HASTY, the Arabidopsis ortholog of exportin 5/MSN5, regulates phase change and morphogenesis

The movement of macromolecules across the nuclear membrane is a central feature of the biology of eukaryotic cells. Nucleocytoplasmic transport is mediated by receptors (karyopherins) that bind cargo molecules and facilitate their transit through the nuclear pore(Chook and Blobel, 2001;Görlich and Kutay, 1999;Kaffman and O'Shea, 1999;Komeili and O'Shea, 2001;Macara, 2001;Nakielny and Dreyfuss, 1999). These nucleocytoplasmic transport receptors include proteins that regulate nuclear import (importins), proteins that regulate nuclear export (exportins),and proteins that import some proteins and export others. The macromolecules transported by various karyopherins have been defined primarily by biochemical methods in yeast and mammals, and often have conserved functions in these two organisms. These studies have shown that some karyopherins regulate the transport of a large number of functionally diverse proteins, whereas others have a more limited range of cargo molecules. Much less is known about the cellular and developmental processes for which these receptors are required,and how their activity is regulated by endogenous or environmental stimuli(Kaffman and O'Shea, 1999). InSaccharomyces cerevisiae, genetic analysis has revealed a requirement for specific karyopherins in both constitutive cellular processes and a variety of signal transduction pathways. Genetic evidence for the involvement of karyopherins in specific cellular and developmental processes in higher organisms is more limited, and thus far has been obtained for only three proteins in Drosophila (Lippai et al., 2000; Lorenzen et al.,2001; Tekotte et al.,2002; Tirian et al.,2000).

Nucleocytoplasmic transport operates by a similar mechanism in yeast and mammals (Chook and Blobel,2001; Görlich and Kutay,1999; Kaffman and O'Shea,1999; Komeili and O'Shea,2001; Macara,2001). Karyopherins interact with cargo proteins at specific sites known as nuclear localization signals (NLS) or nuclear export sequences (NES). Most karyopherins bind directly to these sequences but a few bind via an adaptor protein, such as importin α, the adaptor for importin β. Nuclear transport is driven by the interaction between karyopherins and the small GTPase, Ran. Importins bind cargo molecules in the presence of Ran-GDP and release them upon binding Ran-GTP, whereas exportins bind cargo molecules in association with Ran-GTP and release them when Ran-GTP is hydrolyzed to Ran-GDP. Ran-GTP is kept at a relatively low level in the cytoplasm by the activity of the GTPase-activating protein RanGAP and the nucleotide-exchange factor RanBP1 (RCC), but is present at a relatively high level in the nucleus where these factors are absent. Importins bind cargo molecules in the cytoplasm, where Ran is primarily in its GDP-bound form, and release them in the nucleus, where the concentration of Ran-GTP is relatively high. Conversely, exportins bind Ran-GTP and their cargo molecules in the nucleus,and dissociate from these molecules in the cytoplasm when Ran-GTP is hydrolyzed to Ran-GDP. Other factors that influence cargo binding include the phosphorylation state of the cargo and its association with factors that regulate the accessibility of the NLS or NES.

Although the mechanism of nuclear transport in plants has not been intensively studied, it is likely to be similar to the mechanism that has been described in other organisms (Merkle,2001; Smith and Raikhel,1999). Proper localization of proteins in plant cells depends on the same type of nuclear localization signal(Hicks and Raikhel, 1995) and nuclear export sequence (Haasen et al.,1999; Ward and Lazarowitz,1999) that fulfill this function in other organisms. Furthermore,plants possess homologs of many of the proteins known to be involved in nucleocytoplasmic transport in other organisms. These include homologs of importin α (Ballas and Citovsky,1997; Hubner et al.,1999; Jiang et al.,1998a; Jiang et al.,2001; Smith et al.,1997a; Smith et al.,1997b), importin β (Jiang et al., 1998a), the exportin Crm1/Xpo1(Haasen et al., 1999), Ran1(Haizel et al., 1997), RanGAP(Rose and Meier, 2001) and RanBP1 (Kim et al., 2001). Plant homologs of importin α (Hubner et al., 1999; Jiang et al.,1998a; Jiang et al.,2001) and importin β(Jiang et al., 1998b) have also been shown to mediate nuclear transport in animal cell systems, although with somewhat different properties than the homologous yeast and mammalian proteins.

