HAWAIIAN SKIRT – an F-box gene that regulates organ fusion and growth in Arabidopsis

A fast neutron-mutagenised population of Arabidopsis thaliana Col-0 WT plants was screened for floral phenotypes and a novel mutant, termed hawaiian skirt ( hws ), was identified that failed to shed its reproductive organs. The mutation is the consequence of a 28bp deletion that introduces a premature amber termination codon into the ORF of a putative F-box protein (At3g61590). The most striking anatomical characteristic of hws plants is seen in flowers where individual sepals are fused along the lower part of their margins. Crossing of the abscission marker, Pro PGAZAT :GUS into the mutant reveals that whilst floral organs are retained it is not the consequence of a failure of abscission zone cells to differentiate. Anatomical analysis indicates that the fusion of sepal margins precludes shedding even though abscission, albeit delayed, does occur. Spatial and temporal characterisation, using Pro HWS :GUS or Pro HWS :GFP fusions, has identified HWS expression to be restricted to the stele and lateral root cap, cotyledonary margins, tip of the stigma, pollen, abscission zones, and developing seeds. Comparative phenotypic analyses performed on the hws mutant, Col-0 WT, and Pro 35S :HWS ectopically expressing lines has revealed that loss of HWS results in greater growth of both aerial and below-ground organs whilst over-expressing the gene brings about a converse effect. These observations are consistent with HWS playing an important role in regulating plant growth and development.


INTRODUCTION
Abscission involves the detachment of an organ from the body of a plant and takes place at a site that is predestined for the purpose (Sexton and Roberts, 1982;González-Carranza et al., 1998). The phenomenon may be triggered by a range of environmental stresses including: drought, waterlogging, nutrient deficiency, or pathogen attack (Addicott, 1982;Taylor and Whitelaw, 2001), but is also programmed to occur at discrete stages during plant development such as after leaf senescence, flower fertilisation, or fruit ripening have taken place. Although the precise events that bring about shedding are unclear the process is preceded by an increase in the activity of several wall-loosening enzymes including β 1-4 glucanase and polygalacturonase precisely at the site of abscission (Roberts et al., 2002). It is proposed that the action of these enzymes, coupled with an increase in expansin activity (Belfield et al., 2005;Sampredo and Cosgrove, 2005), may lead to the dissolution of the middle lamella that brings about cell separation (Roberts et al., 2002;González-Carranza et al., 2002).
To dissect further the mechanisms responsible for regulating the abscission process forward genetic strategies have been employed to identify non-or delayedshedding mutants followed by the mapping and characterization of the mutated genes (Patterson, 2001;Roberts et al., 2002). Whilst a number of such mutants have been documented only a few of the genes responsible have been cloned and characterized.
These include jointless (j) that is the consequence of a mutation in a MADS box gene involved in the differentiation of abscission cells in the pedicel of tomato flowers (Mao et al., 2000), and inflorescence deficient in abscission (ida) where the mutated gene has been shown to encode a novel class of ligand that seems to play a key role both in defining the site of cell separation (Stenvik et al., 2006) and in regulating the final step of floral organ shedding in Arabidopsis (Butenko et al., 2003). Recently QTL analysis has led to the identification of a gene (sh4) that has played a critical role in the domestication of rice by bringing about a reduced shattering phenotype (Li et al., 2006). Although the function of the gene is unknown its characteristics suggest that it acts as a transcription factor. Other genes that have been identified to contribute to the abscission process in Arabidopsis include: HAESA that encodes a leucine-rich receptor-like kinase (Jinn et al., 2000); AGAMOUS-LIKE15 (AGL15) a MADS domain transcription factor that if non-functional results in a delay in the time course of abscission (Fernandez et al., 2000;Harding et al., 2003;Lehti-Shiu et al., 2005); and the redundant BLADE ON PETIOLE1 and 2 (BOP1 & BOP2) genes encoding two BTB/POZ domain proteins that when silenced result in a disruption in leaf patterning and floral organ abscission (Hepworth et al., 2005;Norberg et al., 2005).
During the screening of a fast neutron mutagenised population of Arabidopsis thaliana Col-0 WT plants a mutant, termed hawaiian skirt (hws), was isolated that retained its floral organs even after silique desiccation had taken place. In addition to exhibiting a non-shedding phenotype hws also exhibited sepals that are fused for some distance along their margins. Other mutants from Arabidopsis that show varying degrees of sepal fusion include: unusual floral organs (ufo) where the mutated gene has been shown to encode an F-box protein (Levin and Meyerowitz, 1995;Samach et al, 1999); leafy (lfy) which is the result of a mutation in a floral meristem identity gene (Schultz and Haughn, 1991;Weigel and Meyerowitz, 1993); and the cup shaped cotyledon mutants (cuc1, cuc2, cuc3) that arise as a consequence of mutations in putative NAC-transcription factors genes (Aida et al., 1997;Takada et al., 2001;Vroemen et al., 2003). Interestingly, ectopic expression of the microRNA miR164, which has been shown to induce post-transcriptional down regulation of CUC1 and CUC2 (Laufs et al., 2004), leads to the generation of flowers with fused sepals that fail to undergo the normal shedding process (Mallory et al., 2004). A number of other genes have been shown to play an important role in the specification of lateral organ boundaries in leaves or flowers including LATERAL ORGAN BOUNDARIES (LOB) (Shuai et al., 2002); LATERAL ORGAN JUNCTIONS (LOJ) (Prasad et al., 2005); PETAL LOSS (Brewer et al., 2004), the FUSED FLORAL ORGANS loci (FFO1, FFO2, FFO3) (Levin et al., 1998); HANABA TARANU (Zhao et al., 2004); and RABBIT EARS (RBE) (Krizek et al., 2006).
In this report, the mapping and identification of a mutation in a putative F-box gene (At3g61590), which is responsible for bringing about the hws-1 phenotype, is described. Although the transcript of HWS accumulates throughout the plant, reporter gene analysis reveals that expression is restricted to only certain tissues. By comparing and contrasting the phenotypic features of hws-1, an over-expressing HWS line driven by the 35SCaMV promoter, and wild type plants a role for HWS in regulating plant growth and development has been highlighted.

