A Novel Arabidopsis Gene Causes Bax-like Lethality in Saccharomyces cerevisiae*

Overexpression of the mammalian proapoptotic protein Bax induces cell death in plant and yeast cells. The Bax inihibitor-1 (BI-1) gene rescues yeast and plant from Bax-mediated lethality. Using the Arabidopsis BI-1 (AtBI-1) gene controlled by the GAL1 promoter as a cell death suppressor in yeast, Cdf1 (cell growth defect factor-1) was isolated from Arabidopsis cDNA library. Overexpression of Cdf1 caused cell death in yeast, whereas such an effect was suppressed by co-expression of AtBI-1. The Cdf1 protein fused with a green fluorescent protein was localized in the mitochondria and resulted in the loss of mitochondrial membrane potential in yeast. The Bax-resistant mutant BRM1 demonstrated tolerance against Cdf1-mediated lethality, whereas the Δatp4 strain was sensitive to Cdf1. Our results suggest that Cdf1 and Bax cause mitochondria-mediated yeast lethality through partially overlapped pathways.

Proteins encoded by Bcl-2 family interact with each other and either promote or inhibit metazoan apoptosis. A basic local alignment search tool (BLAST) data base search of the Arabidopsis thaliana genome showed no obvious homologues of any crucial regulator of metazoan apoptosis (members of the Bcl-2 family, p53, etc.) (1,2). However, it has been demonstrated that mammalian proapoptotic proteins can kill plant (3,4) as well as yeast (5)(6)(7)(8)(9). Furthermore, the antiapoptotic proteins Bcl-XL, Bcl-2, and Ced9 protect tobacco plants against cell death induced by ultraviolet light irradiation, paraquat treatment, hypersensitive response upon tobacco mosaic virus infection, and fungal pathogens (10,11). The Pseudomonas AvrPtoB effecter protein, conserved among diverse genera of plant pathogens, acts to inhibit Bax-induced cell death (12). This evidence clearly suggests that some mechanisms of cell death are evolutionarily conserved throughout the metazoa and plants.
In yeast, cell death with apoptosis-like features has also been reported after treatment with acetic acid, UV light irradiation, and H 2 O 2 (reviewed in Ref. 13). Madeo et al. (14) reported that a caspase-like gene (Yor197w) identified in Saccharomyces cerevisiae was implicated in cell death induced by H 2 O 2 , acetic acid, and aging. Yeast cells undergoing Bax-induced death exhibit ultrastructural changes that include massive cytosolic vacuolation and apparent disruption of mitochondria (6,8).
Classical yeast genetic approaches have been successfully applied for identification of genes related to the regulation of mammalian apoptosis. Even though the yeast genome lacks some of the molecular machinery responsible for apoptosis in metazoans, it can be powerful tool in the isolation of apoptosis-related genes.
The Bax inhibitor-1 (BI-1) 2 gene of Arabidopsis (AtBI-1) and the Oryza sativa Bax inhibitor-1 gene (OsBI-1), which are plant homologues of mammalian antiapoptotic gene BI-1, were isolated as suppressors of Bax-mediated lethality in yeast (15)(16)(17)(18)(19). Mammalian BI-1 suppresses apoptosis induced by Bax, etoposide, staurosporine, and growth factor deprivation in mammalian cells (15). The BI-1 protein has several transmembrane domains and is thought to be localized in the endoplasmic reticulum (4,15,20). We reported previously that the expression of mammalian Bax in Arabidopsis plants results in apoptotic cell death, which could be suppressed by the coexpression of AtBI-1 (4). Furthermore, rice cultured cells overexpressing AtBI-1 demonstrate a reduced hypersensitive response induced by an elicitor obtained from rice blast pathogen (21). The role of BI-1 in Mlo-mediated resistance to Blumeria graminis was also demonstrated by an overexpression analysis of barley BI-1 (22). Similarly, H 2 O 2 -and salicylic acid-induced cell death of tobacco suspension cells (23) and H 2 O 2 -and heat shock-induced cell death of yeast are also suppressed by overexpression of plant BI-1 gene (18).
