The Mechanism Forming the Cell Surface of Tip-Growing Rooting Cells Is Conserved among Land Plants.

To discover mechanisms that controlled the growth of the rooting system in the earliest land plants, we identified genes that control the development of rhizoids in the liverwort Marchantia polymorpha. 336,000 T-DNA transformed lines were screened for mutants with defects in rhizoid growth, and a de novo genome assembly was generated to identify the mutant genes. We report the identification of 33 genes required for rhizoid growth, of which 6 had not previously been functionally characterized in green plants. We demonstrate that members of the same orthogroup are active in cell wall synthesis, cell wall integrity sensing, and vesicle trafficking during M. polymorpha rhizoid and Arabidopsis thaliana root hair growth. This indicates that the mechanism for constructing the cell surface of tip-growing rooting cells is conserved among land plants and was active in the earliest land plants that existed sometime more than 470 million years ago [1, 2]. Copyright Â

Results and Discussion

The first land plant rooting structures comprised systems of tip-growing filamentous cells (rhizoids). Comparing the genetic mechanism that controls the development of filamentous rooting cells (rhizoids and root hairs) among different groups of land plants allows us to reconstruct the mechanism that controlled the development of these first land plant rooting systems. To identify genes required for the growth of rhizoids in one of the earliest diverging taxa of land plants, we generated a mutant population of 336,000 lines by transforming germinating Marchantia polymorpha spores with the pCAMBIA1300 T-DNA vector and screened for plants with defective rhizoid morphology (DRM). 301 DRM mutants were isolated (Table S1); 165 mutants were crossed to wild-type and the mutant phenotype was inherited in the F1 generation, whereas 136 were not successfully crossed (Table S1). The approximate 1:1 segregation of wild-type to DRM mutant rhizoid phenotypes in the F1 generation for each of the 165 inherited mutants indicated that the DRM phenotypes were caused by single nuclear mutations (Table S2). The DRM mutant rhizoid phenotype co-segregated with the hygromycin resistance encoded by the hygromycin phosphotransferase gene on the T-DNA in 62 of the 165 inherited mutant lines (Table S2). This is consistent with the hypothesis that the insertion of a T-DNA carrying a functional hygromycin resistance gene caused the mutation that resulted in defective rhizoid growth in 37% of the inherited mutants (Table S2).

To identify T-DNA insertion sites, we first generated a draft assembly of the M. polymorpha genome. Because the plants used in the mutant screen grew from spores generated in a cross between wild-type male (Takaragaike-1 [Tak-1]) and female (Tak-2) accessions, DNA was isolated from Tak-1 and Tak-2 plants, pooled, and sequenced. Illumina HiSeq technology was used to generate 84,554,420 short-insert paired-end reads and 32,963,957 long-insert paired-end reads. The draft genome comprised 4,137 scaffolds with a total scaffold length of 206 Mb (Data S1), scaffold N50 length of 376 kb, and estimated coverage of 64× (Data S1). To identify protein-coding genes in this draft genome, we sequenced, assembled, and mapped an M. polymorpha gametophyte transcriptome onto the genome assembly. The transcriptome was generated using pooled RNA isolated from mature dorsal thallus epidermis (excluding midrib region and gemma cups), the meristematic zone, rhizoids, and 0- and 1-day-old gemmae. RNA was sequenced using Illumina HiSeq in 183,475,609 short-insert paired-end reads and assembled into contigs (Data S1); 29,453 gametophyte-expressed contigs were mapped to the genome assembly. The whole-genome shotgun assembly (DDBJ: LVLJ00000000) and transcriptome shotgun assembly (ENA: GEFO00000000 and GenBank: GEFO01000000) have been deposited at the DNA Data Bank of Japan, European Nucleotide Archive, and GenBank.