The HASTY (HST) gene in Arabidopsis was originally identified in a screen for mutations that affect the transition between the juvenile and adult phases of vegetative development(Telfer and Poethig, 1998),and has also been identified in screens for mutations affecting leaf(Berna et al., 1999;Serrano-Cartagena et al.,2000) and floral (Berna et al.,1999; Eshed et al.,2001; Serrano-Cartagena et al., 2000) morphology. We present a detailed analysis of thehst phenotype that reveals a role for this gene in many different aspects of plant development. The basis for this pleiotropic phenotype is suggested by the discovery that HST encodes a protein similar to the karyopherins exportin 5 (Xpo5) in mammals and Msn5p in yeast(Alepuz et al., 1999;Bohnsack et al., 2002;Brownawell and Macara, 2002;Calado et al., 2002;Görlich et al., 1997;Kaffman et al., 1998). Thehst loss-of-function phenotype indicates some of the developmental pathways for which HST is required, and suggests that regulation of nucleocytoplasmic transport may play an important role in these pathways.

All of the genetic stocks described in this paper are in a Columbia (Col)genetic background. Analyses of leaf and inflorescence morphology were conducted using plants grown in pots in Metromix 200 (Scotts). Seedling growth and morphology was studied using plants grown vertically in petri dishes on a medium consisting of 1/2× MS salts (Gibco BRL), 0.5 g/l MES, 5% sucrose,0.8% agar (pH 5.7). Seeds were stratified at 4°C for 2-3 days before being transferred to a growth chamber, and were grown at 22°C under continuous fluorescent illumination unless otherwise noted.

Abaxial trichomes were scored by observing seedlings 10-18 days after planting with a stereomicroscope. Leaf morphology was recorded by photocopying leaves attached to cardboard with double-sided tape. These photocopies were subsequently scanned into a computer and manipulated with Photoshop to produce the images presented here. The rate of leaf initiation was measured using plants homozygous for a LFY::GUS transgene obtained from D. Weigel(Salk Institute). Soil-grown plants were harvested on successive days and stained for GUS activity, and the number of leaf primordia was then counted with the aid of a compound microscope.

GUS activity was observed by incubating seedlings overnight in 1 mM X-gluc,50 mM sodium phosphate buffer (pH 7), 0.5 mM potassium ferricyanide, 0.5 mM potassium ferrocyanide and 0.1% Triton X-100 at 37°C followed by decolorization in 95% ethanol. They were then dehydrated in 100% ethanol,transferred to xylene and mounted in Permount on microscope slides. Specimens expressing 35S::LFY were stained overnight; specimens expressing35S::HST-GUS were stained for 2 hours.

Seedlings were fixed overnight at 4°C in 3% glutaraldehyde in 0.5 M phosphate buffer (pH 5.7), post-fixed in 1% OsO₄ for 2 hours, and then dehydrated and embedded in Spurr's resin. Sections were stained with 1%Methylene Blue in 1% sodium borate. Scanning electron microscopy was performed on flowers fixed in FAA (Jensen,1962).

F₂ progeny of a cross between hst-1 (Col) and wild-type Landsberg erecta (Ler) were screened for kanamycin resistance either by spraying soil-grown plants with 200 μg/ml kanamycin one week after planting, or by sowing F₂ seeds on germination medium containing 50μg/ml kanamycin. Kanamycin-resistant plants that were homozygoushst-1 were allowed to set self seed at 18°C. RFLP polymorphisms were scored using DNA isolated from 10 or more progeny of these recombinant F₂ plants. RLFPs useful for mapping hst-1, as well as allele-specific RFLPs associated with hst mutations, were identified by hybridizing radioactively labeled BAC clones to blots of genomic DNA digested with a variety of restriction enzymes.

HST cDNAs were isolated using standard methods(Sambrook et al., 1989) from size-fractionated cDNA libraries (Kieber et al., 1993) obtained from the Arabidopsis Biological Resource Center. Bluescript plasmids (pBS SK–) were excised from lambda clones using a rapid excision kit (Stratagene) and the protocol supplied by the manufacturer.

RNA for northern analysis was prepared from leaf and floral bud tissues from mature plants, and from root tissue from plants grown in culture. RNA was prepared using TRIzol reagent (GibcoBRL) and poly(A) RNA was isolated using the Oligotex mRNA kit (Qiagen). PolyA RNA (0.75 μg) was run on a 1.2%agarose gel containing 3% formaldehyde and then transferred to Hybond N⁺ membrane (Amersham). Hybridization was carried out according to Sambrook et al. (Sambrook et al.,1989), using a 0.7 kb 5′ fragment of the HSTcDNA.

