Conserved regulatory mechanism controls thedevelopment of cells with rooting functionsin land plantsThomas Ho Yuen Tam, Bruno Catarino, and Liam Dolan1

Department of Plant Sciences, University of Oxford, Oxford OX1 3RB, United Kingdom

Edited by Philip N. Benfey, Duke University, Durham, NC, and approved June 11, 2015 (received for review August 22, 2014)

Land plants develop filamentous cells—root hairs, rhizoids, andcaulonemata—at the interface with the soil. Members of thegroup XI basic helix–loop–helix (bHLH) transcription factors en-coded by LOTUS JAPONICUS ROOTHAIRLESS1-LIKE (LRL) genespositively regulate the development of root hairs in the angio-sperms Lotus japonicus, Arabidopsis thaliana, and rice (Oryza sat-iva). Here we show that auxin promotes rhizoid and caulonemadevelopment by positively regulating the expression of PpLRL1and PpLRL2, the two LRL genes in the Physcomitrella patens ge-nome. Although the group VIII bHLH proteins, AtROOT HAIRDEFECTIVE6 and AtROOT HAIR DEFECTIVE SIX-LIKE1, promote root-hair development by positively regulating the expression of AtLRL3in A. thaliana, LRL genes promote rhizoid development indepen-dently of PpROOT HAIR DEFECTIVE SIX-LIKE1 and PpROOT HAIRDEFECITVE SIX-LIKE2 (PpRSL1 and PpRSL2) gene function inP. patens. Together, these data demonstrate that both LRL andRSL genes are components of an ancient auxin-regulated genenetwork that controls the development of tip-growing cells withrooting functions among most extant land plants. Although this net-work has diverged in the moss and the angiosperm lineages, our datademonstrate that the core network acted in the last common ancestorof themosses and angiosperms that existed sometime before 420mil-lion years ago.

auxin | bHLH | evolution | rhizoids | root hairs

The evolution of rooting structures was a key morphologicalinnovation that occurred when plants colonized the relativelydry continental surfaces of the planet sometime before 470 millionyear ago. The rooting structures of the earliest diverging groups ofland plants comprised systems of rhizoids. Rhizoids are eitherunicellular filaments (in liverworts and hornworts) or multicel-lular (in mosses) and elongate into the growth substrate or airsurrounding the plant. Bryophyte (liverwort, moss, and hornwort)rhizoids develop on gametophytes, the multicellular haploid stagein the life cycle. The evolution of vascular plants was accompaniedby an increase in the morphological diversity of the sporophyte,the diploid multicellular stage of the plant life cycle (1). Roots,multicellular axes derived from meristems with protective caps,were a key innovation that first appeared in the fossil record∼380 million years ago and likely evolved at least twice—at leastonce among the Lycophytes and at least once in the Euphyllo-phyte clade (2). Roots generally grow into the soil and expandthe plant–soil interface into different soil horizons. For example,some root systems extend deep into the soil (taproots), whereasothers proliferate in the nutrient-rich horizons near the soil sur-face. However, unicellular, filamentous protuberances called roothairs, which are morphologically similar to rhizoids, develop onall roots with few exceptions (2–5). Despite the different contextsin which they develop, root hairs and rhizoids carry out similarrooting functions, including nutrient uptake and anchorage (6, 7).Group VIII basic helix–loop–helix (bHLH) transcription fac-

tors encoded by ROOT HAIR DEFECTIVE SIX-LIKE (RSL)genes positively regulate root-hair development in the angio-sperm Arabidopsis thaliana and both rhizoid and caulonema in

the moss Physcomitrella patens (8–12). To test whether the con-servation of RSL function is unique or indicative of a more gen-eral, conserved regulatory mechanism, we determined whetherother components of the gene-regulatory network controlling an-giosperm root-hair development are conserved among land plants.Group XI bHLH transcription factors encoded by LOTUSJAPONICUS ROOTHAIRLESS1-LIKE (LRL) genes are alsokey regulators of root-hair development in angiosperms. LRLgenes encode several proteins that regulate root-hair devel-opment in diverse angiosperms, including ROOTHAIRLESS1(LjRHL1) in Lotus japonicus, AtLRL1, AtLRL2, AtLRL3 inA. thaliana (13, 14), and ROOTHAIRLESS1 (OsRHL1) in Oryzasativa (15). Other LRL genes include OsPTF1, involved inphosphate starvation tolerance in rice (16), and AtUNE12, in-volved in female gametophyte development in Arabidopsis(17). Here we define the function of LRL genes in the mossP. patens and conclude that, together, LRL and RSL genes formthe core of an ancient network that operated in the commonancestor of the mosses and angiosperms that existed in theSilurian Period 415–435 million years ago.

ResultsTwo LRL Genes Are Present in the Genome of the Moss P. patens. TheLRL transcription factors belong to group XI of plant bHLHtranscription factors (18) and share a conserved LRL domain of36 amino acids between the bHLH domain and the C terminus(Fig. 1A). To identify LRL homologs in P. patens, we performedBLAST searches in selected land plant genomes using theAtLRL3 (At5g58010) sequence as a query. Phylogenetic anal-yses identified two LRL genes in P. patens, designated PpLRL1

Significance

This work describes the discovery of an ancient genetic mech-anism that was used to build rooting systems when plants col-onized the relatively dry continental surfaces >470 million yearsago. We demonstrate that a group of basic helix–loop–helixtranscription factors—the LOTUS JAPONICUS ROOTHAIRLESS1-LIKE proteins—is part of a conserved auxin-regulated genenetwork that controls the development of tip-growing cells withrooting functions among extant land plants. This result suggeststhat this mechanism was active in the common ancestor of mostland plants and facilitated the development of early land plantfilamentous rooting systems, crucial for the successful coloniza-tion of the land by plants.

Author contributions: T.H.Y.T. and L.D. designed research; T.H.Y.T. and B.C. performedresearch; T.H.Y.T., B.C., and L.D. analyzed data; and T.H.Y.T., B.C., and L.D. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1To whom correspondence should be addressed. Email: liam.dolan@plants.ox.ac.uk.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1416324112 PNAS | Published online July 6, 2015 | E3959–E3968

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

1, 2

021

http://crossmark.crossref.org/dialog/?doi=10.1073/pnas.1416324112&domain=pdfmailto:liam.dolan@plants.ox.ac.ukhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplementalhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplementalwww.pnas.org/cgi/doi/10.1073/pnas.1416324112

(Pp1s173_83V6.1) and PpLRL2 (Pp1s209_22V6.2), respectively(Fig. 1B). Maximum- likelihood (ML) analysis resolved PpLRL1,PpLRL2, AtLRL1 (At2g24260), AtLRL2 (At4g30980), AtLRL3,AtLRL4 (At1g03040), AtUNE12 (At4g02590), five rice proteinsincluding OsRHL1 (Os06g08500), and four Selaginella moellen-dorffii proteins into a monophyletic group (Fig. 1B, Fig. S1, andDataset S1). The LRL domains of PpLRL1 and PpLRL2 are97.2% identical to each other and are 77.8% and 80.6% identicalto the LRL domain of the AtLRL3 protein, respectively. The twoPpLRL proteins have conserved bHLH domains that are 100%identical to each other (Fig. 1A). The bHLH domains of thePpLRL proteins are 92.4% identical to the bHLH domain ofthe AtLRL3 protein. To infer the phylogenetic relationship of thedifferent proteins containing the LRL domain, we constructedphylogenetic trees using Bayesian and ML methods (Fig. 1C andFig. S2). These trees showed that LRL homologs are present in all32 land plant species sampled (Fig. S3). The LRL proteins com-prise three major groups, designated XIa, XIb, and XIc (Fig. 1Cand Fig. S2). Genes that control the development of root hairs arepresent in group XIa (Fig. 1C, Fig. S2, and Dataset S2). PpLRL1and PpLRL2 are not members of any of these three groups (Fig.1C and Figs. S2 and S3). Together, these results indicate that theLRL genes are a conserved group of transcription factors whoseexistence predates the divergence of mosses and flowering plants.