Mutant isolation and characterization
The hawaiian skirt-1 (hws-1) mutant was isolated, as a result of an inability to shed its floral organs, during a screen of M 2 progeny grown from a fast neutron mutagenised population (dose: 55Gy; Lehle seeds) of M 1 Arabidopsis seeds in the Col-0 (Wildtype -WT) background. Sepals, petals and anther filaments were retained throughout reproductive development and remained in situ even after silique desiccation and dehiscence had taken place. To study flower development in more detail material was harvested from positions throughout the primary inflorescence with the first being where petals were visible. From that position all subsequent flowers were numbered.
A scanning electron microscope (SEM) study of hws-1 flowers revealed that sepals of the mutant were fused, for a distance along the lower part of their margins, and that characteristically the sepal whorl was broader than that seen in WT plants . Whilst the shedding of petals and anther filaments did not take place in the mutant, fine structural analysis of these tissues indicated that abscission of these organs could be detected in both WT and hws-1 flowers ( Figures 1G-J). Cell separation at the sepal bases was also apparent in hws-1 flowers, however, the timing of this was delayed in comparison to WT plants and the sepal whorl was retained even though abscised (Figures 1H and J).
Further SEM analysis revealed that no distinction could be made between the WT and hws-1 plants during early bud development (Figures 2A, B, E and F).
However, by the time the buds have reached stage 10 (Smyth et al., 1990) the line of separation between individual sepals had become demarcated to the bud base in WT but not in hws-1 material ( Figure 2C and G). Initially the region of sepal confluence in hws-1 is restricted to just a few cells; however, further division must take place so that by the time buds have reached stage 12 sepals are fused for nearly 25% of their distance ( Figures 2D and H).
The hws-1 mutant was crossed with a WT plant and a 3:1 segregation of shedding:non-shedding individuals in the F 2 population showed that the phenotype is due to a recessive mutation in a single gene of nuclear origin.