The HST cDNA (AY198396) was placed under the regulation of the CaMV 35S promoter in pCAMBIA 3301. The HST-coding region was amplified from a cloned HST cDNA template using primers 5′-GCGGATCCATGGAAGATAGCAACTCCACG-3′ and 5′-CGGATCCGCTAGCTCATTGTACGAACTCTTCATCC-3′ that included the start and stop codons, respectively. The 5′ primer also included anNcoI site, and subsequent digestion of the PCR product withNcoI generated a HST cDNA with an NcoI end and a blunt end. pCAMBIA 3301 was digested with BstEII to remove theβ-glucuronidase gene, blunt-ended with Klenow enzyme and digested withNcoI. The vector fragment was gel-purified and ligated to theHST cDNA to generate the HST over-expression construct. AHST-GUS fusion construct was generated by amplifying theHST-coding region minus the stop codon from the HST cDNA using primers 5′-GCGGATCCATGGAAGATAGCAACTCCACG-3′ and 5′-GCGGATCCATGGATTGTACGAACTCTTCATCC-3′, which added NcoI sites at each end of the amplicon. NcoI-cut PCR product was then inserted into the NcoI site of vector pCAMBIA 3301, to generate a translational fusion between the 3′ end of the HST gene and the 5′ end of the Escherichia coli uidA gene encodingβ-glucuronidase. Both constructs were used to transform wild-type (Col)and hst-1 (Col) plants using the floral dip method(Clough and Bent, 1998), and transformants were selected by spraying the seedlings produced by these plants with 250 mg/l glufosinate (AgrEvo).

The coding region of Arabidopsis RAN1(Haizel et al., 1997) was amplified from first-strand cDNA products generated by reverse transcription of RNA isolated from Arabidopsis seedlings. The RT-PCR reaction employed a 5′ RAN-specific primer with an EcoRI site,(5′-GGAATTCATGGCTCTACCTAACCAGCAAACCG-3′) and a 3′RAN-specific primer with a BamHI site(5′-CGGATCCTTACTCAAAGATATCATCATCGTC-3′). Upon digestion withEcoRI and BamHI, the PCR product was inserted intoEcoRI/BamHI-digested pGBKT7 (Clontech) to generate theRAN1:BD vector. To generate the HST:BD vector, theHST-coding region was amplified from the HST cDNA using primers flanked by NcoI sites, and was inserted intoNcoI-digested pGADT7. The ΔN-hst:AD vector, which contains an 106 N-terminal deletion of HST, was created in a similar fashion using the primers 5′-AGACCATGGCTCTTAAGAGTCAGTCTGCT-3′ and 5′-GCGGATCCATGGATTGTACGAACTCTTCATCC-3′. The hst-3:AD vector was created by in vitro mutagenesis (Stratagene Quick Change kit) of the HST-AD vector to introduce a mutation present in hst-3. Yeast transformation and subsequent two-hybrid analyses were carried out according to the Matchmaker protocol (Clontech). In addition, we made constructs to test the reciprocal two-hybrid interaction, i.e. RANfused to pGADT7 and HST fused to pGBKT7, but we were unable to recover RAN-GAD transformants, which suggests that this construct somehow affected yeast growth.

All of the results presented in this paper were obtained usinghst-1. Because the phenotype of hst-1 is identical to a likely null allele, hst-6 (Table 1), we believe the phenotype of hst-1 represents the null phenotype of this gene. Previous studies have described several of the effects of hst-1 on plant morphology. These include: a reduction in the size of leaves, sepals and petals; up-rolling of the leaf blade; a reduction in leaf number; accelerated production of abaxial trichomes; disrupted phyllotaxis in the inflorescence; and reduced male and female fertility(Fig. 1)(Serrano-Cartagena et al.,2000; Telfer and Poethig,1998). hst mutations have also been reported to enhance the floral phenotype of kan; pkl double mutants(Eshed et al., 2001). We confirmed and extended these observations, and found that this phenotype can be attributed to defects in several different processes, including cell growth, phase change and the specification of organ polarity.

hst-1 has significant effects on root and hypocotyl development under both dark and light conditions. hst-1 seedlings grown in the dark, on a medium containing 5% sucrose, have an open apical hook and irregularly oriented cotyledons; wild-type seedlings typically have a tight apical hook and appressed cotyledons (Fig. 2A). Light- or dark-grown hst-1 seedlings also have an abnormally short hypocotyl and primary root(Fig. 2A,B,I). The effect ofhst-1 on root length is attributable to both a decrease in the rate of root elongation and premature cessation of root growth(Fig. 2I). hst-1 roots grow at a rate of about 0.9 mm/day, compared with 2.6 mm/day for wild type,and cease elongating by 9 days after planting. This defect in root growth is accompanied by the production of adventitious roots from the base of the hypocotyl (Fig. 2B).hst-1 has no obvious effect on the anatomy of the hypocotyl and root(Fig. 2G,H), which suggests that its effect on root and hypocotyl growth does not result from a change in cell identity, as is the case for several other mutations that produce short roots (Benfey et al., 1993). However, hst-1 does affect the anatomy of the shoot apical meristem(SAM): 3-day-old hst-1 seedlings(Fig. 2F) have a larger and more rounded SAM than do wild-type seedlings(Fig. 2C). The primordia of leaves 1 and 2 are also smaller than normal in 3-day-old hst-1seedlings (Fig. 2F), probably because their initiation is delayed relative to wild type (see below).

