmae cups and ventral rhizoids are produced along a midline (Fig 4F and 4G, and S2 Fig). Wild-type gemmae cups produce a characteristic cup rim that is several cell layers thick at its base and unistratose towards its apex and acts to store and protect dormant gemma (Fig 4K).
Ectopic expression of wild-type MpTPL using proEF1-α resulted in slightly smaller and darker green thalli able to form a bifurcating thallus with normal dorsiventral polarity ( Fig 4H and S10B  Fig). However, gemmae cup formation was disrupted in 80% of pro EF1:MpTPL lines (N = 15), with gemmae cups entirely absent in some cases (Fig 4H and 4L). The loss of gemmae cups did not impair the formation of gemmae, as naked gemma and gemma primordia developed in the positions where gemmae cups would normally develop along the midline (Fig 4K and 4L), reminiscent of pharmacological effects of L-Kyn on the thallus. Intriguingly, pro EF1:MpTPL lines were hypersensitive to 2.5 μM exogenous 2,4-D, similar to knock-down MpIAA lines (Fig 4J).
We created a putative MpTPL loss-of-function allele analogous to the dominant negative Arabidopsis TPL (N176H) and drove its expression with pro EF1 [9]. The most conspicuous phenotype of pro EF1:MpTPL N176H plants was an increased frequency of thallus branching compared to wild type and pro EF1:MpTPL plants (Fig 4I; S2 Table). In this regard pro EF1:MpTPL N176H plants resemble MpARF1 overexpression lines (Fig 4M). Thalli of pro EF1:MpTPL N176H plants produced nearly three apical notches in the time in which wild-type plants produce one apical notch (Fig 4G and 4I). pro EF1:MpTPL N176H plants had diminished production of gemmae cups, and when produced, gemmae cups were aberrant, being narrow and elongate, reminiscent of gemmae cups that develop in the presence of exogenous auxin (Fig 4M). When reproductive development was induced under far-red light pro EF1:MpTPL N176H gametophore stalks showed decreased elongation compared to wild type, resulting in gametophores of short stature and with dorsal protrusions reminiscent of rhizoids (S10C Fig).

Auxin insensitive M. polymorpha plants fail to establish a dorsiventral thallus
In Arabidopsis translational fusions of TPL and D34 of AUX/IAA proteins, such as IAA12/ BODENLOS (BDL; TPL-D34 BDL ), constitutively bind to ARF5/MONOPTEROS (MP) via interactions with D34, keeping MP in a repressive complex even in the presence of auxin [9]. Given the putative role of MpIAA and MpTPL proteins in auxin signaling in M. polymorpha, chimeric translational fusions of MpTPL to D34 of all M. polymorpha loci bearing putative D34 of protein-protein interaction were generated and all fusions were regulated by pro EF1, making sure non of the fusions carried an endogenous miR binding site (S11 Fig; [40]). pro EF1:MpTPL-D34 MpIAA plants exhibited severe dwarfism ( Fig 5A) and failed to establish a dorsiventral thallus. Individuals were composed of a mass of undifferentiated cells and incompletely developed organs with random orientations (Fig 5B). Occasional rhizoids, scales and air chambers were observed, but these structures were not organized relative to one another as in wild-type plants (Fig 5B). Similar phenotypes were observed for pro EF1:MpTPL-D34 MpARF1 (Fig 5C and 5D), pro EF1:MpTPL-D34 MpARF2 (Fig 5E). A majority of these lines only achieved the level of differentiation of the sporeling or prothallus stages of wild-type development ( Fig  5F), with individuals not transitioning further than single cell layer tissues (Fig 5D). Fusions of TPL with D34 of BDL ( pro EF1:MpTPL-D34 BDL ) resulted in similar phenotypes, suggesting a high degree of functional conservation between liverwort and angiosperm D34 IAA protein domains (Fig 5G). Weaker pro EF1:MpTPL-D34 BDL lines had a narrow thallus with fused gemmae cups, as in L-Kyn treatments and in lines expressing the IAA conjugating enzyme iaaL (S12A and S12B Fig).