Crossing of an abscission zone specific gene marker into the hws-1 mutant
To determine whether the non-shedding phenotype of hws-1 was due to a failure of abscission zone differentiation a cross was made between the mutant and a transgenic marker line (Pro PGAZAT :GUS) that expresses the reporter gene β-glucuronidase (GUS) specifically at the site of floral organ separation. PGAZAT (At2g41850) encodes a polygalacturonase that is transcribed immediately prior to organ abscission in Arabidopsis and is thought to contribute to cell wall degradation (González-Carranza et al., 2002). Several homozygous hws-1 lines containing the GUS gene were isolated. Figures 3A and B show that both WT and mutant plants express the Pro PGAZAT :GUS gene at the base of the sepals, petals, and anther filaments indicating that the hws mutation does not seem to impede abscission zone differentiation.
However, although Pro PGAZAT :GUS expression is apparent in Position 8 flowers of WT plants ( Figure 3A) it cannot be detected until Position 10 in hws-1 flowers ( Figure 3B). In addition, Pro PGAZAT :GUS expression is less pronounced in hws-1 plants and by Position 20, in contrast to WT plants, expression of the reporter gene is no longer detectable at the site of floral organ abscission. For comparison, expression of Pro HWS :GUS is shown at different positions throughout floral development ( Figure   3C). However, these observations will not be described in detail until a later part of the results section of this paper.

Other phenotypic features of the hws mutant
The identification of another allele of HWS (hws-2) from the SALK collection (Alonso et al., 2003), during the course of mapping the gene, lead to the confirmation of additional phenotypic characteristics associated with the hws-1 mutation.
A detailed analysis revealed that 28% of flowers in the hws-1 mutant had fused anther filaments, to differing extents along their length ( Figure 4A), compared with 2% in the WT. Occasionally, in hws-1 plants, these filaments were fused to the side of siliques (Figures 4B). The top of the hws-1 silique was consistently broader than WT and the length of the abscission zone region, exposed by manually removing the floral organs, was longer in the mutant ( Figures 4C and D). Some hws-1 siliques comprised more than two valves ( Figure 4E) and dissection of aberrant pods revealed that this was associated with abnormal development of the septum ( Figure 4F). The lamina tissue of the primary cauline leaves of hws-1 plants routinely exhibited fusion to the inflorescence stems ( Figure 4G). Homozygous hws-1 (Col-0 ecotype) plants were crossed with the Landsberg erecta (Ler) ecotype and the F 2 progeny used as a mapping population. An amplified fragment length polymorphism (AFLP)-based genome-wide approach (Peters et al., 2004) was adopted to map the hws mutation to a 3.28 Mb domain at the bottom of chromosome 3. Using cleaved amplified polymorphic sequences (CAPS), simple sequence length polymorphisms (SSLP), and insertion/deletion (InDel) markers this interval was reduced to a region of 56kb containing 18 annotated genes ( Figure 5A). PCR amplification of the At3g61590 genomic region of WT and hws-1 DNA and restriction analyses of the amplified products with the high frequency cutting enzymes, AluI, RsaI and TaqI revealed subtle differences in banding patterns between the two genotypes ( Figure 5C). Sequencing of this region from hws-1 revealed that it contained a 28bp deletion located 966 bp downstream of the translation start of the ORF of At3g61590. The consequence of this deletion is to introduce a frame shift resulting in the introduction of a premature termination amber codon in place of an isoleucine residue and the predicted production of a truncated version of the At3g61590 protein ( Figure 5B).