PpLRL Genes Have Different, but Overlapping, Expression Patterns.We hypothesized that LRL genes controlled the development ofrhizoids and caulonemata and that PpLRL1 and PpLRL2 wouldbe expressed in these cell types. To define where LRL genes areexpressed, we replaced the endogenous PpLRL1 and PpLRL2genes with the coding sequence (CDS) of the UidA glucuroni-dase-encoding gene (GUS) (Fig. S4). Consequently, the GUSexpression in these lines identified regions of the protonema inwhich the PpLRL1 and PpLRL2 promoters are active in re-spective loss-of-function Pplrl1 or Pplrl2 mutant background. InP. patens, the germination of spores is followed by the development

of a uniseriate filamentous network (protonema) comprisingcells with large chloroplasts and transverse cell walls (chlor-onema cells). Subsequently, increasingly longer cells with obliquecell walls (caulonema cells) develop. Caulonema cells are mor-phologically and genetically similar to rhizoids (19). BothPpLRL1 and PpLRL2 were expressed in the central region of theprotonema, which is rich in chloronema and from which caulo-nema develop (Fig. 2 A and B). No expression was detected inthe very edges of the 30-d-old protonema, where caulonema isthe dominant cell type. These observations suggest that PpLRL1and PpLRL2 are active in cells that develop caulonema.PpLRL1 promoter was also active in the rhizoid-forming re-

gion at the base of the gametophore in plants in which thePpLRL1 CDS was replaced by the GUS-reporter gene (Fig. 2 Cand D). However, no GUS expression was detected in rhizoidsthemselves. This expression pattern is consistent with a role forPpLRL1 in the regulation of the earliest stages of rhizoid de-velopment. PpLRL1 promoter was also active in the gameto-phore apex and leaves, suggesting that this gene likely controlsother aspects of gametophore development. PpLRL2 promoteractivity was detected in the epidermal cells of the gametophoreand leaves (Fig. 2 C and D). Unlike PpLRL1, no GUS activitywas detectable in the buds or at the base of the gametophore.Like PpLRL1, there was no detectable expression of PpLRL2promoter in rhizoids.

PpLRL1 and PpLRL2 Are Positive Regulators of Rhizoid Development.To determine whether PpLRL genes function in rhizoid de-velopment, we generated mutants that lack LRL gene functionby homologous recombination (Figs. S5 and S6). To confirm thathomologous recombination had taken place, we characterizedthe genomic organization of the transformants. PCR analyseswere performed to select transformants in which the endogenousPpLRL gene was replaced by the gene-targeting construct (Fig.S5) and thus represent complete loss-of-function alleles. Fromthese, Southern blot analyses were used to further select those

Fig. 1. (A) Schematic representation of gene structure and amino acid alignment of LRL transcription factors. Only the bHLH and LRL domains are shown.(B) Tree of LRL proteins and related groups of bHLH proteins based on ML statistics. This tree resolved PpLRL1, PpLRL2, OsRHL1, AtLRL1, AtLRL2, AtLRL3,AtLRL4, and AtUNE12 within a monophyletic group, which also contains four additional rice and four additional Selaginella moellendorffii genes (Fig. S1).aLRT values are indicated above the nodes. Two other groups of plant bHLHs involved in root-hair and rhizoid development, group VIIIc (1) encoded by theclass I RSL genes, and group VIIIc (2) encoded by the class II RSL genes, are labeled in red and green, respectively. The full ML and Bayesian trees are shown inFig. S1. (C) Unrooted tree showing the relationships of different LRL proteins in land plants using Bayesian statistics. The location of selected LRL proteinsfrom the moss P. patens (Pp), the monocot O. sativa (Os), and the eudicot A. thaliana (At) are indicated in the diagram. The full ML and Bayesian trees areshown in Fig. S2.

E3960 | www.pnas.org/cgi/doi/10.1073/pnas.1416324112 Tam et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

021

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.1416324112.sd01.txthttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF3http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.1416324112.sd02.txthttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF3http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF4http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF6http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF2www.pnas.org/cgi/doi/10.1073/pnas.1416324112

transformants in which there were no additional transgene insertionsin the genome (Fig. S5). Three independent lines for each genewere maintained for phenotypic analyses. With the exception ofPplrl2-1, all single-mutant lines and all three double-mutant linesshowed a single band in the Southern blot analysis, indicatingthat there was a single insertion into the genome (Fig. S5).Pplrl1 and Pplrl2 single-mutant plants developed ∼40% fewer

rhizoids than wild-type (WT) (Fig. 3 A and B), and these rhizoidswere ∼35% and 18% shorter than WT, respectively (Fig. 3C).Pplrl1 Pplrl2 double mutants were completely rhizoidless, and norhizoids developed on these plants (Fig. 3D). Together, thesedata show that the activity of PpLRL1 and PpLRL2 together arerequired for the initiation and development of rhizoids frombuds and gametophores.To determine whether the bHLH and LRL domains of the

LRL proteins—characteristic of group XI plant bHLHs—arerequired for the control of rhizoid development in P. patens, wegenerated additional mutants in which the region containing thebHLH–LRL domain of each protein was deleted (Fig. S6). Theorganization of the genomic DNA in the mutants confirmed thatthe bHLH–LRL domains were deleted (Fig. S6). These mutantswere designated Pplrl1ΔbL and Pplrl2ΔbL, respectively. Defectsin rhizoid and caulonema development in these partial-deletionmutants were similar to the defects in the complete gene-deletion mutants (Fig. 3E). To independently verify that the bHLHand LRL domains are positive regulators of rhizoid develop-ment, we overexpressed a fragment of the PpLRL1 and PpLRL2CDS containing the essential bHLH and LRL domains (Fig. S6).Plants were transformed with constructs in which the bHLH–LRL domain was placed under the control of the CaMV35Spromoter (Fig. S6). Plants transformed with 35S:PpLRL1bL-1,which overexpressed the bHLH–LRL fragment of PpLRL1, de-veloped longer and more intensively pigmented rhizoids thanWT (Fig. 3E). In contrast, rhizoid pigmentation in 35S:PpLRL2bL-1–transformed plants, which overexpressed the bHLH–LRL fragmentof PpLRL2, resembled WT (Fig. 3E and Fig. S6). This difference inrhizoid phenotype in these lines suggests that PpLRL1 plays amore important role than PpLRL2 in rhizoid development.Together, these data are consistent with the hypothesis that thebHLH and LRL domains are responsible for PpLRL genefunction in rhizoid development.