The juvenile and adult phases of vegetative development inArabidopsis are distinguishable by differences in leaf size and shape and by the distribution of trichomes on the leaf blade(Chien and Sussex, 1996;Röbbelen, 1957;Steynen et al., 2001;Telfer et al., 1997;Telfer and Poethig, 1994;Tsukaya et al., 2000). The first four rosette leaves of wild-type plants (juvenile leaves) are round or elliptical, have a smooth margin, a low ratio of blade length to petiole length, lack trichomes on their abaxial surface and have a relatively low density of trichomes on their adaxial surface(Fig. 3;Table 1). The adult phase starts with the production of leaf 5 or 6, and is marked by the production of more highly serrated leaves that have a high blade:petiole ratio and an increased capacity for trichome production on both their abaxial and adaxial surfaces. Between these two phases there is a brief transition phase, which is defined by the production of leaves that are partially covered with abaxial trichomes.

In hst-1, the onset of abaxial trichome production(Fig. 3A;Table 1), the increase adaxial trichome density (Fig. 3B) and the decline in blade:petiole ratio (Fig. 3C) begin with leaf 3 and proceed unusually quickly after this point. To determine if this phenotype reflects a change in timing of the juvenile-to-adult transition, we examined the effect of hst-1 on the rate of leaf and floral initiation (Fig. 3D). For this purpose, we took advantage of plants carrying theLFY::GUS transgene because this gene makes it possible to observe leaf and floral bud primordia at an early stage in their development(Blazquez et al., 1997). Under continuous light (CL), hst-1 seedlings exhibited a slight delay in leaf initiation during germination but went on to produce leaves at the same rate as wild type throughout the rest of vegetative development.hst-1 plants produced their first leaf with abaxial trichomes (leaf 2.6±0.1) 3.8 days after planting (DAP), whereas wild-type plants produced this leaf (leaf 4.8±0.1) 2 days later. Mutant plants also initiated floral buds 2 days earlier than did wild type. Thus, hst-1accelerates both the juvenile-to-adult transition and floral induction. The early flowering phenotype of hst-1 is completely attributable to its effect on vegetative phase change because flower initiation occurred 7 days after the juvenile-to-adult transition in both mutant and wild-type plants.

Although hst-1 accelerates floral induction, it has a more variable effect on bolting and the appearance of the first open flower.hst-1 plants sometimes bolt and produce mature flowers earlier than normal (Table 1), but may flower later than wild-type plants under some conditions. For example, thehst-1 plants illustrated in Fig. 3D produced their first open flower 1 day later than normal (24.3 DAP versus 23.5 DAP) even though they produced floral buds 2 days early. This observation suggests that bolting and floral maturation are either regulated independently of floral initiation in mutant plants, or that hst-1delays bolting and floral maturation after floral initiation has taken place. Given the effect of hst-1 on root and hypocotyl growth, we believe that the latter explanation is more likely.

Unexpectedly, hst-1 has opposite effects on flowering time in short days (SD). Although mutant plants flower slightly earlier than normal in long days (LD; 16 hours light:8 hours dark) or continuous light, they flower much later and with nearly twice as many leaves and bracts as wild-type plants in SD (8 hours light:16 hours dark) (Table 1). We conclude HST normally acts to repress flowering in LD and to promote flowering in SD.

The first two leaves of hst-1 plants are generally flat or only slightly up-rolled, but subsequent leaves curve strongly upward(Fig. 1B). This effect on leaf expansion is associated with a loss of tissue polarity within the mesophyll of the leaf blade. In wild-type leaves, cells in the upper mesophyll are spherical or slightly cylindrical in shape and are densely packed, whereas cells in the lower mesophyll layer are irregular in shape and are separated by large air spaces (Fig. 4A). Although the upper mesophyll layer of hst-1 appears normal, cells in the lower spongy layer are more regular in shape and have less intercellular space than normal (Fig. 4B),which causes this layer to resemble the upper (adaxial) mesophyll layer. A role for HST in the regulation of organ polarity is also apparent in the effect of hst-1 on carpel development. Although the carpels ofhst-1 flowers are quite normal early in the growth of the inflorescence, older inflorescences typically produce flowers with a laterally expanded stigma, unfused carpels and external ovules(Fig. 4D,E). The severity of this defect varies between plants, but occurs in ∼30% of flowers(n=277). The defect in carpel fusion and ovule production is usually limited to the apical end of the carpels although, in rare cases, it may extend along their entire length (Fig. 4E). This phenotype is characteristic of adaxializing mutations,e.g. mutations in members of the YABBY(Bowman, 2000;Siegfried et al., 1999) andKANADI (KAN) (Eshed et al., 2001; Kerstetter et al.,2001) gene families, and suggests that HST may regulate abaxial polarity in carpels as well as in leaves.