The genomic locations of 57 of the 62 T-DNAs linked to DRM mutations were identified by thermal asymmetric interlaced (TAIL) PCR (Data S2). The T-DNA insertion sites of the 57 DRM mutants were distributed among 31 different genes (Figure S1). TAIL PCR was also carried out on DRM mutants that were sterile and could not be crossed, and this resulted in identification of the three alleles of MpCSLD1 and two alleles of MpSCD. Therefore, in total, 33 genes were identified in the mutant screen. Additional mutant alleles in eight of the 33 genes—Mpalba-3, Mpemb2756-2, Mpexl-1, Mppi4ka1-5, Mppti-2, Mpsri1-1, Mpsri3-2, and Mpxut-3—were identified by sequencing DNA flanking T-DNA insertions in sterile DRM mutants. The Mpire mutation was complemented with a transgene expressing the wild-type MpIRE-coding sequence (Figure S3). Phylogenetic analysis was conducted to assign putative functions and identify related genes in Arabidopsis thaliana (Table 1). Trees were constructed with maximum-likelihood statistics using protein sequences predicted from the M. polymorpha transcriptome assembly and published A. thaliana genome (Figure S2; Data S3 and S4). In total, we identified between one and five alleles in 33 genes; multiple independent mutant alleles were identified for 17 genes, and single alleles were identified for 16 genes.

Table 1

Genes Required for Rhizoid Growth

Gene	Predicted Function of Encoded Protein	Closest Arabidopsis Homolog	Mutant Phenotype	No. of Mutant Alleles	In Arabidopsis
					Expression Enriched in Root Hairs	Role in Root Hair Development
Cell Wall Biosynthesis and Integrity Sensing
MpCSLD1	cellulose synthase-like class D protein	AT3G03050	very short rhizoids	3	yes	yes
MpCSLD2	cellulose synthase-like class D protein	AT3G03050	short rhizoids	5	yes	yes
MpPTI	PTI-like serine/threonine kinase	AT2G30740	short rhizoids	2	yes	yes
MpXUT1	xyloglucan-specific galacturonosyltransferase	AT5G41250	very short rhizoids	3	yes	yes
MpGMP	GDP-mannose pyrophosphorylase	AT2G39770	very short rhizoids	1		embryo lethal
MpRHM	rhamnose biosynthesis	AT1G78570	short rhizoids	1	yes	yes
MpTHE	CrRLK1L family receptor-like kinase	AT5G54380	very short rhizoids	1	yes	yes
Vesicle Transport and Cytoskeleton
MpPI4Ka1	1-phosphatidylinositol 4-kinase alpha	AT1G49340	very short rhizoids	6
MpSCD	Rab guanine nucleotide exchange factor	AT1G49040	short rhizoids	2	yes	yes
MpSPI	WD-40 repeat protein	AT1G03060	short rhizoids	3	yes	yes
MpSRI1	Rab guanine nucleotide exchange factor, similar to S. cerevisiae RIC1	AT3G61480	short rhizoids	3	yes
MpWDL	microtubule-binding protein/TPX2 domain-containing protein	AT2G35880	curly rhizoids	3	yes
MpXI	class XI myosin	AT3G12130	short rhizoids	5	yes	yes
MpAP5M	AP-5 complex subunit mu	AT2G20790	short rhizoids	1	yes
MpREN	pleckstrin homology domain/RhoGAP domain-containing protein	AT5G12150	curly rhizoids	1	yes
MpSRI2	calcium-binding EF-hand family protein, similar to S. cerevisiae PAN1	AT1G21630	very short rhizoids	1	yes
MpZWI	calmodulin-binding/microtubule motor	AT5G65930	short rhizoids	1	yes
Others/Unknown Function
MpALBA	alba-like DNA/RNA-binding protein	AT1G76010	short rhizoids	5
MpEMB2756	DUF616-containing protein, ceramidase	AT1G34550	short/few rhizoids	2
MpEXL1	EXORDIUM-like	AT4G08950	short rhizoids	2
MpFBA1	fructose-bisphosphate aldolase	AT4G38970	short rhizoids	4
MpGATA1	class A GATA zinc-finger transcription factor	AT5G25830	short rhizoids	2	yes
MpIRE	AGC kinase	AT5G62310	very short rhizoids	1a	yes	yes
MpSRI3	unknown protein, ceramide metabolic process	AT5G42660	short rhizoids	2
MpTMT	tonoplast monosaccharide transporter	AT3G51490	short rhizoids	2	yes
MpACLB-2	ATP citrate lyase subunit B	AT5G49460	short rhizoids	1
MpCPR	regulator of expression of pathogenesis-related (PR) genes	AT5G64930	short rhizoids	1
MpGDPD	glycerophosphodiester phosphodiesterase	AT3G02040	short rhizoids	1	yes	yes
MpGDPDL	glycerophosphodiester phosphodiesterase-like	AT3G20520	few rhizoids	1	yes	yes
MpPRPL	plastid ribosomal protein large subunit	AT1G07320	very short rhizoids	1
MpSQE	squalene monooxygenase	AT1G58440	short rhizoids	1		yes
MpSRI4	unknown protein, similar to S. cerevisiae EFR3	AT2G41830	short rhizoids	1
MpTZP1	zinc knuckle (CCHC-type) family protein	AT5G49400	short rhizoids	1

See Data S3 for full gene names. See also Figures S1–S3 and Tables S1 and S2.