PpLRL Genes Positively Regulate Caulonema Development. Thetransition from chloronema to caulonema, characteristic of WTprotonema development, did not occur in Pplrl1 Pplrl2 double-

mutant plants (Fig. 4 A and B). Caulonemata did not develop,and, instead, a dense mat of chloronemata constituted the entireprotonema (Fig. 4A). The diameter of the Pplrl1 Pplrl2 double-mutant protonema was approximately half that of the WT, whichcan be explained by the absence of caulonemata, because thesecells have higher growth rates and are longer than chloronemata(Fig. 4 B and C). To verify that the smaller diameter of Pplrl1Pplrl2 mutants was due to the loss of caulonemata, we comparedthe growth rate of Pplrl1 Pplrl2 double mutants with WT (Fig. 4C).In WT plants, chloronema growth rates were 7.94 ± 0.45 μm·h−1,and caulonemal growth rates were 16.86 ± 0.23 μm·h−1. Thegrowth rates of protonema filaments in Pplrl1 Pplrl2 doublemutants were 8.48 ± 0.81 μm·h−1. That is, the growth rates ofprotonema filaments of the Pplrl1 Pplrl2 double mutant were notsignificantly different from those of WT chloronema. These dataindicate that chloronemata, but not caulonemata, develop inPplrl1 Pplrl2 double mutants. Pplrl1 and Pplrl2 single mutantsdeveloped less severe phenotypes than the Pplrl1 Pplrl2 doublemutants. Protonema diameters of Pplrl1 and Pplrl2 mutants were83% and 89% that of WT, respectively (Fig. 4 A and B). To-gether, these data demonstrate that PpLRL1 and PpLRL2 pos-itively regulate the developmental transition from chloronema tocaulonema during the development of moss protonemata.

Auxin Positively Regulates PpLRL Activity. Auxin positively regu-lates the development of rhizoids and caulonemata in P. patens(8, 9, 20–22). To test whether auxin promotes rhizoid and cau-lonema development by positively regulating the expression ofPpLRL genes, we determined the effect of auxin treatment onthe activities of the PpLRL1 and PpLRL2 promoters. P. patensplants in which the endogenous PpLRL CDS were replaced bythe GUS gene (Fig. S4) were grown on Knops-GT medium(defined in Plant Material, Growth Conditions, and PhenotypicAnalyses) for 30 d. No GUS expression was detectable in theprotonema or buds, even when incubated in glucuronidase re-action buffer for 24 h (Fig. 5A). Addition of 1-naphthaleneaceticacid (NAA) to the medium at a final concentration of 100 nMresulted in detectable GUS expression in both protonema andbuds of PpLRL1pro:GUS and PpLRL1pro:GUS plants (Fig. 5A);this indicates that exogenous treatment with auxin increases theactivity of the PpLRL1 and PpLRL2 promoters. Together,these data suggest that auxin positively regulates PpLRL1and PpLRL2 expression.If auxin positively regulates LRL expression, and LRL genes in

turn promote caulonema development, we hypothesized that

Fig. 2. GUS staining pattern in plants transformed with PpLRL1pro:GUS and PpLRL2pro:GUS and untransformed controls (WT). (A and B) PpLRL1 and PpLRL2were expressed in the center of the protonema, composed mainly of chloronema filaments. (C and D) Young buds and gametophores isolated from 4-wk-oldprotonemata grown on KNOPS-GT medium. PpLRL1, but not PpLRL2,was expressed in the early buds. In the gametophores, PpLRL1was expressed in the base,where basal rhizoids develop. It was also expressed in the stem and at the apex. PpLRL2, in contrast, was not expressed in the early buds or the stem of thegametophores, but was expressed in the leaves. [Scale bars: 2 mm (A), 1 mm (B and D), and 200 μm (C).]

Tam et al. PNAS | Published online July 6, 2015 | E3961

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

1, 2

021

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF6http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF6http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF6http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF6http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF6http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF4

Fig. 3. Rhizoid phenotypes of PpLRL loss- and gain-of-function mutants. (A) Rhizoid phenotypes of complete PpLRL loss-of-function mutants. (Scalebars: 2 mm.) (B) Quantification of rhizoid number. (C ) Quantification of rhizoid length. Rhizoid length was determined by measuring the length be-tween the base of the gametophore and the tip of the rhizoid cluster. Both rhizoid number and length were significantly reduced in the Pplrl1 andPplrl2 single mutants. Rhizoids are absent in the Pplrl1 Pplrl2 double mutant. Asterisks represent P values from two-tailed unequal variance (Welch’s) ttests comparing individual mutant lines with WT. **P < 0.01; ***P < 0.000005. Exact P values are as follows. For rhizoid number, Pplrl1 and WT: P = 2.6 ×10−6; Pplrl2 and WT: P = 3.1 × 10−6; and Pplrl1 Pplrl2 and WT: P = 1.5 × 10−8. For rhizoid length, Pplrl1 and WT: P = 1.4 × 10−6; and Pplrl2 and WT: P = 1.6 ×10−3. NA, not applicable (the Pplrl1 Pplrl2 mutant has no measurable rhizoids). Error bars represent SEM (n = 9 for WT, n = 15 for Pplrl1, n = 12 for Pplrl2,and n = 11 for Pplrl1 Pplrl2). (D) No rhizoids developed at the base of the Pplrl1 Pplrl2 double-mutant gametophore. (Scale bars: 500 μm.) (E ) Phenotypesof PpLRLΔbL partial-deletion mutants and 35S:PpLRLbL mutants. The 35S:PpLRL1bL-1 mutant showed an increased number of rhizoids and increasedpigmentation.

E3962 | www.pnas.org/cgi/doi/10.1073/pnas.1416324112 Tam et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

021

www.pnas.org/cgi/doi/10.1073/pnas.1416324112

auxin-induced morphological changes in caulonema would re-quire PpLRL1 and PpLRL2 activity. To test this hypothesis, wecompared the morphology of Pplrl1 Pplrl2 double mutants grownin medium containing 100 nM NAA with double mutants grownwithout added auxin. Pplrl1 Pplrl2 protonema that developedin the presence of 100 nM NAA were morphologically identicalto double mutants grown without auxin (Fig. 5B). That is, Pplrl1Pplrl2 mutants were resistant to auxin. Furthermore, auxin in-duced the development of rhizoids on buds of WT, but rhizoidinduction did not occur on the buds of Pplrl1 Pplrl2 doublemutants (Fig. 5C). Together, these data demonstrate that rhizoidand caulonema development in plants that lack both PpLRL1and PpLRL2 function are resistant to auxin. This result is con-sistent with the model in which auxin-regulated caulonema andrhizoid development require PpLRL1 and PpLRL2 activity.The STYLISH1/SHI transcription factors regulate auxin bio-

synthesis in both A. thaliana and P. patens (23, 24). To testwhether auxin promotes the expression of PpLRL1 and PpLRL2in a PpSHI-dependent fashion in P. patens, we quantified mRNAlevels in the auxin biosynthesis mutants Ppshi1, PpOXSHI1-5,and Ppshi2. We could not detect a difference in PpLRL1 orPpLRL2 mRNA levels between WT and these mutants (Fig. S7

A–C). This result suggests that auxin-regulated expression ofPpLRL1 and PpLRL2 is independent of the STYLISH/SHItranscription factors.