Mapping the hws locus
To determine whether the hws-1 phenotype was a consequence of a mutation in the At3g61590 gene a cross between the mutant and the SALK_088349 line (KO) was performed. All progeny of this cross displayed hws-1 characteristics. To test that these had not arisen by a self-pollination event two plants were analysed by PCR and RT-PCR. The PCR demonstrated that both a gene-specific product (originating from the hws-1 mutant) and an insertion product (originating from the KO) were amplified in each individual but only the former was present in the WT control ( Figure 5D). RT-PCR analysis of RNA extracted from various tissues of the SALK_088349 line and from two F 1 plants (hws-1 x KO) showed no At3g61590 expression in the KO but the presence of a transcript in both progeny ( Figure 5E).
Proof that HWS is encoded by the At3g61590 gene was obtained by complementing the hws-1 mutant with a 3.513kb fragment amplified from WT DNA containing 1291bp upstream of the promoter, plus the 5' and 3' UTRs, and intron and exon of the At3g61590 gene. This segment proved to be of sufficient length to rescue fully the hws-1 mutant ( Figure 5F).

HWS encodes a putative F box protein
The likely function of the HWS gene (At3g61590) is that it encodes an F-box protein.
These proteins, as part of a SCF complex, are proposed to interact with a substrate leading to their degradation by the 26S proteasome (Ni et al., 2004). HWS encodes a protein of 412 amino acids. It has no introns within the open reading frame (ORF) but has an intron of 532bp within the 5'UTR of the gene ( Figure 5B). The truncated version of the mutant protein, predicted to be generated in hws-1 plants, contains the intact F-box region. Information retrieved from the web site PlantsUBQ, which is a functional genomics database for the Ubiquitin/26S proteasome proteolytic pathway in plants (http://plantsubq.genomics.purdue.edu/plantsubq/cgibin/detail.cgi?db=plantsubq&id=163137), indicates that the predicted F-box domain of HWS is located between amino acids 40 to 85. The software also proposes the existence of a transmembrane spanning region between amino acids 120-140 and a low similarity (29.3%) to a Kelch_2 motif in the region between amino acids 290-

338.
Outside the F-box region, the ORF of HWS shows only a low level of sequence similarity with other putative F-box proteins that have been annotated within the Arabidopsis genome. The Arabidopsis gene encoding a protein with highest sequence similarity (approximately 30%) to HWS is UNUSUAL FLORAL ORGANS (UFO).
UFO is a protein that has been shown to be required for normal patterning and growth of the floral meristem (Samach et al., 1999).  Figure   3C).

Ectopic expression of the HWS gene produces smaller seedlings
Transgenic plants were generated where the ORF from the HWS gene was expressed ectopically using a double 35SCaMV (Cauliflower Mosaic Virus) promoter.
Homozygous plants from two lines (8.3, A23.3) were examined in detail as RT-PCR analysis had indicated that these lines most strongly expressed both HWS and the transgene ( Figure 1S) and contained only a single insertion (data not shown). Line 8.3 exhibited a more severe phenotype than line A23.3 in all the experiments that were undertaken. A growth study, carried out two weeks after emergence, revealed that both ectopic expressing lines were substantially smaller than WT (Figures 8A). However, hws-1 plants at the same stage were larger than the control ( Figure 8A). Compared to the WT, the rosette leaves of the over-expressing lines had shorter petioles, narrower and more rounded lamina with serrated borders and displayed a greater degree of hyponastic bending ( Figure 8B). hws-1 had the longest, while the over-expressing line had the shortest, stigmatic papillae. A close-up view of flowers confirmed these observations ( Figure 2S).

Impact of ectopic expression of HWS on flower development
Measurements of dissected sepals and petals from flowers at position 3 revealed that hws-1 had significantly longer and wider sepals and petals compared to the WT or Pro 35S :HWS A23.3 line. The floral bases of the two over-expressing lines were the narrowest (Figures 9 and 3S), whilst hws-1 was the widest compared to the WT (see also Figure 1B and E).

Manipulation of HWS expression affects root growth and seed size
Seeds from hws-1, WT and the two over-expressing lines, were germinated in GM or MS media and root length was measured after a period of two weeks. The mutant exhibited the longest roots while both Pro 35S :HWS lines 8.3 and A23.3 had significantly shorter roots than the WT ( Figure 10A).
The dimensions of mature seeds from the three different genotypes were analysed. The hws-1 mutant was found to have statistically larger (both in length and width) seeds compared to the WT while the over-expressing lines produced the smallest seeds ( Figure 10B).

