Abscisic acid controlled sex before transpirationin vascular plantsScott A. M. McAdama, Timothy J. Brodribba,1, Jo Ann Banksb,1, Rainer Hedrichc,1, Nadia M. Atallahb, Chao Caib,Michael A. Geringerc, Christof Lindc, David S. Nicholsd, Kye Stachowskib, Dietmar Geigerc, and Frances C. Sussmilcha

aSchool of Biological Sciences, University of Tasmania, Hobart, TAS 7001, Australia; bDepartment of Botany and Plant Pathology, Purdue University,West Lafayette, IN 47907; cInstitute for Molecular Plant Physiology and Biophysics, University of Würzburg, D-97082 Wuerzburg, Germany; and dCentralScience Laboratory, University of Tasmania, Hobart, TAS 7001, Australia

Edited by Mark Estelle, University of California at San Diego, La Jolla, CA, and approved September 29, 2016 (received for review May 1, 2016)

Sexual reproduction in animals and plants shares common elements,including sperm and egg production, but unlike animals, little is knownabout the regulatory pathways that determine the sex of plants. Herewe use mutants and gene silencing in a fern species to identify a coreregulatory mechanism in plant sexual differentiation. A key player infern sex differentiation is the phytohormone abscisic acid (ABA), whichregulates the sex ratio of male to hermaphrodite tissues during thereproductive cycle. Our analysis shows that in the fern Ceratopterisrichardii, a gene homologous to core ABA transduction genes inflowering plants [SNF1-related kinase2s (SnRK2s)] is primarily re-sponsible for the hormonal control of sex determination. Further-more, we provide evidence that this ABA–SnRK2 signaling pathwayhas transitioned from determining the sex of ferns to controllingseed dormancy in the earliest seed plants before being co-opted tocontrol transpiration and CO2 exchange in derived seed plants. Bytracing the evolutionary history of this ABA signaling pathway fromplant reproduction through to its role in the global regulation ofplant–atmosphere gas exchange during the last 450 million years,we highlight the extraordinary effect of the ABA–SnRK2 signalingpathway in plant evolution and vegetation function.

OST1 | fern | stomata | evolution | sex determination

The phytohormone abscisic acid (ABA) plays a critical role ineveryday plant function by translating the hydration status of

plant tissue into a chemical signal that can activate metabolic re-sponses (1). Two prominent processes regulated by ABA signalingin seed plants are highly distinct in terms of target tissue andphysiology: one involving seed dormancy (2) and the other the reg-ulation of leaf transpiration (3). In seeds, the ability to delay ger-mination until conditions are suitable for growth is controlled by anantagonism between ABA (which promotes dormancy) and gibber-ellin (GA; which breaks dormancy and promotes germination) (2).In leaves, ABA regulates transpiration by activating anion channelsin the guard cells on either side of tiny pores on the leaf surface(stomata), causing cell turgor loss and pore closure (4, 5). Despiteinvolving distinct target organs, both processes share componentsof a common ABA-signaling cascade, including the ABA receptorcomplex, comprised of PYRABACTIN RESISTANCE1 (PYR1),PYR1-like (PYL), and REGULATORY COMPONENTS OFABA RECEPTORS (RCAR) proteins, TYPE 2 PROTEINPHOSPHATASES (PP2Cs), andmembers of theOPEN STOMATA1(OST1) subclade of SNF1-related kinase2 (SnRK2) kinases (4,6–9). Characterizations of ABA-deficient and ABA-insensitivemutants across a diversity of angiosperm species have confirmedthe universal involvement of this ABA-signaling cascade forstomatal control and seed dormancy within the angiospermclade (7, 9, 10). However, an absence of available mutants in non-angiosperm vascular plants has led to uncertainty about theevolution of this critical signaling pathway and the genes involved(cf refs. 11 and 12).To provide insight into the question of how ABA signaling

evolved in seed plants, we used the model fern species, Ceratopterisrichardii, as a representative member of a plant lineage basal to the

seed plants (13). Our specific aim was to determine the role of theABA–SnRK2 signaling pathway in plants that evolved before seedsas a means of understanding the evolution of key SnRK2-relatedABA signaling processes in vascular plants. It is known that ABAand SnRK2s are both present in early land plants, yet the onlyfunction clearly associated with ABA action in ferns involves sexdetermination of the free-living, haploid gametophyte generationof C. richardii (14). In many fern species, antheridiogen [a modifiedform of GA that is converted to bioactive GA4 once importedinto fern cells (15)] is secreted by hermaphroditic gametophytes,causing their immature neighbors to develop as males (14, 15).Experiments with C. richardii demonstrate that exogenous ABAcompletely blocks the sex-determining effect of antheridiogen, suchthat no male gametophytes develop when grown in the presence ofABA and the antheridiogen of C. richardii (ACE) (16). The strongsimilarity between the antagonism of ABA and ACE observed inC. richardii (16), and the ABA-GA antagonism known to regulateseed dormancy in angiosperms (2), raises an interesting possibil-ity that the ABA-GA signaling system in ferns may have beenco-opted to control seed dormancy in the earliest seed plants.We sought to investigate the genetic ancestry of ABA signaling

in the vascular plant lineage, using both forward and reversegenetic approaches to target core ABA-signaling genes in the fernC. richardii. We use ABA sensitivity mutants (17) to demonstratehow evolution has altered the downstream targets of ABA sig-naling while preserving the core components of the transductionpathway virtually unchanged.