PpLRL1 and PpLRL2 Are Required for Low-Phosphate-InducedDevelopment of Caulonema. We hypothesized that PpLRL geneswould be required for the adaptive response of protonema tolow phosphate; growth of protonema in medium containinglow phosphate promotes the transition from chloronema tocaulonema (25). To test this hypothesis, we compared thephenotypes of Pplrl1 Pplrl2 double mutants with WT in mediawith different phosphate content. Low phosphate promotes thechloronema-to-caulonema transition in WT protonema, but theprotonema morphology of Pplrl1 Pplrl2 double mutants wasidentical in replete and no phosphate (Fig. 6 A and B). That is,the double mutant protonema was insensitive to changes inexternal phosphate concentration, and caulonema did not de-velop. These results suggest that the morphological changesthat take place as part of the adaptive response to low phos-phate require PpLRL1 and PpLRL2 activity. However, wecould not detect a difference in PpLRL1 and PpLRL2 mRNAlevels in different phosphate conditions (Fig. S7D), suggesting

Fig. 4. Protonema phenotypes of PpLRL loss-of-function mutants. (A) Caulonema phenotype of complete PpLRL loss-of-function mutants. Caulonema de-velopment was defective in the Pplrl1 and Pplrl2 single mutants, and no caulonemata developed in the Pplrl1 Pplrl2 double mutants. [Scale bars: 2 mm(column 1) and 500 μm (column 2).] (B) Quantification of protonema development in terms of green (chloronema-rich) and nongreen (caulonema-rich)contributions to diameter. Asterisks represent P values from two-tailed unequal variance (Welch’s) t tests comparing individual mutant lines with WT. *P <0.05; **P < 0.01. The exact P values are as follows. For chloronema, Pplrl1 and WT: P = 0.052; Pplrl2 and WT: P = 0.10; and Pplrl1 Pplrl2 and WT: P = 7.3 × 10−6.For caulonema, Pplrl1 and WT: P = 0.0011; and Pplrl2 and WT: P = 0.074. Error bars represent SEM (n = 8 for WT and Pplrl1; n = 7 for Pplrl2; and n = 6 for Pplrl1Pplrl2). (C) Growth rate of caulonema and chloronema filaments in the loss-of-function mutants. In all of the mutant lines, the growth rate of chloronemafilaments was not significantly different from WT, but the growth rate of caulonema was significantly reduced. Asterisks represent the P values from two-tailed unequal variance (Welch’s) t tests comparing individual mutant lines with WT. *P < 0.05; **P < 0.01. The exact P values are as follows: for chloronema:P > 0.25 for all mutant lines compared with WT; for caulonema, Pplrl1 and WT: P = 0.017; and Pplrl2 and WT: P = 0.0024. Error bars represent SEM (n = 3 forchloronema; n = 5 for Pplrl1 and Pplrl2 caulonema; and n = 3 for WT caulonema).

Tam et al. PNAS | Published online July 6, 2015 | E3963

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

1, 2

021

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF7

that the induction of caulonemal development in low-phos-phate media does not increase the expression of LRL genes inP. patens.

PpLRL and PpRSL Genes Act Independently in P. patens. The groupVIII basic helix–loop–helix transcription factors, AtROOT HAIRDEFECTIVE6 (AtRHD6) and AtROOT HAIR DEFECTIVE6-LIKE1 (AtRSL1), promote the development of root hairsby positively regulating AtLRL3 expression (13, 14). To de-termine whether there are regulatory interactions betweenPpRSL and PpLRL genes, we compared the expression of PpLRL1and PpLRL2 in Pprsl1 Pprsl2 double mutants and WT, and theexpression of PpRSL1 and PpRSL2 in Pplrl1 Pplrl2 double mu-tants with WT. We found no evidence of transcriptional regu-latory interactions between PpLRL and PpRSL genes in P. patens(Fig. S7 E and F). Together, these data suggest that the LRL andRSL genes are transcriptionally independent of each other inP. patens.

DiscussionThe colonization of the land by streptophyte plants sometimebefore 470 million years ago was a pivotal event in the history ofthe Earth. Structures that anchor plants to their growth substrateand provide access to mineral nutrients, water, and the soil mi-croflora were key adaptations to life on the relatively dry con-tinental surfaces. Recent phylogenetic analyses suggest that therhizoidless algae of the Zygnematales or Colecochaetales (or aclade consisting of Zygnematales and Coleochaetales) are sisterto the land plants (26–29). If a clade consisting of Zygnematalesand Coleochaetales is sister to land plants, then there are twoequally parsimonious models for the evolution of rhizoids, eachinvolving two changes: (i) rhizoids evolved independently in thecommon ancestor of land plants and in the more distantly relatedalgal lineages such as the Charales or (ii) rhizoids evolved in thecommon ancestor of Charales, Zygnematales, Coleochaetales,and land plants, but were lost in the common ancestor of theZygnematales and Coleochaetales.

Fig. 5. Auxin-regulated caulonema and rhizoid development require PpLRL1 and PpLRL2 function. (A) Four-week-old protonema grown with 0 μM (−NAA)or 0.1 μM (+NAA) NAA. Auxin treatment induced the activity of the PpLRL1 and PpLRL2 promoters. [Scale bars: 5 mm (columns 1 and 4) or 500 μm (columns 2,3, 5, and 6).] (B) The Pplrl1 Pplrl2 mutants are partially insensitive to auxin. (Scale bars: 5 mm.) (C) Close-up of a gametophore from 2-wk-old protonemagrown with 0.1 μM NAA. Auxin failed to induce rhizoid development in the Pplrl1 Pplrl2 double mutants. (Scale bars: 500 μm.)

E3964 | www.pnas.org/cgi/doi/10.1073/pnas.1416324112 Tam et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

021

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF7www.pnas.org/cgi/doi/10.1073/pnas.1416324112

We show here that LRL genes positively regulate the devel-opment of the tip-growing rhizoids and caulonema in the mossP. patens. LRL genes also positively regulate the developmentof root hairs in angiosperms (13–15). Given that similar proteinscontrol the development of filamentous rooting cells at theplant–soil interface in mosses and angiosperms, we conclude thatLRL genes positively regulated the development of tip-growingrooting cells in the common ancestor of mosses and angiospermsthat existed sometime before 420 y ago (30).Basic helix–loop–helix proteins encoded by RSL genes also

form a network that controls filamentous rooting cell develop-ment. The class I RSL genes, AtRHD6 and AtRSL1, promoteroot-hair development in A. thaliana; root hairs do not developon Atrhd6 Atrsl1 double loss-of-function mutants (10). AtRHD6and AtRSL1 positively regulate the expression of the class II RSLgenes AtRSL2, AtRSL3, and AtRSL4 and the LRL gene AtLRL3;these in turn modulate late stages of root-hair differentiation inA. thaliana (12). AtRSL4 is a class II RSL gene that is necessaryand sufficient for growth. Modulation of AtRSL4 expression byclass I RSL genes, environment, and auxin determine root-haircell size. AtRSL4 controls cell size by controlling the expressionof genes that encode proteins involved in tip growth (12). Class IRSL genes in P. patens—PpRSL1 and PpRSL2—positively reg-ulate rhizoid and caulonema development (8–10). Class II RSLgenes—PpRSL3, PpRSL4, PpRSL5, and PpRSL6—regulatecaulonema development, but have much more subtle functionsthan class I RSL genes (11).