The interaction between hst-1 and the adaxializing mutationkan-11 (Kerstetter et al.,2001) is consistent with the hypothesis that HST regulates adaxial/abaxial polarity. We found that hst-1; kan-11 double mutants had more highly up-rolled cotyledons and leaves(Fig. 4F) and produced significantly more abaxial trichomes on leaves 1 and 2 than did either single mutant. In a 16 mm² area of leaf 1 or 2, kan-11 had an average of 2.2±0.4 abaxial trichomes and 10.3±0.5 adaxial trichomes, hst-1 had 0.3±0.1 abaxial and 12.5±0.7 adaxial trichomes, and hst-1; kan-11 seedlings had an average of 11.9±8 abaxial and 11±0.5 adaxial trichomes. Because hst-1;kan-11 seedlings did not exhibit an increase in adaxial trichome density(and, in fact, had equal numbers of trichomes on both surfaces of the leaf),we attribute the synergistic effect of these mutations on abaxial trichome production to a loss of leaf polarity, rather than to an overall increase in the capacity for trichome production.

Mutations in SQUINT (SQN), the Arabidopsisortholog of cyclophilin 40, are phenotypically similar to hstmutations in that they slightly delay leaf initiation (albeit at a different stage in shoot development), accelerate vegetative phase change and produce aberrant phyllotaxy in the inflorescence as well as abnormal patterns of carpel development (Berardini et al.,2001). To determine the genetic relationship between these two genes, we examined the phenotype of hst-1, sqn-1 plants. Double mutant seedlings had a much more severe phenotype than either single mutant(Fig. 5). In addition to being significantly smaller than both sqn-1 and hst-1, double mutants produced large numbers of abaxial trichomes starting with leaf 1. By contrast, hst-1 and sqn-1 did not produce abaxial trichomes until leaf 3, and had only a few abaxial trichomes on this leaf. We conclude that SQN and HST either operate cooperatively in the same regulatory pathway or operate in parallel pathways.

hst-1 maps to chromosome 3, ∼0.8 cM south of GAPC(Telfer and Poethig, 1998). Recombinants between GAPC and hst-1 were isolated by screening 650 F₂-hst plants from a cross of hst-1(Col) to Ler. Recombinants to the south of hst-1 were identified by screening for kanamycin-resistant hst plants in F₂ families from a cross of hst-1 to DsTn 108, an Ler line containing a kanamycin-resistant transgene located 3 cM south of GAPC (Bancroft and Dean,1993). Existing RFLP markers from this region were mapped relative to hst-1 using this collection of recombinant lines. The most tightly linked marker, lambda clone m262, detected an Xho1 polymorphism that was not recombinant in any of these lines(Fig. 6). m262 was hybridized to the IGF bacterial artificial clone (BAC) library to identify larger clones from this region. Two such clones, F1D17 and F13L14, identifiedEcoRV and HindIII RFLPs on either side of hst-1,demonstrating that these BACs span the HST locus. These and other overlapping BACs were then hybridized to Southern blots of DNA from all existing hst mutant alleles in order to identify allele-specific polymorphisms. The clone F9J1 revealed a 5.5 kb EcoRV band that was present in wild-type Col plants but absent in hst-9(Fig. 6). This 5.5 kb fragment hybridized to two restriction fragments in EcoRV digests ofhst-9 DNA, which implies that the mutation in hst-9 is located in this fragment and is likely to be an inversion(Fig. 6). Hybridization of thisEcoRV fragment to a seedling cDNA library produced cDNAs corresponding to a single gene. The longest of these cDNAs (AY198396) is 3840 bp, consists of 22 exons and encodes a protein of 1202 amino acids with a predicted molecular weight of 133 kDa (Fig. 7A). Hybridization of this cDNA to poly(A)-enriched RNA revealed a single 4 kb transcript in leaves, roots and floral buds(Fig. 7B), an expression pattern that is consistent with the fact that HST has a mutant phenotype in all of these organs.

All of the hst alelles we analyzed had mutations in the predictedHST transcript (Fig. 7A). hst-1 is a single base transversion that changes the conserved GT to GA at the splice donor site of the 12th intron. RT-PCR reactions with primers that flank this intron demonstrated that it is not spliced from pre-mRNA in hst-1 (data not shown), resulting in an in-frame stop codon 31 bp after the exon-intron junction. hst-3 is a 3 bp deletion in exon 1 that replaces amino acids D³⁶S³⁷with an A. hst-6 contains a G to A transition that results in a stop codon at position 107 in the predicted protein. This fragment represents only 9% of the full-length protein, which suggests that hst-6 is likely to be a null allele. hst-7, an X-ray-induced allele, has a 7 bp deletion in exon 13 that introduces a stop codon in the open reading frame 12 bp after this mutation. hst-8 has a 24 bp in-frame deletion in exon 9.hst-9 is a 300 kb inversion with breakpoints in intron 15 and in the last exon of At3g05770, a transcribed gene of unknown function.