Significance

Since the dawn of land plants, the phytohormone abscisic acid(ABA) has played a critical role in regulating plant responses towater availability. Here we seek to explain the origins of the coreABA signaling pathway found in modern seed plants. Using thecharacterization of mutants and gene silencing in a fern species,we find that the same hormone signaling components are used insex determination of ferns as are used for the control of seeddormancy and transpiration in seed plants. Ferns are shown to lackdownstream functionality of stomatal components, suggestingthat the origins of the core ABA signaling pathway in seed plantsmay lie in the sexual differentiation of ferns.

Author contributions: S.A.M.M., T.J.B., J.A.B., R.H., N.M.A., D.G., and F.C.S. designed research;S.A.M.M., J.A.B., N.M.A., C.C., M.A.G., C.L., K.S., and F.C.S. performed research; S.A.M.M., J.A.B.,N.M.A., D.S.N., and F.C.S. contributed new reagents/analytic tools; S.A.M.M., J.A.B., D.G., andF.C.S. analyzed data; and S.A.M.M., T.J.B., J.A.B., R.H., D.G., and F.C.S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The sequence reported in this paper has been deposited in the GenBankdatabase (accession no. KU556808–KU556812, KT238910–KT238912, KT285524,KT238907).1To whom correspondence may be addressed. Email: timothyb@utas.edu.au, banksj@purdue.edu, or hedrich@botanik.uni-wuerzburg.de.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1606614113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1606614113 PNAS Early Edition | 1 of 6

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

July

23,

202

0

Results and DiscussionWe identified five ABA-insensitive mutant alleles in a mutagen-esis screen of C. richardii (SI Appendix, Table S1). Unlike wild-typeplants, gametophytes of these mutants are unable to perceiveABA, and thus do not develop as hermaphrodites when grownon media containing both ACE and ABA (SI Appendix, Fig. S1).Further analysis revealed that these mutant alleles represent fourindependent loci, which we named GAMETOPHYTES ABAINSENSITIVE ON ACE1 (GAIA1; three alleles), GAIA2 (twoalleles), and GAIA3 and GAIA4 (one allele each), accordingly(SI Appendix, Table S1). The identity of these loci was furtherinvestigated, using our Trinity RNA-Seq de novo transcriptomeassembly for C. richardii to screen homologs of ABA-signalinggenes characterized in angiosperms as possible candidates.

SnRK2s and Sex Determination. Through our analysis, we identifieda C. richardii homolog within the OST1 subclade of SnRK2s (SIAppendix, Fig. S2) that was significantly altered in each of the threegaia1 mutant alleles relative to wild-type (Fig. 1A and SI Appendix,Fig. S3). In the gaia1-1 mutant, an A217G transition affects aninvariant residue within a region of the kinase domain that is

important for protein conformation and kinase activity (17). In thegaia1-2 mutant, a duplication of 134 bp affects the terminal portionof the ABA box, a domain that is conserved among members of thestrongly ABA-responsive OST1 subclade and is necessary and suf-ficient for target protein binding (18). No portion of this gene couldbe isolated from gaia1-3, using PCR-based techniques (SI Appendix,Fig. S3), suggesting a significant deletion or rearrangement. On thebasis of these distinct, functionally significant mutations, we con-cluded that this gene corresponds to the GAIA1 locus and namedthe geneGAIA1, accordingly. To confirm that these mutations werethe cause of the ABA-insensitive phenotype, we silenced the ex-pression of GAIA1 in wild-type gametophytes by RNA interference(RNAi) (19). Unlike wild-type gametophytes expressing GAIA1,gametophytes in which GAIA1 expression was silenced were in-sensitive to ABA, not switching to hermaphrodites (small meristics)when grown on media containing both ABA and ACE, but de-veloping only as males (Fig. 1C and SI Appendix, Fig. S4). ABAinsensitivity in the gaia1mutants or wild-type plants in whichGAIA1expression is silenced is not surprising, given the well-documentedABA-insensitive phenotypes described in all angiosperm mutants ofOST1 subclade genes characterized so far (6, 7). By characterizing

Fig. 1. GAIA1, a C. richardii homolog of OST1, regulates ABA signaling for gametophyte sex determination. (A) C. richardii sporophyte. (B) Gametophytephenotype for wild-type (Hn-n) and gaia1 mutant alleles grown on media containing ACE and 10 μM ABA, with diagrams showing GAIA1 gene and predictedprotein structure for each allele, indicating the nature of mutations. No portion ofGAIA1 could be isolated from gaia1-3 (SI Appendix, Fig. S3). (Scale bars, 200 μm,100 bp, or 10 aa; gene/protein diagrams share the same scale.) (C) Phenotypes of three representative gametophytes bombarded with only 35S::DsRed2 acting ascontrols and three GAIA1 RNAi gametophytes cobombarded with both 35S::DsRed2 and 35S::hpGAIA1. Knocking down the expression of GAIA1 prevents thedevelopment of a new hermaphroditic prothallus (indicated by arrows in the controls) when plants are transferred from media containing only ACE to mediacontaining both ACE and 5 μMABA. (Scale bars, 200 μm.) (D) Phylogram of C. richardii (Cr),Oryza sativa (Os), and A. thaliana (At) SnRK2-type protein kinases, withVicia faba (Vf) AAPK also included. The OST1-type subgroup is shaded in blue with the names of ABA-signaling proteins associated with stomatal control in bluetext, proteins acting in seed dormancy in green, and proteins linked to both processes in blue and underlined in green. CrGAIA1 is shown in red. Branches withbootstrap values <50% obtained from 1,000 trees have been collapsed. Representative Arabidopsis SnRK1 and SnRK3 proteins are included as outgroups.Phylogenetic analysis is based on the sequence alignment in SI Appendix, Dataset. For sequence details, see SI Appendix, Dataset. (E) Alignments showing residuesin the αC-helix of the kinase domain and ABA box conserved among OST1 homologs and between SnRK2 proteins. Shading indicates degree of conservation(black = 60%, dark gray = 50%, and light gray = 40%, yellow = residues affected in gaia1 mutants). Protein names are colored as in D.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1606614113 McAdam et al.