Our results demonstrate that the LRL and RSL genes representcore components of a conserved gene-regulatory network that waspresent in the common ancestor of mosses and angiosperms.However, although we showed that the genetic components ofthis gene-regulatory network have been conserved, the detailedpatterns of transcriptional regulation within this network remainto be resolved. For example, AtRHD6 and AtRSL1 (class I RSLgenes) regulate the expression of AtLRL3 in A. thaliana (13, 14),but we could not find evidence that PpRSL genes control PpLRLgene expression in P. patens. If there is a genuine absence oftranscriptional regulation between LRL and class I RSL genes inP. patens, then two alternative evolutionary scenarios can besuggested. Either (i) class I RSL genes acquired the ability toregulate the expression of LRL genes in the lineage that gave riseto angiosperm, or (ii) RSL-regulated LRL expression was anancestral trait that was lost in the moss lineage.The mechanism of auxin-regulated caulonema and rhizoid

development in P. patens involves key components of the auxin-signaling pathway that is conserved among land plants (8, 9,11, 20–23, 31). We showed here that auxin positively regulatesthe expression of PpLRL1 and PpLRL2 and that the PpLRLgenes are necessary for auxin-induced caulonema and rhizoiddevelopment. Therefore, the observed GUS activity in thePpLRL1pro:GUS and PpLRL2pro:GUS protonema may be dueto higher auxin concentration in the center of the protonema.Previously, it was shown that auxin promotes caulonema andrhizoid development by positively regulating the expression ofthe class I PpRSL genes, PpRSL1 and PpRSL2 (8, 9). We showedhere that auxin positively regulates PpLRL expression. However,because we could not find evidence that class I PpRSL genesregulate the expression of PpLRL genes, or that PpLRL genesregulate the expression of PpRSL genes in P. patens, these datasuggest that auxin positively regulates class I PpRSL and PpLRLgenes independently.According to the classical P. patens literature (22, 32), game-

tophores form only on caulonema. Our results suggest that budformation can occur in chloronema. The morphology, chloro-plast density, and growth rate of the Pplrl1 Pplrl2 protonemaresemble WT chloronema, but buds develop from this mutant.Similarly, the caulonema-less Pprsl1 Pprsl2 loss-of-function mu-tants (8, 33) also develop buds from chloronema. Together, thesedata suggest that the bud-formation process in P. patens is notlimited to caulonema, but can develop from chloronema in certaingenetic backgrounds.LRL genes are not general regulators of tip growth. There are

three types of tip-growing cells in P. patens: chloronema, caulo-nema, and rhizoid. The caulonemata are cytologically similar torhizoids (33–36). PpLRL genes positively control both rhizoidand caulonema development, but not chloronema development;chloronema development is indistinguishable from WT in Pplrl1Pplrl2 double mutants. These results are consistent with theobservation that caulonemata and rhizoids also have similar re-sponses to auxin and low phosphate, and these responses aredifferent from the response of chloronema to auxin (8, 9, 25).Together, these data suggest that the PpLRL genes are specificregulators of the development of tip-growing cells with a rootingfunction. Similarly, RSL genes are specific regulators of caulo-nema and rhizoid development and do not regulate chloronemadevelopment (8–10). Therefore, we concluded that LRL andRSL genes specifically regulate rhizoid and caulonema devel-opment and are not simply general regulators of tip growth. Thisconclusion is supported by the finding that the tip-growth de-fect in Atrsl1 Atrsl2 double mutants is restricted to root hairs inA. thaliana; pollen tube development is identical to WT in thesedouble mutants (10).LRL transcription factors are ancient and were present in the

last common ancestor of mosses and angiosperms that was extantat least 420 million years ago. Bayesian analysis indicates that

Fig. 6. PpLRL1 and PpLRL2 function are required for low phosphateadaptive response. (A) Pplrl1 Pplrl2 double mutants are insensitive to lowphosphate. Protonemata were grown in high (+P) and low (−P) phosphatemedium for 2.5 wk. (Scale bars: 5 mm.) (B) Low phosphate increased cau-lonema development significantly in WT protonemata. *P = 0.034 [two-tailed unequal variance (Welch’s) t tests)] but had no discernable effect onPplrl1 Pplrl2 mutants. Error bars represent SEM (n = 20 for WT+P and Pplrl1Pplrl2 +P; n = 16 for WT-P; and n = 12 for Pplrl1 Pplrl2 −P).

Tam et al. PNAS | Published online July 6, 2015 | E3965

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

1, 2

021

three monophyletic clades (XIa, XIb, and XIc) constitute themajority of land plant LRL proteins. However, PpLRL proteinsare not members of groups XIa, XIb, or XIc. Groups XIb andXIc each contain a single gymnosperm LRL protein that is sisterto the other angiosperm LRL proteins in each of these mono-phyletic groups, suggesting that groups XIb and XIc originatedbefore the divergence of gymnosperms and angiosperms (Figs.S2 and S3). The evolution of group XIa LRL protein is morecomplex (Figs. S2 and S3), but the presence of a group XIa LRLproteins in Amborella and other angiosperm taxa suggests thatthis group was present in the common ancestor of extant an-giosperms. Only group XIa LRL proteins are known to regulateroot-hair development in angiosperms. These results support ourconclusion that the role of LRL proteins in regulating filamen-tous rooting cell development is ancient. Further sampling ofLRL genes in well-annotated, high-quality assemblies of mon-ilophyte (ferns and horsetails) and gymnosperm genomes willallow the elucidation of the evolutionary history of LRL genesduring land plant evolution.There is evidence that PpLRL1 plays a greater role than

PpLRL2 in rhizoid development. First, the PpLRL1 and PpLRL2expression patterns are different. The strong expression ofPpLRL1 in chloronema cells, young buds, and the base of thegametophore is consistent with its role in controlling early stagesof caulonema and rhizoid development. In contrast, PpLRL2promoter activity was not detected in either young buds or thebase of gametophores where rhizoids develop, suggesting thatPpLRL2 regulates rhizoid development via a nonautonomousmechanism. Second, Pplrl1 mutants have a stronger rhizoid de-fect than Pplrl2 single mutants. Third, overexpression of thePpLRL1bL partial transcript led to an increased number ofrhizoids, whereas overexpression of the PpLRL2bL partialtranscript had no discernable effect on rhizoid development.Nevertheless, the activity of both proteins is required for thedevelopment of rhizoids because only the double mutant doesnot form rhizoids. Together these results suggest that, althoughboth PpLRL1 and PpLRL2 function in rhizoid development(because the double mutant does not form rhizoids), PpLRL1plays a greater role than PpLRL2.

ConclusionLRL genes are part of an ancient gene-regulatory network thatcontrols the development of tip-growing filamentous cells withrooting functions—rhizoids, caulonema, and root hairs—at theinterface between land plants and the soil. This auxin-regulatednetwork was active early in land plant evolution and was presentin the last common ancestor of the angiosperms and mosses. Thisancient gene-regulatory network diverged since P. patens andA. thaliana last shared a common ancestor (Fig. 7).

Materials and MethodsSequence Retrieval and Alignment. To identify potential LRL proteins inP. patens, we performed BlastP searches using AtLRL3 (At5g58010) aminoacid sequence as a query on Phytozome (Version 9.1) (37) with a E-valuethreshold of 1e-6, and we retrieved all of the resulting sequences. We usedonly the primary transcripts where alternative transcripts exist. CDS retrievedwere translated to amino acids and then aligned by using Mafft with theL-INS-I option (38). The alignment was trimmed with TrimAl (Version 1.3) (39)as implemented in Phylemon (Version 2.0) with automated parameters (40).The resulting alignment consisted of 139 sequences and 54 columns fromfour species.