In the present study, we devised a yeast genetic screening that allowed us to isolate plant genes responsible for cell death. For this purpose, we utilized the plant cell death suppressor AtBI-1 and found that overexpression of Cdf1 gene of Arabidopsis could induce apoptosis-like cell death in yeast. The effects of Cdf1 in yeast are partially similar to the mammalian Bax-induced lethality.
with appropriate supplements at 30°C. Yeast transformation was performed by the lithium acetate method.
Screening of the Arabidopsis cDNA That Causes Growth Defect in Yeast-Screening of the cDNA library was based on the rescue of a growth defect in yeast by AtBI-1 gene, which is a suppressor of Baxmediated yeast lethality. The Arabidopsis cDNAs that cause yeast growth inhibition and whose effects are inhibited by AtBI-1 were identified. Thus, wild type yeast strain W303-1A, possessing galactose-inducible AtBI-1 gene (NMV4-AtBI-1, constructed using pYX112-AtBI-1 as described in Ref. 16), was transformed with Arabidopsis cDNA library (in pYX112 expression vector, TPI promoter (24) and plated onto the SD-galactose solid medium in which AtBI-1 is expressed. The transformants were replicated to a SD-glucose plate to obtain a clone that survives on the galactose-containing medium, but not on glucosecontaining medium. The plasmid was recovered from each candidate clone and re-introduced into a W303-1A to confirm the reproducibility. The isolated cDNA (Cdf1) was connected under the GAL1 promoter, and the resultant plasmid, pTS-Cdf1, was used for further analysis.
Plasmid Construction and Spot Assay-The coding regions of Cdf1 and the mammalian apoptotic gene Bax, tagged with SphI recognition sequence using the PCR method, were introduced into a SphI-digested pTS909 expression vector (GAL1 promoter, a Trp-marked, 2-m replicon; a kind gift from Dr. Y. Kikuchi, University of Tokyo, Japan). For the expression of the Cdf1-GFP fusion protein, the same fragment was cloned into the pTS909-GFP plasmid. To express the mammalian antiapoptotic gene Bcl-2, the coding region tagged with the EcoRI recognition sequence was ligated to the EcoRI site of the expression vector pYX112 (TPI promoter, a Ura-marked, 2-m replicon). The expression vectors pYX112-AtBI-1 and pYX112-OsBI-1 used in this study have been described previously (16). The mouse Bax and rat Bcl-2 genes were kindly provided by Dr. J. C. Reed (The Burnham Institute).
In the spot assay, yeast cells grown in SD-glucose medium for 1 day were adjusted to A 600 of 0.1, and diluted to various concentrations (A 600 ϭ 0.1, 0.05, 0.01, 0.005, 0.001, and 0.0005). The aliquot (5 l) from each dilution was spotted onto SD-glucose or SD-galactose medium and incubated for 2 days (glucose medium) or 3 days (galactose medium) at 30°C.
Cytological Analysis-Microscopic examination was conducted using a fluorescent microscope system (DMRD; Leica, Wetzlar, Germany). For the detection of GFP fluorescence, yeast strains cultured in SD-glucose medium for 1 day were switched to SD-galactose medium for 4 h and examined at a 488-nm excitation wavelength. The plasmids pYX142-mt-gfp and p414GPD-mt-RFP, used for the observation of mitochondrial morphology, were kindly provided by Dr. K. Okamoto (University of Utah School of Medicine). To visualize active mitochondria with a transmembrane potential, yeast cells were stained with the fluorescent probe MitoTracker Red (650 nM; Invitrogen) for 5 min and examined under a fluorescent microscope at a wavelength of 568 nm. Reactive oxygen species (ROS) generation was monitored by treatment with 2,7-dichlorodihydrofluorescein diacetate (0.1 mM; Funakoshi, Tokyo, Japan) for 5 min and observed under a fluorescent microscope.
The number of viable cells was counted by the addition of Evans blue (0.05%, Nakalai, Kyoto, Japan), which penetrates only dead cells and results in blue staining of the cellular contents (25). The percentage of dead cells under each treatment was determined by scoring several hundred cells under the microscope.
For electron microscopic analysis, yeast cells cultured on the SD-galactose plate for 3 days were fixed by the freeze-substitution method, treated with 4% osmium tetraoxide (OsO 4 ), and embedded in Spurr's resin. Serial sections stained with uranyl acetate and lead citrate (UA/ Pb) were observed using the electron microscope (Hitachi H-7600; Hitachi High Technologies, Tokyo, Japan).