In contrast to the severe developmental defects described above, pro EF1:MpTPL-D34 MpARF3 plants (Fig 5H) did not exhibit altered patterning in a manner comparable to other chimeric genes tested. pro EF1:MpTPL-D34 MpARF3 thalli had a convoluted dorsal surface (Fig 5H and S12C  Fig), with thallus margins curled inwardly towards the midline, especially close to the apical notches, resulting in a more compact thallus than that of wild type (Fig 5H). pro EF1: MpTPL-D34 MpARF3 thalli produced ventral rhizoids and dorsal air chambers and gemmae cups. However, pro EF1:MpTPL-D34 MpARF3 plants grew bigger than other TPL-D34 fusions (S12C Fig). As MpTPL acts as a repressor of auxin transcriptional response, we drove expression of the chimeric proteins with the MpTPL regulatory sequences. pro MpTPL:MpTPL-D34 MpIAA lines had a similar, but slightly milder, phenotype than pro EF1:MpTPL-D34 MpIAA lines: plants failed to pattern a flattened dorsiventral thallus and also formed a cluster of organs similar to wild- type prothalli (Fig 5F). Some pro MpTPL:MpTPL-D34 MpIAA lines however were able to form a single gemma cup able to produce gemmae (S12D and S12E Fig). The gemmae from these cups had distorted shapes with marginal extensions, and lacked the two apical notches conspicuous on wild-type gemmae (S12E Fig). One prediction of constitutively expressing MpTPL-D34 chimeric proteins is a downregulation of activator ARF activity even if MpIAA proteins are degraded due to the presence of auxin, i.e. the transgenic plants should be auxin insensitive. When grown in the presence of exogenous auxin pro EF1:MpTPL-D34 MpIAA plants occasionally produce ectopic rhizoids compared to controls (Fig 5I-5K), while expression of MpTPL-D34 MpIAA using pro MpTPL produced auxin insensitive plants as assayed by the lack of ectopic rhizoid formation (Fig 5L). These results indicate that expression of the chimeric MpTPL-D34 MpIAA protein is capable of creating auxin insensitive plants if it expressed in the appropriate cells at appropriate levels.

A minimal auxin signaling network is sufficient to pattern a complex body plan
Phylogenetic analyses indicate the M. polymorpha genome possesses orthologues encoding components of the auxin transcriptional response machinery as it has been characterized in other land plants [72,73,74]. In contrast to other land plant lineages only single orthologs of each of the three classes of ARF and single orthologs of the AUX/IAA, TIR1/AFB, and TPL gene families are present in M. polymorpha. Phylogenetic analyses also suggest that the last common ancestor of land plants had single orthologs of these classes of genes, implying that M. polymorpha has retained the ancestral condition. Single orthologs of other regulatory gene families, e.g. the class III and IV HD-ZIP transcription factor families, in the last common ancestor of land plants have also been proposed and may be a common feature of the ancestral land plant [75,76]. As the genome of the filamentous charophycean alga Klebsormidium does not encode any of the canonical auxin transcriptional response machinery [77], its origin must lie either at the base of the land plant lineage or along the charophycean algal grade leading to land plants, as is the case for several other transcriptional regulators [78].
In flowering plants, the complexity and the robustness of the auxin transcriptional response has been hypothesized to be generated by a large number of signaling network paralogs encoding components of differing affinities [10,24]. However, the relative transcriptional simplicity of the auxin response system in M. polymorpha is sufficient to pattern its complex thallus with several tissue and cell types. While the common ancestor of mosses and flowering plants is also predicted to have possessed only single orthologs of the genes in question, relatively recent gene duplications in the lineage leading to the moss P. patens has resulted in this species possessing multiple closely related paralogs [74]. As both M. polymorpha and P. patens have arguably similar levels of complexity at the gametophytic level, our findings support the idea that the relationships between auxin signaling components-rather than the number of paralogsare the key to establish a complex body plan. Regarding the sporophyte generation, it is still possible to speculate that increases in the numbers of paralogs of the auxin genetic toolkit could be correlated with increased morphological complexity but this requires further sampling of land plant genomes, in particular those of hornworts.