Dow

nloa

ded

by g

uest

on

July

23,

202

0

the GAIA1 gene, we provide very strong evidence that the ABA–SnRK2 signaling pathway in C. richardii is engaged in the regulationof sex determination of the fern gametophytes.

SnRK2s and Spore Germination. The phylogenetic position of GAIA1within the ABA-signaling SnRK2s and the strong ACE–ABA in-teraction associated with this C. richardii signaling pathway sug-gests a common ancestry with the ABA-signaling pathway that wasused, possibly as early as 300 million years ago, to regulate seeddormancy in early seed plants (8, 20). This hypothesis is supportedby the fact that gaia1 mutants showed reduced sensitivity to ABAin terms of spore germination relative to wild-type plants (Fig. 2A).The high levels of ABA that we found are required to inhibit sporegermination in C. richardii correspond to very high endogenouslevels of ABA present in fresh spores, levels that decline rapidlyupon germination (SI Appendix, Fig. S5). Interaction betweenABA and ACE to control spore germination was also observed,with the double-mutant of gaia1-3 and the ACE-insensitive her3mutant (21) found to restore a wild-type germination sensitivity to

ABA (Fig. 2A). The recent discovery of the modulatory effect ofABA and diterpenes (GA precursors) on spore germination inthe moss Physcomitrella patens (22) is intriguing, because if thisprocess in mosses is also regulated by an SnRK2, then propaguledormancy may be the ancestral function of this core ABA sig-naling pathway in land plants.

SnRK2s and Stomatal Control. To gain further insight into the evo-lution of SnRK2-related ABA signaling, we examined the possibilitythat stomatal closure, another major ABA-regulated process inseed plants that operates via SnRK2s, might also be dysfunctional inthe ABA-insensitive gaia1 mutant plants. Declining leaf watercontent in seed plants is known to result in rapid ABA synthesis,triggering stomatal closure via an SnRK2-mediated activation ofS-type anion channels (SLACs) (4). Examination of stomatal re-sponses to humidity and desiccation in C. richardii wild-type andgaia1 mutant plants revealed identical behavior in all genotypes,whereby stomata closed relatively quickly in response to either areduction in atmospheric water vapor content or leaf drying (Fig. 2B and C). Thus, we can conclude that the OST1 subclade geneGAIA1 is not involved in ABA signaling for stomatal closure inC. richardii.Angiosperm SnRK2s that function in ABA-dependent stomatal

control require guard cell-specific expression patterns in leaves(6, 8). We examined the localization of expression of GAIA1 andthe other C. richardii SnRK2 gene that fell within the OST1 sub-clade associated with ABA signaling in seed plants (SI Appendix,Fig. S2), PARALOG OF GAIA1 (PGAI). Unlike the expressionof stomatal-associated OST1 subclade genes in angiosperms,neither PGAI nor GAIA1 show guard cell specificity in expressionbut, rather, these genes are expressed in all tissues (SI Appendix,Fig. S6).A second, more specific test of the functionality of the SnRK2

pathway for stomatal ABA signaling in C. richardii was conductedto confirm an apparent lack of functionality suggested by non-specific gene expression. We tested the ability of the fern SnRK2sto activate the SLACs responsible for anion efflux during ABA-driven stomatal closure (4). We examined the ability of GAIA1and PGAI kinases to activate SLACs from both the fern andArabidopsis (SLAC1; Fig. 3 and SI Appendix, Fig. S9). Using aXenopus oocyte expression system, together with the two-elec-trode voltage-clamp technique and bimolecular fluorescencecomplementation (BiFC), we found that both of the nativeC. richardiiOST1 subclade SnRK2s were able to physically interact

Fig. 2. The germination of ABA-insensitive gaia1 mutant spores is lessinhibited by ABA compared with wild-type, whereas stomatal responses tochanges in leaf water status are normal. (A) The percentage of spores ger-minating when sown on media containing increasing concentrations of ABArelative to the percentage of spores germinating without ABA, including thedouble-mutant gaia1-3 and ACE-insensitive her3 mutant (n = 3). (B) Meanresponse of stomatal conductance (n = 3) to a reversible step change in VPD.Change in VPD is denoted by vertical lines; data during equilibration of thegas exchange chamber after the VPD change have been removed. (C) Meanresponse of stomatal conductance (n = 3, ±SE) to leaf excision. The vertical linedenotes leaf excision.