Northern Blot Analysis-Total RNAs were isolated from various tissues (leaf, stem, flower, and root of 3-week-old Arabidopsis plants, 3-day-old suspension cells, and mature or senescent leaves of 8-weekold plants) using the guanidinium thiocyanate method as described previously (16). Total RNAs, electrophoresed on a denaturing 1.2% agarose gel, were transferred onto a nylon membrane (Biodyne B; Pall Biosupport Division, Port Washington, NY) and then hybridized to the 32 Plabeled, 3Ј-untranslated region fragment of Cdf1 in 10% dextran sulfate, 1 M NaCl, 1% SDS, and 0.1 g/ml heat-denatured salmon testis DNA. Washing was performed under high stringency conditions (0.1ϫ SSC, 0.1% SDS at 65°C). The membrane was analyzed with a BAS1500 imaging plate scanner (Fuji Film, Tokyo, Japan).

Screening of Plant Genes with Growth Suppression Effect Inhibited by
AtBI-1-BI-1 is an antiapoptotic factor that can inhibit Bax-induced cell death in mammalian, plant, and yeast cells (15,16). To screen plant genes that can induce Bax-like cell death, we utilized the yeast system. Wild type yeast strain W303-1A harboring NMV4-AtBI-1, which produces AtBI-1 under the control of galactose-inducible GAL1 promoter, was transformed with the Arabidopsis cDNA library (TPI promoter) and plated on the SD-galactose medium. To obtain clones that were growth-inhibited on glucose-containing medium, these transformants were duplicated onto the SD-glucose plate. After 2 days of incubation at 30°C, clones that survived on galactose-containing medium but were growth inhibited on the glucose medium were picked up. Plasmids were isolated from the candidate clones, followed by DNA sequencing. Through successive screening of 2 ϫ 10 4 transformants, a single clone containing Arabidopsis cDNA (At5g23040) was isolated and named Cdf1 (cell growth defect factor-1). We found two more possible Cdf homologues in the Arabidopsis genome, At3g51140 and At2g20929. As shown in Fig. 1, Cdf-related genes were conserved in plant species. The Cdf1 encoded a polypeptide of 258 amino acids and had an estimated molecular mass of 28.8 kDa. The Cdf1 protein showed sequence homology to At3g51140 (32.9%), At2g20920 (22.4%), and Synechocystis slr1918 (23.5%). These proteins have no reported homology with any known animal protein and do not contain any recognizable protein motifs. Hydrophobicity analysis revealed that Cdf1 and At3g5r1140 contained three transmembrane domains, whereas other Cdf-related proteins, including Synechocystis slr1918 and At2g20920, possessed four transmembrane domains.
Cdf1 Induces Yeast Cell Death in a Bax-like Manner-To confirm the activity of Cdf1 as a cell growth suppressor, isolated cDNA was connected under the GAL1 promoter, and plasmid pTS-Cdf1 was constructed. Yeast strains transformed with pTS, pTS-Cdf1, and pTS-Bax were cultured in SD-glucose medium for 1 day and subjected to the spot assay ( Fig. 2 A). Cdf1 showed yeast growth inhibitory effects.
To clarify the intracellular localization of Cdf1 protein in yeast, a plasmid containing Cdf1 fused to the N terminus of GFP was constructed. The yeast cells expressing Cdf1-GFP demonstrated similar growth defect as those expressing Cdf1 alone (data not shown). After 4 h of culture in the SD-galactose liquid medium, yeast cells expressing both Cdf1-GFP and mt-RFP were collected and observed under a fluorescent microscope. As shown in Fig. 2B, the fluorescence signals of Cdf1-GFP and mt-RFP were co-localized and showed punctate structures. To make clear the mitochondrial shape in yeast expressing Cdf1, yeast cells expressing Cdf1 and mt-GFP were also observed under a fluorescent microscope. As shown in Fig. 2C, without Cdf1 or Bax expression the mitochondria kept an elongated shape. On the other hand, Cdf1 or Bax caused dramatic morphological change in mitochondria. We then visualized functional mitochondria with a transmembrane potential using MitoTracker Red, a vital mitochondrial fluorescent probe (Fig. 2B). In GFP-expressing yeast cells, MitoTracker Red clearly stained active elongated mitochondria. In contrast, the expression of Cdf1-GFP caused the loss of mitochondrial membrane potential (Fig. 2D. To evaluate whether Cdf1 induces cell death in yeast, we counted the number of dead cells by Evans blue staining. As shown in Fig. 3, A and C, ϳ20% of cells started to die after 14 h of Cdf1 expression. One of the key mechanisms that trigger the Bax-mediated lethality in yeast is production of ROS (26). To detect ROS generation, 2,7-dichlorodihydrofluorescein diacetate was used. This reagent diffuses through cell membranes and is subsequently enzymatically deacetylated by intracellular esterases to the non-fluorescent molecules; then it is oxidized by ROS to the highly fluorescent dichlorofluorescein (27). To determine whether Cdf1-induced growth defect in yeast is also accompanied by ROS generation, Cdf1-expressing yeast cells cultured in SD-galactose medium were stained with 2,7-dichlorodihydrofluorescein diacetate. As shown in Fig. 3, B and C, fluorescence was noted in ϳ50% of cells expressing Cdf1 or Bax after 9 h of culture. The majority of cells of the control strain with an empty vector showed no fluorescence. As shown in Fig. 3C, the ROS generation preceded dead cell stimulation.