Conserved auxin transcriptional machinery in M. polymorpha
Several lines of evidence indicate a conserved role for MpIAA, MpARF1 and MpTPL in auxin response. First, these genes are orthologous to those that have been shown to act downstream of auxin in developmental and physiological processes in other land plants (S4-S6 Figs). Second, similar phenotypic defects are seen in M. polymorpha when expressing MpTPL-D34 translational fusions with D34 derived from either the Arabidopsis AUX/IAA protein BDL, or the M. polymorpha MpIAA, MpARF1 or MpARF2 proteins (Fig 3D-3F). Third, pro MpTPL: MpTPL-D34 IAA M. polymorpha plants are insensitive to exogenous auxin, similar to the corresponding plants in Arabidopsis [9]. Fourth, MpTPL-D34 alleles and loss-and gain-of-function alleles of MpIAA, MpARF1 and MpTPL recapitulate phenotypes observed when M. polymorpha is grown in the presence of either exogenous auxin or auxin biosynthesis inhibitors (Fig 1), as is discussed in detail below. Fifth, reduction of MpIAA transcripts or ectopic expression of MpARF1 and MpTPL results in auxin hypersensistivity (Figs 2, 3 and 4J). Sixth, loss-of-function alleles of MpARF1 result in auxin hyposensitivy [69,79]. Taken together, these data strongly suggest that the MpTPL, MpARF1, and MpIAA genes encode components of an auxin transcriptional response system in M. polymorpha.

The auxin transcriptional response machinery in land plants
Our data in M. polymorpha support aspects of the model proposed for Arabidopsis explained in detail in the introduction. That indistinguishable phenotypes were observed when any of the chimeric proteins MpTPL-D34 MpIAA , MpTPL-D34 MpARF1 , or MpTPL-D34 MpARF2 were constitutively expressed suggests MpARF1 can homodimerize as well as heterodimerize with MpIAA and MpARF2 in planta. Y2H and in BiFC experiments on these proteins also suggest that they can all interact with one another, with the strongest interactions occurring between MpIAA and MpARF1 [67]. Furthermore, that pro TPL:MpTPL-D34 MpIAA plants are insensitive to auxin indicates that interactions between MpIAA, MpARF1, and MpARF2 mediate the activation of gene expression in response to auxin in M. polymorpha. In contrast, constitutive expression of MpTPL-D34 MpARF3 chimeric proteins does not inhibit establishment of dorsiventrality or differentiation of tissues, suggesting MpARF3 does not interact effectively with MpARF1 and MpIAA proteins, pointing to a possible role of class C ARFs regulating gene expression in an auxin independent manner. Consistent with this hypothesis, is the observation that MpARF3 can't homodimerize and its interaction with MpIAA is weaker than the interactions between MpIAA and either MpARF1 or MpARF2 [67].
Auxin patterns the development of many tissues in M. polymorpha A large body of literature, primarily of pharmacological experiments whereby M. polymorpha has been grown in the presence of exongenous auxins, anti-auxins, and auxin transport inhibitors, has indicated auxin is involved in rhizoid formation, establishment of dorsiventral polarity in gemmalings, polar thallus regeneration, gemma and apical cell dormancy, gemmae cup development, air pore development, and spacing of organs during branching [40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,80]. As outlined below, our pharmacological and genetic studies confirm and extend the known roles of auxin in the development of multiple structures during several stages of M. polymorpha development. For the sake of clarity, the different experiments and their putative effects performed in this study are summarized in S4 Table. Dorsiventral polarity. Auxin distribution influences establishment of the dorsiventral body plan. Consistent with earlier reports, growth of gemmae from germination on auxin containing media resulted in rhizoid formation from both dorsal and ventral surfaces. In addition, transfer of gemmae with established dorsiventrality to auxin plates caused several other developmental abnormalities, such as distorted growth of air chambers, air pores, and gemmae cups, but not changes in dorsiventrality (Fig 1A-1H). This is consistent with the observation that dorsiventrality is established within the first 24 hours of gemmaling growth and is irreversible after this time [32,45,81,82,83,84,85].
Rhizoids. Rhizoid formation is promoted by auxin in receptive tissues. When gemmae are germinated in the presence of exogenous auxin, cells on both the dorsal and ventral surfaces have the potential to differentiate into rhizoids, however, once dorsiventrality is established, dorsal cells lose the capacity to form rhizoids in response to high levels of auxin. However, in gemmalings with established polarity, auxin induces production of additional central and ventral rhizoids and they grow to greater lengths (S1B Fig). Conversely, when gemmae are grown in the presence of auxin biosynthesis inhibitors, fewer, and sometimes shorter, rhizoids develop from the central surface (Fig 1K and 1L). Finally, pro MpTPL:TPL-D34 IAA plants do not form rhizoids even in the presence of exogenous auxin.