Fig. 3. Fern and lycophyte SnRK2s are unable to activate native S-type anion channels. Mean whole-oocyte current measurements at −100 mV in chloride-based standard medium of S. moellendorffii, C. richardii, and A. thaliana SLACs coexpressed with OST1 subclade SnRK2s in Xenopus oocytes (n ≥ 4, ±SE). Shownbelow are example whole-oocyte currents of representative SLACs either expressed with a native OST1 subclade SnRK2 or, in the case of S. moellendorffii,SLAC1b alone. For data in nitrate-based standard medium (SI Appendix, Fig. S9), all SnRK2 proteins physically interacted with SmSLACs, except SmSLAC1d(SI Appendix, Fig. S8). SmSLAC1a was represented by two allelic variants: SLAC1a.1 and SLAC1b.1.

McAdam et al. PNAS Early Edition | 3 of 6

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

July

23,

202

0

with and activate Arabidopsis SLAC1 (Fig. 3 and SI Appendix, Fig.S8). These findings are similar to SnRK2s from other basal non-vascular species, such as algae, liverworts, and mosses (23). A dif-ferent result, however, was observed when the action of C. richardiiOST1 subclade SnRK2s was tested using native C. richardii SLACs.Although both PGAI and GAIA1 were able to interact with nativeC. richardii SLACs (SI Appendix, Fig. S8), coexpression of thesenative SnRK2-SLAC pairs yielded weak or immeasurable macro-scopic anion channel currents (Fig. 3 and SI Appendix, Fig. S9). Ourdata from C. richardii suggest that unlike all angiosperm speciesexamined thus far, the ABA-signaling pathway for stomatal closurethrough SnRK2-mediated SLAC activation does not appear to beoperating in the fern C. richardii.A recent report of activation of a moss (P. patens) SLAC protein

by a native SnRK2, albeit weakly, could suggest that the evolutionof ABA-SnRK2-SLAC signaling for stomatal closure predates thedivergence of ferns (23), and that the absence of native anionchannel activation by SnRK2s in C. richardii represents a uniqueloss, similar to the loss of a stomatal response to blue light in thislineage (24). To address this possibility, we examined the func-tionality of SnRK2–SLAC combinations in a representative of thelycophyte clade, the earliest diverging clade of vascular plants.Thus, all native SnRK2-SLAC combinations in a sequenced rep-resentative of the lycophyte clade (Selaginella moellendorffii) wereidentified. The S. moellendorffii genome has three OST1 subcladeSnRK2s and four SLAC1 homologs. We found that the nativeS. moellendorffiiOST1 subclade SnRK2s were able to interact withall but one of the SLACs from S. moellendorffii (SI Appendix, Fig.S8), as well as Arabidopsis SLAC1. However, similar to the situa-tion in C. richardii, none of the S. moellendorffii OST1 subcladeSnRK2s were found to activate native S. moellendorffii SLACs(Fig. 3). Only one S. moellendorffii SLAC protein was capable ofconducting anions, but this activity was not further induced bycoexpression with SnRK2s (SI Appendix, Fig. S10). Thus, weconclude that the elements necessary for SnRK2 activation ofanion channels are not present in the most basal vascular plantlineage. That the two most basal vascular land plant clades lack anSnRK2 activation of endogenous SLACs provides strong evidencefor the conclusion that this ABA signaling pathway for stomatalclosure during water stress was absent in the common ancestor ofall vascular land plants (Fig. 3) and remains absent in extant fernsand lycophytes.It is possible that some form of ancestral ABA signaling via

SnRK2s may have evolved in mosses, as proposed by Lind et al.(23), before being subsequently lost in the earliest vascular landplants and reemerging in seed plants, but this seems an im-probable scenario. Several pieces of evidence suggest that ABA-mediated stomatal closure reported in the moss species P. patensis not analogous to that observed in seed plants. The first is thatP. patens does not exhibit the guard cell-specific expression of anSnRK2 (22, 25), which is required for a stomatal response toABA (SI Appendix, Fig. S6) (26). Questions about a lack offunctional homology between the stomata of bryophytes and vascularplants (27, 28) arise; given the recent revelation that mosses aresister to liverworts (29). The high number of losses, or possiblegains, of stomata across bryophyte lineages (27) make any in-terpretation about the function of the first land plant stomatafrom species in a single moss family difficult.

ConclusionThe linkage established here between ABA perception andSnRK2s in ferns provides direct evidence that this pathway, whichis critical for the regulation of seed dormancy, stomatal aperture,and other dehydration responsive processes (e.g., dehydrin in-duction) (30) in seed plants, extends back at least 360 millionyears to the emergence of ferns. We propose that the origins ofthe ABA signaling pathway critical for regulating seed dormancyare present in the fern lineage that evolved 60 million years before

the first fossil evidence of seed dormancy (20). This signaling systemin ferns primarily provides a means to regulate the degree of out-breeding in populations by modifying the sex ratio of the gameto-phyte reproductive stage (31). The connection between ABA andstomatal control via the specific activation of transmembrane anionschannels (4) appears to be a more recent innovation that did notevolve until after the divergence of ferns and seed plants (Fig. 4).This latest engagement of the ABA signaling pathway for stomatalclosure occurs in the earliest seed plants and, through their success,has become one of the most influential signaling pathways on Earth.