Electron microscopic examination of Cdf1-and Bax-expressing cells revealed extensive disorganization of intracellular structures. Numerous cells showed morphological change of nuclei, accumulation of autophagic bodies, and increased intracellular vacuolization (Fig. 3D). In contrast, intracellular structures of control cells possessing empty vector were homogenous in shape and cytoplasmic electron density.
BI-1 Suppresses Cdf1-induced Lethality-As a first step toward determining the role played by this protein, we tested whether the mammalian Bcl-2 and plant BI-1 (OsBI-1 and AtBI-1) genes could suppress the Cdf1-induced growth defect in yeast. The expression plasmids containing Bcl-2, AtBI-1, and OsBI-1 (pYX-Bcl-2, AtBI-1, and OsBI-1) were transformed into a yeast strain containing galactose-inducible Cdf1 plasmid (pTS-Cdf1). As shown in Fig. 4A, when plated on the galactose medium the cells transformed with AtBI-1 and OsBI-1 demonstrated partial recovery of cell growth, whereas the cells transformed with empty vector and Bcl-2 did not. As shown in Fig. 4B, AtBI-1 suppressed the cell death caused by Cdf1 or Bax expression, whereas the ROS generation was not suppressed. Furthermore, BI-1 suppressed the loss of mitochondrial membrane potential as shown by the stainability with MitoTracker Red. In contrast, morphological change of mitochondria was not prohibited by the BI-1 expression (data not shown).
Cdf1-mediated Lethality in Bax-resistant Mutants-To further analyze the Cdf1-mediated yeast cell death, Bax-resistant yeast strains (BRM1 and ⌬atp4) were used. An unknown genetic mutation in BRM1 results in resistance against Bax-induced lethality (28). It was also demonstrated that disruption of the ATP4 gene caused tolerance to Bax (28,29). Using these strains, a growth assay was performed to clarify whether Cdf1-and Bax-mediated effects are similar in terms of the cell death inducer in yeast. Wild type strains (EGY48 for BRM1 and W303-1A for ⌬atp4) and mutant strains were transformed with pTS- Bax or pTS-Cdf1, respectively. Yeast cells cultured in SD-glucose for 1 day were transferred to the SD-galactose medium for 8 h, and their growth (A 600 ) was determined. As shown in Fig. 5A, the BRM1 strain was resistant against both Bax-and Cdf1-induced growth defect. In contrast, the ⌬atp4 strain demonstrated tolerance against Bax but sensitivity to Cdf1-mediated effect (Fig. 5B). The same results were obtained by the Evans blue staining method (data not shown).
Expression Pattern of Cdf Gene in Plant Tissues-To confirm the expression of Cdf1 in plants, mRNA accumulation was examined in several tissues obtained from 3-week-old Arabidopsis plants. Cdf1 showed higher expression in the above ground parts (leaf, stem, and flower) (Fig. 6A). We then examined the possible involvement of this gene in the leaf senescence stage. Leaf senescence is also known to be one of the stages of programmed cell death in plant species (reviewed in Ref. 30). Northern blot analysis of total RNAs isolated from mature and senescent leaves of 8-week-old plants showed enhanced expression of Cdf1 in the senescent leaves (Fig. 6B).