Gemmae and gemmae cups. Gemmae cups are one of the most sensitive tissues to disruptions in auxin homeostasis. When grown in the presence of auxin biosynthesis inhibitors gemmae cup and gemmae formation is delayed (Fig 1I and 1J). Over-expression of MpTPL (as in pro EF1: MpTPL lines), which presumably results in a dampening of auxin transcriptional response, also results in a reduction in gemmae cup growth (Fig 4H, 4K and 4L). In contrast, when thalli with established polarity are treated with exogenous auxin elongated tubular cups are formed (Fig 1E-1H). The tubular gemmae cups often show a gravitropic response opposite to the usual negative gravitropism previously described for gemma cup growth [55], and consistent with the notion that auxin transduces a gravity stimulus or inhibits a phototropic response [86,87]. Excess auxin may be transported into gemmae cups, or alternatively, auxin-conjugating enzymes (such as GH3) may metabolize auxin in most other tissues except the gemmae cup walls.
Branching. Auxin influences the rate of branching during thallus growth. Growth of the M. polymorpha thallus occurs via divisions of a single apical cell with four cutting faces [70,71]. Branching is dichotomous (S2 Fig), but the precise mechanism, e.g. via division of the apical cell into two apical cells or de novo formation of nearby apical cells, is not known. Application of the auxin biosynthesis inhibitor L-Kyn and expression of a bacterial auxin-conjugating enzyme (iaaL) appear to slow the rate of branching. L-Kyn in particular, causes a disruption of the growth phase that spatially separates apical notches (Fig 1M, 1N and 1R). Thalli often have gemmae cups produced by two adjacent apices fused into a single structure, suggesting fields of cells can act coordinately in the development of a single gemmae cup, indicating M. polymorpha exhibits a remarkable plasticity during organogenesis. In contrast, an increase in the rate of branching results from over-expression of MpARF1 and MpARF1 ΔD34 (Fig 3A and 3C; S3 Table) as well as the putative reduction in MpTPL activity caused by the dominant negative construct pro EF1:MpTPL N176H (Fig 4I; S2 Table). Because the overexpression of an MpIAA-insensitive MpARF1 ΔD34 protein results in an enhanced phenotype compared to the one observed in MpARF1 overexpressing lines, it is unlikely that these lines have secondary loss-of-function effects as a result of increased MpIAA transcription [11,12,20,21,88]. Thus, MpARF1 facilitates activities as diverse as cell expansion (expansion of rhizoids) and pluoripotency reprograming (bifurcation).
Developmental stage transitions. Auxin is required for gametophyte developmental stage transitions. Several different MpTPL-D34 lines, especially in the cases of pro EF1: MpTPL-D34 MpARF1 , and pro EF1:MpTPL-D34 MpARF2 , phenotypically resemble wild type at the prothallus stage (Fig 5B-5F), with a multitude of undifferentiated unistratose lobes composed (Fig 5D). Additionally, the phenotypes of strong pro EF1:iaaL lines are reminiscent of wild-type sporelings, prior to transitioning into a prothallus (Figs 1O, 1P and 5F). These results suggest that auxin is required for the morphological transitions between successive developmental stages during M. polymorpha gametophyte development-from sporeling to prothallus, which can be viewed as a transition from unorganized growth to 2-dimensional patterning, and from prothallus to thallus, which is a transition from 2-dimensional to 3-dimensional patterning. The requirement of auxin for developmental transitions has been observed in another bryophyte lineage, the mosses, where auxin is necessary to produce 3-dimensional gametophores from protonemal filaments [25,64]. The ubiquitous requirement for auxin for morphological transitions during development of these basal lineages of land plants suggests that the common ancestor of embryophytes likely utilized auxin as an internal cue to promote morphological transitions. Given the origin of the auxin transcription response machinery was at or near the origin of land plants and that a key innovation of land plants is development of 3-dimensional bodies from an apical cell of at least three cutting faces, it is plausible that the incorporation of auxin response into developmental programs facilitated the evolution of complex land plant body plans.