To extend our analysis and resolve the relationship among LRL proteins,sequences were retrieved from BLAST searches on genomes on Phytozome(Version 9.1) (37), the Pinus taeda v1 transcriptome on the Dendrome project(dendrome.ucdavis.edu/resources/blast/), the L. japonicus genome assemblybuild 2.5 (www.kazusa.or.jp/lotus/index.html), and the Amborella tricho-carpa genome (41). Only sequences containing unambiguous LRL domainswere included. The sequences were aligned with Mafft. The alignment wastrimmed with TrimAl (Version 1.3) (39) as implemented in Phylemon (Version

2.0) with automated parameters (40).The final amino acid alignment, con-taining 164 sequences from 32 species and 121 columns, was used in sub-sequent analyses. This alignment was also used to determine the similarity ofthe bHLH and LRL domains (as shown in Fig. 1) of selected LRL proteins, byusing the Ident and Sim program in the Sequence Manipulation Suite (42).

Phylogenetic Analyses. The best-fit model of amino acid evolution for thetrimmed alignments were determined by Prottest (Version 3) (43), andthe JTT+Γ (for Dataset S1) or the JTT+Γ+I (for Dataset S2) model was chosen.The ML trees were calculated with PhyML (Version 3.0) (44), with the SH-Likeapproximate likelihood-ratio test (aLRT) for branch support (45) and the bestof nearest neighbor interchange and subtree pruning and regrafting fortree improvement. Bayesian analyses were carried out with MrBayes (Ver-sion 3.2) (46) for >15 million generations. One tree was saved every 100generations, and the first 25% of trees were discarded. A 50-majority-ruletree was generated from the remaining trees. The default settings wereused for gap treatment—i.e., as unknown character in PhyML and as missingdata in MrBayes.

Plant Material, Growth Conditions, and Phenotypic Analyses. The GransdenWT strain (47) of P. patens (Hedw.) Bruch & Schimp was used in this study.Unless otherwise stated, all plants were grown in a SANYO MLR-351growth chamber at 25 °C and a 16-h light/8-h dark regime with a pho-tosynthesis photon flux density of ∼40 μmol m−2·s−1. Plants were grownon a modified KNOPS medium (referred to as “minimal medium” or“KNOPS-GT”) (47): 0.8 g/L CaN2O6·4H2O, 0.25 g/L MgSO4·7H2O, 0.0125 g/LFeSO4·7H2O, 0.055 mg/L CuSO4·5H2O, 0.055 mg/L ZnSO4·7H2O, 0.614 mg/LH3BO3, 0.389 mg/L MnCl2·4H2O, 0.055 mg/L CoCl2·6H2O, 0.028 mg/L KI,0.025 mg/L Na2MoO4·2H2O, 25 mg/L KH2PO4 buffer (i.e., 1.84 mM atpH 6.5 with KOH), and 7 g/L agar (Formedium AGA03). Then, 5 g/L glucoseand 0.5 g/L ammonium tartrate dibasic (C4H6O6·2H3N) were added as sup-plements for routine subculture of macerated protonema tissues (KNOPSmedium). Sterile cellophane discs (80-mm 325P Cellulose discs; A.A. Pack-aging) were put on top of the medium to facilitate phenotypic analysis andcollection of tissues for subculture.

PEG-Mediated Transformation of P. patens. PEG-mediated transformation wasperformed as described (10, 48). Antibiotic selection was performed byadding 50 μg/mL G418 disulfate or 25 μg/mL hygromycin B to thegrowth medium.

Southern Blot Analysis. Genomic DNA was extracted from protonema tissuesby using the cetyltrimethylammonium bromide (CTAB) method and thendigested overnight with the appropriate restriction endonuclease. A total of1 μg (for Pplrl2, Pplrl1 Pplrl2, and Pplrl2pro:GUS) or 2 μg (for Pplrl1 and

Fig. 7. Models indicating the transcriptional regulation between auxin, LRL,and class I RSL genes in rhizoid and caulonema development in P. patens (A)and root hair development in A. thaliana (B).

E3966 | www.pnas.org/cgi/doi/10.1073/pnas.1416324112 Tam et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

021

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF3http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF3http://dendrome.ucdavis.edu/resources/blast/http://www.kazusa.or.jp/lotus/index.htmlhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.1416324112.sd01.txthttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.1416324112.sd02.txtwww.pnas.org/cgi/doi/10.1073/pnas.1416324112

PpLRL1pro:GUS) of digested genomic DNA, as quantified with a NanoDropND1000 spectrometer (Thermo Scientific), was subjected to agarose gelelectrophoresis before being transferred to a positively charged nylonmembrane. Southern blots were carried out with the digoxigenin (DIG)system for filter hybridization with DIG-labeled probe generated by PCRaccording to the DIG Application Manual for Filter Hybridization (Roche).DIG-labeled DNAMolecular Weight Marker VII (Roche) was used throughoutthe analysis. Primers for probe synthesis are given in Table S1.

Generation of P. patens GUS-Reporter Lines. The GUS-reporter lines weregenerated by replacing the endogenous PpLRL1 or PpLRL2 gene with theuidA gene and a hygromycin-resistance cassette from the plasmid pBHrev-GUS (pBHSNR-GUS) (9); this strategy for constructing the reporter lines isillustrated in Fig. S4. To generate the GUS reporter lines for PpLRL1, a 871-bpfragment upstream and a 888-bp fragment downstream of the predictedPpLRL1 CDS (Pp1s173_83V6) were cloned into the AscI–NcoI and BamHI site,in the correct orientation, of the pBHrev-GUS plasmid, respectively. Similarly,for the GUS-reporter lines for PpLRL2, a 903-bp fragment upstream and a880-bp fragment downstream of the predicted CDS (Pp1s209_22V6) werecloned into the AscI–NcoI and BamHI–HindIII site of the pBHrev–GUS plas-mid, respectively. These generated the pBHrev-PpLRL1proGUS and pBHrev-PpLRL2proGUS constructs (Fig. S4). The plasmids were linearized with AscI(for pBHrev-PpLRL1proGUS) and AscI + HindIII (for pBHrev-PpLRL2proGUS)before PEG-mediated transformation. Stable transformants were selected byPCR analyses and then further selected with Southern blot analysis (Fig. S4).Three independent lines, which show the same GUS-staining pattern, wereused for each genotype.

Generation of P. patens Loss- and Gain-of-Function Mutants. The plasmidspBHrev (pBHSNR) and pBNRF (10) were used to generate the loss-of-functionconstructs in this study. To generate the Pplrl1 complete knockout mutants,an 812-bp DNA fragment upstream and a 819-bp fragment downstream ofthe PpLRL1 were cloned into the SalI–BamHI and MluI–NcoI sites of thepBHrev plasmid, respectively (Fig. S5). Similarly, to generate Pplrl2 knockoutmutants, a 697-bp fragment upstream and a 655-bp fragment downstreamof the PpLRL2 locus were cloned into the SalI–BamHI and SpeI–AscI site ofthe pBNRF plasmid, respectively (Fig. S5). These generated the pBHrev-PpLRL1KO and pBNRF-PpLRL2KO constructs. The plasmids were linearizedwith SalI + NcoI (for pBHreb-PpLRL1KO) and SalI + AscI (for pBNRF-PpLRL2KO)before PEG-mediated transformation. The Pplrl1 Pplrl2 double mutants weregenerated by retransforming Pplrl1-3 with the pBNRF-PpLRL2KO construct.Stable transformants were selected by PCR analyses and then further selectedwith Southern blot analysis (Fig. S5). Three independent lines, which showedthe same phenotype, were used for each genotype. Two independent lines ofpartial gene-deletion mutants (Pplrl1ΔbL and Pplrl2ΔbL), where only the bHLHand LRL domains are deleted, were generated similarly (Fig. S6) and werePCR-verified.