The Mpire mutation was complemented by a transgene expressing the wild type MpIRE gene.

Of the 33 characterized DRM genes, five—MpCELLULOSE SYNTHASE-LIKE CLASS D 1 (MpCSLD1), MpCSLD2, MpXYLOGLUCAN-SPECIFIC GALACTURONOSYLTRANSFERASE 1 (MpXUT1), MpGDP-MANNOSE PYROPHOSPHORYLASE (MpGMP), and MpRHAMNOSE BIOSYNTHESIS 1 (MpRHM1)—encode proteins that are predicted to function in the synthesis of cell wall polysaccharides. Consistent with the assigned functions, each of these DRM mutants—Mpcsld1, Mpcsld2, Mpxut1, Mpgmp, and Mprhm1—develops shorter rhizoids than wild-type, and Mpcsld1 and Mpxut1 mutant rhizoids also burst at their tips (Figure 1; Table 1). Closely related A. thaliana orthogroup members—AtCSLD3, AtXUT1, and AtRHM1—are expressed in root hairs and required for root hair growth because Atcsld3, Atxut1, and Atrhm1 mutants develop short root hairs [3, 4, 5, 6, 7]. A role for AtGMP1 (AT2G39770) in root hair development has not yet been defined. This is most likely because loss of AtGMP1 function is lethal and mutants do not survive to the stage where root hairs develop [8]. Taken together, these data demonstrate that the same molecular mechanism for wall synthesis operates in M. polymorpha rhizoids and A. thaliana root hairs.

Figure 1

Phenotypes of Mutants with Defects in Cell Wall Biosynthesis and Cell Wall Integrity Sensing

Genes encoding proteins involved in cell wall biosynthesis and integrity sensing are required for rhizoid elongation.

(A) Mpcsld1, Mpcsld2, Mpgmp, Mprhm, Mpthe, and Mpxut1 mutants develop shorter rhizoids than wild-type (Tak-1 and Tak-2); 21-day-old gemmalings. Scale bar, 5 mm.

(B) Defects in cell wall synthesis result in the rupture of the rhizoid tip in Mpcsld1 and Mpxut1 mutants. MpTHE is required for cell wall integrity sensing in elongating rhizoids, because Mpthe rhizoids rupture at their tip. Arrowheads mark the site of brown staining at rhizoid tips indicative of cell wall rupture; 2-day-old gemmalings. Scale bar, 100 μm.

See also Figure S4.

The sensing of cell wall integrity requires a signal transduction cascade that has been defined in A. thaliana. Receptor kinases in the Catharanthus roseus RECEPTOR KINASE 1-LIKE (CrRLK1L) subclass are required for cell wall integrity sensing in tip-growing cells [9, 10, 11]. Maximum-likelihood phylogenetic trees were constructed using CrRLK1L protein sequences from A. thaliana and M. polymorpha (Figure S2). There are 17 CrRLK1L family members in A. thaliana [12], which include AtTHESEUS (AtTHE) and AtFERONIA (AtFER). There is a single member of this family in M. polymorpha that we designated MpTHE because it is more similar to AtTHE than to any other A. thaliana protein in this family (Figure S2). Mpthe mutants develop short and irregularly shaped rhizoids, indicating that the MpTHE protein is required for rhizoid elongation (Figure 1B). To test the hypothesis that the MpTHE protein controls cell wall integrity in M. polymorpha, we examined rhizoid tips in Mpthe mutants for evidence of bursting. The tips of Mpthe rhizoids are brown as a result of rhizoid rupture during elongation and do not reach the same length as wild-type (Figure 1B). The defective elongation and cell integrity phenotypes of Mpthe and Atfer mutant rhizoids and root hairs, respectively, suggest that CrRLK1L proteins carry out similar functions in cell wall integrity signaling in both species and in their last common ancestor. The AtMARIS PTI kinase is a receptor-like cytoplasmic kinase (RCLK) required for cell wall integrity signaling, and acts downstream of CrRLK1L proteins in A. thaliana [13]. M. polymorpha has a single PTI protein that is sister to a group of six A. thaliana PTI proteins that includes MARIS. Mutants with defective MpPTI function develop short rhizoids (Figure 1A), just as mutants that lack AtMRI activity develop short root hairs in A. thaliana. Taken together, these data suggest that at least some of the components associated with wall integrity sensing—RLCK and CrRLK1L proteins—have been conserved since M. polymorpha and A. thaliana last shared a common ancestor.