DISCUSSION
Expression of mammalian apoptotic gene Bax or Bak in yeast results in cell death with an apoptotic phenotype, which is suppressed by the coexpression of antiapoptotic genes including Bcl-2 and Bcl-XL (6,7,9). Yeast-based screening for Bax-suppressors using the human cDNA library have resulted in the isolation of BI-1 gene (15). Using similar screening system, several plant genes were isolated as suppressors of Bax-induced cell death in yeast (31)(32)(33)(34).
AtBI-1 was isolated as a plant homologue of mammalian antiapoptotic gene BI-1 (16). The functions of BI-1 proteins are conserved across evolution and include cell death suppression activity for Bax-mediated lethality even in yeast and plant cells (4). Moreover, the AtBI-1 protein protects tobacco BY-2 cells from cell death induced by H 2 O 2 or salicylic FIGURE 2. Comparison of the growth-suppression activities of Cdf-related genes in yeast. A, spot assay of yeast expressing Cdf1. The mouse Bax gene was also used as a control for the cell growth defect. Yeast cells transformed with plasmids containing galactose-inducible Cdf1 (pTS-Cdf1) and Bax (pTS-Bax) were cultured in glucose-containing SD medium. After 1 day of continuous shaking, the A 600 of each culture was adjusted to 0.1, and each dilution was spotted onto SD solid medium containing glucose or galactose. Photographs were taken 2 days (Glucose) or 3 days (Galactose) after incubation at 30°C. B, distribution of Cdf1-GFP and mt-RFP in yeast. The yeast cells expressing Cdf1-GFP and mt-RFP were observed under a fluorescence microscope after 4 h of culture in SD-galactose medium. The yeast cells expressing GFP and mt-RFP were also observed as a control. C, mitochondrial morphology in yeast expressing Cdf1 or Bax. The yeast cells possessing plasmid pYX142-mt-gfp were transformed with pTS-Cdf1 or pTS-Bax, and the fluorescence of mt-GFP localized in the mitochondria after 4 h of culture in SD-galactose medium was observed. D, MitoTracker Red staining of yeast expressing Cdf1-GFP. Transformants cultured in SD-glucose liquid medium for 1 day were transferred to SD-galactose medium and cultured for 4 h. Active mitochondria were stained with the fluorescent probe MitoTracker Red. DIC, differential interference contrast image. acid (23). Reasoning that at least some of the functions of BI-1 protein appear to be conserved in yeast and plant, we undertook a classical genetic approach designed to identify plant genes that cause yeast cell death, which would be suppressed by BI-1. Through successive screening, Arabidopsis Cdf1 was isolated as a possible cell death inducer in yeast.
Yeast cells expressing Cdf1 showed the yeast lethality, which included Bax expressing cell-like morphological changes observed under the electron microscope. Furthermore, dissection of the intracellular distribution of mt-GFP demonstrated the abnormal morphological change of mitochondria in Cdf1-expressing yeast cells. Mitochondria appear to play a central role in the induction of apoptosis (35). The Bax protein associates with mitochondrial membranes and confers a lethal phenotype when expressed in yeast and plant (3, 36 -38). Recently, it was reported that fragmentation of tubular mitochondria into short punctate structures is a common early feature of apoptotic mammalian cells (47,48) and yeast (49). In our recent work, mitochondrial morphological change was also observed under ROS stress, leading to plant cell death (38,50). Such change might be one of the features of mitochondrially originated cell death. To assess the status of mitochondrial function upon the expression of Cdf1 in yeast strains, the cells were treated with MitoTracker Red, which detects the mitochondrial membrane potential. Yeast cells expressing Cdf1exhibited reduced mitochondrial membrane potential. Overexpression of Bax in mammalian and yeast systems is associated with the transition of mitochondrial permeability and the loss of membrane potential (39). Dissipation of the mitochondrial membrane potential may be the result of either the disruption of the integrity of the mitochondrial outer membrane or the opening of the permeability transition pore caused by Cdf1 expression.