To generate the 35S:PpLRL1bL and 35S:PpLRL2bL mutants, partial CDS forthe bHLH–LRL region of the PpLRL genes (Fig. S6) were first amplified by RT-PCR and subcloned into the pCR8/GW/TOPO Vector (Invitrogen), thentransferred to the p108GW35S vector through the Gateway LR Clonase IIsystem (Invitrogen) for moss overexpression (11). This construct includes the108 locus to facilitate insertion into the neutral 108 locus in the P. patensgenome. The levels of the PpLRL1 and PpLRL2 bHLH–LRL partial transcriptin 35S:PpLRLbL lines were determined by quantitative reverse transcrip-tion-polymerase chain reaction (qRT-PCR) (Fig. S6).

GUS Staining. Unless otherwise stated, protonema inocula were grown onKNOPS medium with cellophane for 2.5 wk and then directly incubated inGUS staining solution: 100 mM NaH2PO4 (pH 7.0 with NaOH), 0.5 mM K3Fe(CN)6, 0.5 mM K4Fe(CN)6·3H2O, 0.05% Triton X-100, 1 mM 5-bromo-4-chloro-

3-indolyl-β-D-glucuronic acid, and cyclohexyl ammonium salt (X-GlcA; Mel-ford) dissolved in N,N-dimethylformamide at 37 °C for 24–48 h. Tissues werethen destained by incubation in 70–100% (vol/vol) ethanol for at least 24 h.No staining was observed in WT plants.

Time-Lapse Microscopy. Filaments at the edge of the protonema growing oncellophane disk onminimal mediumwere photographed every 6 min under aLeica DFC310 FX camera mounted on a Leica M165 FC stereomicroscope.Images were analyzed with ImageJ (49). The increase in filament lengthevery 30 min was used to calculate the growth rate.

qRT-PCR. Ten-day-old macerated protonema growing on minimal mediumwas used for all qRT-PCRs. Approximately 100 mg of fresh protonema thathad been gently squeezed to remove excess water was frozen in liquid ni-trogen before RNA extraction with the RNeasy Plant Mini Kit (Qiagen). Theamount of RNA was quantified with a NanoDrop ND1000 spectrometer(Thermo Scientific) or a Qubit 2.0 Fluorometer (Invitrogen) before treatmentwith Turbo DNase (Ambion, Life Technologies). First-strand cDNA synthesiswas performed with the SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen, Life Technologies) by using oligo(dT). The following amountof DNase-treated RNA was used in the cDNA synthesis: 3.5 μg for detectingPpLRL levels in Ppshi1, Ppshi2 and WT; 9.1 μg for detecting PpLRL levels inPpOXSHI1-5 and WT; 1.3 μg for detecting PpLRL levels in WT treated withhigh and low phosphate; 2.7 μg for detecting PpRSL levels in Pplrl1 Pplrl2double mutants and WT; and >1 μg for detecting PpLRL levels in Pprsl1Pprsl2 double mutants and WT. The resulting cDNA reaction mixture(∼20 μL) was diluted 6.5-fold, and 5 μL of the diluted cDNA mixture was useddirectly in a 20-μL qRT-PCR. qRT-PCR was performed with the SYBR GreenPCR Master Mix (Applied Biosystems) and the Applied Biosystems 7300 Real-Time PCR system. Three biological replicates, each with three technicalreplicates, were used for each plant sample. The cycling conditions were asfollows: 50 °C for 2 min; 95 °C for 10 min; then 40 cycles of 95 °C for 15 s and60 °C for 1 min. Dissociation curve analyses were performed at the end ofthe cycles to ensure the formation of only a single amplicon for each rep-licate. LinRegPCR (50) was used to estimate the baseline, threshold, andprimer efficiencies. The starting concentration of the target transcript in asample, expressed in arbitrary fluorescence units (N0), was computed fromthe following formula: N0 = threshold/[(mean efficiency of each primerpair)̂ Cq] (50). PpGAPDH (11) was used as the internal reference for all ex-periments, except for the determination of bHLH–LRL partial transcripts inthe 35S:PpLRL1bL and 35S:PpLRL2bL lines, where PpAdePRT was used as theinternal references (51). Primers used are given in Table S1.

Auxin and Low-Phosphate Treatment. For auxin treatments, plants weregrown on KNOPS-GT supplemented with 0, 0.1, or 1 μM NAA dissolved indimethyl sulfoxide. High-phosphate treatment was carried out by growingplants in KNOPS-GT medium supplemented with 1 mM Mes. The low-phos-phate medium (−P) was prepared by replacing the KH2PO4 buffer in theKNOPS-GT medium with 0.918 mM K2SO4.

ACKNOWLEDGMENTS. We thank Dr. Katarina Lundberg from Prof. EvaSundberg’s group (Swedish University of Agricultural Sciences) for providingthe Ppshi1, Ppshi2, and the PpOXSHI1 lines (23); Dr. Krzysztof Szczyglowski(Agriculture and Agri-Food Canada) for providing the Atlrl mutant lines forobservation; Nuno Pires for technical assistance and scientific input at thebeginning of the project; and Helen Prescott and Lida Chen for laboratorysupport. T.T.H.Y. was supported by a Clarendon Fund Scholarship andthe Overseas Research Students Awards Scheme; B.C. and L.D. were sup-ported by the PLANTORIGINS Marie Curie Network and the EVO500 ERC-Advanced Grant.

1. Niklas KJ, Kutschera U (2010) The evolution of the land plant life cycle. New Phytol

185(1):27–41.2. Raven JA, Edwards D (2001) Roots: Evolutionary origins and biogeochemical signifi-

cance. J Exp Bot 52(Spec Issue):381–401.3. Jones VAS, Dolan L (2012) The evolution of root hairs and rhizoids. Ann Bot (Lond)

110(2):205–212.4. Datta S, et al. (2011) Root hairs: Development, growth and evolution at the plant-soil

interface. Plant Soil 346:1–14.5. Kenrick P (2002) The origin of roots. Plant Roots: The Hidden Half, eds Waisel Y,

Eshel A, Beeckman T, Kafkafi U (Marcel Dekker, New York).6. Duckett JG, Ligrone R, Renzaglia KS, Pressel S (2014) Pegged and smooth rhizoids in

complex thalloid liverworts (Marchantiopside): Structure, function and evolution. Bot

J Linn Soc 174:68–92.

7. Gahoonia T, Nielsen NE (2003) Phosphorus (P) uptake and growth of a root hairless

barley mutant (bald root barley, brb) and wild type in low- and high-P soils. Plant Cell

Environ 26:1759–1766.8. Jang G, Dolan L (2011) Auxin promotes the transition from chloronema to caulonema

in moss protonema by positively regulating PpRSL1and PpRSL2 in Physcomitrella

patens. New Phytol 192(2):319–327.9. Jang G, Yi K, Pires ND, Menand B, Dolan L (2011) RSL genes are sufficient for rhizoid

system development in early diverging land plants. Development 138(11):

2273–2281.10. Menand B, et al. (2007) An ancient mechanism controls the development of cells with

a rooting function in land plants. Science 316(5830):1477–1480.11. Pires ND, et al. (2013) Recruitment and remodeling of an ancient gene regulatory

network during land plant evolution. Proc Natl Acad Sci USA 110(23):9571–9576.

Tam et al. PNAS | Published online July 6, 2015 | E3967

PLANTBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

1, 2

021

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF4http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF4http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF4http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF6http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF6http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=SF6http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1416324112/-/DCSupplemental/pnas.201416324SI.pdf?targetid=nameddest=ST1

12. Yi K, Menand B, Bell E, Dolan L (2010) A basic helix-loop-helix transcription factorcontrols cell growth and size in root hairs. Nat Genet 42(3):264–267.