Phenotype of the hws mutant
The hws-1 mutant was originally isolated as a consequence of an inability to shed its sepals, petals and anther filaments. Indeed, these organs are retained throughout silique growth and development and remain at the base of the silique even after desiccation and dehiscence is complete. A close examination of the flowers has revealed that the mutation results in the fusion of the sepals along their basal margins.
Ectopic expression of the microRNA miR164 results in the generation of flowers with similar characteristics (Mallory et al., 2004). This miRNA is thought to act by targeting for degradation the mRNA of the CUP-SHAPED COTYLEDON genes CUC1 and CUC2 (Laufs et al., 2004). In addition to exhibiting fused sepals, hws-1 has a prevalence to generate fused anthers, multi-valved first-formed siliques and fusion of the cauline leaf lamina to the inflorescence stem. These observations suggest that, like CUC1 and CUC2, the HWS gene might act to regulate lateral boundary development and that the degree of penetrance of some of its phenotypic characteristics could be due, in part, to redundancy between HWS and other peptides.
An examination of the early stages of development in WT and hws-1 plants indicates that floral morphology is initially indistinguishable, however, whilst sepal separation proceeds to the base of the bud in WT this is terminated prematurely in hws-1 material. Thus the hws phenotype is not the consequence of postgenital fusion of the sepals but due to their failure to undergo complete separation.
To test whether differentiation of the floral abscission zones was taking place in hws-1 we crossed the gene marker Pro PGAZAT :GUS into the mutant. PGAZAT (At2g41850) encodes a polygalacturonase that is expressed specifically within the abscission zone cells of Arabidopsis and has been proposed to play a role in middle lamella degradation (González-Carranza et al., 2002). The results show that hws-1 is not compromised in terms of abscission zone (AZ) differentiation, however, the onset and duration of Pro PGAZAT :GUS expression is delayed and reduced respectively in the mutant. These revelations, together with our SEM and LM observations, indicate that the non-shedding phenotype of hws-1 is not the consequence of a failure of cell separation to take place at the base of petals, anther filaments, and sepals. Whilst the fusion of the sepal margins, encircling the separated petals and anthers, provides a structural barrier to preclude organ shedding in hws-1 our data indicate that the timing of cell separation in the mutant is delayed.

HWS gene encodes an F-box protein
Mapping and characterization of the hws-1 locus has revealed that the mutant phenotype is a consequence of a 28bp deletion in the ORF of a gene (At3g61590  As the phenotypic characteristics of the hws-1 mutant were primarily restricted to the shoot and reproductive tissues it was a surprise to discover that the transcript of the HWS gene could be detected by RT-PCR at high levels throughout the plant. Reporter gene analysis, using either GUS or GFP, revealed that although expression is evident in many tissues HWS promoter activity is restricted to specific cell types or regions of an organ. In roots, the pattern of expression is limited to the vascular tissues and the cells that comprise the lateral root cap. In leaves and stems expression is rarely detected in the vasculature but is associated with the margins of cotyledons. Floral tissues strongly express HWS with GUS activity being detected throughout development in sepals, the distal end of the stigma, anther filaments and pollen. Expression is also evident in the abscission zones of cauline leaves and floral organs. maturation. This spatial and temporal pattern of expression is similar to that observed by genes that have been proposed to contribute to the process of cell separation (González-Carranza et al., 2002). Although anatomical analysis indicates that floral organ shedding is precluded by fusion of the sepals in hws plants, rather than as a consequence of a failure to undergo cell separation, a comparison between the mutant, over-expressing lines and WT plants demonstrates that HWS does, in addition, have a direct influence on the timing of abscission. A study of the mutant material crossed with the Pro PGAZAT :GUS gene also supports the assertion that HWS is necessary for cell separation to proceed at the 'normal' rate.