Organized tissue patterning. In many pro EF1:iaaL lines there is little, if any, organized tissue growth even though the plants continue to grow for months, suggesting that a reduction in auxin levels, that while not inimical to life, results in an almost complete loss of tissue patterning in M. polymorpha (Fig 1). Auxin insensitive pro MpTPL:MpTPL-D34 IAA and pro EF1: MpTPL-D34 IAA plants can produce many of the tissues found in a wild-type thallus, including air chambers, air pores, gemmae cups and gemmae. However, the organization of the tissues relative to one another and of the tissues themselves is severely disrupted (Fig 5B-5G). Such distortions of tissue patterning, (e.g. shapes of air chambers, air pores, gemmae cups), are also observed when thalli are grown in the presence of exogenous auxin or auxin biosynthesis inhibitors. Taken together, these data suggest that auxin, or processes downstream of auxin are required for not only developmental transitions, but also for the patterning of tissues relative to each other and development most tissue types in the M. polymorpha gametophyte.
In a classic experiment nearly a century ago, geotropic sensitivity to decapitated roots was restored by placing either another root tip or, alternatively, a shoot tip on the decapitated stump, thus demonstrating that auxins produced in the shoot and root tips do not have specificity of action, but rather act in a context dependent manner [89]. Given that auxin acts in the patterning of most tissue and cell types of the M. polymorpha gametophyte and the lack of redundancy in its components of the auxin transcriptional response machinery, there appears to be no tissue or cell type specificity of action with respect to auxin signaling components. Rather, it seems that since the origin of land plants, auxin acted as a facilitator rather than a determiner of cell fate, with the specific output of ARF action depending upon the milieu of other transcription factors present in a particular cell [90,91].

Plant growth conditions
Sporangia from an Australian population of Marchantia polymorpha (from now on referred to as wild type) were collected from a southeastern Melbourne field location (37°57'48.36"S, 145°6 '20.41"E). Sporangia were sterilized in 1x sterilization solution (2% sodium hypochlorite, 0.1% TritonX-100) for 1 minute and subsequently washed with water. Sporangia were then dried and stored at 4°C.
Wild-type spores were suspended in water and plated on petri dishes with Gamborg's B5 basal medium (B5; Austratech,). Dormant gemmae from adult plants were subcultured for pharmacological assays. Spores were also grown in liquid 0M51C media [92] for transformation (see transformation section below).

Phylogenetic analyses
Nucleotide sequence alignments for seven species with complete genomes (Physcomitrella patens, Selaginella moellendorffii, Picea abies, Pseudotsuga menziesii (used only for TPL), Amborella trichopoda, Solanum lycopersicum and Arabidopsis thaliana) plus the isolated M. polymorpha genes (see below) were created manually using Se-Alv2.0a11 [93]. Ambiguous alignment regions were trimmed for phylogenetic inferences. Bayesian phylogenies were performed using Mr. Bayes 3.2.1 [94]. A nucleotide TPL phylogeny with 3111 characters and 23 taxa was performed assuming variable rates between first, second and third positions (prset ratepr = variable). The analysis was run for 500,000 generations, sufficient for two simultaneous runs to converge with statistical significance (P <0.01). A nucleotide AUX/IAA phylogeny with 309 characters and 49 taxa was performed assuming variable rates between first, second and third positions (prset ratepr = variable). The analysis was run for 5,000,000 generations, sufficient for two simultaneous runs to converge with statistical significance (P <0.01). For the ARF phylogeny, a nucleotide matrix with embryophyte sequences (44 taxa, 996 characters) were run in Mr. Bayes for 750,000 generations assuming variable rates between first, second and third positions (prset ratepr = variable). Each analysis was run until the standard deviation of split frequencies dropped below 0.01. For all phylogenies, 25% of trees were discarded in the burn-in phase.

Cloning and plasmids
All primer sequences are available in S5 Table. The MpEF1 promoter was amplified with primers MpEF1-NdeI-F and MpEF1-SalI-R, cloned into pCRII-TOPO (Invitrogen) and sequenced. pro EF1 was subcloned into the shuttle vector BJ36 using NdeI and SalI. The resulting plasmid was called pro EF1 BJ36 and was used to create constitutive expression constructs described below.