13. Karas B, et al. (2009) Conservation of lotus and Arabidopsis basic helix-loop-helixproteins reveals new players in root hair development. Plant Physiol 151(3):1175–1185.

14. Bruex A, et al. (2012) A gene regulatory network for root epidermis cell differenti-ation in Arabidopsis. PLoS Genet 8(1):e1002446.

15. Ding W, et al. (2009) A transcription factor with a bHLH domain regulates root hairdevelopment in rice. Cell Res 19(11):1309–1311.

16. Yi K, et al. (2005) OsPTF1, a novel transcription factor involved in tolerance tophosphate starvation in rice. Plant Physiol 138(4):2087–2096.

17. Pagnussat GC, et al. (2005) Genetic and molecular identification of genes required forfemale gametophyte development and function in Arabidopsis. Development 132(3):603–614.

18. Pires N, Dolan L (2010) Origin and diversification of basic-helix-loop-helix proteins inplants. Mol Biol Evol 27(4):862–874.

19. Duckett JG, Schmid AM, Ligrone R (1998) Protonemamorphogenesis. Bryology for theTwenty-First Century, eds Bates JW, Ashton NW, Duckett JG (Maney, Leeds, U.K.).

20. Lavy M, Prigge MJ, Tigyi K, Estelle M (2012) The cyclophilin DIAGEOTROPICA has aconserved role in auxin signaling. Development 139(6):1115–1124.

21. Prigge MJ, Lavy M, Ashton NW, Estelle M (2010) Physcomitrella patens auxin-resistantmutants affect conserved elements of an auxin-signaling pathway. Curr Biol 20(21):1907–1912.

22. Ashton NW, Grimsley NH, Cove DJ (1979) Analysis of gametophytic development inthe moss, Physcomitrella patens, using auxin and cytokinin resistant mutants. Planta144(5):427–435.

23. Eklund DM, et al. (2010) Homologues of the Arabidopsis thaliana SHI/STY/LRP1 genescontrol auxin biosynthesis and affect growth and development in the moss Phys-comitrella patens. Development 137(8):1275–1284.

24. Eklund DM, et al. (2010) The Arabidopsis thaliana STYLISH1 protein acts as a tran-scriptional activator regulating auxin biosynthesis. Plant Cell 22(2):349–363.

25. Wang Y, Secco D, Poirier Y (2008) Characterization of the PHO1 gene family and theresponses to phosphate deficiency of Physcomitrella patens. Plant Physiol 146(2):646–656.

26. Wickett NJ, et al. (2014) Phylotranscriptomic analysis of the origin and early di-versification of land plants. Proc Natl Acad Sci USA 111(45):E4859–E4868.

27. Finet C, Timme RE, Delwiche CF, Marlétaz F (2010) Multigene phylogeny of the greenlineage reveals the origin and diversification of land plants. Curr Biol 20(24):2217–2222, and erratum (2012) 22(15):1456–1457.

28. Timme RE, Bachvaroff TR, Delwiche CF (2012) Broad phylogenomic sampling and thesister lineage of land plants. PLoS ONE 7(1):e29696.

29. Wodniok S, et al. (2011) Origin of land plants: Do conjugating green algae hold thekey? BMC Evol Biol 11:104.

30. Clarke JT, Warnock RCM, Donoghue PCJ (2011) Establishing a time-scale for plantevolution. New Phytol 192(1):266–301.

31. Sakakibara K, et al. (2003) Involvement of auxin and a homeodomain-leucine zipper Igene in rhizoid development of the moss Physcomitrella patens. Development130(20):4835–4846.

32. Cove D, Bezanilla M, Harries P, Quatrano R (2006) Mosses as model systems for the

study of metabolism and development. Annu Rev Plant Biol 57:497–520.33. Menand B, Calder G, Dolan L (2007) Both chloronemal and caulonemal cells expand

by tip growth in the moss Physcomitrella patens. J Exp Bot 58(7):1843–1849.34. Vidali L, Bezanilla M (2012) Physcomitrella patens: A model for tip cell growth and

differentiation. Curr Opin Plant Biol 15(6):625–631.35. Rounds CM, Bezanilla M (2013) Growth mechanisms in tip-growing plant cells. Annu

Rev Plant Biol 64(1):243–265.36. Pressel S, Ligrone R, Duckett JG (2008) Cellular differentiation in moss protonemata: A

morphological and experimental study. Ann Bot (Lond) 102(2):227–245.37. Goodstein DM, et al. (2012) Phytozome: A comparative platform for green plant

genomics. Nucleic Acids Res 40(Database issue, D1):D1178–D1186.38. Katoh K, Standley DM (2013) MAFFT multiple sequence alignment software version 7:

Improvements in performance and usability. Mol Biol Evol 30(4):772–780.39. Capella-Gutiérrez S, Silla-Martínez JM, Gabaldón T (2009) trimAl: A tool for auto-

mated alignment trimming in large-scale phylogenetic analyses. Bioinformatics

25(15):1972–1973.40. Sánchez R, et al. (2011) Phylemon 2.0: A suite of web-tools for molecular evolution,

phylogenetics, phylogenomics and hypotheses testing. Nucleic Acids Res 39(Web

Server issue, suppl 2):W470–W474.41. Amborella Genome Project (2013) The Amborella genome and the evolution of

flowering plants. Science 342(6165):1241089.42. Stothard P (2000) The sequence manipulation suite: JavaScript programs for analyz-

ing and formatting protein and DNA sequences. Biotechniques 28(6):1102–1104,

1104.43. Darriba D, Taboada GL, Doallo R, Posada D (2011) ProtTest 3: Fast selection of best-fit

models of protein evolution. Bioinformatics 27(8):1164–1165.44. Guindon S, et al. (2010) New algorithms and methods to estimate maximum-likeli-

hood phylogenies: assessing the performance of PhyML 3.0. Syst Biol 59(3):307–321.45. Anisimova M, Gascuel O (2006) Approximate likelihood-ratio test for branches: A fast,

accurate, and powerful alternative. Syst Biol 55(4):539–552.46. Ronquist F, et al. (2012) MrBayes 3.2: Efficient Bayesian phylogenetic inference and

model choice across a large model space. Syst Biol 61(3):539–542.47. Ashton NW, Cove DJ (1977) The isolation and preliminary characterisation of auxo-

trophic and analogue resistant mutants of the moss, Physcomitrella patens. Mol Gen

Genet 154(1):87–95.48. Schaefer DG, Zrÿd JP (1997) Efficient gene targeting in the moss Physcomitrella pat-

ens. Plant J 11(6):1195–1206.49. Schneider CA, Rasband WS, Eliceiri KW (2012) NIH Image to ImageJ: 25 years of image

analysis. Nat Methods 9(7):671–675.50. Ruijter JM, et al. (2009) Amplification efficiency: Linking baseline and bias in the

analysis of quantitative PCR data. Nucleic Acids Res 37(6):e45.51. Le Bail A, Scholz S, Kost B (2013) Evaluation of reference genes for RT qPCR analyses

of structure-specific and hormone regulated gene expression in Physcomitrella patens

gametophytes. PLoS ONE 8(8):e70998.

E3968 | www.pnas.org/cgi/doi/10.1073/pnas.1416324112 Tam et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

021

www.pnas.org/cgi/doi/10.1073/pnas.1416324112