MethodsMutant Isolation. Hn-n, the wild-type strain of C. richardii used in this study, isdescribed by Hickok et al. (32). The antheridiogen (ACE)-insensitive hermaphro-ditic (her) mutants are described by Banks (21), and the two ABA-insensitivemutants HαA48 and HαA104 (referred to here as gaia1-3 and gaia4, respectively)are described by Hickok (33). Gametophyte and sporophyte growth conditions,ethyl methyl sulfate mutagenesis, and crosses were performed according to themethods of Banks (21). To identify new gaiamutants that are insensitive to ABA,106 her13 or wild-type spores were mutagenized with ethyl methyl sulfate andplated on medium containing ACE and 10 μM ABA. gaia mutants were selectedas large hermaphrodites or male gametophytes from the ethyl methyl sulfatemutagenized populations after 21 d of growth. When grown on the samemedia, wild-type and her13 spores develop as small gametophytes with amulticellular meristem and no antheridia, and are referred to as small meristics(SI Appendix, Fig. S1). Of the 23 putative gaia her13 hermaphrodites selectedand crossed by wild-type sperm, three (gaia1-1, gaia1-2, and gaia2-1) producedsporophytes with a gametophyte progeny that segregated large hermaphrodites

Fig. 4. Reconstructed evolution of the interactions between ACE/GA (yel-low), ABA (red), SnRK2s (green), and S-type anion channels (blue). Precursorsof GA and ABA interact to modulate spore dormancy in mosses (22). It is notyet known whether an SnRK2 is involved in this process (dotted green line),or indeed whether these hormones and signaling pathways influence phe-notypes in the most basal extant vascular plant lineage, the lycophytes(dotted lines). SnRK2s signal the ACE–ABA regulatory antagonism of bothgametophyte sex determination, as well as spore dormancy in ferns; thissignaling interaction was later adopted to regulate seed dormancy in theearliest seed plants; however, it may persist in regulating sex determinationin angiosperms (43). Seed plants were the first group of land plants to evolveABA signaling through SnRK2s and S-type anion channels to regulate sto-matal behavior.

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1606614113 McAdam et al.

Dow

nloa

ded

by g

uest

on

July

23,

202

0

(her13 gaia), small meristics (HER13 GAIA), and males (HER13 gaia) in a 1:2:1ratio on medium containing ACE and 10 μM ABA, indicating segregation ofthe mutant phenotype as a single Mendelian trait, independent of her13(SI Appendix, Table S1). When two males (gaia2-2 and gaia4) selected fromthe mutagenized wild-type population of spores were switched to her-maphrodites and crossed by wild-type sperm, these produced sporophyteswith a gametophyte progeny that segregated small meristics (GAIA) andmales (gaia) in a 1:1 ratio on a medium containing ACE and 10 μM ABA, alsoindicating that these gaia phenotypes segregate as single Mendelian traits(SI Appendix, Table S1). All seven gaia mutants develop as normal hermaph-rodites in the absence of ACE.

To determine the linkage relationships among the gaia mutants, eachmutant was crossed to gametophytes of all other gaia mutants. Whengrown in the presence of ACE and 10 μM ABA, segregation of the gameto-phyte progeny derived from these crosses as 3:1 males:small meristics in-dicated that the two gaia mutations were not linked, whereas all malesegregants indicated loci were tightly linked and likely allelic. In somecrosses, a hermaphroditic her gaia double mutant was used as the femaledonor, and in these instances, the expected ratio of progeny grown in thepresence of ACE and 10 μM ABA was 3:3:2 normal hermaphrodites:males:small meristics if the gaia mutations were not linked; or 1:1 male:her-maphrodite if the gaia mutations were completely linked. Linkage analysisrevealed that the seven gaia mutants fall into four linkage groups: gaia1(gaia1-1, gaia1-2, and gaia1-3), gaia2 (gaia2-1 and gaia2-2), gaia3, and gaia4(SI Appendix, Table S1).

Gene Silencing by RNA Interference. Eight-day-old Hn-n (wild-type) gametophytesgrown on media containing ACE were bombarded using a PDS 100 HeliumSystem (BioRad), as described by Rutherford et al. (19), with one exception:1.3 mM tungsten M-20 microcarriers (BioRad) were coated with plasmids puri-fied using a NucleoBond Xtra Midi Plus kit (Macherey-Nagel). Gametophyteswere bombarded with a 35S::DsRed2 plasmid (described in ref. 34) or cobom-barded with the 35S::DsRed2 plasmid plus the hairpin-forming GAIA1 mRNAplasmid (called 35S::hpGAIA1). 35S::hpGAIA1 was made by amplifying two300-bp PCR GAIA1 fragments from cDNA, using primer pairs listed in SI Ap-pendix, Dataset, and by amplifying the Ricinus communis Catalase Intron 1 fromthe 35S:irint (19). These three fragments were then cloned into pFF19 (35) be-tween the 35S promoter and poly(A)+ addition site, using the In-Fusion HDCloning Plus CE kit (Clontech). To test the efficiency of cobombardment andRNAi, gametophytes were cobombarded with 35S::DsRed2 plus 35S::irintCrChl(19), which targets the C. richardii Protoporphyrin IX magnesium chelatase generequired for chlorophyll biosynthesis (SI Appendix, Fig. S4). Targeting this geneby RNAi results in colorless gametophytes (19). One day after bombardment,gametophytes with a DsRed2 fluorescent cell were transferred to ACE-containingplates. One day later, one-half of the gametophytes were transferred to ACE (noABA) and the other half to ACE and 5 μM ABA plates that were then scored andphotographed 11 d after bombardment.