Ten of the 33 genes identified in this screen encode proteins involved in vesicle transport or cytoskeleton function (Table 1; Figure 2) [14, 15, 16, 17, 18, 19, 20, 21, 22, 23]. Close homologs of nine are highly expressed in root hairs, and three of these are required for root hair growth in A. thaliana. Mutations in the gene encoding the plant-specific class XI myosin, MpXI, result in the development of short rhizoids (Table 1; Figure 2), just as triple and quadruple myosinXI mutants develop short root hairs in A. thaliana [16, 17]. MpREN is a predicted ROP-GAP protein (Table 1; Figure 2) [18], and because ROPs control microfilament dynamics in A. thaliana [24, 25, 26, 27] it is likely that this protein modulates microfilament dynamics during rhizoid growth. Microtubules are involved in growth direction control in tip-growing cells [28]. Consistent with this role is the observation that Mp wave dampened-like (Mpwdl) mutants developed wavy-shaped rhizoids similar to oryzalin- or Taxol-treated root hairs (Table 1; Figure 2) [28]. Mpzwi rhizoids are shorter than wild-type, indicating that the ZWICHEL-like (ZWI) kinesin motor is required for rhizoid growth (Table 1; Figure 2). Whereas homologs of MpWDL and MpZWI have not been shown to be required for root hair growth in A. thaliana, the expression of AtZWI and two AtWDL genes (AT2G35880 and AT4G32330) is enriched in root hairs (Arabidopsis eFP Browser 2.0 [29, 30]) compared to other root cells, suggesting that they are active during root hair growth.

Figure 2

Phenotypes of Mutants with Defects in Cytoskeleton Function and Membrane Trafficking

Rhizoid elongation is defective in mutants with defects in cytoskeleton organization and function (Mpren, Mpscd, Mpxi, Mpzwi, Mpwdl) and membrane trafficking (Mpap5m, Mppi4kα1, Mpspi, Mpsri1, Mpsri2); 21-day-old gemmalings. Scale bar, 5 mm. See also Figure S4.

Three genes were identified that encode proteins predicted to be involved in endocytosis but that have not been functionally characterized in green plants to date. Their predicted function is based on the roles of similar proteins in yeast and mammals (Table 1; Figure 2). MpSRI2 (Mp SHORT RHIZOIDS2) encodes an EF-hand-containing protein that is similar to S. cerevisiae PAN1. PAN1 is required for association of the ARP-actin polymerization complex with clathrin-coated vesicles during endocytosis in yeast [19]. Mp SHORT RHIZOIDS1 (MpSRI1) encodes a protein similar to S. cerevisiae RIC1, which is a guanine exchange factor involved in activating Ypt6p GTPase and required for trafficking from early endosomes to the Golgi late in the endocytosis pathway [23]. MpAP5M is predicted to encode the subunit mu of the adaptor protein 5 (AP5) complex. AP5 is a tetrameric protein complex that coats vesicles acting as a cargo adaptor complex and is likely to be involved in endocytosis, but its precise function remains to be defined in any organism [20].