To determine the effect of overexpressing HST, the HSTcDNA was fused to the CaMV 35S promoter in pCAMBIA 3301 and the resulting construct was introduced into hst-1 plants by Agrobacterium transformation. This transgene was capable of completely rescuing the mutant phenotype of hst-1 (Table 1), demonstrating that it encodes a functional HST protein. Wild-type plants transformed with this construct, and which expressed high levels of HST mRNA (data not shown), were completely normal. In order to ensure that the absence of phenotypically aberrant plants was not due to loss-of-function mutations in the transgene, an insertion that completely rescued the hst-mutant phenotype(Table 1) was crossed into a Col genetic background. Col plants carrying a single copy of this transgene were not obviously different from their wild-type siblings: transgenic plants(n=13) had 4.6±0.2 juvenile leaves and 5.5±0.3 adult leaves, whereas their wild-type siblings (n=29) had 4.4±0.1 juvenile leaves and 5.5±0.2 adult leaves. This result suggests either that plants are insensitive to levels of the HST protein above the wild-type amount, or that HST is post-transcriptionally regulated.

HST encodes a protein consisting of 1202 amino acids (133 kDa)with approximately 12 regions that have similarity to the HEAT repeats typically found in karyopherins in the importin β family(http://www.embl-heidelberg.de/~andrade/papers/rep/search.html)(Andrade et al., 2001). BLAST searches of non-redundant databases at GenBank revealed that HST is similar in both size and amino acid sequence to the mammalian karyopherin exportin 5(Xpo5) (1204 amino acids, 136 kDa) and the orthologous protein Msn5p (1224 amino acids, 142 kDa) from Saccharomyces cerevisiae(Fig. 7C). Arabidopsisgenes encoding importin β-like proteins were identified by searching theArabidopsis genome database(http://arabidopsis.org/wublast/index2.html)for proteins with similarity to members of this family in yeast and mammals.Arabidopsis has at least 17 predicted members of the importin βfamily. These Arabidopsis proteins are much more similar to mammalian proteins than to their yeast homologs. A cladogram illustrating the relationship between human importin β-like proteins and predicted members of this family in Arabidopsis is shown inFig. 8. HST is the most closely related protein in Arabidopsis to Xpo5 and Msn5p, and is more similar to Xpo5 than it is to any other protein in Arabidopsis. We conclude that HST is the Arabidopsis ortholog ofXpo5/MSN5.

Like many other karyopherins(Görlich and Kutay, 1999;Komeili and O'Shea, 2001;Macara, 2001;Nakielny and Dreyfuss, 1999),the interaction between Xpo5/Msn5p their cargo molecules is regulated by Ran(Brownawell and Macara, 2002;Bohnsack et al., 2002;Calado et al., 2002). To test if HST interacts with Ran we took advantage of the yeast two-hybrid assay. For this purpose, the coding region of Arabidopsis RAN1 was fused to the DNA-binding domain of GAL4 in the plasmid pGBKT7 to produceRAN1:AD, and the HST coding region was fused to theGAL4 activation domain in pGADT7 to produce HST:BD. As negative controls, we produced a construct containing a 106 amino acid N-terminal deletion of HST, encompassing the putative Ran-binding domain (ΔN-hst:BD), and a construct containing thehst-3 mutation (hst-3:BD), which lies in this region. Cells containing the HST:BD and RAN1:AD constructs grew under stringent conditions, whereas strains containing either hst-3:BD orΔN-hst:BD and RAN1:AD grew very poorly(Fig. 9). This result is consistent with the results of a similar experiment performed with Xpo5(Brownawell and Macara, 2002),and suggests that HST has Ran-binding activity.

To determine the subcellular distribution of HST, we produced transgenicArabidopsis plants expressing a HST-GUS fusion protein under the regulation of the CaMV 35S promoter. hst-1 plants containing the transgene were phenotypically wild type(Table 1), demonstrating that the fusion protein was functional. GUS activity was concentrated in the nucleus and the surrounding cytoplasm in root hairs and trichomes, and in cells in the stele and stipules (Fig. 10). In trichomes, which are large cells with large polyploid nuclei, GUS staining was clearly localized in the vicinity of the nuclear membrane and was always most intense on the apical side of the nucleus(Fig. 10B). This is in contrast to the widespread distribution of GUS activity in cells transformed with a control 35S::GUS vector (Fig. 10C). This perinuclear distribution is similar to that of other karyopherins (Fornerod et al.,1997; Görlich et al.,1997; Kudo et al.,1997; Rout et al.,1997) and suggests that HST has a role in nucleocytoplasmic transport. The polarized distribution of HST in trichomes is particularly interesting because it suggests that asymmetric nucleocytoplamic transport may contribute to trichome cell polarity.