Phylogenetic Analysis. Full-length coding sequences for C. richardii SnRK2 andSLAC homologs were identified by tBLASTn search, using Arabidopsis thalianaOST1 and SLAC1 and SLAH1-4 protein sequences, respectively, as queriesagainst an unpublished Trinity RNA-Seq de novo transcriptome assembly (36)generated from gametophyte tissue. Gene identity was confirmed by recip-rocal BLAST searches against A. thaliana at TAIR (www.arabidopsis.org).Genes from other species were identified from publicly available databases,as outlined in SI Appendix, Dataset. For full sequence details, see SI Appendix,Dataset. For each alignment (SI Appendix, Dataset), full-length amino acidsequence was aligned using ClustalX Version 2.7.000; distance-based methodswere used for phylogenetic analyses in PAUP* 4.0b10 (paup.csit.fsu.edu/).

qRT-PCR. Tomeasure the expression of genes of interest in wild-type C. richardiiplants across life-history stages and sporophyte tissues, spores were harvestedafter 24 h of imbibing in water, gametophytes after 14 d of growth in liquidmedium, crozier tissue was taken from the youngest still unfurled primordialleaf, vegetative leaf tissue was taken from the newest most fully expandedleaves without sporangia, reproductive leaf tissue was taken from the newestmost fully expanded leaves with sporangia, and roots were harvested fromaerial asexual propagules. Total RNA was extracted using the Agilent PlantRNA Isolation Mini Kit (Agilent Technologies), and RNA concentrations de-termined using a NanoDrop 8000 (Thermo Scientific). Reverse transcription wasconducted in 20 μL with 1 μg total RNA, using the Tetro cDNA synthesis kit(Bioline) according to the manufacturer’s instructions. RT-negative (no enzyme)controls were performed to monitor for contamination. First-strand cDNA wasdiluted five times, and 2 μL was used in each real-time PCR. Reactions usingSYBR green chemistry (SensiFAST; Bioline) were set up with a CAS-1200N robotic

liquid handling system (Corbett Research) and run for 50 cycles in a Rotor-GeneQ (Qiagen). Two technical replicates and at least two biological replicates wereperformed for each tissue type. All primer details are given in SI Appendix,Dataset. Expression of each gene of interest was examined relative to theBestKeeper index calculated from three reference genes evaluated using Best-keeper and geNorm, and found to be stably expressed in these samples (CrAPT,M = 0.657; CrUBC9, M = 0.661; CrTBP, M = 0.601), using previously outlinedmethods (37–39).

Leaf Gas Exchange. Stomatal responses to a reversible, mild transition in vaporpressure deficit (VPD) and after leaf excision were assessed in the most recentfully expanded nonreproductive leaves of wild-type and gaia1 mutant plants,using a portable infrared gas analyzer (Li-6400; LI-COR Biosciences). Mea-surements were conducted on one leaf each from three potted plants of WTand gaia1-1, gaia1-2, and gaia1-3 individuals grown under controlled glass-house conditions and shaded natural light, supplemented and extended to a16-h photoperiod by sodium vapor lamps, ensuring a minimum 150 μmolquanta m−2·s−1 at the pot surface and 23 °C/16 °C day/night temperatures.Environmental conditions in the cuvette of the gas analyzer were controlledfor the duration of the experiment at an air temperature of 23 °C, light in-tensity of 500 μmol quanta m−2·s−1, and VPD regulated initially at 1 kPa, usinga portable dewpoint generator (Li-610; LI-COR Biosciences). Leaf gas exchangeand cuvette conditions were logged every 30 s. Leaves were allowed toequilibrate to the conditions inside the cuvette until leaf gas exchange hadreached a maximum and stabilized. After stabilization, VPD was increased to1.5 kPa and maintained for 20 min, after which it was returned to 1 kPa for afurther 20 min. Once the VPD transition was complete and leaf gas exchangehad again stabilized for at least 5 min, leaf tissue inside the cuvette was ex-cised, and leaf gas exchange was monitored for a further 20 min.

Inhibition of Spore Germination by ABA. Approximately 2,000 spores of wild-type, gaia1 and gaia1-3 her3 double mutants were plated on standard me-dium containing both ACE and 0, 10 20, 50, 100, and 200 μM ABA and grownfor 14 d, according to the methods of Banks (21). After 14 d, the numbers ofgerminated and nongerminated spores per plate were scored. For each ge-notype, the percentage of spores that germinated at each increasing con-centration of ABA was calculated relative to the percentage of spores thatgerminated on media containing 0 μM ABA.

CRNA Generation. Full-length coding sequence fromOST1 and SLAC homologsof S. moellendorffii, C. richardii, and A. thaliana were isolated, using primersoutlined in SI Appendix, Table S3, and cloned into pNB1uYN and pNB1uYCexpression vectors by uracil excision–based cloning (40). For functional anal-ysis in oocytes, cRNA was prepared with the mMESSAGE mMACHINET7 transcription kit (Ambion). Oocyte preparation and cRNA injection wereperformed as described by Becker et al. (41). For oocyte BiFC and electro-physiological experiments, 10 ng SLAC:YFPCT (vector pNB1uYC) and 10 ngOST1-homolog:YFPNT (vector pNB1uYN) cRNA were injected.