Three further genes were identified with no demonstrated function in green plants. MpSRI3 encodes a protein found in green plants but whose function has not been defined; it is similar to AtEMB2756, which mutates to an embryo lethal phenotype (Table 1; Figure S4) [31, 32]. MpSRI4 is similar to the yeast EFR3 gene that codes for a protein required for the localization of phosphatidylinositol-4-phosphate (PI4) kinase alpha at the plasma membrane, but no similar genes have been functionally characterized in green plants (Table 1; Figure S4) [33]. Consistent with this hypothetical role is the observation that rhizoid development is also defective in mutants in which the PI4 kinase alpha is defective (Table 1). ALBA proteins are nucleic-acid-binding proteins that form chromatin in archaea and bind RNA in a number of animal parasites [34, 35, 36]. Not only is MpALBA required for rhizoid development because Mpalba mutants develop short rhizoids but we discovered that loss-of-function alba mutants in A. thaliana develop shorter root hairs than wild-type (Table 1; Figure 3). This indicates that ALBA proteins are required for tip growth in both M. polymorpha and A. thaliana, and therefore are likely to be required for tip growth in rhizoids or root hairs throughout the land plants.

Figure 3

ALBA Proteins Are Required for Rhizoid and Root Hair Elongation in M. polymorpha and A. thaliana, Respectively

(A) Gene structure of MpALBA, with the T-DNA insertion sites for each allele indicated with a triangle. L and R indicate the location of the left and right borders of the T-DNA, respectively. Boxes represent exons; gray shows UTRs, and black shows coding sequences.

(B) Mpalba mutant rhizoids are shorter than WT (Tak-1); 21-day-old gemmalings. Scale bar, 5 mm.

(C) Maximum-likelihood tree of ALBA proteins from M. polymorpha and A. thaliana. Nodes are marked with approximate likelihood ratio test (aLRT) values. The scale bar represents the average number of substitutions per site.

(D) A. thaliana T-DNA insertion mutants of AT1G76010 (Atalba1) or AT1G20220 (Atalba2) develop shorter root hairs than wild-type (Col-0); 5-day-old seedlings. Scale bar, 1 mm.

(E) Root hairs of plants homozygous for Atalba1 (AT1G76010) and Atalba2 (AT1G20220) are significantly shorter than those of WT (Col-0). Asterisks indicate significant difference from WT (t test, n > 40, p ≤ 0.001). Error bars indicate SD.

These data demonstrate that genes in the same orthogroups control the synthesis of new cell surface in liverwort rhizoids and angiosperm root hairs. This conservation suggests that this mechanism acted during the growth of the first land plant rooting structures at or soon after the colonization of the land by streptophytes. These data also indicate that some of these genes—such as THE and PTI—were co-opted during the evolution of pollen tubes, one of a suite of traits that evolved during the evolution of the seed plant life cycle. Some genes previously shown to be involved in root hair growth have not been identified in this screen. This may be because the screen was not carried out to saturation and other rhizoid development genes remain to be discovered. Furthermore, many Arabidopsis gene families or subfamilies that contain genes implicated in root hair growth were present as a single-copy gene in M. polymorpha. Therefore, the M. polymorpha homologs of some genes involved in root hair growth are likely to have more general developmental roles than their Arabidopsis counterparts, and consequently result in severe growth defects or lethality when mutated. Moreover, it is likely that the function of some genes involved in root hair growth diversified in the lineage leading to the tracheophytes after the divergence of the last common ancestor of liverworts and angiosperms. Such divergence of function is supported by the observation that the phenotypes of some loss-of-function mutants in genes from the same orthogroup are different in M. polymorpha and A. thaliana. These data are consistent with the hypothesis that the evolution of the land plant body and life cycle involved a core set of genes with conserved functions that were active in the earliest land plants and underwent duplication followed by neofunctionalization. These novel functions programmed the development of novel structures and contributed to increased life cycle diversity during the subsequent radiation of land plants.

Author Contributions

S.H. screened 150,000 T-DNA lines and identified 38 tagged mutants; V.A.S.J. screened 105,000 lines and identified 8 tagged mutants; G.M. screened 81,000 lines and identified 15 tagged mutants; G.M. and H. Proust isolated DNA for genome sequencing and RNA for RNA sequencing; C.C. and A.J.H. assembled the genome under the guidance of S.K.; D.S.-M. helped with the coding; A.J.H. constructed the transcriptome under the guidance of S.K.; C.C. worked with G.M. to determine co-segregation of 4 mutants and carried out TAIL PCR; H. Prescott established all M. polymorpha growth and transformation protocols; S.H., V.A.S.J., and L.D. wrote the paper with much input from G.M. and comments from other authors; genes were grouped according to the classification established by S.H.; and L.D. conceived and designed the project.