Role of HWS in plant development
Although the principal characteristic of hws-1 plants is their non-shedding phenotype the isolation of a null allele (hws-2) of the gene has enabled us to dissect other  and increased growth of aerial organs and seeds (Schruff et al., 2006). Further work is on going to identify the mechanism by which HWS exerts its effect on growth and ascertain whether this may be mediated through an auxin-signalling pathway.

Plant material and growth conditions
Arabidopsis thaliana ecotypes Col-0 (Col) and Landsberg erecta (Ler), the hws-1 mutant, SALK lines for the 18 genes in the 56kb segment which were obtained from the NASC stock centre, diverse crosses and the mapping population were grown in a glasshouse with a temperature of 23±2ºC in plastic pots containing Levington M3 containing 0.01% (v/v) Triton X-100 for 5 min followed by ethanol (70% v/v). Seeds were then rinsed 5 times in sterile water before being placed, using a pipette tip, on top of the MS agar Petri dishes. Plates were sealed with micropore tape (3M) and maintained at 4ºC for 72h to stratify them and to facilitate uniform germination. Plates were then transferred to a growth room with a temperature or 23±1ºC under 16h light 8h dark.
The position of the flower was determined from the first site where petals were visible. From that position all subsequent flowers were numbered.

Scanning Electron Microscopy
Inflorescence apices and flowers were taken from Columbia-0 (WT) and hws-1 plants.
Buds were staged in accord with Smyth et al., (1990)   Primers from the region corresponding to the At3g61590 gene which amplified a genomic PCR segment of 1271bp are At361590forcDNA: 5'GCTCTTGAGAATGGAAGCAGAAAC 3' and At3g61590Rev: 5'CAGACCCATTTGCTTCTTCATTGC 3' PCR reactions were performed in a 50µl reaction following red taq (ABgene) manufacturer instructions. Conditions for amplification were: 3 min at 94ºC, 30 cycles of 94ºC for 15sec, 61ºC for 1 min, 72ºC for 2 min, 7 min 72ºC, 4ºC and the PCR products were run in a 1% agarose gel. The digestion of PCR products was performed in 20 µl reactions containing: 500 ng of PCR product, 10X Reaction Buffer, 2µg of BSA, and 0.5 µl of each enzyme, incubated at 37ºC (RsaI and AluI) or 65ºC (TaqI) and run in a 3% agarose gel. The A plant from the Salk_088349 KO line was used to cross with the hws-1 line and PCR analyses were performed to identify the presence of the two T-DNA insertions and the genomic fragment from the hws-1 mutant from the two parental lines; information about primers is described previously. To test the T-DNA insertion the LBb1 of pBIN-pROK2 for SALK lines primer was used 5' GCGTGGACCGCTTGCTGCAACT 3' (http://signal.salk.edu/tdnaprimers.2.html; Alonso et al., 2003) PCR conditions were the same as for the PCR amplification of At3g61590 segment described earlier.