Oocyte Recordings. In two-electrode voltage-clamp studies, oocytes wereperfused with Tris/Mes buffers. The standard solution contained 10 mM Tris/Mes (pH 5.6), 1 mM Ca (gluconate)2, 1 mM Mg (gluconate)2, 50 mM NaCl orNaNO3

−, and 1 mM LaCl3. Osmolality was adjusted to 220 mosmol kg−1 withD-sorbitol. The standard voltage protocol was as follows: starting from aholding potential (VH) of 0mV, 50ms single-voltage pulses were applied in 20mVdecrements from +70 to −150 mV. Instantaneous currents were extractedright after the voltage jump from the holding potential of 0 mV to thetest pulses.

BiFC Experiments. Expression of BiFC constructs in oocytes was performed asdescribed by Geiger et al. (4). For documentation of the oocyte BiFC results,images were taken with a confocal laser scanning microscope (Leica DM6000CS; Leica Microsystems CMS GmbH) equipped with a Leica HCX IRAPO L25×/0.95W objective. Images were processed (low-pass filtered and sharpened)identically with the image acquisition software LAS AF (Leica MicrosystemsCMS GmbH).

Quantification of ABA Levels in Spores and Leaves. Germinating spores or game-tophytes were filtered from liquid culturemedia, or leafmaterial fromunstressedplants was harvested and immediately weighed (±0.001 g), covered in cold(−20 °C) 80% methanol in water (vol·vol−1) with 250 g·L−1 (m·v−1) of addedbutylated hydroxytoluene and transferred to −20 °C. ABA then was extractedfrom tissue at room temperature, and 15 ng [2H6]ABA (National ResearchCouncil of Canada) was added to each sample, as an internal standard. ABA

McAdam et al. PNAS Early Edition | 5 of 6

PLANTBIOLO

GY

Dow

nloa

ded

by g

uest

on

July

23,

202

0

was purified by ether partitioning and quantified by ultraperformance liquidchromatography tandemmass spectrometry, according to the methods of ref. 42.

ACKNOWLEDGMENTS. We thank John Ross for poignant advice on ABAquantification. This work was supported by Australian Research Council

Grants DE140100946 (to S.A.M.M.) and DP140100666 (to T.J.B.); KingAbdullah Institute for Nanotechnology, King Saud University (R.H. andD.G.); Deutsche Forshungsgemeinshaft grant within SFB/TR166 project B8and FOR964 project 4 (to R.H. and D.G.); and National Science FoundationGrant IOS1258091 (to J.A.B.).

1. Leung J, Giraudat J (1998) Abscisic acid signal transduction. Annu Rev Plant PhysiolPlant Mol Biol 49:199–222.

2. Finch-Savage WE, Leubner-Metzger G (2006) Seed dormancy and the control ofgermination. New Phytol 171(3):501–523.

3. McAdam SAM, Brodribb TJ (2012) Stomatal innovation and the rise of seed plants.Ecol Lett 15(1):1–8.

4. Geiger D, et al. (2009) Activity of guard cell anion channel SLAC1 is controlled bydrought-stress signaling kinase-phosphatase pair. Proc Natl Acad Sci USA 106(50):21425–21430.

5. Schroeder JI, Allen GJ, Hugouvieux V, Kwak JM, Waner D (2001) Guard cell signaltransduction. Annu Rev Plant Physiol Plant Mol Biol 52(1):627–658.

6. Li J, Wang X-Q, Watson MB, Assmann SM (2000) Regulation of abscisic acid-inducedstomatal closure and anion channels by guard cell AAPK kinase. Science 287(5451):300–303.

7. Mustilli A-C, Merlot S, Vavasseur A, Fenzi F, Giraudat J (2002) Arabidopsis OST1protein kinase mediates the regulation of stomatal aperture by abscisic acid and actsupstream of reactive oxygen species production. Plant Cell 14(12):3089–3099.

8. Fujii H, Verslues PE, Zhu J-K (2007) Identification of two protein kinases required forabscisic acid regulation of seed germination, root growth, and gene expression inArabidopsis. Plant Cell 19(2):485–494.

9. Lin Q, et al. (2015) The SnRK2-APC/C(TE) regulatory module mediates the antagonisticaction of gibberellic acid and abscisic acid pathways. Nat Commun 6:7981.

10. Koornneef M, Jorna ML, Brinkhorst-van der Swan DLC, Karssen CM (1982) The iso-lation of abscisic acid (ABA) deficient mutants by selection of induced revertants innon-germinating gibberellin sensitive lines of Arabidopsis thaliana (L.) heynh. TheorAppl Genet 61(4):385–393.

11. Ruszala EM, et al. (2011) Land plants acquired active stomatal control early in theirevolutionary history. Curr Biol 21(12):1030–1035.

12. Brodribb TJ, McAdam SAM (2011) Passive origins of stomatal control in vascularplants. Science 331(6017):582–585.

13. Kenrick P, Crane PR (1997) The origin and early evolution of plants on land. Nature389(6646):33–39.

14. Banks JA, Hickok L, Webb MA (1993) The programming of sexual phenotype in thehomosporous fern Ceratopteris richardii. Int J Plant Sci 154(4):522–534.