We show that HST is the Arabidopsis ortholog ofXpo5 in mammals and MSN5 in yeast, and describe the phenotype of a loss-of-function mutation in this gene. HST is one of at least 17 importin β-like genes in Arabidopsis; a similar number exist in yeast and humans(Görlich and Kutay, 1999;Chook and Blobel, 2001). This remarkable conservation of gene number, as well as evidence that at least two members of this family are capable of mediating nucleocytoplasmic transport(Haasen et al., 1999;Jiang et al., 1998b), suggests that the function of these proteins has been largely conserved in plants. Although we have yet to demonstrate that HST regulates nucleocytoplasmic transport, the observation that it interacts with Ran, and its perinuclear distribution, is consistent with this function.

The potential cargo molecules of HST are difficult to predict because its orthologs in yeast and mammals have quite different functions. In yeast, Msn5p exports phosporylated forms of several different transcription factors(Blondel et al., 1999;Boustany and Cyert, 2002;DeVit and Johnston, 1999;Gorner et al., 2002;Kaffman et al., 1998;Komeili et al., 2000) and imports RPA (Yoshida and Blobel,2001), a protein involved in DNA replication and repair. Loss-of-function mutations in MSN5 cause defects in the processes in which these factors are involved, demonstrating that Msn5p has an essential function. The function of Xpo5 is more controversial. Brownawell and Macara(Brownawell and Macara, 2002)concluded that Xpo5 is a general exporter of double-stranded RNA-binding proteins (dsRNA-BP) because it binds to and exports the dsRNA-BP interleukin enhancer-binding factor 3 (ILF3), and it interacts in vitro with the dsRNA-binding domains of several other proteins. Other investigators(Bohnsack et al., 2002;Calado et al., 2002) have demonstrated that Xpo5 exports tRNAs and regulates the export of the translation factor eELF1a indirectly, through the association of eELF1a with tRNAs. The interaction between Xpo5 and ILF3 is also regulated by dsRNA(Brownawell and Macara, 2002)but the functional significance of this observation remains unclear. Brownawell and Macara (Brownawell and Macara, 2002) conclude that binding of ILF3 to dsRNA occurs in the cytoplasm and promotes the dissociation of ILF3 from Xpo5, whereas Bohnsack et al. (Bohnsack et al., 2002)propose that the interaction between Xpo5 and ILF3 is actually mediated by dsRNA and is essential for the export of ILF3. These latter investigators propose that Xpo5 primarily regulates RNA export, not protein export. This conclusion is supported by the recent results of Gwizek et al. (Gwizek et al.,2003), which indicate that Xpo5 preferentially exports RNAs that possess a 20-nucleotide minihelix with a base-paired 5′ end and an unpaired 3′ end. These investigators suggest that the primary cargo molecules of Xpo5 may be siRNAs involved in gene silencing and the double-stranded precursors of miRNAs.

Whatever the cargo molecules of HST may be, the phenotype of hst-1suggests that they include factors required for plant growth as well as factors that have more specific regulatory functions. A general requirement for HST is evident in the relatively small size and reduced growth rate of essentially all the organs of mutant plants. HST is also required for the growth and/or organization of the shoot apical meristem (SAM), as demonstrated by the aberrant morphology of the SAM in mutant seedlings and the transient delay in leaf initiation after germination. We tested the sensitivity ofhst-1 seedlings to auxin, cytokinin and abscisic acid but observed no major effect of these hormones on root or hypocotyl growth (M.-Y.P. and R.S.P., unpublished). Thus, the hormonal basis, if any, for this growth defect is unclear. An alternative possibility is that this growth defect results from a change in cell identity. Mutations that disrupt radial patterning in the root, for example, have an effect on root growth similar to that ofhst-1 (Benfey et al.,1993). However, we do not think this explains the effect ofhst-1 on root and hypocotyl elongation because hst-1 has no major effect on the cellular organization of these structures. We are intrigued by the evidence that Xpo5 exports tRNAs and other double-stranded RNAs because defects in the export of this class of RNAs would be expected to have widespread effects on growth, like those observed in hstmutants.