The common denominator in most cell death models that involve yeast and plants is accumulation of ROS. Madeo et al. (26) demonstrated that Bax expression induces ROS generation in yeast. Release of ROS from mitochondria to cytosol alters the cellular redox potential, which may be an important determinant of cell death in yeast. We tested whether Cdf1 expression also triggers ROS generation and accumulation. Experiments showed that ϳ50% of cells fluoresce after 9 h of induction, whereas almost no fluorescent cells were found in the control. These results suggest that the expression of Cdf1 can induce the formation of ROS and that Cdf1 toxicity may be mediated at least in part through the generation of oxidative stress. against Cdf1-induced lethality. Yeast cells possessing pTS-Cdf1 were transformed with pYX yeast expression plasmids containing OsBI-1, AtBI-1, and Bcl-2, respectively. Transformants cultured in SD-glucose medium for 1 day were used for the spot assay. Photographs were taken 2 days (Glucose) or 3 days (Galactose) after incubation at 30°C. B, the effect of AtBI-1 on Cdf1-induced ROS generation and lethality. ROS generation was examined in each yeast stain by 2,7-dichlorodihydrofluorescein diacetate staining at 9 h after culture in SD-galactose medium. Evans blue staining for evaluation of dead cells was performed at 14 h after culture in SD-galactose medium. Data shown are mean Ϯ S.E. of three experiments.  Total RNAs (20 g/lane) isolated from several tissues were fractionated in an agarose-formaldehyde denaturing gel, blotted to a membrane, and hybridized with specific probes. B, accumulation of Cdf1 transcripts during Arabidopsis leaf senescence. Total RNAs (10 g/lane) isolated from mature leaves and senescent leaves were used for the RNA gel blot analysis. RNA gel blots were quantitated using the BAS1500 imaging plate scanner and expressed as relative values.
The precise pathways in which Bcl-2 family members act as regulators of apoptosis are still not fully elucidated (40). The interaction between pro-and anti-apoptotic Bcl-2 proteins determines the fate of a cell, i.e. to live or to die. A major target site for these proteins is the mitochondrion (41)(42)(43). To prevent apoptosis, Bcl-2 has been proposed to interact with Bax (5). Furthermore, Bcl-2 protein, as well as the proapoptotic protein Bax, can form ion channels in synthetic membranes in vitro (44,45). On the other hand, the BI-1 protein, which is localized on the endoplasmic reticulum membrane, inhibits Bax-induced lethality downstream of the mitochondria (23). Specifically, AtBI-1 inhibits Baxinduced lethality, but not ROS generation, in plants. Similar trends were shown in the present study (Fig. 4). Furthermore, both Bax-and Cdf1induced lethality were suppressed by AtBI-1, whereas Cdf1-induced lethality was not inhibited by Bcl-2, suggesting the overlap of the Cdf1and Bax-mediated cell death pathways.
The Bax-resistant mutant BRM1 was isolated from a screening of mutagenized yeast wild type strain EGY48 (28). Furthermore, the same group identified a subunit of the F 0 F 1 ATPase of the inner mitochondrial membrane as a yeast cell component necessary for Bax-induced lethality (28). Using such yeast mutant stains, we evaluated Bax-and Cdf1-mediated yeast lethality. A cell growth defect was not detected in either Bax-or Cdf-expressing BRM1 cells. In contrast, Cdf1-overexpression was lethal to the ⌬atp4 strain. As a conclusion, Bax and Cdf1 may provide clues regarding the different entry points into the mitochondrion-mediated yeast cell death pathway.
As in mammals, plant mitochondria are also the major sites of ROS production (46). At present, it remains obscure as to how overexpression of Cdf1 initiates cell death in yeast. We assume that overproduction of Cdf1 in yeast would affect mitochondrial function, which may trigger cell death. The subcellular distribution of Cdf1 and Cdf-related proteins has not been determined in Arabidopsis, but their amino acid sequences and putative function are indicative of organelle localization such as mitochondria or plastids. Both organelles are known to be involved in the ROS production in plants. To clarify the function of Cdf1 protein in plants, the cellular localization of this protein in plant is intriguing. In our results, accumulation of Cdf1 mRNAs was noted during the leaf senescence stage, which is one of the programmed cell death stages in plants regulated by strict mechanisms. The sequence similarity suggests that Arabidopsis has two more Cdf-related genes in its genome (Fig. 1). Further studies are necessary for isolation and expression analysis of other Cdf-related genes in order to determine the function of Cdfs in plant. Such research should address whether Cdf-mediated mechanisms are operational in plant cells.