15. Tanaka J, et al. (2014) Antheridiogen determines sex in ferns via a spatiotemporallysplit gibberellin synthesis pathway. Science 346(6208):469–473.

16. Hickok LG (1983) Abscisic acid blocks antheridiogen-induced anteridium formation ingametophytes of the fern Ceratopteris. Can J Bot 61:888–892.

17. Ng L-M, et al. (2011) Structural basis for basal activity and autoactivation of abscisicacid (ABA) signaling SnRK2 kinases. Proc Natl Acad Sci USA 108(52):21259–21264.

18. Vilela B, et al. (2013) The maize OST1 kinase homolog phosphorylates and regulatesthe maize SNAC1-type transcription factor. PLoS One 8(2):e58105.

19. Rutherford G, Tanurdzic M, Hasebe M, Banks JA (2004) A systemic gene silencingmethod suitable for high throughput, reverse genetic analyses of gene function infern gametophytes. BMC Plant Biol 4:6.

20. Mapes G, Rothwell GW, Haworth MT (1989) Evolution of seed dormancy. Nature337(6208):645–646.

21. Banks JA (1994) Sex-determining genes in the homosporous fern Ceratopteris.Development 120(7):1949–1958.

22. Vesty EF, et al. (2016) The decision to germinate is regulated by divergent molecularnetworks in spores and seeds. New Phytol 211(3):952–966.

23. Lind C, et al. (2015) Stomatal guard cells co-opted an ancient ABA-dependent desic-cation survival system to regulate stomatal closure. Curr Biol 25(7):928–935.

24. Doi M, Kitagawa Y, Shimazaki K (2015) Stomatal blue light response is present inearly vascular plants. Plant Physiol 169(2):1205–1213.

25. Chater C, et al. (2011) Regulatory mechanism controlling stomatal behavior conservedacross 400 million years of land plant evolution. Curr Biol 21(12):1025–1029.

26. Li J, Assmann SM (1996) An abscisic acid-activated and calcium-independent proteinkinase from guard cells of fava bean. Plant Cell 8(12):2359–2368.

27. Pressel S, Goral T, Duckett JG (2014) Stomatal differentiation and abnormal stomatain hornworts. J Bryol 36(2):87–103.

28. Field KJ, Duckett JG, Cameron DD, Pressel S (2015) Stomatal density and aperture innon-vascular land plants are non-responsive to above-ambient atmospheric CO2concentrations. Ann Bot (Lond) 115(6):915–922.

29. 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.

30. Li J, et al. (2002) Modulation of an RNA-binding protein by abscisic-acid-activatedprotein kinase. Nature 418(6899):793–797.

31. Schneller JJ, Haufler CH, Ranker TA (1990) Antheridiogen and natural gametophytepopulations. Am Fern J 80(4):143–152.

32. Hickok LG, Warne TR, Fribourg RS (1995) The biology of the fern Ceratopteris and itsuse as a model system. Int J Plant Sci 156(3):332–345.

33. Hickok LG (1985) Abscisic acid resistant mutants in the fern Ceratopteris: Character-ization and genetic analysis. Can J Bot 63:1582–1585.

34. Indriolo E, Na G, Ellis D, Salt DE, Banks JA (2010) A vacuolar arsenite transporternecessary for arsenic tolerance in the arsenic hyperaccumulating fern Pteris vittata ismissing in flowering plants. Plant Cell 22(6):2045–2057.

35. Timmermans MCP, Maliga P, Vieira J, Messing J (1990) The pFF plasmids: Cassettesutilising CaMV sequences for expression of foreign genes in plants. J Biotechnol 14(3-4):333–344.

36. Haas BJ, et al. (2013) De novo transcript sequence reconstruction from RNA-seq usingthe Trinity platform for reference generation and analysis. Nat Protoc 8(8):1494–1512.

37. Pfaffl MW, Tichopad A, Prgomet C, Neuvians TP (2004) Determination of stablehousekeeping genes, differentially regulated target genes and sample integrity:BestKeeper–Excel-based tool using pair-wise correlations. Biotechnol Lett 26(6):509–515.

38. Vandesompele J, et al. (2002) Accurate normalization of real-time quantitativeRT-PCR data by geometric averaging of multiple internal control genes. Genome Biol3(7):RESEARCH0034.

39. Pfaffl MW (2001) A new mathematical model for relative quantification in real-timeRT-PCR. Nucleic Acids Res 29(9):e45.

40. Nour-Eldin HH, Hansen BG, Nørholm MHH, Jensen JK, Halkier BA (2006) Advancinguracil-excision based cloning towards an ideal technique for cloning PCR fragments.Nucleic Acids Res 34(18):e122.

41. Becker D, et al. (1996) Changes in voltage activation, Cs+ sensitivity, and ion per-meability in H5 mutants of the plant K+ channel KAT1. Proc Natl Acad Sci USA 93(15):8123–8128.

42. McAdam SAM (2015) Physicochemical quantification of abscisic acid levels in planttissues with an added internal standard by ultra-performance liquid chromatography.Bio Protoc 5(18):e1599.

43. Golenberg EM, West NW (2013) Hormonal interactions and gene regulation can linkmonoecy and environmental plasticity to the evolution of dioecy in plants. Am J Bot100(6):1022–1037.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1606614113 McAdam et al.

Dow

nloa

ded

by g

uest

on

July

23,

202

0