It is unlikely that HST transports only factors with housekeeping functions because of the developmental specificity of many aspects of its mutant phenotype. In a previous study, we showed that hst-1 accelerated abaxial trichome production and increased the sensitivity of plants to the floral promoter LFY. This leads to the conclusion that HST normally acts to promote the expression of the juvenile phase and/or to repress the expression of the adult phase (Telfer and Poethig,1998). Because the effect of hst-1 on abaxial trichome production can also be attributed to a change in leaf polarity, we examined its effects on two non-polarized heteroblastic traits: adaxial trichome density and leaf shape (Fig. 3). We found that hst-1 accelerates changes in both of these traits, which supports our previous conclusion about its role in phase change. This conclusion is also supported by the fact that the early flowering phenotype of hst-1 is completely accounted for by its effect on the duration of the juvenile and intermediate phases of shoot development; under CL, it has little or no effect on the duration of the adult phase. This specific effect on the duration of the juvenile phase is unusual. Other flowering time mutations whose effect on vegetative phase change has been studied have either no effect on the duration of the juvenile phase(Martinez-Zapater et al.,1995; Melzer et al.,1999; Scott et al.,1999; Telfer et al.,1997; Weigel and Nilsson,1995) or affect the duration of both the juvenile and adult phases of development (Gomez-Mena et al.,2001; Soppe et al.,1999; Telfer et al.,1997). hst-1 is also unusual in that it has a completely opposite effect on flowering time in SD. Under these conditions, mutant plants display accelerated vegetative phase change but are extremely late flowering. One possible interpretation of this phenotype is that HST has different functions in vegetative phase change and floral induction. Early in shoot development, HST promotes the juvenile vegetative phase and reproductive incompetence, whereas during the adult phase it acts to promote flowering under SD but plays no direct role in floral induction under LD. These functions may be carried out by different molecules that require HST for their nucleocytoplasmic transport.

How might HST regulate phase change? Because overexpression of HSThas no effect on the timing of vegetative phase change or floral induction, it is unlikely that these processes are regulated by a change in the transcription of HST. One possibility is that phase change is regulated by post-transcriptional modulation of the amount or activity of HST,as occurs in the case of CRM1 during Xenopus embryogenesis(Callanan et al., 2000). A second possibility is that phase change is regulated by a change in the amount or character of the cargo molecules transported by HST, not by changes in the activity of HST itself. In this scenario, HST would have a permissive function, rather than a regulatory one. We favor the latter hypothesis becausehst-1 has a phenotype in every part of the plant, implying that HST is active at all times and in all tissues. However, the possibility that HST is post-transcriptionally regulated in certain tissues cannot be excluded.

Along with promoting the production of trichomes on the abaxial surface of basal leaves in the rosette, hst-1 reduces the lobing of cells in lower mesophyll layer of the leaf, causes up-rolling of the leaf blade, and produces carpels that are partially unfused and bear ovules on their external(adaxial) margin. hst-1 also interacts synergistically withkan to increase abaxial trichome production and enhances the polarity defects in kan; pkl carpels(Eshed et al., 2001). These features are typical of adaxialized mutants(Bowman, 2000;Eshed et al., 2001;Kerstetter et al., 2001;McConnell et al., 2001), and suggest that HST may be required for the specification of abaxial cell identity in both leaves and floral organs. However, whether HST regulates leaf polarity independently of its role in phase change is unclear. Although adaxial-abaxial polarity is an inherent feature of all lateral organs, the way in which this polarity is expressed varies in different organs and at different stages of shoot development. For example, the first two leaves of the rosette are usually flat, subsequent rosette leaves curve downward,inflorescence leaves are flat, and sepals, petals and carpels curve upward(Griffith et al., 1999). Similarly, trichomes are restricted to the adaxial surface of juvenile leaves,are present on both surfaces of adult leaves and are confined to the abaxial surface of leaves in the inflorescence. Many polarized traits in other organisms are also expressed phase specifically(Kerstetter and Poethig,1998). This phenomenon implies that the mechanism that establishes the adaxial-abaxial polarity of an organ is regulated at some level by the mechanism that specifies the phase identity of the organ. Thus, although it is possible that HST regulates organ polarity independently of its role in phase change, it is also possible that the effect of HST on organ polarity is actually mediated by its effect on phase change.

The basis for the mutant phenotype of HST will only become apparent, of course, when its cargo molecules have been identified. This pleiotropic phenotype suggests that HST transports many different molecules,some of which have regulatory functions. We are particularly interested in the possibility that HST regulates the transport of tRNAs and other double-stranded RNAs and their associated proteins. This hypothesis is not only suggested by function of Xpo5, but is also consistent with the observation that mutations in PAUSED, the Arabidopsisortholog of the tRNA export receptor exportin t, are phenotypically similar tohst (C.H. and R.S.P., unpublished)(Telfer et al., 1997). Further studies should reveal the types of molecules transported by HST and add to our understanding of the role of nucleocytoplasmic transport in plant growth and development.

We are grateful to the Arabidopsis Biological Resource Center, Joe Ecker(Salk Institute) and Detlef Weigel (Salk Institute) for materials used in this study. Randall Kerstetter provided helpful advice and Hui Sun provided valuable technical assistance. The comments of Jeongsik Yong are also gratefully acknowledged. This work was supported by grants from NIH to K.M.B. and C.H., and by grants from NIH and NSF to R.S.P.