Cloning of MpSHI promoter and pro MpSHI:GUS. A 3981bp promoter/UTR fragment of the single SHI/STY ortholog in M. polymorpha, MpSHI, was amplified using primers ME1 and ME2. The fragment was cloned in pCRII-TOPO, creating pME68, and subsequently transferred to pRITA, which has MCS upstream of a GUS reporter gene and transcriptonal terminator sequence [95] using NdeI and KpnI, creating pME82. A pro MpSHI:GUS NotI fragment from pME82 was later subcloned to pSKF HART. pSKF HART, created by Dr. Sandra Floyd, is a binary vector based on pBART (used for Arabidopsis transformation), but in contrast to the BASTA/Glufosinate resistance gene in pBART, HART has two CaMV35S promoters in tandem driving an hptII gene (Hygromicin resistance). An NdeI/KpnI pro MpSHI fragment from pME68 was subcloned into pro EF1 BJ36, replacing the EF1 promoter, resulting in plasmid pME159.
EF1:iaaL and MpSHI:iaaL constructs. A bacterial iaaL CDS was amplified from plasmid iaaL-BJ36 with primers ME83 and ME84. The resulting fragment was cloned in pCRII-TOPO and sequenced. iaaL was subcloned into pro EF1 BJ36 and pME159 using KpnI and HindIII. The resulting pro EF1:iaaL and pro MpSHI:iaaL fragments were then subcloned into pSKF HART.
Cloning and mutagenesis of MpTPL. A 1kb MpTPL fragment was cloned from wild-type M. polymorpha thallus cDNA using a nested PCR approach with degenerate primers TPL-F1, TPL-F2, TPL-R1 and TPL-R2, flanking a conserved region in TPL transcripts. The obtained fragment was cloned in pCRII-TOPO and sequenced. For 5'RLM-RACE outer primers (UPM and TPLR4) and inner primers (NUP and TPLR3) were used. For 3'RACE TPLF3, TPLF4 and a GAGA primer were used. All fragments were cloned in pCRII-TOPO and sequenced. A fulllength transcript was amplified using primers MpTPL-SalI-F and MpTPL-EagI-R and cloned and sequenced in pCRII-TOPO. This clone was named 1416G. PCR based mutagenesis was used to create MpTPL N176H . An initial set of two PCR fragments were amplified using primers M13F and MpTPL(N176H)-R (fragment 1), and primers MpTPL(N176H)-F and M13R (fragment 2), using 1416G as template. Both fragments 1 and 2 were used as template in a subsequent PCR reaction and amplified with primers M13F and M13R. The PCR product (named 1626) was cloned in pCRII-TOPO and sequenced.
Fusions using the endogenous MpTPL promoter were made by excising the MpTPL promoter with NdeI and MluI and subcloning it into pro EF1:MpTPL-D34 MpIAA BJ36 and subsequently subcloning into pSKF HART using NotI.
MpTPL reporters. Promoter sequences were obtained by assembling partial 454 reads from the first M. polymorpha genome drafts provided by JGI. Once a putative assembly was made, experimental corroboration of promoters were performed using PCR. A 3.6Kb fragment of the MpTPL promoter was amplified using primer MpTPLproF5-NdeI and MpTPLproR1--SalI, and cloned and sequenced in pCRII-TOPO (named 3685). The MpTPL promoter was subcloned into 3xVENUSN7BJ36 using NdeI and SalI. 3xVENUSN7 has 3 copies of YFP Venus with a C-terminal Nuclear localized sequence [96]. pro MpTPL:3xVENUSN7 BJ36 was subcloned into pSKF HART using NotI.
Cloning of MpARF1. MpARF1 was amplified using RACE with nested PCRs. For 5'RACE, primers UPM, MpARF(miR167)R4-out, NUP and MpARF(miR167)R3-in were used. For 3'RACE, GAGA and ARFmiR167F were used. Full length transcripts were amplified with MpARF1(BamHI)-F1 and MpARF1(EcoRV)-R1 and cloned and sequenced in pCRII-TOPO (Invitrogen). The full-length transcript was subcloned into pE2B using BamHI and EcoRV. The resulting plasmid was subcloned into a modified version of pSKF HART with attR1 and attR2 sites that allow a Gateway LR recombination (Invitrogen). The dominant MpARF1 ΔD34 allele was created by assembly PCR, erasing the internal NotI site with a silent mutation using Universal M13 Primers and primers MpARF1-mNotI-F2 and MpARF1-mNotI-R2. A second round of PCR using primers MpARF1-ATG-F-KpnI and MpARF1-DeltaD34-R-HindIII removed domains D34 of MpARF1. The resulting fragment was sequenced and subcloned into EF1:BJ36 using KpnI/HindIII. All constructs were subcloned into HART binary vector using NotI.