Reverse Transcriptase PCR analysis of expression
Total RNA from roots, buds, flowers, rosette leaves, stem, young siliques and old siliques from WT; and a mix of tissues from progeny plants 1 and 2 from the cross between the hws-1 mutant with Salk_088349 line (hws-2) and from a mix of flowers from several developmental positions from F 1 over-expressing lines, was extracted using a modified method described by Han and Grierson (2002) where the removal of carbohydrates was performed by differential precipitation of RNA using 4M LiCl at 20ºC for 1h instead of using the CTAB buffer. RNA was quantified with a spectrophotometer and its quality was analysed by visualization in a 1% (w/v) agarose gel.
Expression analyses were determined using the SuperScript™ II Reverse Transcriptase kit from Invitrogen according to the manufacturer's instructions. First strand cDNA synthesis was performed in a 20 µl reaction containing 2 µg of total RNA, 1 µl of 500 µg.ml -1 oligo (dT) and 2 µl of 5 mM dNTPs and 13 µl of water.
PCR reactions were performed in 25 µl following red taq (ABgene) manufacturer instructions; the PCR conditions for amplification were: 3 min at 95ºC, 30 cycles of 95ºC for 1 min, 50-58ºC for 1-2 min (depending on primer combination and size of expected band), 72ºC for 1-2 min, and 7 min 72ºC, 4ºC. PCR products were run in a 1% agarose gel. The forward and reverse primers used in tissues from: wild type, the progeny from crosses, and endogenous gene for over expressing lines, were 590 5'UTR: 5' CTTCTCTCATCCTCGCGCTTGCTCTCTC 3' and www.plantphysiol.org on August 22, 2017 -Published by Downloaded from Copyright © 2007 American Society of Plant Biologists. All rights reserved.
At3g61590Rev (previously described) which give a genomic band of 2,213bp and a cDNA of 1668bp.Use of these primers allows the identification of genomic contamination.
To amplify the transgene of over-expressing lines the primer pKT735Sprom 5' GAGGAGCATCGTGGAAAAAG 3' from the 35S promoter and At3g61590Rev were used. The amplified band from this primer combination is 1,677bp.
Ubiquitin (At4g05320) primers UBQ10For, 5'-TAAAAACTTTCTCTCAATTCTCTCT-3' and UBQ10Rev, 5'-AAGCTCCGACACCATTGACAA-3' were used to evaluate the amounts of RNA levels in all tissues; these primers amplify a 1,555bp from genomic DNA and 1251bp from cDNA; in addition, control reactions without reverse transcriptase were performed using the same conditions to confirm absence of genomic contamination in all RT-PCR reactions performed.

Plasmid construction and plant transformation
All the constructs generated originated from WT genomic DNA extracted (Qiagen, DNAeasy Plant Mini kit) and amplified with the proof reading pfx DNA polymerase (Invitrogen) following the manufacturers instructions. All the PCR products were subcloned in P-GEM T-Easy from Promega, unless otherwise specified, digested and fused to the binary vectors at the multiple cloning site or at the site of digestion, as described below.
To perform the complementation test of the hws-1 mutant; a genomic segment from At3g61590 containing 1,291bp of the promoter region, 419bp from the 5' UTR, 532bp from the intron, 1,236bp from the ORF and 181bp from the 3' UTR using the primers: 590compfor 5' CCTCCAGTTTCAGAATCCGACC 3'and 590comprev 5' CCTCCAGTTTCAGAATCCGACC 3' was amplified from DNA of Columbia-0 wild type. Using this PCR product as template, the following primers containing a SalI site and a BamHI site in the forward and reverse primer respectively (highlighted in bold) were used: 590CompSalFor 5' GCAGTCGACGGCACTAAGGAGCAATGTG 3' 590CompBamRev 5' GCCGGATCCTCCAGTTTCAGAATCCGAC 3'. The PCR parameters used were: 94ºC for 5 min, followed by 35 cycles of 94ºC 15 sec and 68ºC 3.5 min, and a final elongation step at 68ºC for 7 min. To generate the β-glucuronidase reporter lines, a segment containing the promoter of the HWS (At3g61590), the 5'UTR and the intron (2,242bp), was also amplified by PCR.
The primers used to amplify this genomic segment were the 590CompSalFor described previously and the 590PrBamRev 5' GCCGGATCCTCTCAAGAGCCTCTGAAAC 3'.
With a SalI and a BamHI site at the forward and reverse primers respectively (highlighted in bold). The PCR parameters used were: 94ºC for 5 min, followed by 35 cycles of 94ºC 15 sec and 68ºC 2.5 min, and a final elongation step at 68ºC for 7 min.
To generate the GFP lines, the Pro HWS :GUS construct was modified by digesting and replacing the β-glucuronidase gene with the GFP ORF also digested from the MOG vector used previously for the PGAZAT gene (González-Carranza et al., 2002). The enzymes used were BamHI and SacI (Fermentas) following manufacturer's instructions. Once the HWS::GUS construct was digested; the reaction was dephosphorylated with alkaline phosphatase (Promega) following the manufacturer's instructions. The digestions were visualised in a 1% agarose gel and the linearized/dephosphorylated vector and the GFP segment was gel-extracted using a phenol: chloroform extraction method. The two segments were ligated using T4 DNA ligase from Promega following the manufacturer's instructions.
The over expressor construct was generated by amplifying the ORF from the HWS gene using the primers: 59035SBamHATGfor 5' GCGGGATCCCTCTTGAGAATGGAAGCAG 3' and 59035SsacIrev 5' CGTGAGCTCCCAGTTTCAGAATCCGACC 3' with a BamHI and a SacI site at the forward and reverse primers respectively (highlighted in bold). The PCR parameters used were the same as for the Pro HWS :GUS construct. This segment was sub-cloned in a MOG402 engineered vector containing two copies of the 35SCaMV promoter.
Escherichia coli DH5α cells were transformed and positive colonies were selected by PCR and the integrity of all plasmids generated was confirmed by sequencing, these plasmids were electroporated into Agrobacterium tumefaciens C58 strain and grown to an OD 600 of 0.5-0.8. WT Arabidopsis plants were transformed using the "floral dip" method described by Clough and Bent (1998 kanamycin resistance and plants were grown and analysed for rescue of the hws phenotype. T 2 seeds were collected from individual lines for the GUS and GFP reporter lines and the over-expressor lines, and screened for kanamycin resistance to identify at least six homozygous lines to check for consistency of expression. Histochemical analysis of β-glucuronidase (GUS) activity GUS staining, washing and mounting for the different tissues analysed was performed as described in (González-Carranza et al., 2002). The material was analysed and photographws taken using either a Zeiss Stemi SV6 Stereo microscope with a Kodak MDS290 digital camera or a Nikon microscope with a Leica DFC320 camera using the IM50 Leica software (Leica Microsystems Imaging Solutions Ltd, Cambridge, UK).

Confocal analyses of GFP expression
GFP fluorescence in the transgenic lines was examined using a Leica (TCS SP2) laser scanning confocal microscope equipped with argon krypton and green HeNe lasers and an AOBS scan head system (Leica Microsystems, Bannockburn, IL). GFP was excited at 488 nm with the argon ion laser. Images were recorded using the Leica CONFOCAL software.       (A) Mapping strategy used to identify the HWS locus using an AFLP approach to analyse the DNA from 33 F 2 hws-1 mutant plants. The initial location of the locus was identified to a region of 3.28 MB at the bottom of chromosome 3 between the markers SM148,6 and SM239_119,5. Further use of CAPS and SSLP on DNA from 156 F 2 hws-1 mutant plants allowed the reduction of the region to a segment of 0.4MB. With the use of InDels on DNA from 600 F 2 hws-1 plants, the region was reduced to a 56Kb segment between InDel 470113 and 469675. This region contained 18 candidate genes. Insertional SALK lines of these 18 genes were analysed for hws-1 characteristics and the DNA from region At3g61590 was amplified by PCR and products were subject to digestion by restriction enzymes.

ACKNOWLEDGMENTS
(B) Structure of the HWS gene. The gene has a 5'UTR of 419bp interrupted by an intron of 532bp, an ORF of 1236bp and a 3'UTR of 181bp. The hws-1 mutation is a consequence of a 28bp deletion 966bp downstream from the ATG that introduces a premature amber nonsense codon. Individuals from the SALK_088349 line (hws-2) that phenocopies the hws-1 mutation, have two T-DNA insertions inserted in opposite directions 475bp and 491bp downstream from the ATG (shown). two bands were amplified with vector border primer and each one of the specific primers, indicating that this line has two insertions.
(E) RT-PCR in root, bud/flower and rosette tissues from the SALK_088349 line (hws-2) and a mix of the same tissues in two progeny plants from the hws-1 mutant and SALK_088349 (hws-2). Ubiquitin (At4g05320) primers were used to demonstrate equal amounts of mRNA amplified.