Cloning of amiRs. amiRs 268 bp in length were designed in silico with flanking KpnI and HindIII sites and synthesized at GeneScript (USA). amiRs were subcloned into EF1:BJ36 using KpnI and HindIII and later into HART or KART as described in previous sections.

RLM-RACE
RNA was extracted using QIAGEN Plant RNAeasy Kit. Total RNA was ligated to an RNA oligo (First Choice RLM-RACE) using T4 RNA ligase at 37°for one hour. Samples were precipitated using Phenol:Chlroform and eluates were used as a template for cDNA synthesis using Bioscript Reverse Transcriptase (Bioline), according to manufacturers instructions. PCRs were performed using Ex-Taq (TAKARA) using nested PCR using 5'RACE-Out-F and 5'RACE-in-F (First Choice RLM-RACE) as forward primers. To detect cleaved MpIAA transcripts, RLM-MpIAA-R3 and RLM-MpIAA-R1 were used as outer and inner reverse primers, respectively.

Branching rates measurements
Primary sporeling-transformants expressing either pro EF1:MpTPL N176H or a Hyg Res control T-DNA were simultaneously plated and notches were counted by eye after 25 days of growth in B5 media (N = 17).

Plant transformation and culture
M. polymorpha transformation was performed essentially as described previously [92]. M. polymorpha spores were grown in 25ml liquid 0M51C media for 9 to 11 days prior to Agrobacterium co-cultivation. Agrobacterium strain GV3001 was transformed with binary vectors and cultured for 2 days at 28°C in LB media whereafter the bacteria was pelleted and cultured for an additional 4 hours in 10ml 0M51C media with 100μM acetorsyringone (Austratech). 500ul bacterial suspension and acetosyringone (100μM final concentration) was added to 25ml spore suspension. Spores and bacteria were co-cultivated for 48-60 hours. Primary transformants (T1 generation) were selected on Gamborgs B5 plates with 10μg/ml Hygromicin (Sigma-Aldrich) and/or G418 (Sigma-Aldrich) and 200μM Timentin (Austratech). After 10-15 days, survivors were re-plated to a second round of selection. After approximately 1 week, survivors were transferred to unsupplemented B5 media or stored at 4°C for later use. Gemmae from the T1 generation (G1 generation) or gemmae from the G1 generation (G2) were used for all experiments, except in cases where no viable gemmae were formed in T1 plants.

Microscopy
Plants were observed in a Lumar.V12 dissecting microscope (Zeiss) and photographed with an AxioCam HRc (Zeiss). Fixation for SEM was performed in FAA (ethanol 50%, acetic acid 5%, Formaldehyde 10%) overnight at 4°C. Samples were dehydrated in ethanol before critical point drying in a CPD 030 (Baltec). Samples were sputter coated with gold using a SCD 005 (Baltec). Scanning Electron Microscopy was performed with microscope S-570 (Hitachi) at 10kV and photos were digitalized using SPECTRUM Software (Dindima). Summary of a 20-day-long time-lapse series of wild-type thallus with, initially, a single apical notch (Day 0; white asterisk). By day four, the first gemmae cup is visible (c). By day five, thallus bifurcation can be observed, with two apical notches evident (light blue asterisks). The first gemmae cup has by now been displaced away from the growing point (meristem). The two daughter notches separate by the growth of a mid-lobe (L). By day twelve, new gemmae cups (c') can be seen behind each of the two daughter notches. By day sixteen, a third branching event is visible by eye and the one new gemmae cups (c') have matured and start producing gemmae (dark blue asterisks). The asterisks in the central Fig (summary) show the path of a single apical notch following two branching events and 20 days of growth. The path follows the midline of the thallus. The motion through time is due to growth of the thallus tip. Scale bars are 1mm.